Sie sind auf Seite 1von 78

( )

Think board to
go broad Pg 14
EDN.comment: Can’t
Issue 12/2009 see the forest ... Pg 6
Baker’s Best Pg 16
Peering into ultrasound
machines Pg 18
Design Ideas Pg 43

VOICE O F T HE ENGINEE R Tales from the Cube Pg 74


Page 30

Page 21

Page 39


Cut development time as much as 50% while adding breakthrough performance,
power and cost benefits. The ISE Design Suite 11 enables new domain-specific design
for Targeted Design Platforms, allowing you to focus on building design differentiation.
Download a free evaluation at

© Copyright 2009 Xilinx, Inc. XILINX, the Xilinx logo, and ISE Design Suite are trademarks of Xilinx in the United States and other countries. All other trademarks are the property of their respective owners.
Enter xx at
“I need a function generator that generates efficiency, too.”

Signal quality is just the beginning.

You’re facing faster design cycles and more complicated testing
protocols. PC connectivity is more critical than ever, too. Agilent
function/arbitrary waveform generators help shorten your test times
with legendary signal quality and purity — along with state-of-the-art
I/O connectivity. Plus they’re bench-friendly, rack-able, reliable and easy
to use. That’s Agilent. That’s smart.

Function/Arbitrary Waveform Generator Series

33210A – 10MHz $1,195*
33220A – 20MHz $1,860*
33250A – 80MHz $4,571*

Add a FREE External Reference

worth up to $495**

Agilent Authorized Distributors

© 2009 Agilent Technologies, Inc. Allied Electronics Davis Instruments

*Pricing in USD and subject to change
**Limited time offer 800-433-5700 800-358-5525
Little-known flash-
memory features
protect data and IP

Features from block
locking to encrypt-
Accelerating ed-password-access
consumers’ NAS mechanisms can prevent uninten-
Diagnostic ultrasound tional disruption, malicious dam-
adoptions: assessing
gets smaller, faster, age, or copying.
your product options by Bill Stafford, Numonyx
and more useful

Network storage is

a notably bright spot The signal path in ultra-
in the otherwise-blah sound machines is a A SPECIAL EDN SECTION
consumer-electronics economy. multichannel transmit-
Carefully select and cull hardware, ter-receiver system with blazingly 55
software, and their jointly imple- fast data rates. Engineers need to
weigh a host of options in design- 57 Innovation: genius, practice,
mented features to ensure prod-
uct success. by Brian Dipert, ing these complex machines. or luck?
Senior Technical Editor by Paul Rako, Technical Editor 59 Application engineers:
serving the customer

67 Improving on PCB design
Dilbert 11

10 Scope-based tools support 12 Nonvolatile-RAM module

complete debugging and com- swaps batteries for super-
pliance testing of Superspeed capacitors
USB 3.0
12 Chip allows isolated USB
10 Analog Devices, Infineon col-
13 POE ICs offer increased
laborate on air-bag systems
power efficiency for PDs
11 1- to 4-GHz, 20G-sample/sec
DSOs and MSOs integrate 13 Tool bundles target FPGAs
protocol analysis 14 Voices: Avnet’s Jeff Ittel:
Think board to go broad

5V 5V 5V 5V 5V
43 Create a swept-sine function in LabView with just one virtual instrument
Q5 BC337
Q4 BC337


Q1 BC337
Q2 BC337
Q3 BC337

44 Charlieplexing at high duty cycle

D12 D13 D14 D15


48 Serial port tests digital circuits

D21 D23 D24 D25

D31 D32 D34 D35


D41 D42 D43 D45
51 DAC calibrates 4- to 20-mA output current
D51 D52 D53 D54


51 Alarm tells you to close the refrigerator door

JUNE 25, 2009 | EDN 3
The Newest
New Products from:

Cortex™ ARM® STM8S 8-bit Microcontroller
8051 PSoC® Encore!®

Z8® Flexis™ AVR® MSP430 HCS08 MSP430x5xx Microcontroller

PIC® Z80® ST7
HC12 ST6 A
HC08 M


The ONLY New Catalog Every 90 Days

Experience Mouser’s time-to-market

advantage with no minimums and same-day
shipping of the newest products from more
than 390 leading suppliers.

The Newest Products

For Your Newest Designs (800) 346-6873

Over A Million Products Online
contents 6.25.09

S U R FD TfHR ) Nr s
o -Hm
(An s
S Ua
Tr a n sd u c t o r s
& In
18 74
D E PA R T M E N T S & C O L U M N S
6 EDN.comment: Can’t see the forest ...
16 Baker’s Best: Is your amplifier offset way out of whack? matter!
18 Prying Eyes: Peering into ultrasound machines
72 Product Roundup: Motion, Integrated Circuits
74 Tales from the Cube: Weather or not: All signs point to “no”

online contents from

Check out these Web-exclusive articles: MORE EDN INNOVATORS
profile .19"ht.
“In-design” physical verification
Obstacles and opportunities
Once you finish reading the special
is “on-time” physical verification
Innovators section in this issue, head • Audio Transformers
It’s no longer practical to wait until the end
of the physical design of an IC to do physical
to the Web to read our earlier • Pulse Transformers
Innovators supplements that high-
verification. Physical-design engineers who
create chips at the 45-nm node and beyond
light other • DC-DC Converter
face a difficult task. The time-tested flows
and their in- Transformers
used at previous nodes are no longer viable
novations in
to maintain productivity at today’s advanced
• MultiPlex Data Bus
➔ design. Transformers
➔ www.
RFIDs power themselves • Power &
RFIDs have enabled a significant improve- EMI Inductors
ment in many very costly business systems
ranging from warehousing to tagging your diately
g imme
luggage at the airport. The secure contact- full Catalo s .c o m
See Pic
o’s r o n ic
less-IC variants have found their way into our ico e le c t
credit cards and passports, and their future w w w.p
application almost certainly includes the The latest tech news, delivered every
replacement of the checkout counter at the or send direct for free PICO Catalog
afternoon. Call Toll Free 800 431-1064
local grocery chain. ➔ Fax 914-738-8225
➔ E Mail:
EDN ® (ISSN#0012-7515), (GST#123397457) is published biweekly, 24 times per year, by Reed Business Information, 8878 Barrons Blvd, Highlands
Ranch, CO 80129-2345. Reed Business Information, a division of Reed Elsevier Inc, is located at 360 Park Avenue South, New York, NY 10010. Tad
Smith, Chief Executive Officer; Mark Finkelstein, President, Boston Division. Periodicals postage paid at Littleton, CO 80126 and additional mailing offic-
es. Circulation records are maintained at Reed Business Information, 8878 S Barrons Blvd, Highlands Ranch, CO 80129-2345. Telephone (303) 470-
PICO Electronics,Inc.
143 Sparks Ave. Pelham, N.Y. 10803-1837
4445. POSTMASTER: Send address changes to EDN ®, PO Box 7500, Highlands Ranch, CO 80163-7500. EDN ® copyright 2009 by Reed Elsevier
Inc. Rates for nonqualified subscriptions, including all issues: US, $179.99 one year; Canada, $229.99 one year (includes 7% GST, GST#123397457);
Mexico, $229.99 one year; air expedited, $399.99 one year. Except for special issues where price changes are indicated, single copies are avail-
able for $10 US and $15 foreign. Publications Agreement No. 40685520. Return undeliverable Canadian addresses to: RCS International, Box
697 STN A, Windsor Ontario N9A 6N4. E-mail: Please address all subscription mail to EDN ®, 8878 S Barrons Blvd,
Highlands Ranch, CO 80129-2345. EDN ® is a registered trademark of Reed Elsevier Properties Inc, used under license. A Reed Business Information Delivery - Stock to one week
Publication/Volume 54, Number 12 (Printed in USA).
JUNE 25, 2009 | EDN 5

Jim Williams, staff scientist at Lin-

ear Technology, also discusses the
changing role of application engineers
(see “Application engineers: serving
the customer,” EDN, this issue, pg 59).
Williams observes that modern appli-
cation engineers are fundamentally
systems engineers. “In a well-run an-
alog company, [components are] sec-
Can’t see the forest ... ond- or third-tier parts of the job,” he
says. “The fundamental responsibility
of an analog-application engineer is
nalog design is all about trade-offs, so system-design ex-

to understand his customer’s problem
pertise is vital to any company offering analog chips. For and provide a solution.”
example, Dave Kress, director of technical marketing at Semiconductor companies are tak-
Analog Devices, points out that his company is trying ing an interest in systems design that
to get a better understanding of the entire system design is essential to continued innovation
so it can solve design problems that span multiple chips and success. For example, Qspeed
Semiconductor’s products are faster
and subcircuits. Meanwhile, Doug Bailey, vice president of marketing than other silicon diodes and cheap-
at Power Integrations, says that his er than silicon-carbide diodes. Yet the
company’s customers expect help with Qspeed parts are three times more ex-
thermal design and EMI (electromag- pensive than legacy slow-recovery sili-
netic-interference) characterization, con diodes. If Qspeed thought of itself
functions that were previously the do- as just a diode company, this approach
main of system engineers. would make it tough to sell parts into
Texas Instruments is also dedicated cost-sensitive applications. The com-
to understanding system design to bet- pany looks past the two pins of its di-
ter serve its customers. Matt McKin- odes, however, and sees the total im-
ney, marketing-communications man- pact of better performance.
ager, mentions that TI offers reference “It’s an example of Andrew Smith, power-marketing
designs that solve the system-power manager at Power Integrations, points
and analog-signal-path problems of vendors’ not looking out that the company looks at the
customers using the company’s DSPs
or embedded microprocessors.
at the whole receiver- whole power system, designing con-
troller chips that operate the PFC
Bob Thomas, technical leader at signal chain.” (power-factor-correction) section of
Cisco Systems, confirms the impor- —John Scampini the power system, as well as the pri-
tance of vendors’ having system-lev- mary isolated LC (inductor-capacitor)
el cognizance. “A vendor shouldn’t John Scampini, director of strategic power stage for generating low-volt-
just toss a part on our desk and walk marketing at Maxim Integrated Prod- age system power.
away,” says Thomas. “The vendor ucts, points out that some vendors Ed Lam, vice president of market-
has to get to know our system and its recommend amplifiers that would ing and engineering at Advanced
unique characteristics so [the vendor] limit the SNR (signal-to-noise ratio) Analogic Technologies, is also look-
can add something special to our de- of subsequent analog-to-digital stag- ing at the entire system—from the
signs.” Steve Abe, a senior software es, meaning that the output-referred wall plug to the final analog output.
engineer at the company, mentions noise of the variable-gain amplifiers The components in this path might
that chip manufacturers should help would be larger than the noise floor be the power in an LCD television,
develop the firmware to initialize the of the ADC that it was feeding. “It’s ending in a Class D audio amplifi-
chips in Cisco’s designs. “The vendors an example of vendors’ not looking er, or the charging system in a cell
have an intimate understanding of the at the whole receiver-signal chain,” phone. “The system is the thing,” he
chips, and we look to them to help us he says (see “Diagnostic ultrasound says.EDN
... keep the chip working in our sys- gets smaller, faster, and more useful,”
tem environment,” he says. EDN, this issue, pg 21). Contact me at

6 EDN | JUNE 25, 2009

More Q. Less Cu


These tiny new air core inductors

have the highest Q and current handling
in the smallest footprint.
Coilcraft’s new SQ air core inductors have unmatched Q
factors: most are above 200 in the 1-2 GHz range! That’s 3
times higher than comparably sized 0805 chip coils.
And with their extremely low
DCR, they can handle 4 to 8 times
more current: up to 4.4 Arms.
SQ air core inductors are perfect
for your LC filter and RF impedance
matching applications. They come in
15 values ranging from 6 to 27.3 nH,
Q factors are 3X higher than all with 5% tolerance.
standard chip inductors
These coils
are significantly smaller than exist-
ing air core inductors. We reduced
the footprint by using close-wound
construction and keeping the leads
close to the body. The square shape
cuts the height to as low as 1.5 mm
and creates flat top and bottom sur-
faces for easy automated handling
and stable mounting.
See how the ultra-high Q and
The square shape and narrow footprint current handling of Coilcraft’s
reduce board space by 60-75% over new SQ air core inductors can
conventional air core inductors.
maximize the performance of
your next design. For complete specifications and free
evaluation samples, visit

RCOMoPLHIANST 800/322-2645
it’s everybody’s business
to get answers as
as yesterday.
Does your enterprise software help give you the insight
to keep up with the pace of demand?
Mark Finkelstein,
Russell E Pratt, 1-781-734-8417; Frances T Granville 1-781-734-8439;
fax: 1-303-265-3131;

Judy Hayes, 1-925-736-7617; EDITORIAL/WEB PRODUCTION Diane Malone, Manager
VICE PRESIDENT, EDITORIAL DIRECTOR 1-781-734-8445; fax: 1-303-265-3024
Karen Field, 1-781-734-8188; Steve Mahoney,
Rick Nelson, 1-781-734-8418;
Production/Editorial Coordinator
1-781-734-8442; fax: 1-303-265-3198
Melissa Annand,
PRECISION Newsletter/Editorial Coordinator From a full spectrum of pin receptacles
1-781-734-8443; fax: 1-303-265-3279 Regardless of whether your components fly with the Blue
Adam Odoardi, Prepress Manager
Ron Wilson, 1-510-744-1263; Angels or control industrial robots, good design should
1-781-734-8325; fax: 1-303-265-3042 never be limited by a lack of options. Mill-Max Mfg. Corp.
Jim Williams, Staff Scientist, offers hundreds of high-precision receptacle styles with
Amy Norcross
Linear Technology an unprecedented range of options and features:
fax: 1-720-356-9161; • 35 pre-tooled contact styles. Dan Strassberg, • A full selection of 3-, 4- and 6-finger contact designs.
Contact for contributed technical articles
Nicholas Cravotta, • .008" to .102" (0,20 - 2,59 mm) pin acceptance.
Mike O’Leary • Accepts round, square and rectangular
1-781-734-8307; COLUMNISTS component leads.
Howard Johnson, PhD, Signal Consulting
fax: 1-303-265-3021; • Styles for soldering, swaging or press-fitting. Bonnie Baker, Texas Instruments
Pallab Chatterjee, SiliconMap
Paul Rako, Technical Editor PRODUCTION
1-408-745-1994; Dorothy Buchholz,
Group Production Director
PCs, AND PERIPHERALS 1-781-734-8329
Brian Dipert, Senior Technical Editor Joshua S Levin-Epstein,
1-916-760-0159; Production Manager
fax: 1-303-265-3187; 1-781-734-8333; fax: 1-781-734-8096 EDN EUROPE
MICROPROCESSORS, DSPs, Graham Prophet, Editor, Reed Publishing Large Variety of Styles
AND TOOLS The Quadrant, Sutton, Surrey SM2 5AS
Robert Cravotta, Technical Editor +44 118 935 1650;
1-661-296-5096; fax: +44 118 935 1670;
fax: 1-303-265-3116; EDN ASIA
NEWS Luke Rattigan, Chief Executive Officer
Suzanne Deffree, Managing Editor
1-631-266-3433; Kirtimaya Varma, Editor-in-Chief
Margery Conner, Technical Editor William Zhang, Knock-out Bottom OFP ®
1-805-461-8242; Publisher and Editorial Director
fax: 1-805-461-9640; Jeff Lu, Executive Editor
Martin Rowe, EDN JAPAN
Senior Technical Editor, Katsuya Watanabe, Publisher
Test & Measurement World Ken Amemoto, Editor-in-Chief
Available on Tape and Reel
EDN. 225 Wyman St, Waltham, MA 02451. Phone 1-781-734-8000.
Address changes or subscription inquiries: phone 1-800-446-6551; fax: 1-303- Mill-Max receptacles. Plug into the simple,
470-4280; For a free subscription, go to www.getfreemag.
com/edn. Reed Business Information, 8878 S Barrons Blvd, Highlands Ranch, CO 80129-
reliable connectivity solution.
2345. Include your mailing label.
To view our Design Guide, new product offerings
and order free samples, visit ®


Scope-based tools support complete debugging FEEDBACK LOOP

and compliance testing of Superspeed USB 3.0 “Brilliant anec-
dote. … A classic

ektronix’s new DPO (digital-phosphor- set for characterizing, debugging, and auto-
oscilloscope)/DSA (digital-signal-ana- mating compliance testing of Superspeed
case of every-
lyzer) 7000B family is now available devices—to its TekExpress oscilloscope-based thing is possible
with a USB-TX option, which, with the press serial-data-test framework, which also supports but not always
of a button, allows you to automatically and several other ultra-high-speed serial buses. In feasible.”
rapidly validate USB (Universal Serial Bus) 3.0 addition, the manufacturer has introduced a full —EDN reader and engineer
Superspeed transmitters. These devices trans- set of USB 3.0 fixtures that enable you to per- “Harry,” in EDN ’s Feedback
mit data more than 10 times as fast as high- form transmitter, receiver, and cable tests. Loop, at
speed devices that conform to the previous The increased bandwidth of the Superspeed CA6651606. Add your
(2.0) version of the USB standard. Tektronix mode brings with it critical signal-fidelity chal-
has added the option—a comprehensive tool lenges. Tektronix points out that, whereas
other industry offerings provide only norma-
tive measurements in accordance with the
USB-IF (Implementers’ Forum) electrical-test
specification, the USB-TX option supports all
measurements, including normative and infor-
mative tests, such as SSC (spread-spectrum
clocking), slew, and voltage levels. A unique
plug-style fixture enables probing close to the
silicon to provide a true representation of the
signal. The list price for the USB-TX option
is $5000. Prices for the DPO/DSA 7000B
With the addition of the new USB-TX option,
one button press is all it takes for 7000B scopes begin at $53,300.
oscilloscopes to begin extensive validation —by Dan Strassberg
testing on USB 3.0 Superspeed transmitters. 컄Tektronix Inc,

Analog Devices, Infineon collaborate on air-bag systems

Analog Devices and Infineon Technologies have collaborated to drivers. Both companies are now offering the results of the
produce next-generation automotive-air-bag systems. In the im- collaboration and continuing to sell their product lines. A typi-
mediate future, the two companies will produce a hardware plat- cal product-development cycle in this market is approximately
form using parts from each company’s current offerings. Design- two years, according to the companies. They claim that starting
ers will be able to adopt this platform to begin development of with their hardware platform could save six months of develop-
code using a well-defined software interface. Later, the two com- ment time.
panies will align their plans to ensure interoperability. Analog Devices has shipped more than 500 million acceler-
Analog Devices brings its MEMS (microelectromechanical- ometers and gyroscopes in more than 15 years and is the largest
system)-air-bag-sensor technology to the mix, and Infineon supplier of MEMS-based inertial sensors to the automotive indus-
contributes most of the rest of the chip set, including sensor try. Infineon has supplied more than 600 million air-bag compo-
interfaces, microcontrollers, power-management chips, CAN nents in the last 15 years.—by Graham Prophet
(controller-area-network) and LIN (local-interconnect-net- 컄Analog Devices,
work) transceivers, pressure sensors, and air-bag-squib-firing 컄Infineon,

10 EDN | JUNE 25, 2009

1- to 4-GHz, 20G-sample/sec DSOs ior of protocol-rich serial buses
without finding and connecting
and MSOs integrate protocol analysis additional test equipment.
The digital inputs accept

gilent Technologies four-channel mode, the ana-
has expanded its MSO log channels offer a standard The 15-in. many Agilent logic-analyzer
(mixed-signal-oscillo- memory depth of 10M sam- displays accessories. In addition, gen-
scope) and DSO (digital-stor- ples, with options for as many eral-purpose measurement
age-oscilloscope) portfolios as 1G samples in the two-
offer 57% great- and analysis features allow
with the addition of the six- channel mode. On any of the er display area you to customize the instru-
model Infiniium 9000 series. three analog-only models, you than do 12-in. ments by adding software. To
All of the units offer true ana- can in minutes enable the logic provide insight into common
log bandwidths of 1, 2.5, or channels without returning the
displays. serial buses, FPGAs, and RF
4 GHz on four channels and scopes to the factory or recali- measurements, 25 optional
optional built-in logic and pro- brating them. sitive XGA (extended-graph- application packages allow
tocol analysis on 16 digital All models come in easy-to- ics-array) LCDs. for rapid verification of a unit
channels. All analog channels carry, 9-in.-deep, 25-lb pack- The units are also the first under test’s compliance with
offer both 50⍀ and 1-M⍀ ages and feature screens that, scopes to integrate both logic- industry standards.
inputs and acquire a maxi- according to the company, are analysis channels and a proto- Debugging software in-
mum of 20G samples/sec in the industry’s largest: 15-in., col-analyzer viewer for buses cludes protocol triggering and
the two-channel mode. In the 1024⫻768-pixel, touch-sen- such as PCIe (peripheral-com- viewing for PCIe and USB;
ponent-interconnect express) serial decoding and trigger-
and USB (Universal Serial ing for I2C (inter-integrated-
Bus), according to the com- circuit), SPI (serial-peripher-
pany. These features simplify al-interface), CAN (control-
setting up and using the instru- ler-area-network), RS-232,
ments to test complex designs, and UART (universal-asyn-
encouraging greater instru- chronous-receiver/transmit-
ment usage and increased ter) interfaces; rapid core-as-
return on investment. sisted debugging of designs
Regardless of the scope’s with Xilinx (
analog bandwidth, the inte- and Altera (
grated digital channels acquire FPGAs; and an InfiniiScan ca-
digital signals at rates as high pability for triggering on infre-
as 2G samples/sec and trigger quent signal anomalies that
on a rich set of mixed analog/ meet visual criteria. Compli-
digital conditions for accurate ance-testing features include
analysis of timing relationships communication-mask testing
among control signals and data and compatibility with DDR-1,
buses. A protocol viewer for -2, and -3 interconnects; Ether-
PCIe and USB lets you extend net; and USB 2.0.
the scope’s debugging and The 15-in. displays offer
The 20G-sample/sec Infiniium 9000 scope’s 15-in., touch-sensi-
tive XGA LCD proves helpful in mixed-signal debugging, letting testing capabilities to obtain 57% greater display area than
you view 16 logic-analysis channels and four analog waveforms. rapid insights into the behav- do 12-in. displays. The larger
displays provide the space
users need to comfortably
DILBERT By Scott Adams view large numbers of digital
and serial signals in addition to
analog traces. Infiniium 9000
series base prices range from
$19,900 for a 1-GHz-band-
width DSO to $41,500 for a
4-GHz MSO.
—by Dan Strassberg
컄Agilent Technologies,

JUNE 25, 2009 | EDN 11

Nonvolatile-RAM module swaps cally saves the SDRAM data
to NAND-flash memory using
batteries for supercapacitors the energy it stores in its inter-
nal supercapacitors.

ystems often need The AgigaRAM product
some form of nonvol- encompasses the Bali and
atile RAM to protect Capri product families. Bali
against memory losses after comes in 4- to 64-Mbyte den-
a power interruption or loss. sities, incorporates 100-MHz
Battery-backed RAM has been SDRAM with 200-Mbyte/
the most common form of sec peak transfers, requires a
high-density nonvolatile mem- 3.3V power supply for 4 to 32
ory, but it is bulky, involves long Mbytes or 5V for 64 Mbytes,
charging times for the battery, and operates over a 0 to 70⬚C
has a relatively low-tempera- operating range. It comes in
ture operating environment a 200-pin SO-DIMM pack-
that requires low-temperature age or a mezzanine card and
operation, and has a limited sells for approximately $4 per
number of cycles. The Bali version of the AgigaRAM nonvolatile-memory system megabyte. The Capri comes in
Agiga Tech’s new AgigaRAM combines synchronous DRAM with flash memory. If the system 256-Mbyte to 2-Gbyte capac-
nonvolatile-system technol- loses power, the module automatically saves the SDRAM data ities, uses a high-speed DDR-
ogy addresses these issues to NAND-flash memory using the energy it stores in its internal 800 interface, and sells for
by combining synchronous less than $1 per megabyte.
DRAM with flash memory. 2 Gbytes with transfer rates SDRAM during normal oper- —by Margery Conner
The approach provides mem- equivalent to those of DRAMs. ation, but, if the system loses 컄Agiga Tech, www.agiga
ory capacity of 4 Mbytes to The AgigaRAM relies on its power, the device automati-

Analog Devices’ new You can implement a
ADuM4160 bidirectional fully isolated USB inter-
isolator provides UL face by combining the
(Underwriters Laboratories) ADuM4160 data isolator
1577-rated, 5-kV isolation with the ADuM5000 iso-
for 1.5- or 12-Mbps USB lated dc/dc-converter IC.
(Universal Serial Bus)- Both parts use Analog
data lines. The isolator Devices’ iCoupler isolation
has propagation delay technology, which employs
comparable with that of a on-chip transformers made
standard hub and cable. during die metallization,
It operates at 250V rms along with a Kapton poly-
for IEC (International imide film, to provide the
Electrotechnical voltage isolation. The high-
Commission) 60602-1 quality isolation makes the
CSA (Canadian Standards parts suitable for medical,
Association), 600V scientific, and industrial ap-
rms under IEC 60950- You can create a fully isolated USB 2.0 interface by combin- plications.
1 CSA, and 846V peak ing the ADuM4160 data isolator with the ADuM5000 dc/dc The ADuM4160 is avail-
for VDE (Verband der converter. able in a 16-pin, wide-body
Elektrotechnik, Elektronik SOIC with a suggested
und Informationstechnik) or 3.3V power and provides current is 7 mA at 1.5-Mbps retail price of $4.89 (1000).
certification. IEC 60601-1 isolated control of the data rates and 8 mA at 12- ADuM4160EBZ evaluation
medical-safety approvals pullup resistor, allowing Mbps rates; idle current is modules are also avail-
are pending. the peripheral to control 2.5 mA. The unit operates able.—by Paul Rako
The device operates from connection timing. The over a ⫺40 to ⫹105⬚C tem- 왘Analog Devices, www.
the 5V USB power supply maximum upstream-supply perature range.

12 EDN | JUNE 25, 2009

POE ICs offer increased TOOL BUNDLES
power efficiency for PDs TARGET FPGAs
rather than on-chip isolation.

kros Silicon canni- space. PDs require isolation Xilinx this year introduced
ly plays the efficien- between the power-manage- The TPS23754 supports 13 or two families of Virtex-6
cy card—for both en- ment circuitry for the system 26W PD applications and pro- and Spartan-6 FPGAs that
ergy and space—with its new devices, such as VOIP-phone vides dc/dc-converter topolo- enable targeted-design
AS1854 POE (power-over- electronics and display, and gies that can achieve more- platforms. The company is
Ethernet) PD (powered-device) the primary-side electronics than-90% power-conversion now adding domain-spe-
IC. Akros claims that the inte- and sensor-control circuits. efficiency. The TPS23756 en- cific-tool support to the
gration of parts into the 1854 POE systems require “mas- ables PDs to accept auxiliary targeted-platform concept.
PD-controller platform saves ter” PSE (power-sourcing input voltages as low as 12V, Xilinx’ ISE (integrated soft-
more than 75% in board space equipment) that provides pow- allowing the use of popular and ware environment) Design
and 25% in BOM (bill-of-ma- er to networked equipment inexpensive 12V wall adapters. Suite 11.1 offers interop-
terials) costs. The chip enables through an Ethernet cable. For The TPS23757 supports less- erable domain-specific
higher power efficiency with its an IEEE 802.3at draft version, than-13W systems that re- design flows and tool con-
GreenEdge synchronous recti- the PSE must allocate as much quire the higher-efficiency dc/ figurations for logic, DSP,
fication, which the company as 25.5W per PD and 4.5W of dc topologies. The TPS23754 embedded processing, and
claims can yield greater-than- cable loss. The AS1854’s PSE is available now, and the system-level design. Each
8% improvement in efficiency can determine the PD’s pow- TPS23756 will be available in of the four editions of the
in the 6 to 8W “sweet-spot” er usage and cable losses. the third quarter. Each comes design-tool suite includes
range for VOIP (voice-over-In- In this way, you need not rely in an HTSSOP-20 package for a bundle of tools that will
ternet Protocol) phones, cur- on a worst-case power bud- $1.90 (1000). The TPS23757, match different FPGA
rently the most common ap- get. Budgetary pricing for the also available in the third quar- users’ needs.
plication for POE. GreenEdge AS1854 is $4.69. ter, will sell for $1.65 (1000). For example, the logic
digital-isolation technology al- Texas Instruments has also —by Margery Conner edition—for classic logic
so provides 2-kV on-chip iso- introduced a family of three 컄Akros Silicon, www.akros and connectivity-function
lation, eliminating the need for devices supporting POE PDs. design—includes the ISE
optoisolators that reduce sys- The TPS2375X relies on ex- 컄Texas Instruments, www. simulator, the PlanAhead
tem reliability and add cost and ternal optoelectronic isolators design-and-analysis tool,
the ChipScope debug-
ging- and serial-I/O-tool
kit, and a catalog of IP
(intellectual property).
Similarly, the DSP edition
includes Xilinx’s System
Generator for DSP, plus
related tools and DSP-
specific IP. The embedded
edition supports both
hardware and software
designers, and its IP in-
cludes the MicroBlaze
soft processor. Xilinx
licenses the packages us-
ing the Flexnet system,
and all packages come
configured for global use
over a network. For more
information on these
products, go to www.edn.
—by Graham Prophet
The AS1854 POE platform chip from Akros Silicon includes on-chip 2-kV isolation.

JUNE 25, 2009 | EDN 13

Again, it’s not only a latest-
technology, high-end LED, but
drivers, thermal management,
optics, everything involved to
be able to deliver the concept

VOICES there. For us, that’s a really nice

testimonial to the fact that sys-
tem knowledge and innovative
Avnet’s Jeff Ittel: technologies are going to be

Think board to go broad two key factors, and those are

two things we think we bring to
the market.

eff Ittel serves as senior vice president of business devel-
opment and marketing at Avnet Electronics Marketing ket. You’ve got to have driv-
Americas, a division of components distributor Avnet Inc ers, you’ve got to have thermal While we are on the topic
( Previously, Ittel served for two years as presi- management, and you’ve got to of power-saving LEDs, why
dent of the former Avnet Cilicon division of Avnet Electronics have optics on virtually every is “green” more than a
Marketing, where he and his management team led success- single application. To be able buzzword for components
ful efforts throughout a previous industry downturn. Ittel spoke to deliver the product and our distributors?
with EDN about the changing role of distributors and the support ability to have expertise in each When you look at the
they can offer to board-level designing engineers and suppliers. one of those areas, coupled A way industries have
Excerpts of that interview follow. For the full interview, go to www. with leading suppliers in each been forced to go with ROHS
edn.com090625pb. one of those areas, allows us to [restriction-of-hazardous-
An Avnet exec once told Why is that? bring a solution to an engineer substances] directives and
me that “designing across You’ve got to have prod- that is more than just the latest everything, the new technolo-
the board” was the biggest A uct that is going to be high-brightness LED. ... It’s an gies that come out are more
opportunity for distribu- around for a while for manu- exciting area that requires sys- energy-efficient; they are more
tors. Do you agree? facturability—for lifetime. The tem-level knowledge. environmentally sustainable
Yes, I do. As a distributor, guys that are designing in one and cost-effective. When you
A we see across a whole room and building in another There’s now a $10 million look at that [scenario], ... every-
technology or commodity, so room are becoming fewer and stimulus prize for the first body wins. There’s less throw-
we’ve developed some exper- farther between. They are out- company or engineer that away [product], and product life
tise in which products might sourcing their manufacturing. can design a screw-in LED- is lengthened. Our customers’
optimize certain applications It’s easy if they are outsourcing based replacement for the customers, especially if they
in certain uses. Probably one in the same city to a customer standard Edison 60W light work with government, are tell-
of the biggest things we see we already call on; it’s a little bulb. The new bulb would ing us they are required to start
more and more now is that we more difficult when they are have to be dimmable and designing more green, and that
are not just asked to go in and outsourcing out of their state include LEDs, as well as [trend] is not just smaller, faster,
design in a chip. Customers or region. More and more is driver circuitry, and have and cheaper; it’s all the different
want [to know] how every- getting outsourced overseas. efficacy greater than 90 things I just mentioned: energy
thing else works together. And most [customers] have lumens per watt. This is no efficient, sustainable, cost-
We’ve been asked and are a combination of all of these easy feat. effective, that type of thing.
required to provide more over- [scenarios]. They want to be It’s a great recogni- Customers really have to keep
all solutions. That’s all from able to pick someone who can A tion just for the future up on it. There’s a lot of new
a design point. From a sup- support the supply chain and of LEDs for somebody to put technology coming out. They
ply-chain point, if the prod- move wherever they want to that much money out there. It can stick with what they know
uct is designed-in off our line move it and keep the support says that this is a real technol- and try to figure this all out, or
card, it’s much easier to sup- ongoing. ogy, and we encourage you they can reach out to people
port from our vantage point, to try to adapt to it. For us, who have maybe more market-
whichever way that customer Can you share an example it’s also very clear that, even ing insights and technical train-
wants to be supported in the of designing across the full in your description, it’s not as ing and that can offer support.
supply chain or how they set it board? simple as just plugging in an That’s an area we like to think is
up. Supply chains are getting Lighting is a good LED. It is much more complex. one of our places in the whole
more and more complex, and A example of a solution It is not just a component sale. supply and design chain.
your design decisions affect that is much more than having It is a system sell to be able —interview conducted
your supply-chain decisions a neat, high-tech LED that is to deliver the end state that and edited by
more than ever these days. the brightest one on the mar- people are looking for there. Suzanne Deffree

14 EDN | JUNE 25, 2009


R A Q ’ s

Rarely Asked Questions

Strange stories from the call logs of Analog Devices

What shall we do with an unused op-amp?

(Sung to the tune of “What shall we do with the drunken sailor?”)

Contributing Writer
James Bryant has
Q. In the RAQ on op-amps as com- been a European
parators you mentioned an unused Applications Manager
op-amp in a quad as being possible with Analog Devices
justification for using it as a com-
parator. What do you do if you’ve an since 1982. He holds
unused op-amp and don’t need a degree in Physics
a comparator? and Philosophy from
the University of Leeds.
A. This is a trickier than it looks. If an op-
He is also C.Eng., Eur.
amp is overdriven, the output stage will
saturate at one of the supply rails, and the Eng., MIEE, and an FBIS.
op-amp will consume excess power. Many In addition to his passion
common configurations of an unused up- What we should do is connect the device for engineering, James
amp will overdrive it. as a follower (output to inverting input)
is a radio ham and holds
and connect the non-inverting input to
If the terminals are all left unconnected, a potential somewhere between the the call sign G4CLF.
there is a real risk that stray electrostatic supply rails. With a dual-supply system,
fields will cause an input to go outside the ground is ideal, but connecting to the
supply rails. This can cause latch-up and positive or negative supply of a single sup-
destroy the whole chip. Even if latch-up ply system will cause saturation and the
does not happen, a dc field may cause resulting power waste if the offset voltage Have a question
saturation and power waste. In addition, has the wrong polarity. The “potential involving a perplex-
the amplifier may amplify an ac field and, somewhere between the supply rails” may ing or unusual analog
if overdriven, will heavily modulate its own be any point in the circuit with a suitable
problem? Submit
supply current and cause crosstalk to other potential, since the loading caused by the
op-amp input is minimal. For diagrams see your question to:
amplifier(s) on the chip.
the linked article.
Some users connect one input to the posi-
tive supply and the other input to the neg- Or you might use it as a buffer amplifier
ative supply. This again saturates the out- in a part of your system that does not
put and wastes power; it may also exceed need one but might perform slightly
better if it had one. For Analog Devices’
the differential input voltage rating and
damage the device. Even if damage does Technical Support,
not occur, some input stages draw several What shall we do with the unused Call 800-AnalogD
tens of milliamps under these conditions, op-amp? (X3)
wasting even more power. Early in the morning.

Grounding both inputs, or shorting them Hook-up as a buffer with a dc input, (X3)
together at some other potential, also Early in the morning.
causes the output stage to saturate, since
the offset voltage of an op-amp is never
exactly zero; shorting them together and
not biasing them has the same latch-up To Learn More About
risks that we have already mentioned. Op-Amps as Comparators

JUNE 25, 2009 | EDN 15


charge) cells (Figure 1). A more ac-

curate descriptor for this current error
is “input-leakage current.” The mag-
nitude of leakage current with JFET
or CMOS amplifiers is less than 1 pA
at 25⬚C. This specification is indepen-
dent of the common-mode voltage
and the magnitude-amplifier power.
Almost all amplifiers have ESD cells
Is your amplifier offset for protection from an ESD event, but
you will never see ESD-leakage cur-
way out of whack? rent in bipolar amplifiers. The input-
bias current swamps out the picoam-
ave you ever spent a great deal of time selecting the per- pere leakage current from the ESD

fect operational amplifier for your circuit, only to find cells.
Input-bias and input-leakage cur-
that the offset voltage is wrong at the manufacturer’s rent can change over temperature.
bench-specified input? What if you find that it is more However, depending on the opera-
than 10 times higher than specification in your applica- tional-amplifier design, the bipolar in-
tion circuit? Do you send the chip in for failure analysis put-bias current can be fairly stable.
or just toss the chip out and have another look at your list of amplifiers? The JFET and CMOS input amplifi-
ers may not be, however. Because the
As an alternative, I suggest that you try to explain the offset error by re-
leakage current is from the reverse-bi-
examining your amplifier’s specifications. ased ESD diodes, the leakage current
If you are using your amplifier as the same as the base current of the NPN increases approximately two times per
key component in a transimpedance or PNP transistors at the input of the 10⬚C change.
amplifier, an analog filter, a sample- amplifier. The magnitude of the bipo- In ensuring that the input-leakage
and-hold circuit, an integrator, a ca- lar amplifier’s input-bias current rang- current remains low with JFET and
pacitance transducer, or any other cir- es from a few nanoamperes for low- CMOS amplifiers, you must under-
cuit with high-impedance components power devices to hundreds of nano- stand the impact of your PCB (print-
around your amplifier, you might find amperes for higher-power devices. ed-circuit board) on the picoampere
that the amplifier’s input-bias current The term “input-bias current” loses levels of current. For instance, a small
creates an offset-voltage error through its meaning when you look at JFET or amount of dust, oil, or water mole-
the resistors in your circuit. CMOS input amplifiers. With these cules can increase leakage current and
In the bipolar-amplifier days, the types of amplifiers, the current sink- masquerade as input-bias current. The
term “input-bias current” was an accu- ing or sourcing from the amplifier’s in- good news is that, if you exercise spe-
rate descriptor, and it still is. A bipo- put pins is actually the leakage current cial care, you can build a PCB that
lar amplifier’s input-bias current is the from the input-ESD (electrostatic-dis- will adhere to a 1-pA performance
The most effective way you can re-
duce or minimize the effects of in-
INPUT- put-bias or input-leakage current is to
check your circuit configurations. As

you examine your circuits, look at the
voltage characteristics of each node
⫹_ OUT and make sure that you understand
LEAKAGE LEAKAGE INPUT- the impact of all of the current paths
in your circuit.EDN
Bonnie Baker is a senior applications engi-
neer at Texas Instruments and author of
A Baker’s Dozen: Real Analog Solu-
Figure 1 Input-bias or -leakage current creates a voltage drop across R F. tions for Digital Designers. You can
reach her at

16 EDN | JUNE 25, 2009

+ Read an expanded version of this
article at
+ Go to for
past Prying Eyes write-ups.

Peering into ultrasound machines

ltrasound finds use in noninvasive

U imaging in cardiac, obstetric, gyne-

cologic, and other diagnostic areas.
What’s inside this tool that is playing an
The front-end processing controls and
performs beam-forming, which involves
steering and focusing the phased array
expanding role in today’s medical world? of sound waves. Steering involves
sweeping the angle and direction of the
beam to focal points with a precom-
The systems often operate in the 2- to 20-MHz puted depth. Focusing involves exciting
frequency range. An ultrasound system transmits multiple piezoelectric elements in the
a phased array of sound waves through a linear transducer with precisely time-delayed
transducer so that the waves constructively pulses so that the sound waves con-
combine at a focal point. As the generated sound verge at each focal point along a scan
waves propagate toward the focal point, they line. The receiver’s beam-forming abil-
undergo a slight change in direction and produce ity detects the time delay, phase, and
a reflected sound wave each time they cross amplitude of the reflected sound wave
through matter of different densities. A variable at each focal-point location to recon-
controlled amplifier scales the reflected sound struct the flight path of the waves using
waves before an ADC samples and passes the a delay and sum algorithm to support
data to the front-end processing of the system. creating an ultrasound image.

The midprocessing end lacks a common FRONT END

definition, but it can include filter, detec- TRANSDUCER AND
tion, and compression processing on a ANALOG FRONT END
scan line of beam-formed data. The filter
processing is typically bandpass filtering
to reduce noise. The detection process-
ing extracts a signal from the envelope PREPROCESSING

of the signal. A complex rotator demodu-

lates the signal in baseband; lowpass
filtering then eliminate side lobes. The
system may perform additional lowpass PROCESSING CONVERSION
filtering with decimation or interpolation
before presenting this data for back-end
processing. INPUT/OUTPUT

The back-end processing focuses on forming quality

images for display from the received data. A scan conver- Display-mode-dependent processing is vendor-spe-
sion interpolates raw-data coordinates to displayed-data cific and plays a critical role in producing images on
coordinates. The raw data can be in Cartesian coordinates the display. This type of processing can involve the
for linear probes or polar coordinates for curvilinear or combination of overlays, such as color Doppler and
phased-array probes. Frame-smoothing techniques reduce color-flow imaging. It can also include support for
noise without blurring the image’s edges. Edge-detection the user interface, including menus, help, and display
techniques can help remove this blurring. Doppler pro- options. Images courtesy Zonare Medical Systems
cessing focuses on measuring and displaying shifts and (; for more information about
motion of structures, such as blood flow, in the data. A CW ultrasound, see “Diagnostic ultrasound gets smaller,
(continuous-wave)-Doppler system is analog and is highly faster, and more useful,” EDN, this issue, pg 21.
sensitive and selective so that it can estimate velocities.

18 EDN | JUNE 25, 2009

Accelerating the pace of engineering and science

MATLAB? Over one million people around
the world speak MATLAB.
Engineers and scientists in every field
from aerospace and semiconductors
to biotech, financial services,
and earth and ocean sciences
use it to express their ideas.
Do you speak MATLAB?

Electrons gain 42 billion

electron volts in a plasma
wakefield accelerator.
Provided by Stanford Linear
Accelerator Center.


©2008 The MathWorks, Inc.

The language of technical computing



This output of a CW-Doppler sys-

evelopers of diagnostic-ultrasound machines

tem (above left) does not show an
must carefully consider these devices’ de- image of your body. It represents
sign and their intended applications, making blood flow over time. You can
trade-offs among such factors as SNR (signal- measure maximum velocity and
observe the negative periods of
to-noise ratio), channel count, selection of blood flow as arterial valves close.
ADCs, Doppler versus conventional technol- A cardiogram signal is superim-
ogy, PW (pulsed-wave) versus CW (continu- posed below the ultrasound signal
ous-wave) approaches, power consumption, (courtesy Analog Devices).
cable selection, and cost. With advancements By using a 2-D transducer array
in electronics, smaller and faster machines are emerging that are (above right) or wobbling a linear
array with a stepper motor, you
applicable in a variety of new applications. Portable and 3- and 4-D can derive a 3-D ultrasonic image
machines are also making inroads in this expanding field. (courtesy Texas Instruments).
The diagnostic-ultrasound technique spheres, accurately delivering drugs to
creates images of organs and measures their intended targets. Other applica-
blood flow within patients’ bodies (Fig- tions include chemopotentiation, which
ure 1). The technology is now also find- helps chemotherapy drugs find and de-
ing use in therapeutic applications, such stroy cancerous tumors, and veterinary
as targeted drug delivery. In this sce- medicine, in which patients cannot ver-
nario, the machine emits a pulse wave- bally describe symptoms.
form that couples with microspheres in You can divide diagnostic ultrasound
a patient’s blood; the pulses break these into two broad categories: convention-

JUNE 25, 2009 | EDN 21

al and Doppler. Conventional ultra-  Physicians use ultrasound thera- mitting pulse on one point in the body
sound, also known as B (body) mode peutically for tasks such as breaking (Reference 1). Applying delay to the
or B (beam) scan, represents the classic up kidney stones and diagnostically receiving side makes the reflected waves
use of technology: to peer into organs of for peering into patients’ bodies. adhere to that point.
the body, perhaps to determine the sex Ultrasound machines can have 16 to
 You can divide diagnostic ultra-
of a fetus. Doppler ultrasound relies on 512 channels; those with fewer channels
sound into conventional B (body)-
the Doppler effect, which takes its name may be portable, whereas you may need
wave, PW (pulsed-wave) Doppler,
from Austrian mathematician Chris- and CW (continuous-wave)
wheeled carts to move larger machines
tian Doppler, who proposed it in 1842. Doppler. with many channels. Having many sig-
The Doppler effect describes the change nal channels affects not only the cost
 The transmitting path applies
in the frequency of sound waves for an and size but also the power consumption
observer relative to the source of the digital pulses or DACs to high-volt- of these machines.
waves. You can imagine this effect when age amplifiers. Beware of manufacturers’ channel-
you think of the sound an ambulance si-  The receiving signal path is simi- count claims, however, because they
ren makes as it approaches you, when lar to that of a phased-array radar confuse some would-be purchasers.
it is right next to you, and as it moves system. Some manufacturers use signal-pro-
away from you. PW, or color, Doppler  Spatial resolution is a function
cessing techniques to increase the ap-
encodes velocity information as colors parent channel count and then claim
of the channel count and the SNR
on a display. CW, or spectral, Doppler (signal-to-noise ratio) of the signal
double the number of analog channels
broadcasts a continuous wave of sound path. in the front end. Others multiplex the
into the body and measures the phase analog channels across a wider-chan-
shift of the returning signals. This tech- nel-count transducer. For example, they
nique yields only velocity information, For this reason, physicians typically use may switch a 16-channel-wide analog-
with no profile of where the velocities CW Doppler as an add-on to a con- input section across a 64-channel trans-
occur. ventional ultrasound-signal path. The ducer, perhaps moving one element at a
All ultrasound systems are able to use physician can find an artery or a vein time to “paint” an entire beam-formed
beam-forming and beam-steering ap- with conventional ultrasound and then area across the transducer length. Some
proaches—signal-processing techniques closely examine its blood flow with the manufacturers also provide more trans-
for directional-signal transmission or high resolution of CW techniques. mitters than receivers, an approach that
reception. Physicians can employ PW- Doppler systems employing beam- can provide acceptable but less accurate
Doppler technology to map a patient’s steering use linear arrays of piezoelec- results than using one analog channel
interior organs and blood flow, for ex- tric transducers that steer the beams and for each transducer element. To increase
ample. CW systems must be aimed at a provide for increased spatial resolution. spatial resolution, you can use more an-
certain point because they have no way By delaying the pulses on the center of alog channels than elements and focus
of knowing where the blood is flowing. a linear array, they can focus the trans- between elements, according to Lee D
Dunbar, vice president of market inno-
vation at SonoSite, a maker of portable
ultrasound equipment.
A recent development in this market
is 3-D ultrasound, in which the trans-
ducer is a 2-D, rather than a linear, ar-
ray. This approach allows the system
software to create a 3-D view into a
patient’s body. (For more on 3-D ultra-
sound, see “Peering into ultrasound ma-
chines,” EDN, this issue, pg 18.) A 4-D
ultrasound adds a time dimension to a
3-D image. It can show a moving image
of organs and structures inside the body.
Ultrasound images also rely on the
SNR of the analog-signal path. If there
is less dynamic range in the signal chain,
the ultrasound system cannot discrimi-
nate among the reflected signals. The
Figure 1 Portable ultrasound machines allow doctors to get early information about analog chips’ high noise floor would
trauma and emergency conditions. US Army medics donated this SonoSite machine to mean that conventional ultrasound
the Hawija Hospital in Iraq. Major Charles Buck of the 25th Infantry Division shows an could not peer as far into the body and
Iraqi physician how to use the machine (courtesy Sergeant Sean Kimmons, US Army). would not resolve small features, where-
as CW-Doppler ultrasound would not

22 EDN | JUNE 25, 2009

pick up slow or subtle varia- AMP
tions in blood flow. By trad-
ing off channel count versus HIGH-VOLTAGE ANTIALIASING FILTER
provide the system software RECEIVING SWITCHER AMP
with enough information to
make accurate images. CLOCKS
As in most analog-signal
paths, the first amplifier in TRANSDUCER
the receiver path controls AMP ADC
SNR. You can increase SNR CW
by broadcasting a larger pulse PHASE
into the body, but you must AMP ADC
stay within the safe limits of
energy. The ADC in ultra-
sound systems must provide
enough SNR to accommo- MULTIPLE
date the amplifier chain in TO-CURRENT CROSS-POINT SHIFTER WITH
front of it. In modern sys- DELAY LINES I/O MIX DOWN TO
tems, this task requires a 12- RF INPUTS I DATA
bit ADC for the pulsed path FROM PROBE
and at least an 18-bit unit for
the CW path. Similarly, you
can increase SNR in an an- Figure 2 The analog front end of an ultrasound machine is a complex transmitter-receiver system
alog-signal chain by burning similar to those of phased-array radar systems (courtesy Analog Devices).
more power in the front-end
amplifiers, but this approach
affects battery-power consumption. frequency. Ultrasound cannot travel ultrasound a similarity to MIMO (mul-
Equally daunting, if you build a system through air, so physicians typically apply tiple-input/multiple-output) wireless sys-
with hundreds of channels, you need a gel to the skin where they are using the tems. The same ADC might find use in
the digital processing and software to transducer. Specialized transducers work either system, according to Allan Evans,
do something with all the information. inside the body to provide closer exami- vice president of marketing for fabless-
This approach also consumes power. You nation of the esophageal, reproductive, semiconductor company Samplify Sys-
can use a “slice-count” technique—that or digestive tracts. The ultrasound signal tems. The transmitting section emits a
is, increase the number of channels—to does not penetrate air-filled cavities in pulse in the same way that a sonar or a
improve not only the resolution but also the body, including lungs and intestines; radar system does. The input section is
the speed of imaging and the volume of it can transmit images only from organs a receiver similar to a radar receiver, ex-
tissue you can examine. and from measuring blood flow. cept that it works at the speed of sound
A medical-ultrasound-signal path A medical-ultrasound system is in instead of the speed of light. The input
looks similar to that of a phased-array many ways similar to other communica- path of a CW-Doppler-ultrasound sys-
radar installation, except that the ul- tion systems (Figure 2). The presence of tem has I (in-phase) and Q (quadrature)
trasound-signal path operates at much multiple transmitters and receivers gives demodulation that is familiar to the de-
lower excitation frequencies. signers of base stations and
Unlike a radar receiver, a mod- cellular phones.
ern ultrasound unit may be a The transducers usually
portable device that fits into the comprise quartz-piezoelec-
palm of your hand and runs on tric materials. Researchers
batteries. These machines trans- are also developing MEMS
mit a 2- to 17-MHz ultrasound (mi cro electromechanical-
frequency into a patient’s body; system) CMUTs (capaci-
the lower the frequency, the tive-micromachined-ultra-
deeper the machine can peer in- sonic transducers) to create
to the body. The round-trip at- and receive the sound puls-
tenuation of an ultrasonic signal es. One advantage of these
in the body is 1.4 dB/cm-MHz. devices is that they can
Doppler systems can discrimi- Figure 3 The transmitted pulse from an ultrasound system make repetitive linear and
nate the velocity of only those affects the quality of the image the software produces (cour- 2-D arrays. “These CMUTs
particles that are smaller than tesy Texas Instruments). [eliminate] the whole in-
the wavelength of the excitation terconnect problem,” says

JUNE 25, 2009 | EDN 23

John Scampini, director of strategic
marketing at Maxim Integrated Prod-
ucts. In addition, designers can include
some circuitry on the transducer itself
because it is a silicon die. Unfortunate-
ly, these devices are currently in the re-
search phase.
In general, most ultrasound machines
have no signal-processing abilities in
their transducer heads. Instead, all of
the channels connect through coaxial
cables that must withstand 150V trans-
mitting pulses and must have low stray
Figure 4 Rather than make the analog front end on one die, Texas Instruments com-
capacitance to carry the received sig-
bines a silicon-germanium die and a CMOS die into one package.
nals without undue attenuation. Gore,
a participant in the ultrasound-imaging-
equipment area and a range of other con- struments, for example, ensures that its entire signal path. In addition, the am-
sumer and electronics markets, makes a TX734 pulser transmitter has no undue plifier must have a sufficient power sup-
ribbon-based cable that is electrically ringing as the burst ends (Figure 3). In ply so that there is no clipping of the sig-
equivalent to but thinner than 75⍀ co- CW-Doppler applications, you use one- nal. Input signals can be 0.5V p-p from
axial cable. This thinner cable targets half of the transducer array as transmit- hard reflections, such as the bone in
use in sonographic ultrasound whose op- ters and the other half as receivers. the patient’s body that is closest to the
erators experience repetitive-stress in- The receiver side of a medical-ultra- transducer. Furthermore, should the in-
juries from manipulating heavy cables. sound system shares many characteris- puts overload, the low-noise amp must
Gore’s thinner and more flexible cables tics of communications receivers. Be- quickly and predictably exit saturation.
solve this problem for the operators of cause a 150V transmitting pulse would Manufacturers frequently use a sili-
these machines. Also, machines that overload any amplifier system, the first con-germanium or a CMOS process for
create virtual channels by multiplexing block is a switch that protects the am- low-noise amps, but silicon germani-
16 analog channels across a 64-element plifiers from the transmitting pulse and um provides better linearity and a low-
transducer may include multiplexing then quickly switches in the transducer er noise figure (Reference 3). You can
circuitry in their transducer heads—a so that it can receive the sonic echoes. operate amplifiers employing this pro-
benefit because the cable needs to carry These switches are distinct from systems cess at a higher supply voltage, ensuring
only 16 analog channels and can be four that may be multiplexing 16 analog that you do not overload the amplifier.
times thinner and one-fourth the cost of channels into a 64-element transducer. Another benefit of silicon germanium
a 64-channel cable. All ultrasound systems need the receiv- is that it yields a low flicker-noise cor-
Ultrasound transmitting side of an ul- er switches to prevent overloading in ner, an important criterion in CW-Dop-
trasound system needs to deliver high- the signal path. The switches are often pler ultrasound. Even though the flicker
voltage pulses to the transducer ele- current-fed diode bridges, much like noise is worse at dc, it still modulates the
ments. The ultrasound machine delays the front end of a sampling oscilloscope signal when it passes through the I- and
the pulses at the center of the linear (Reference 2). Q-demodulator sections. Thus, the flick-
transducer so that the sound waves con- The first gain block is a fixed-gain, er noise results in a broader “skirt,” the
verge on a point inside the body. For low-noise amplifier that typically runs widening of the base of the frequency-
this reason, each channel needs its own at 19 dB and often has an active-termi- domain-receiver spectrum. This phase
pulser and perhaps a DAC. The voltag- nation network to minimize noise. The noise obliterates small Doppler signals
es are typically 70V or higher, and peak amplifier’s noise limits the SNR of the and prevents the display of slow and sub-
output currents reach ⫾2A. Frequencies tle changes in blood flow.
range from 2 to 17 MHz, and burst times On the other hand, using large CMOS
are 1 ␮sec to several microseconds. The transistors reduces voltage noise. Ana-
waveshapes are often just simple digital log Devices, for example, claims that its
pulses, but more advanced machines in- AD9272 low-noise amplifier has lower
clude DACs to tailor the pulse shapes. terminated noise than any other ana-
In this case, the transmitter-driver chip log front end. CMOS always excels at
is not a simple MOSFET array but a current-noise specifications because the
high-voltage amplifier. These DAC- input-bias currents are so low to begin
generated pulses can be useful in ther- with. “CMOS has the advantage of hav-
apeutics, in ultrasounds using contrast Figure 5 Ultrasound systems are becom- ing very low input-noise current,” says
agents, and in chemopotentiation. The ing available in handheld form factors Corey Peterson, project leader of the
quality of the burst waveshape also fac- (courtesy Signostics Medical). AD927x-product line at Analog Devic-
tors into the image quality. Texas In- es. “For higher probe impedance, the in-

24 EDN | JUNE 25, 2009

put-noise current can be as significant as erate at 50M samples/sec so that the an- front of the converter because the loop
the input-noise voltage.” He also notes tialiasing filters have sufficient roll-off filters in the converters inherently pre-
that the company’s triple-well process and flatness to prevent higher-frequen- vent aliasing. With CTSD converters,
prevents digital-substrate currents from cy harmonics from mixing back into the the internal loop amplifiers control the
interfering with the analog signals and signal and reducing quality. With active bandwidth, so you cannot undersample
that using large input transistors can re- filtration, however, you also must speed them. As a result, a 50M-sample/sec
duce flicker noise to acceptable levels. up overload recovery. Passive networks, CTSD converter can operate only at
Although the higher flicker-noise cor- on the other hand, need no recovery that sampling rate; it doesn’t work at
ner of CMOS is unavoidable, it is often period. You should characterize your 25M or even 40M samples/sec. Another
acceptable if you do not use CW mode SAW devices for overload, but they also caveat has to do with overload recov-
and if the machine is a low-performance recover nearly instantaneously. ery. Because pipelined converters carry
portable model. The analog signal ends at the ADC, their samples in separate sections of a
After passing through the low-noise which is typically a 12-bit, 50M-sample/ pipeline, they inherently recover well
amplifier, the signal splits off between sec pipeline device. Older systems use 8- from overload—often during one sam-
a CW-Doppler path and a B-mode path bit converters for B-mode scans, but the ple. CTSD-ADC designers must tack
and conventional color-Doppler path. overload recovery with clamping net-
The conventional path starts with a works onto the internal integrator, an
PGA (programmable-gain amplifier)
DOZENS OF CHAN- approach that typically reduces SNR
that typically adds 40 dB of gain. The NELS OF 12-BIT DATA and increases power consumption while
PGA varies dynamically as it receives the part is in an overloaded state. This
each pulse; thus, the signals that reflect
WOULD USE UP ALL situation is not a serious problem be-
from deeper in the body receive more OF THE PINS ON EVEN cause the ADC typically does not re-
gain. This approach allows a channel THE LARGEST FPGA. main in saturation for more than a few
SNR of 110 dB and extends the effective cycles.
70-dB dynamic range of a 12-bit ADC. Because dozens of channels of 12-bit
As with a low-noise amp, bandwidths data would use up all of the pins on even
for PGAs are approximately 20 MHz. the largest FPGA, the converters use
One vital characteristic of PGAs is re- serial outputs, such as LVDS (low-volt-
covery from overload. You can expect age-differential signaling). Parts that
hard tissue or bone to reflect a large sig- have eight ADC channels in one pack-
nal that overloads the PGA. The system age and multiple channels on one pair
can sense this signal and reduce the gain of differential outputs further reduce the
at that point in the receiving time, but pin count in beam-former FPGAs. Be
the faster the PGA recovers from over- aware that SNR specs can be deceiving.
load, the sooner the system will have ac- requirements of pulsed Doppler have “An ADC can have very poor near-car-
curate information to process. Overload driven bit counts to 12. Another factor rier SNR and still having excellent full-
can also create frequency modulations to consider is harmonic-imaging mode. Nyquist-band SNR specifications,” says
that are indistinguishable from blood Compressed tissue under an ultrasound Maxim’s Scampini, who also warns that
flow in a Doppler system. pulse reflects back second harmonics, flicker noise in the ADC reference can
Amplifier and ADC vendors measure and a 12-bit converter allows you to cause poor near-carrier SNR.
overload recovery differently, and even sense these harmonics and compensate The other analog path in a medical
ADC vendors differ among themselves for the tissue compression, allowing for ultrasound front end is for CW Doppler.
in how they evaluate overload recovery. better resolution in the scan. Although This path needs greater SNR than the
This measurement is not so much about SAR (successive-approximation-regis- 110 dB that the B-mode and PW-Dop-
specmanship but about the probabilistic ter) converters provide better SNR fig- pler paths provide. “Bear in mind that
nature of the overload regimens the sys- ures, pipelined converters are accept- the transmission signal is continuous and
tem must undergo. “You have to work able. The VGA rather than the ADC is also being received,” says Samplify’s
with customers carefully and closely determines the SNR of the signal chain. Evans. “You can also imagine that the
to translate what is a system phenom- How quickly the converter recovers refection from a blood vessel wall would
enon of overload into a spec for a part,” from overload conditions may be more be far larger than the Doppler scat-
says Scott Pavlik, a marketing manager important than SNR, however. tering of the blood carried inside the
in the health-care segment at Analog Ultrasound systems can alternative- vein. In the CW Doppler, you are try-
Devices. ly use CTSD (continuous-time-sigma- ing to maintain a dynamic range of 154
An antialiasing SAW (surface-acous- delta) converters, which are available dBm [decibels referenced to milliwatts]/
tic-wave) or a passive filter follows the from National Semiconductor and An- Hz. The thermal noise floor of 50⍀ is
PGA. Although the 20-MHz bandwidth alog Devices. Systems using this topol- ⫺174 dBm/Hz.”
is high for active filters, amplifiers are ogy have low power consumption for Because the CW-Doppler technique
continuously pushing this boundary. It is their speed and SNR. CTSD convert- does not use pulses, you cannot disregard
common for the ultrasound ADC to op- ers also require no antialiasing filter in the large signals by blanking them out

JUNE 25, 2009 | EDN 25

or attenuating them for a certain time. front end, you would need to add only
Instead, the technology picks off the ul- A FUNDAMENTAL the transmitting functions, the input
trasound signal after the low-noise-amp switches, and the amplifiers and ADCs
section and sends them to a demodula- CHOICE IN BUILDING for the CW-Doppler function. Texas In-
tor section that performs analog-signal AN ULTRASOUND SYS- struments’ AFE5804 analog front end
processing. Doppler signals represent integrates the amplifiers and ADC in
changes in phase, so the I and Q de-
TEM IS HOW MUCH one package but uses a low-noise amp
modulation that advanced-communica- INTEGRATION TO USE. and VGA on a silicon-germanium pro-
tions techniques use is also applicable to cess and mounts it as a separate die into
mixing down the 2-MHz CW signal to the same package as the CMOS ADC
baseband. This demodulation allows the (Figure 4). The company also offers the
system to discriminate between blood AFE5851, which integrates the VGA
flow toward the sensor and blood flow and ADC but omits the low-noise am-
away from the sensor. The I and Q com- plifier, so you can use an external part.
ponents then go to a higher-resolution, “Customers want in the long term to
slower-sampling ADC. Demodulation put the electronics into the transducer,”
has moved down the signal from the fre- says Veronica Marques, strategic-mar-
quency of the transmitting pulse to the tem is how much integration to use. keting manager for the medical-business
frequency of blood flow in your veins. Analog Devices this year won an EDN unit at Texas Instruments. “If customers
Suitable ADCs are 16- or 18-bit devices Innovation Award for its latest ultra- want to integrate the low-noise amp in-
sampling at 150k or 100k samples/sec, sound front end (Reference 4). The to the transducer, they can still use a TI
respectively. The I and Q components company’s eight-channel AD9276 in- integrated analog front end,” she says.
then move to the spectral-processing cludes the low-noise amp, the PGA, Choosing the integrated front end locks
subsection in the beam-former FPGA. antialiasing filters, and an ADC—all you into one vendor, making it impos-
on one CMOS-silicon die. Leveraging sible to upgrade separate blocks, such as
INTEGRATION TRADE-OFFS Analog Devices’ expertise in commu- the PGA or the VGA.
A fundamental choice you will have nications, the part also integrates the I Maxim, on the other hand, offers
to make in building an ultrasound sys- and Q demodulators. To make an analog high-performance, silicon-germanium,

Design Services

The leader in wireless product development

26 EDN | JUNE 25, 2009

low-noise amps and then lets you choose F O R M O R E I N F O R M AT I O N (electromagnetic interference), and the
PGAs and ADCs from the company or Analog Devices Signostics Medical compression routines also reduce fre-
from a competitor. The MAX2038 in- www.signostics quency peaks and keep the EMI spec-
tegrates an amplifier with the quadra- trum more like pink noise. In addition SonoSite
ture mixer. The MAX2078 integrates products/medical/oem/ to the chips, the company can provide
the amplifier, VGA, filters, and CW- imaging Tektronix you the IP (intellectual property) for
Doppler mixers across eight channels. If Maxim Integrated your FPGA or DSP to decompress the
you want to differentiate your product
Texas Instruments signal. One interesting development
by its software or user interface, using National
has been the use of lossy compression,
an integrated analog front end makes Semiconductor
which provides medical images that are
sense, especially if your design has space
25th Infantry Division
indistinguishable from those from non-
and power constraints. SonoSite, on Samplify Systems compressed systems.
the other hand, uses a proprietary ana- UMC Other factors you should consider
log front end because the results of the when using integrated chips are their
cost and design time show up in the cesses are better suited to those blocks availability and their obsolescence.
displayed images. All of the company’s (Reference 5). Samplify’s Evans notes Due to the fine-line CMOS content
analog front ends are custom-designed. that a low-power, 12-bit, 50M-sam- in a multichannel ADC system, many
“The chip makers are interested in le- ple/sec converter will work across sev- of the analog-semiconductor vendors
veraging our technologies,” says the eral products—from cart-carried sys- farm the fabrication out to TSMC (Tai-
company’s Dunbar. “But they still have tems to portables. “You can then mix wan Semiconductor Manufacturing
a way to go.” and match low-noise-amp/VGA front Co), UMC (United Microelectronics
Another integration factor involves ends, depending on your power/perfor- Corp), or other large digital-CMOS-fab
the digital-data side of the front end. mance trade-offs,” he says. Samplify’s vendors. This approach reduces your
Samplify Systems, for example, which 16-channel ADCs have built-in loss- chances of ending up on an allocation
also this year won an EDN Innovation less and lossy data compression, reduc- list when the boom in semiconductor
Award, does not integrate the low-noise ing the cost, the complexity, and the demand inevitably comes. Equally con-
amp and PGA into the ADC converter data rates to the beam-forming chip. cerning is obsolescence. It is not benefi-
because high-voltage or low-noise pro- The reduced data rates have lower EMI cial to design in a proprietary chip for

View Online Now at PO ER
Get Practical Answers To Your
Most Important Power Questions
View Online NOW!
Vicor's PowerTechtorial Series concentrates
on important, real-world technical issues
in power system design.
Questions, posed by power designers worldwide,
are answered through concise, expert instruction.
Each techtorial, on or on CD, is
accompanied by relevant technical articles and
application notes.
Electromagnetic Interference: Gain access to view an ever-growing number
NEW How to Get the Lowest Noise
of Vicor PowerTechtorials. You'll be informed
as each NEW techtorial becomes available.
NEW Thermal & Mechanical Considerations As always, technical assistance with your power
system design is there, worldwide, from
NEW Input Overvoltage Protection Vicor Applications Engineering at 800-927-9474.
Improving Output Filtering

Vicor DC-DC Converter Theory Overview

cost, space, and power considerations + Go to R E FE R E NCE S
if the vendor plans to stop producing and click on Feedback Loop to post
1 Brunner, Eberhard, “How Ultrasound
the chip. Semiconductor vendors know a comment on this article. System Considerations Influence Front-
that the medical-device market differs End Component Choice,” Analog Dia-
from the consumer-electronics market. + For more technical articles, go to logue, Volume 36, No. 3, May—July,
Customers for medical devices must get 2002,
FDA (Food and Drug Administration) Dialogue/archives/36-03/ultrasound.
approvals and expect chip runtimes of cycles have decreased from five years to 2 “Sampling Notes,” Tektronix, 1964,

a decade or more. Although it is more less than two years. A product that finds
difficult to win a medical-ultrasound use in aircraft in a battlefield will need lotr?ct=TI&cs=afs&ci=14824&lc=EN.
socket, once a company gets that sock- approvals from the FDA, Federal Com- 3
Rako, Paul, “Silicon germanium: fast,
et, the part can flourish for years. Even munications Commission, CE (Confor- quiet, and powerful,” EDN, Sept 18,
in a recession, people still become sick. mité Européenne), FAA (Federal Avia- 2008, pg 27,
Therefore, all the analog-semiconduc- tion Administration), and Department CA6594090.
tor companies keep their proprietary of Defense. You may have to design 4 “AD9272 eight-channel ultrasound

chips in production for as long as cus- systems that work in a decompression receiver,” EDN, March 30, 2009, www.
tomers need them. chamber to evaluate Navy SEAL (sea/
Remember that the architecture of air/land) special forces or Air Force per- 5
“AM1610/05/00 12-bit ADCs,”
an ultrasound system involves an ana- sonnel at altitudes of 30,000 feet. EDN, March 30, 2009,
log trade-off (Figure 5). You must de- You can understand why engineers article/CA6631770.
cide whether CW-Doppler capability is devote their entire careers to design-
important, given that it requires so many ing analog front ends for medical-ul- You can reach
dedicated analog circuits. You must trasound systems. “There are a lot of Technical Editor
weigh multiplexing 16 channels into 64 people that get into it and then get ad- Paul Rako at
against the performance this approach dicted to it,” says SonoSite’s Dunbar. 1-408-745-1994
will yield. You must trade off silicon ger- “All of us are addicted to ultrasound. and paul.rako@
manium versus CMOS, quartz versus You are building something that helps
MEMS, and DSPs versus FPGAs. Design somebody.”EDN

NEW Panel Meters

We’re METER MAD at Our design engineers have been working like crazy to bring you the very best
in digital panel meter technology. Here are just a few of their latest designs!
Murata Power Solutions!
DMS-20RM Series DMS-20PC Series ACA-20RM-ALM Series
True-rms AC Low-Power Blue True-rms AC
Voltmeters with and Green LED Ammeters with
Isolated Inputs Voltmeters Alarm Function
t AC mains or DC powered t Only available from Murata t User-settable alarm function,
t 2000V isolation Power Solutions factory or field adjustable
t True-rms AC measurements of t Draws 12mA from 5V supply t Flashing display indicates
complex waveforms or <0.5mA in “standby” mode overcurrent fault conditions
t Large, easy-to-read red display t 12-pin DIP package

For specifications or a sample, visit our website:

11 Cabot Boulevard, Mansfield, MA 02048-1151, USA Tel: 800 233 2765 Fax: 508 339 6356

28 EDN | JUNE 25, 2009

Why is Agilent the fastest growing
oscilloscope manufacturer?

Because we listen to you.

To build our scopes, Agilent carefully examines the challenges you face. Then we
deliver products that solve your problems in imaginative ways. Like the multi-chip
module that enables Infiniium’s industry-leading signal integrity. And the ASIC that
underlies InfiniiVision’s patented MegaZoom deep memory giving you the industry’s
best signal visibility. You’ll find innovations like these in each of our scopes —
that’s why more and more engineers are choosing Agilent over other scope brands.*

Agilent 20 MHz to >90 GHz real-time and sampling scopes


See why more and more engineers choose Agilent.

Download our catalog



© Agilent Technologies, Inc. 2008 *Prime Data August 2008 CAGR T&M Report u.s. 1-800-829-4444 canada 1-877-894-4414



Accelerating consumers’ NAS adoptions:


OPTIONS t least one potentially posi-

NAS would also act as a backup repository for all the comput-
tive counterpoint—the NAS ers on the network. Translating this vision into reality, howev-
(networked-attached-stor- er, requires that home-NAS suppliers deliver an easily justifi-
able price for the target market; an easy-to-grasp and compel-
age) server—shines among ling sales pitch; an easy way for consumers to both integrate
the abundance of predomi- the NAS within their networks and subsequently access it
nantly negative economic from diverse devices; sufficient speed in storage, retrieval, oth-
news about the technology er processing functions, and network bandwidth; and a care-
sector, particularly consumer fully crafted set of features and cosmetics.
electronics. People continue In the more than 12 years that I’ve been dabbling in home-
office NAS, I’ve seen abundant evidence of both evolution
to take still and video pictures, listen to music, and maturation in the consumer-NAS-product category. Ac-
and download movies—maybe even more so companying these trends, both diamonds and lumps of coal
than in the past—because they’re now staying have emerged across the dozens of products I’ve used (see side-
home and looking to entertainment as a means bar “Hardware-test beds”). Therefore, this article aims to pro-
of distracting themselves from their recession-re- vide not just a snapshot of current system and silicon-and-soft-
ware building blocks but also a forecast of how the NAS cat-
lated woes. More of them are also now working egory might further develop, with the guidance of historical
from home-based offices rather than in the cu- precedents, product capabilities, and customer expectations.
bicles of times past, when large enterprise servers
and IT (information-technology) personnel met and managed THE NETWORK TETHER
their corporate-storage needs. Further, an increasing percent- Begin the architecture definition of your next NAS design
age of their homes contain reasonably robust networking set- from the outside, focusing first on its LAN interfaces. Wired
ups, enabling various LAN (local-area-network) clients, such Ethernet is the most common LAN-tethering approach—with
as computers, game consoles, media extenders, and printers, to good reason. Because NAS normally operates in a “headless”
not only share a common Internet connection but also inter- fashion—that is, without the need for a keyboard, a mouse,
communicate. and a display—it can easily locate nearby the router and con-
All of these trends suggest the allure of a consolidated nex- nect to it over Category 5, 5e, or 6 cable. Wired-Ethernet con-
us in consumers’ residences for both professional and personal nections are comparatively robust and speedy. And your cus-
content that multiple LAN clients could simultaneously ac- tomers can leverage some other networking technology by us-
cess. Ideally, this centralized storage would implement a RAID ing an external bridge adapter.
(redundant array of independent disks), which would protect However, for aesthetics, operating noise, or other reasons,
the NAS from the failure of any one hard-disk drive, and the your customers might instead want to hide the NAS in some

30 EDN | JUNE 25, 2009

out-of-the-way location, such as a clos- unit, power-line networking might be a
et. In these cases, you’ll want to first en- AT A G L A N C E tempting alternative-LAN-interface ap-
 Only power users can justify pay-
sure that you’ve educated your custom- proach. Not every power outlet is a valid
ing for blazing-fast network tethers.
ers on the need for consistent NAS ac- power-line-networking candidate, how-
cess to sufficient supplies of cool, ambi-  Mirrored drives increase a sys- ever, and performance and reliability
ent air. Consumers probably won’t want tem’s size, weight, power consump- also vary with time of day and time of
to string Ethernet cable around their tion, and price, but they also prevent year (Reference 2). So, at least for now,
homes, so they might be willing to pay customers’ data loss and frustration. ignore the temptation to integrate this
extra for an integrated alternative net-  Ensure that your selected oper- feature and stick with an external Eth-
work-access technology. ating system and applications have ernet-to-power-line-bridge adapter. The
Wi-Fi is probably not only the first robust features and are interoper- three contending power-line-network-
approach that would come to mind but able, but hide advanced features ing technologies—HomePlug AV, UPA
also the leading candidate by virtue of from neophyte users. (Universal Powerline Association), and
its pervasive presence in modern rout-  ARM and x86 appear to be the
HD-PLC (high-definition power-line
ers. Keep in mind that, even in its lat- dominant CPU architectures for
communications)—exhibit no serious
est-generation 802.11n form, it’s like- consumer-tailored network storage signs of interest in pursuing interopera-
ly to be a lower-performance approach in the future by virtue of their ubiq- bility, so if you “bet on the wrong horse,”
than wired Ethernet, however. Perform- uity and intense industry focus. you’ll add cost to your design and gain
ance isn’t the only reason to focus on nothing (Reference 3).
 Cost-effective and power-thrifty
802.11n. Because 802.11g and other Similarly, I don’t currently recommend
hardware has proved valuable for in-
wireless predecessors are now mature, that you integrate either a HomePNA
depth hands-on evaluations.
they won’t provide justification for a (Phoneline Networking Alliance) or a
substantive price premium. MOCA (multimedia-over-coaxial) trans-
Speed aside, Wi-Fi is also less reliable entific/medical)-band options, as well as ceiver. Neither technology is sufficiently
than wired Ethernet, due to RF (radio- how elaborate and expensive to make the mature to be in use by much of your po-
frequency) interference and other issues, unit’s MIMO (multiple-input/multiple- tential customer base. And the need to
so you’ll need to ensure that the NAS output)-antenna array (Reference 1). connect such a NAS to a phone-line- or
recovers from dropped network connec- Because the NAS is ac-powered, coaxial-cable-based network tether is too
tions in a user-friendly and data-preserv- thereby requiring a nearby wall outlet location-restrictive for many homes.
ing manner regardless of what operating no matter where your customer puts the
mode it’s in at the time. And the need to HOW MANY DRIVES?
provide the NAS with both the WLAN Unless your target customer is a power
(wireless-LAN) SSID (service-set-iden- user and particularly considering that a
tifier) and encryption-key information notable percentage of LAN clients will
before it can make the Wi-Fi connection likely be accessing the NAS over low-
is a challenging setup requirement for a bandwidth Wi-Fi connections, it may be
headless-system design. Finally, you need difficult to justify the incremental cost
to decide whether to support both the of 1-Gbps GbE (gigabit-Ethernet) LAN
2.4- and the 5-GHz ISM (industrial/sci- transceivers versus conventional 10/100-
Mbps alternatives. If a built-in Wi-Fi or
100-Mbps wired-Ethernet bottleneck
constrains the NAS transfer-rate speed,
there’s little rationale for a performance-
tailored RAID 0, RAID 5, or similar
multidrive-striped-storage architecture
behind the network PHY (physical)-
layer IC. Conversely, if you believe your
target customer will see tangible value
in GbE or multistream, bonded-channel
(a) (b) 802.11n-networking capabilities, you
should seriously consider corresponding-
ly beefing up your design’s drive array.
Including more than one drive in your
Figure 1 Linksys’ first stab at a full-blown network-storage appliance for consumers, the design typically costs more unless you’re
NAS200 (a), had insufficient performance potential and, therefore, too few features. A comparing, say, a leading-edge 2-Tbyte
switch from x86 to a Marvell ARM-based CPU substantially boosted the capabilities of drive with two more mature 1-Tbyte
follow-on Media Hub devices (b). alternatives in a concatenated arrange-
ment (Reference 4). Using a multidrive

32 EDN | JUNE 25, 2009

design also means that the NAS will greater likelihood of needing to employ 7200-rpm alternatives. Thanks to dense
need a larger system form factor, gen- a noisy system fan. As such, serious- bit-packing PMR (perpendicular-mag-
erate more heat, and, therefore, have a ly consider 5400-rpm drives instead of netic-recording) techniques, the slower-

I’ve been dabbling with network storage as long as I’ve • Ximeta’s NetDisk.
been with EDN. Check out this list of hardware I’ve tried One other recent hardware evaluation begs for more
out over the years, which, given my imperfect memory, in-depth discussion. I mated a 1-GHz Via Technologies
may not be comprehensive: C7 CPU-based EPIA SN mini-ITX motherboard to a
• Addonics Technologies’ NASU2; Casetronic Travla C137 enclosure (Figure A). I custom-
• ADS Technologies’ NAS (network-attached-storage) ized the C137 to hold dual 3.5-in. hard-disk drives from
drive kit; both Seagate and Western Digital for mirrored storage.
• Buffalo Technology’s LinkStation, LinkStation Pro, Because many of the NAS systems on the earlier list
and TeraStation; use modified Linux distributions, I focused this evalua-
• D-Link’s DNS-323 and DSM-604H; tion on Windows Home Server, which Microsoft derived
• Intel’s SS-4200E; from Windows Server 2003, and FreeNAS, which its
• Linksys’ NAS200, NMH305 Media Hub, and NSLU2; open-source developers based on FreeBSD (Berkeley
• Maxtor’s Shared Storage II; Software Distribution) and which they recently and
• Netgear’s ReadyNAS X6, ReadyNAS NV, ReadyNAS conveniently upgraded to Version 0.69.1. In its “embed-
NV⫹, ReadyNAS NVX, and SC101 Storage Center; ded”-mode option, FreeNAS runs completely from flash
• Toshiba’s Magnia SG10; memory. Although the EPIA SN offers an integrated
• Tritton Technologies’ T-NAS; CompactFlash card slot, I instead installed FreeNAS on
• Via Technologies’ Artigo A2000; a USB (Universal Serial Bus)-flash drive.
• various Windows and Mac OS X-based computers, Both operating systems ran smoothly, although I
along with an Apple Power Mac G4 Cube running OS admittedly didn’t load them up with abundant add-ons,
10.3 Server; and and they had comprehensive feature sets. Via’s core-
logic chip set on the EPIA SN supports hardware-accel-
erated RAID (redundant array of inexpensive disks), a
key advantage when you partner it with a power-thrifty
but performance-strapped CPU. However, Microsoft’s
proprietary mirroring approach in Windows Home
Server couldn’t tap into its features.
Similarly, I couldn’t track down BSD drivers that would
enable me to use the EPIA SN’s core-logic chip set to
offload the C7 CPU from handling some or all of the
FreeNAS software-RAID algorithms. As with many other
enthusiast-driven open-source projects, FreeNAS offers
scant and incomplete documentation. The user inter-
face, although feature-rich, is correspondingly complex
and unintuitive. Should you decide to use FreeNAS as
the foundation of your next NAS design, I’d encourage
you to focus some tangible effort in polishing these
areas and, per the open-source license, to return your
results to the organization so that it can incorporate
your improvements.
One other minor frustration involved the EPIA SN
BIOS (basic input/output system). My version of the
board contained the initial Version 1 firmware image.
(b) Via offers a newer Version 2.01 BIOS for downloading
from its Web site, but the only corresponding update
Figure A Combining the 1-GHz fanless-CPU version of Via utility the company provides runs under DOS. I couldn’t
Technologies’ EPIA SN mini-ITX board (a) with a two-drive even execute it in command-line mode under Windows
customized version of Casetronic’s Travla C137 enclosure Home Server. As soon as I track down an old copy of
(b) and hard-disk drives from Seagate and Western Digital DOS, along with a floppy drive to install it, I’ll be able to
forms a stable and robust test bed for several NAS operat- upgrade my BIOS.
ing systems.

JUNE 25, 2009 | EDN 33

spinning drives still deliver robust trans- terparts will be advocating that your cus- convince your customers of the value of
fer rates, and they consume notably less tomers should use the NAS as a single- redundancy within the NAS, yes? This
current. Despite the downsides of using point-of-storage contact for all of their topic brings up a bigger issue regarding
multiple drives, avoid selecting a non- precious—often irreplaceable—digital how the NAS market may evolve in the
mirrored-drive architecture unless the data: music libraries, photographs, vid- future. Today’s NAS suppliers include
customer will use the NAS exclusively eos, financial records, and the like. Un- traditional hard-disk-drive companies,
for connected-computer backup. In the less you use a RAID 1, RAID 5, or other such as Seagate and Western Digital;
backup-only case, if the NAS drive fails, mirrored-drive arrangement, such as In- traditional network-equipment vendors,
your customers will likely be able to frant’s (now Netgear’s) proprietary and such as Cisco’s Linksys division, D-Link,
swap in a replacement drive before any flexible X-RAID, an inevitable drive and Netgear; and start-ups, such as Data
backed-up computer’s drive also fails. failure will render that information per- Robotics. Hard-disk-drive companies are
Think about it: Your marketing coun- manently irretrievable. You can surely understandably more loath than compa-
nies in the other two categories to ad-
mit to the inevitable impermanence of
NAS ADAPTERS AND ALTERNATIVES drives. Also, is there a future NAS-sup-
Although a full-blown NAS (net- disk drives that they might want
plier role for traditional consumer-elec-
work-attached-storage) system to migrate from a single-computer
tronics companies, such as Samsung or
might be the preferable approach tether to full network access.
for newcomers, some consum- Linksys was among the first to
All this talk about hard drives inev-
ers already own USB (Universal address this market with its NSLU2
itably brings up the topic of the solid-
Serial Bus)-based external hard- network-storage link (Figure A).
state-drive alternative (Reference 5).
Advocates of open-source soft-
These drives are increasingly becoming
ware have built on the code that
available in cost-effective capacities that
Linksys provides to turn the NSLU2
make them compelling hard-disk alter-
into a full-featured server appli-
natives for client computers. However,
ance. More recently, Addonics
the bulk-storage nature of NAS makes
Technologies released the NASU2
it likely that it will continue as a hard-
NAS adapter, which is substan-
drive candidate at least for the next few
tially smaller than the NSLU2 and
years. Near-term pragmatism aside, in-
includes built-in Bittorrent, FTP
creased flash-memory density and lower
(file-transfer-protocol), and print-
prices are indisputable trends, particu-
server capabilities for both LAN
larly since the advent of multilevel-cell-
(local-area-network) and WAN
storage techniques. As such, solid-state
(wide-area-network) users.
drives’ increased reliability and perform-
A NAS adapter isn’t the only way
ance, decreased power consumption
to get a USB-inclusive hard-disk
and heat dissipation, and silent opera-
drive online. An increasing number
tion will likely in the future encourage
of routers, for example, embed USB
their adoption in NAS at hard drives’
ports that accomplish a comparable
function. Some routers even provide
WAN access to the drive either free
or with a $99-per-year MobileMe
If you constrain your NAS brain-
(a) storming to only networked bulk stor-
membership for Apple’s routers.
Carefully select the file system the
age, you might at first glance think that
networking equipment uses to for-
any of a number of operating systems
mat the drive. The FAT (file-alloca-
could suffice. Dig a bit deeper, though,
tion-table) system, for example, has
and you’ll quickly realize that it’s more
the advantage of operating-system
difficult to solve this problem. First, a
ubiquity. However, even the newest
tangible percentage of your users will
FAT32 variant exhibits a 4-Gbyte-
likely want to be able to carve up the
file-size limit, along with an operat-
available capacity into more than one
Figure A Linksys’ NSLU2 (a) jump- ing-system- and cluster-size-depen-
shared-storage resource, with per-share
started the USB NAS-adapter cat- dent volume limit. These limitations
access rights, such as disabled, read-
egory, which has now expanded to may be too constraining, depending
only, or read/write, that customers will
include products such as Addonics’ on how your customers use network
define on a per-user and -group basis.
NASU2 (b), a more fully featured out- storage.
They’ll access the networked storage
of-the-box offering. from LAN clients running various oper-
ating systems and therefore with various

34 EDN | JUNE 25, 2009

supported file-access protocols, such as (see sidebar “NAS adapters and alterna- virtualized on another operating system
AFP (Apple-filing protocol), NFS (net- tives”). Windows currently integrates (Reference 6).
work-file system), and SMB/CIFS (serv- no comparably robust backup features, Next consider the laundry list of oth-
er-message block/common Internet-file although both Microsoft and third par- er NAS capabilities that your potential
system). They’ll also want both configu- ties can subsequently augment the oper- customers might value and, therefore,
ration and subsequent access to work in ating system with such capabilities. You pay extra for. These features include on-
a way that doesn’t force them to com- also might want to include Rsync sup- the-fly encryption during storage and
prehend and grapple with the underly- port for Unix clients. Keep in mind that subsequent decryption during read-back
ing complexity. users often want to back up files that the of information archived on the NAS,
LAN-client backup, another com- operating system or an application run- along with USB (Universal Serial Bus)
monly requested NAS feature, is simi- ning on it is currently using. As a case ports for printer serving, augmented
larly challenging to implement in a si- study of the concept, although the Con- storage capacity, and networked access
multaneously robust and trouble-free nector client-side software for the Win- to scanners and other USB peripherals.
manner. Apple OS 10.5’s built-in Time dows Home Server NAS operating sys- Your customers might also want auto-
Machine capability, for example, initial- tem generally runs well, it’s not without matic network discovery through proto-
ly supports full backups and subsequent- limitations and quirks. It automatically cols such as UPnP (universal plug and
ly supports incremental backups to AFP- wakes up PCs once a day, even if they’re play) and Apple’s Bonjour—that is, Ze-
cognizant storage media. However, Ap- on standby at the time, but it sometimes roconf. Media streaming is also on the
ple officially sanctions backups only to fails to put them back to sleep once list. Protocols such as UPnP AV (audio/
its own Time Capsule hardware, which backup completes. Automatic wakeup video) and DLNA (Digital Living Net-
combines a router and a hard-disk drive also doesn’t work if Windows is running work Alliance) enable this feature both

The relatively archaic 1-GHz Via the system with noise levels remain- scalar, out-of-order architecture that
Technologies processor I used for ing below 26.8 dB” (Reference A). Via includes in its Nano CPU and
this project seemed to have suf- Don’t forget about Intel’s cost- implements in its VB8002, the first
ficient horsepower for entry-level effective, single- and dual-core Nano-based and media-server-tai-
NAS (network-attached-storage) Atom CPUs and mini-ITX boards lored mini-ITX board.
applications. However, as you beef that the company based on them.
up your software with concurrently HyperThreading’s virtual-multi- R E FE R E NCE S
running utilities, you might find core support delivers even more A Artigo A2000 Barebone Storage
that your design begins to slow horsepower in some configurations Server, Via Technologies, www.via.
down. If so, consider Via’s Artigo (Reference B). Also, if you’re doing
A2000, which includes a 1.5-GHz C7 PVR (personal-video-recorder) artigo/a2000.
CPU in a NAS-friendly dual-3.5-in. applications, such as video encod- B Dipert, Brian, “The price of falling
hard-disk-drive case (Figure A). The ing before archiving or transcoding prices: evaluating value-oriented x86
A2000 also features “a quiet, ball- before streaming, you might want to CPUs,” EDN, Jan 8, 2009, pg 30, www.
bearing fan, [which] silently cools consider using the three-way-super-

(a) (b) (c)

Figure A Via’s Artigo A2000 NAS-tailored design embeds a 1.5-GHz CPU and a nearly silent system fan (a), whereas Intel’s single-
and dual-core Atom boards support a more modern CPU architecture (b). For stringent performance demands, consider Via’s VB8002
board, which the company based on its most recent Nano CPU, complete with a three-way-superscalar, out-of-order architecture (c).

JUNE 25, 2009 | EDN 35

across the LAN and over a WAN (wide- in automatic Windows Update capabili-
area-network) connection. Firewall-sur- WARN YOUR ty of Microsoft’s Windows Home Server
mounting technologies, such as UPnP is one notably robust implementation of
and NAT-PMP (network-address-trans- CUSTOMERS OF the concept; Apple’s Time Capsules al-
lation/port-mapping protocol), support PENDING PROBLEMS so regularly check for updates and alert
the WAN connection. users to their availability.
Customers might also pay for addi-
tional file-access and update protocols, WHILE THE OWNERS CPUs AND SUCH
such as FTP (file-transfer protocol) and Innumerable factors drive your selec-
Bittorrent, including built-in servers for
CAN STILL RECTIFY tion of a CPU architecture, the features
them. They might even want approach- THE SITUATION. within that architecture, and a perfor-
es such as Telnet and TFTP (trivial mance option of that feature set, in-
FTP). Dynamic DNS (domain-name- cluding the software suite’s demands,
service) and NTP (network-time-proto- the system’s BOM (bill-of-materials)-
col) support may also be on customers’ cost expectations, and the availability
wish lists, along with a Web server, both of highly integrated and application-
for convenient user access to the NAS optimized IC variants. In addition, con-
settings and for enabling the NAS to customers of pending problems with sider not only architecture-tailored soft-
serve generic Web pages through both their NAS while the owners can still ware from your company but also that
HTTP (hypertext transfer protocol) rectify the situation. For example, you of third-party software you might want
and HTTPS (HTTP-secure) over the can send automatic e-mails to inform us- to license, along with additional utili-
LAN and WAN. Also consider iTunes, ers of high temperatures, which may in- ties that your customers may want to
SqueezeCenter (formerly, SlimServer), dicate clogged or otherwise failing fans install after the purchase. For example,
and other media servers; iSCSI (Inter- and vents, along with SMART (self- many enthusiasts have developed free-
net-small-computer-system-interface) monitoring/analysis/reporting technol- ly downloadable add-ons for Windows
support for optional SAN (storage-ar- ogy)-drive-diagnostic results that ex- Home Server on the We Got Served
ea-network) access; workgroup-tailored ceed predetermined thresholds. To get Web site.
servers, such as DNS and e-mail, the lat- those e-mails to your customers, though, Two examples highlight the diver-
ter complete with spam filtering; mas- you also need to support spam-blocking gent paths that companies have taken
ter-browser capabilities for Windows safeguards that ISPs (Internet-service in this regard. First, look at Cisco’s Link-
peer-to-peer workgroups; and direct at- providers) now put in place. These po- sys division (Figure 1). The company in
tachment to OTA (over-the-air), cable, tential roadblocks include nonstandard January 2007 introduced the NAS200,
and satellite television tuners for both SMTP (simple-mail-transfer-protocol) which employed RDC Semiconductor’s
live-TV viewing and record-and-play- ports, user-name and password authenti- R3210 CPU, implementing the i486 mi-
back features using network extenders. cation at the SMTP server, and SSL (se- croprocessor-instruction set. However,
Although both open-source and pro- cure-sockets-layer) capabilities. the NAS was so performance-strapped
prietary software exists to implement Keep in mind, too, that you must sup- that it couldn’t support either SMTP-
these capabilities, each incremental port no-brainer updates to the NAS server authentication or SSL cognizance
concurrent task puts incremental de- BIOS (basic input/output system) or EFI for e-mail alerts; it also could not use its
mand on memory, processing, and other (extensible-firmware-interface) code, op- USB port to implement a print server.
system resources. Incremental functions erating system, and applications, both to Similarly, the company initially shipped
also threaten to exponentially increase patch vulnerabilities and bugs and to up- the NAS200 with support for only the
the complexity of the perceived cus- grade features in the field. It would be na- journaling-inclusive XFS (extended file
tomer experience with the end result, ive to assume that your customers will system).
along with the probability that func- remember to regularly search for, down- Journaling support is desirable in typi-
tions will negatively interact with each load, and install service packs. The built- cal consumer environments, in which
other. With several of the NAS devices a UPS (uninterruptible-power supply)
that I’ve tested over the years, multiple + See the “Accelerating consumers’ doesn’t feed the NAS, which can, there-
applications have insisted on using the NAS adoptions” posts at fore, abruptly shut down in the middle
same TCP (transmission-control-pro- briansbrain for supplemental information of a media write. But journaling and
tocol) and UDP (user-datagram-pro- on this article’s topics. other advanced-file-system capabilities’
tocol) ports, and other programs have algorithm processing also steal CPU
blocked the NAS from putting its hard- + Go to cycles. In response to user complaints
disk drives in spin-down mode, there- and click on Feedback Loop to post about slow accesses, Linksys added op-
by precluding consequent power-con- a comment on this article. tional support for nonjournaled ext2
sumption reductions and operating-life + For more technical articles, go to
(second extended file system) through
extensions. a firmware upgrade. Marvell’s beefier
Speaking of operating life, warn your ARM-based 88F5182 Orion SOC (sys-

36 EDN | JUNE 25, 2009

tem on chip) powers the newer Media ed variants. ReadyNAS Pro leverages a
Hub NAS line, which has substantially dual-core Intel x86 CPU. It currently
more capabilities than its NAS200 pre- occupies the high end of the company’s
decessor. Similarly, the latest iterations product line, complete with six-drive
of Buffalo Technology’s LinkStation and support. Although Netgear doesn’t
TeraStation NAS products have started comment on future product plans, it’s
to use ARM processors; a mix of MIPS not a stretch to imagine the company’s
and PowerPC CPUs initially fueled embrace of the x86 extending through-
these products. out the ReadyNAS line in the future
Linksys migrated away from x86, but (see sidebar “x86 enhancements”). As a
Netgear seems to be going in the op- longtime ReadyNAS X6 user, I’ve been
posite direction. The company in May frustrated at the long delays between the
2007 acquired Infrant Technologies and debut of new versions of PacketVideo’s
its ReadyNAS product line. Infrant be- TwonkyMedia DLNA server on con-
gan life as a silicon supplier of the Leon ventional platforms and its availabil-
SPARC-compatible CPU for NAS. For ity on Leon-based hardware. Adopting
reasons that likely involved a dearth of a mainstream-CPU foundation would
stand-alone-IC sales to others, the com- probably shorten those delays.EDN
pany later switched gears and decided to
become a system supplier selling Leon- R E FE R E NCE S
based NAS. Netgear no longer manufac- 1 Dipert, Brian, “802.11n: compli-
tures the initial 600, X6, and NV prod- cated and about to become even
uct lines; Duo, NV⫹, and NVX systems messier,” EDN, May 28, 2009, pg
continue to use Leon. 6,
In late 2008, however, Netgear rolled post/1970043397.html.
out the ReadyNAS Pro, available in 2 Dipert, Brian, “Power line: more reli-

both enterprise- and consumer-target- able than wireless? You’ve got to be

kidding ... yes?” EDN, Feb 23, 2009,
Addonics Technologies MIPS Technologies
3 Dipert, Brian, “Power line: Does mar-
ADS Technologies Multimedia Over ket success necessitate a function and Coax Alliance feature reset?” EDN, Feb 23, 2009,
Advanced RISC
Machines (ARM) Netgear
4 Dipert, Brian, “Hard-disk drive
Apple PacketVideo price tracks: competitive impacts,”
Buffalo Technology EDN, April 7, 2009,
Casetronic RDC Semiconductor html. 5 Dipert, Brian, “Solid-state drives
Data Robotics Samsung challenge hard disk,” EDN, Nov 13,
D-Link Seagate (Maxtor) 2008, pg 36, CA6611643.
FreeBSD 6 Dipert, Brian, “Virtualization: sili- Sony con and software salvation or techno-
FreeNAS logical tower of Babel?” EDN, Oct 2, Toshiba 2008, pg 34,
HomePNA Alliance Tritton Technologies CA6598366.
Via Technologies
Linksys You can reach
Western Digital Senior Technical Editor
com Brian Dipert
Marvell Technology
Group Ximeta at 1-916-760-0159,,
Microsoft and
Moscone Center, July 26-31, 2009
San Francisco, California, USA

Connections for Design Success

Now, more than ever, knowing which design tools, IP, test, and manufacturing processes to use makes all the difference for design success.
No other event gives you the opportunity to interact with the community of researchers, designers, and tool developers who enable the
electronic design process. Invest your time in efficiency and productivity – attend DAC.

Only DAC offers:

• New DAC User Track: Over 90 technical paper and poster
presentations on design tool use and methodology – by design
engineers for design engineers! Learn how your peers have solved
challenging problems with novel approaches and solutions.
• Research and innovations from leading EDA and design technologists
with over 150 papers, eight panels, three keynotes and nine special
sessions on multicore designs, system-level and HW/SW codesign,
physical design, power analysis/low-power design, verification
and more.
• Conveniently located in San Francisco – close to the East Bay, Silicon
Valley and SFO.
• Worldwide attendance from developers, designers, researchers, academics,
managers and engineers from leading electronics companies and universities.
• A vibrant exhibition with over 200 companies with solutions covering the
electronic design eco-system from concept to silicon. Check out the new
IC Design Central Pavilion featuring IP, design and foundry services, PLUS
the FREE sessions in the DAC Pavilion and the Exhibitor Forum.
Make DAC your one stop this year for networking, technical insights,
and to find all the products and services you need to stay competitive.
DAC – Where electronic design meets.

Register today!


Little-known flash-memory
features protect data and IP
ou design a system, and somebody messes it up. a flash device from a PCB (printed-circuit board).

The damage is sometimes unintentional. For Finally, identify whether the threat is unintentional or in-
example, a service provider may install its soft- tentional. Unintentional alterations, such as those that bugs
ware on your device and corrupt your original in software cause, are typically easier to prevent because the
code. On the other hand, hackers and IP (intel- cause of the problem is not elusive or persistent. If the attack
lectual-property) thieves go out of their way to comes from a hacker or a thief, quantify how much effort the
overwrite, copy, or clone data stored in your systems. What- attacker is willing to make. The amount of time and money a
ever the cause, the resulting damage or theft represents no less hacker is willing to spend affects how much security the de-
of a problem. It’s not surprising that designers need a way to sign requires. With these data points, determine which flash-
protect system integrity. What may be surprising is that within security features provide the right level of protection against
its bits and blocks, flash memory holds the key to protecting the source and intent of the attack. For example, if you must
firmware and even hardware designs. protect the design from data corruption from an Internet at-
Flash devices offer a number of data-protection measures, tack, block locking provides moderate protection, and OTP
each with its own advantages for read, write, or erase protec- (one-time-programmable) blocks provide the best protection
tion. The security options add layers of security to slow down (Figure 1). If an IP thief aggressively targets the design by re-
would-be hackers and thieves and provide protection from un- moving the flash device and attempting to read the data us-
intentional modifications. Some flash-security features don’t ing a PROM (programmable-read-only-memory) programmer,
even add cost to the final design, and, although the strongest protecting the design may warrant paying more for flash-data-
flash-protection features may cost more than standard flash, encryption features (Figure 2).
they are far more affordable than a nonflash-hardware-encryp-
tion engine, hidden operations, authenticated operations, or DETAILED FEATURE REVIEW
software-encryption applications. From block locking to advanced encrypted-password access,
Manufacturers and even devices from the same manufac- you can choose the features that address the type and source
turer offer different features. Designers must
select the right flash device for the final ap-
plication after considering a number of fac- EFFECTIVE OTP BLOCKS
tors, such as the built-in security options, ENCRYPTED-PASSWORD ACCESS
performance, density, size, and cost. HIDDEN OPERATIONS
Evaluating the options starts with identi- PHYSICAL PASSWORD PROTECTION
fying the problem that you want to solve. ATTACK

Features perform specific functions, and

some come with added cost. First, deter- BLOCK LOCKING
mine what you must protect amid the bits,
data, and code. For example, you might STANDARD FLASH
need to protect electronic-system serial
numbers, security keys, boot code, or finan- EFFECTIVE EFFECTIVE
cial information for services access, such as SOFTWARE OR ONLINE DISRUPTION
for pay TV. Once you know what you need
to protect, determine whether a software or Figure 1 Security alternatives offer differing levels of protection against physical
a physical disruption is likely to affect those attack and software disruption, whether due to malicious corruption or unintentional
bits, data, or code. A software attack may data alteration. Note that adding such features as encrypted-password access,
come from the Internet or a system applica- authenticated operations, hidden operations, software-encryption applications, and
tion, for example. A physical attack, for in- hardware-encryption engines increases system cost.
stance, could involve the direct removal of

JUNE 25, 2009 | EDN 39

of an expected attack, with acceptable cost MOST HIDDEN OPERATIONS
impact on the design. One such feature is HARDWARE-ENCRYPTION ENGINE
password access. Some flash devices offer
password-access features that slow down ENCRYPTED-PASSWORD ACCESS

thieves, creating barriers that make the de- AUTHENTICATED OPERATONS

sign a less desirable target for copying or

cloning. IP thieves must quickly and eas- PHYSICAL ATTACK
ily copy system data. Password access adds
time, cost, and effort to low-overhead clon-
ing operations. Password access locks ei- VOLATILE-BLOCK

ther the entire array or selected blocks in OTP SPACE

the main array from program, erase, or read
access, depending on the flash device. You LEAST
can individually set each block to the de-
sired protection level. Before the system
leaves the factory for the end customer, the Figure 2 Various flash-security features can protect against cloning and intellectual-
manufacturer must store a 64-bit password property theft. Note that adding such features increases system cost.
in the password area of the flash device and
program a matching password into the sys-
tem microcontroller or other hidden storage. turer more time in the market before the clone can compete
When the system receives a command to read, modify, or for revenue, and a deterrent may create enough delay for the
erase data in the protected blocks, the system processor looks original manufacturer to refresh its design before the cloner
for a match between the number in the microcontroller and has an opportunity to produce a viable product. This built-in
the one in the flash device. If the passcode is not valid with flash feature offers a cost-effective method for countering the
both the microcontroller and the flash device, the would-be revenue impact of IP loss.
hacker cannot read or modify the data. If the system detects Encrypted-password access for higher-level IP security is an-
a matching passcode, a user can read or modify individual other method of protecting IP. A 64-bit password slows down a
blocks. Depending on the flash device, designers can choose cloner or a service thief, but an encrypted password adds a sig-
from various protection modes, including read, modify, and nificantly higher level of data protection. A few flash devices
substitution prevention. include an encrypted-password feature through an algorithm
Password protection is both a service-theft deterrent and an that manufacturers implement in silicon. The manufacturer
IP-copying and -cloning deterrent. Duplicated flash chips can encrypts passwords that pass from the flash and the microcon-
provide premium services to users who don’t pay for them, troller to the processor. The processor deciphers the passwords
representing lost revenue for the service provider. Password- using the algorithm and confirms a match. A bus snooper, on
based read protection is a simple, cost-effective way to thwart the other hand, can only read—but not decipher—the en-
attempts to distribute pirated flash chips that enable access crypted password. Without the unencrypted password, the
to premium services. If the designer uses password protec- flash chip is unreadable, and the IP remains protected. Flash
tion on the flash device, he leaves the pirate with inoperable devices with an encrypted password typically cost more than
chips. When a would-be service thief attempts to read the those without one because manufacturers implement the al-
data stored on the flash device, the device attempts to vali- gorithm on silicon, which adds to the component cost. How-
date the 64-bit password. Without the password, the device ever, the cost of the flash chip may be nominal compared with
returns only values of zero, rendering the copied chip inoper- lost service revenue. For products with long refresh cycles,
able. Even if the thief can snoop, discover the password, and encrypted passwords are essential tools in preventing clones
copy the data from the chip, the 64-bit password in the pi- from reaching the market before the manufacturer updates its
rated chip will not match the password in the microcontroller model.
in the system into which the thief inserted it, again making
the chip inoperable. OTP FLASH
In the case of cloning, IP thieves must replicate and produce Some flash devices include a system-level OTP area in
a design before an updated version of the original makes the which you can permanently lock the bits after programming.
clone obsolete. Flash-memory-password protection can create Once you lock the bits, a hacker cannot program or erase the
a significant enough delay to make the cloner seek an easier blocks the bits are mapped to. The OTP has one factory-pro-
target because the flash protects the hardware signature of the grammed segment with a unique, unchangeable number. The
system. A flash device with a 64-bit password limits access to other segment is blank so that the designer can program it.
legitimate sources. Without the password, an IP cloner who Flash devices with OTP typically come in varying configura-
uses a PROM programmer to read the flash chip will read back tions of segments with as many as 2112 bits.
only zeros. Several flash devices include an additional OTP feature that
As with the service-theft deterrent, the cloner can resort allows you to permanently lock blocks in the regular memo-
to bus-snooping to discover the password, but that step adds ry array. This implementation of OTP prevents modifications
time, cost, and effort. The delay gives the legitimate manufac- that disrupt system integrity. For instance, a service provider

40 EDN | JUNE 25, 2009

typically adds its own code to a set-top bypass a router, for example, the mali-
box after delivery from the manufac- cious code will reach the flash device,
turer. A designer cannot know how the and the device will check the pin for the
service provider’s code will affect the block that stores the boot code. Finding
system. To protect the erasure or altera-
tion of boot code, the designer can store
the voltage at a level that doesn’t allow
modification, the chip does not execute S
the boot code in an OTP block and set the malicious code, and the router con-
it to be permanently locked. Then the tinues to work. Volatile- and nonvola-
service provider’s code cannot write or
erase to the locked block, thereby pre-
tile-block-locking features use software
commands to lock and unlock blocks, DE
venting those inconvenient customer- protecting data from inadvertent mod-
tech calls. ification. In volatile-block locking, bits
Most flash devices include some form in a volatile array are mapped to main-
of hardware-write-protection capability memory-array blocks. You can individ-
that can prevent programming and era- ually modify, set, and clear these vola-
sure of either a block or the entire de- tile-protection bits. However, they can DC-DC
vice. Hardware-write protection works
by setting a pin to a certain voltage, ei-
protect only those blocks that you have
not locked with nonvolatile-array bits.
ther through hard-wiring or by toggling When you cycle the system power or
a bit from an I/O pin on a processor.
Before executing a modify command,
reset the hardware, the volatile-protec-
tion bits revert back to their original un-
the flash chip checks the pin that cor-
responds to program/erase protection. If
locked or locked state.
Nonvolatile-block locking keeps
the pin is not at the correct voltage for blocks locked or unlocked, as the de- Supplies
modification, the chip will not execute signer defines, even after a power cycle
the command, and the code or data will or reset. A nonvolatile-protection bit
not change. is mapped to and can individually lock
When a valid voltage is present in each block. You can clear nonvolatile-
the program-supply voltage, you can protection bits through a clear-bits com-
modify the blocks in the main array. If mand or an erase command. You can use
you ground the program-supply voltage, nonvolatile-block locking to ensure that • Expanded Operating
you cannot program or erase the blocks. blocks remain locked against inadver- Temperatures -55 to +85C
When you ground the supply voltage, tent overwrites even after an unexpect- • Vibration, Method 204, Cond. D
attempts to program or erase will fail, ed power cycle or reset occurs.
and grounding sets the appropriate sta- Flash-security features vary by manu- • Shock, Method 213, Cond. I
tus-register fail bit. facturer and by device. You should con- • Altitude, Method 105, Cond. D
Another version of hardware-write sider these features, along with density,
protection protects the highest or low- performance, technology, lithography,
• Environmental Screening
est block against program and erase op- cost, size, and packaging. These factors • Specification Review
erations. To protect the highest or low- all work together to enable a design, • Custom Models Available
est block, set VPP /WP⫽VIL, where VPP and you can use them to protect its per-
is the program-supply voltage, WP is formance and position in the market.
• 400 Hz and Now -
800 Hz AC-DC Models
write protection, and VIL is the input- Flash-security features offer an afford-
low voltage. In this situation, the block able, secure alternative to protecting IP,
is in lock-down mode, and you cannot content, data, or system integrity.EDN s of Stand
Thousand DC
to 10,000 V
modify it. To remove a lock-down situa-
tion, set VPP /WP⫽VIH, where VIH is the AU T H O R ’ S B I O G R A P H Y Models 2V Watts
input-high voltage. In this case, you can Bill Stafford is the director of segment mar- uts - 0 .75 to 2,000
lock or unlock the block. keting at Numonyx, where he develops ap-
plication strategies and product require- Call toll free
HARDWARE PROTECTION ments for the embedded-flash-memory mar- 800-431-1064
The hardware-based approach pro- ket and enables the flash-memory ecosys- for PICO Catalog
vides an inexpensive layer of protec- tem. A 25-year veteran in the electronics Fax 914-738-8225
tion against the malicious code that field, he has experience in product engineer-
slithers in through the Internet. Mali-
cious code cannot modify or erase data
ing, field quality, and marketing for flash
memory, PCBs, systems, and aircraft com- PICO Electronics,Inc.
143 Sparks Ave, Pelham, NY 10803-1837
stored on a flash device that is locked ponents. Before joining Numonyx, Stafford E-Mail:
at the hardware level unless it first re- spent 22 years at Intel Corp after time with
sets the pin voltage. If a hacker tries to the US Department of Defense. See full Catalog immediately
w w w. p i c o e l e c t r o n i c s . c o m

JUNE 25, 2009 | EDN 41

45V Synchronous Buck



Actual Size
Demo Circuit

Efficient, Rugged & Easy

Linear Technology introduces the LTC®3642, the first in a new family of synchronous high voltage micropower monolithic
step-down converters. It addresses applications in 4mA to 20mA industrial control loops, security systems and general-purpose
conversion where harsh conditions exist. The LTC3642 provides an extremely compact 45VIN synchronous step-down solution
with transient protection to 60VIN and an I Q of only 12μA. This device provides improved performance in size, efficiency,
ease of design, and excellent thermal performance.

Features LTC3642 Efficiency Info & Free Samples

• Input Voltage Range: 4.5V to 45V, 100

with 60V Transient Capability 95 VIN = 10V 1-800-4-LINEAR
• Low Quiescent Current: 12μA 90
VIN = 15V
• 50mA Output Current
Efficiency (%)

• Adjustable Peak Current Limit
• 3.3V, 5V & Adjustable Versions
VIN = 24V
• Only 3 Externals Required
• 3mm x 3mm DFN or MSOP-8E
Packages 65
• LTC3631: 20mA, 45VIN* 60
0.1 1 10 100
• LTC3632: 100mA, 45VIN* , LTC, LT and LTM are registered trademarks of Linear
Technology Corporation. All other trademarks are the property of
Load Current (mA)
their respective owners.
*Future Products



Create a swept-sine function

D Is Inside
in LabView with just one virtual
44 Charlieplexing at high duty
instrument cycle
Sean McPeak, University of California—San Diego 48 Serial port tests digital circuits

51 DAC calibrates 4- to 20-mA

 Swept sine waves are useful
when you want to test a prod-
The LabView software calculates
an array of numbers that represent output current
uct over a wide frequency range. A the swept-sine-wave time series at 51 Alarm tells you to close the
large research project included the re- each sample point as the frequency refrigerator door
quirement to determine wave propaga- either increases or decreases, depend-
tion in the open ocean. This applica- ing on the direction of the sweep. You 왘To see all of EDN's Design
tion required the generation of a swept must handle the frequency change of Ideas, visit
sine wave to drive an acoustic trans- the output on a point-by-point ba- ideas.
ducer. Although many waveform gen- sis. The basic form of the equation is
erators have a built-in function for this Y(I)⫽V⫻sin((A⫻I2)/2⫹B⫻I), where
requirement, you must program it your- Y(I) is the amplitude of the swept sine
self if you want to implement a swept wave as a function of the sample point, and stop frequencies, you must change
sine with a multifunction data-acqui- I is the integer that steps through the the unit to cycles per sample. You ac-
sition card. You can create a swept- time series, V is the peak voltage, and complish this task by dividing the f1
sine function in National Instruments’ A and B are variables. You define A as and f2 frequencies in hertz by the sam-
( LabView with just one 2⫻␲(fSTOP⫺fSTART)/N, and you define ple rate. You determine the sample rate
VI (virtual instrument). Using this B as 2⫻␲fSTART, where N is the number by deciding how smooth of a transition
function, you can control start and stop of samples, fSTART is the normalized start you want to represent your swept sine
frequencies, sample rate, and the over- frequency, and fSTOP is the normalized wave. A good rule of thumb is to have
all duration of the sweep (Figure 1). stop frequency. To normalize the start at least 10 samples/cycle at the high-

Figure 1 With just one LabView virtual instrument, you can control start and stop frequencies, sample rate, and the overall
duration of the sweep.

JUNE 25, 2009 | EDN 43

est frequency. When setting the sample Nyquist criteria, given the sample rate this simple program. One idea would
rate, you need to take into account the and highest frequency you require. be to use the output array, which is
overall frequency span you are sweep- This approach drives a simple Bool- nothing more than the time series
ing and the duration of the sweep itself. ean variable to alert the user about representing a predescribed frequency
It is also helpful to compare the results whether the design meets the Nyquist sweep, to feed a loop that would drive
and performance of the LabView data- criteria. You set the “for” loop to run a data-acquisition module. The out-
acquisition-system implementation of for the total number of samples you put of the module should accurately
the swept sine wave with those of an want to calculate. You determine this represent the frequency sweep, as long
AWG (arbitrary-waveform generator). value by multiplying the duration in as the module’s output sample rate is
You use two methods of comparison. seconds by the sample rate in samples the same as the sample rate you use for
First, you compare the output of both per second. To guarantee that the loop generating the frequency-sweep time
the data-acquisition and the AWG processes all of the generated samples, series. You should then be able to track
swept sine wave on a spectrum analyz- you must add one, because the loop the output samples and, when they are
er. Second, you run them both through stops at N⫺1. complete, reverse the frequency-sweep
an audio-amplifier/speaker system and You implement the output function array. You then again feed this new
simply listen to the output. This meth- in the “for” loop with simple algebraic flipped array to the data-acquisition
od is useful in determining sweep rate, operators and the sine block. The out- module. Depending on the maximum
duration, and stop and start frequen- put is an array that reaches the perim- and minimum frequencies, sweep dura-
cies. This type of comparison is valid eter of the “for” loop. It is important tion, and sample rate you use, it may be
only if the frequencies involved are to enable indexing at this node. This difficult to flip the array and configure
in the audible range. The LabView VI action allows the circuit to individu- the module quickly enough to not miss
employs simple array manipulation and ally handle each element in the array a sample. In that case, you can prefill a
uses a “for” loop. The input duration at the output of the “for” loop. You can frequency-sweep array for a set number
is in seconds, the sample frequency is also add a simple gain stage to set the of passes.
in samples per second, and the start- peak-to-peak value to whatever point These modifications allow the
ing and ending frequencies are in hertz. you want. Finally, you use the “rotate- sweep to continue back and forth for
Dividing the sample rate immediately 1D-array-block” case structure to flip a set period. Another improvement
converts the start and end frequencies the array if the ending frequency is would be to add some real-time FFT
to cycles per sample. A maximum/min- lower than the starting frequency. This (fast-Fourier-transform) capability so
imum block takes the normalized end- approach handles cases in which you that the user can see the sweep in the
ing and starting frequencies as its inputs want a frequency sweep that starts out frequency domain. This approach also
and uses the maximum output of the in a higher frequency and descends to adds an increased level of insurance
input pair. You use this method to de- a lower frequency. that the circuit properly meets the
termine whether your design meets the You can easily modify and expand sweep definition.EDN

ICs using the technique (Reference

Charlieplexing at high duty cycle 2). Allen used the high-impedance
third input state available to most mi-
Luke Sangalli, Digital Designs, Melbourne, Australia
croprocessors for turning off LEDs in a
A few articles have re- matrix; the LEDs’ respective
 cently appeared describ- TABLE 1 NO. OF LEDs AND DUTY CYCLES microprocessor pins’ high or
ing novel ways to increase the No. of No. of Charlieplexing Standard multiplexing low states individually turn
number of LEDs a micropro- pins LEDs duty cycle (%) duty cycle (%)/no. of pins on these LEDs. Using this
cessor can drive with a limit- Three Six 33.3 50/five
method, you can drive nine
ed number of pins (Reference seven-segment LED displays
1). The standard multiplex- Four 12 25 33/seven using only nine microproces-
ing technique made popular by Five 20 20 25/nine sor pins rather than the usual
multidigit seven-segment dis- Six 30 16.6 20/11 17. For N pins, you can indi-
plays has, in pin-scarce designs, vidually address N⫻(N⫺1)
given way to “Charlieplexing.” Seven 42 14.2 16.6/13
LEDs using Charlieplexing.
Charlie Allen devised this Eight 56 12.5 14.2/15 One of the gripes people
technique while working at Nine 72 11.1 12.5/17 often level at Charlieplexing
Maxim (, regards its poor duty cycle. A
10 90 10 11.1/19
which has since introduced previous Design Idea com-

44 EDN | JUNE 25, 2009

HB LED drivers for the widest
range of lighting applications

Flexible and robust with wide input-voltage and output-power ranges

Maxim’s HB LED drivers reduce the size, cost, and complexity of lighting designs. Our portfolio includes
29 ICs covering the full range of linear and switch-mode topologies. All devices operate over -40°C to
+125°C, support wide input-voltage ranges, and integrate advanced protection features.

Architectural and street lighting


Offline lighting
Solar-powered lighting

EV kits
SPI is a trademark of Motorola, Inc.


For free samples or technical support, visit our website.

Innovation Delivered is a trademark and Maxim is a registered trademark of Maxim Integrated Products, Inc. © 2009 Maxim Integrated Products, Inc. All rights reserved.
pares the standard multiplexing meth- view, consider that, by the time you get
od with Charlieplexing (Reference 3). BY THE TIME YOU GET to 90 LEDs, the PCB (printed-circuit-
Using Charlieplexing, the maximum board) real estate and cost of the addi-
duty cycle for a 20-LED display is only TO 90 LEDs, THE PCB tional 10 transistor/resistor sets pale in
5%. The poor duty-cycle figure is due REAL ESTATE AND comparison to the display itself.
not to the method, however, but rath- Examining the circuit in detail,
er to the driving capability of the mi-
COST OF THE you’ll notice that it has five micropro-
croprocessor and the parasitic-leakage 10 TRANSISTOR/ cessor pins, P1 through P5, available, for
paths. A single pin cannot usually sink RESISTOR SETS PALE a total of N⫻(N⫺1)⫽20 LEDs. When
the current a number of LEDs require P3, for example, is high, the emitter of
to effectively light up, so these designs IN COMPARISON TO Q3 is at approximately 4.4V, and you
often require one source pin and one THE DISPLAY ITSELF. can turn off D13, D23, D43, or D53 if you
sink pin to light only one LED at any make P1, P2, P3, or P5 low. Any pin that
time. However, adding a transistor or you set to input, or high impedance,
two resistors allows you to circumvent five pins instead of nine (Table 1). alternatively turns off the correspond-
these issues. One of the drawbacks of adding the ing LED. When P1 and P4 are low, P3 is
If you rearrange the LEDs in the fa- transistor and resistors to each column high, and P2 and P5 are in high-imped-
miliar cross-point array and add a tran- is that you need additional components ance states. With P3 high, transistor Q3
sistor to each column to carry the com- to achieve a reasonable LED bright- biases on, all the other transistor bases
mon current, you’ll see the duty cycle ness when a large number of LEDs is are either low, which ensures that no
of the Charlieplexing method does not involved. This approach, however, is a current will flow, or high-impedance,
differ much from standard multiplexing better alternative to using a costly IC which supplies no current into the base
(Figure 1). For a 20-LED, five-column and no worse than standard multiplex- to allow the transistor to conduct. All
matrix, each LED remains on for 20% of ing or “Gugaplexing,” which also re- the diodes in the third column can turn
the time compared with 25% for stan- quires additional transistors and resis- on, but only D13 and D43 have a path
dard multiplexing, but now using only tors. From a cost and benefits point of directly to ground through P1 and P4,
which are low and through the 100⍀
5V 5V 5V 5V 5V
current-limiting resistors.
D23 and D53 connect to the high-im-
470 pedance input pins and can conduct
Q5 BC337
470 only through the 100⍀ resistors at-
Q4 BC337 tempting to turn on Q2 and Q5. Be-
Q3 BC337 cause of their forward-voltage drop—
470 typically, 2.2V—the emitters of Q2 and
Q2 BC337
Q5 will be less than 1.6V, as the fol-
Q1 BC337 lowing equation shows: 5VCC⫺0.6V
D12 D13 D14 D15 (Q3)⫺2.2V (D23 or D53)⫺0.6V (Q3 or
Q5)⫺ILED⫻100⍀⬍1.6V, where ILED is
100 the current of the LEDs. This scenario
D21 D23 D24 D25 does not allow any LED in Column 2 or
Column 5 to light up to any level that
100 would have an undesirable effect.EDN
D31 D32 D34 D35
100 1 Lancaster, Don, Tech Musings,
D41 D42 D43 D45 August 2001,
100 2 “Charlieplexing—Reduced Pin-
D51 D52 D53 D54 Count LED Display Multiplexing,” Ap-
plication Note 1880, Maxim, Feb 10,
100 2003,
3 Gupta, Saurabh, and Dhananjay V

Figure 1 Arranging LEDs in a cross-point array and adding a transistor to each Gadre, “Multiplexing technique yields
column show that the duty cycle of Charlieplexing is similar to that of standard a reduced-pin-count LED display,”
multiplexing. EDN, Oct 16, 2008, pg 68, www.

46 EDN | JUNE 25, 2009

HB LED drivers
Industry’s forand
smallest thesmartest
range of lighting applications
touch-screen controllers

More features than the competition, yet in a smaller package

The MAX11800/MAX11801 are resistive touch-screen controllers (R-TSCs) available in TQFN and ultra-small WLP
packages. These devices have advanced operation modes and other digital functionality that reduce bus loading of
the system/applications processor, plus numerous advantages including the reduction of necessary board space and
the ability to connect to any host μP. Ideal applications include mobile phones, MP3 players, personal navigation devices
(PNDs), UMPCs, office printers, gaming consoles, and digital photo frames.


X+ Complete R-TSC reduces

High-speed, 25MHz Y+ PCB footprint by 62%
SPI™ interface for high X- MUX SAR ADC AVG
data throughput AUX
and connection to any μP

reduces interrupt SCL/CLK
overhead by up to 90%
Only 1.6mm x 2.1mm!


Part Interface Temp Range (°C) Package (mm x mm) Price† ($)
MAX11800ETC+ 12-TQFN (4 x 4) 1.55
MAX11800EWC+ 12-WLP (1.6 x 2.1) 1.51
-40 to +85
MAX11801ETC+ 12-TQFN (4 x 4) 1.50
MAX11801EWC+ 12-WLP (1.6 x 2.1) 1.46
SPI is a trademark of Motorola, Inc.
*Patent pending.
†1000-up recommended resale. Prices provided are for design guidance and are FOB USA. International prices will differ due to local duties, taxes, and exchange rates. Not all pack-
ages are offered in 1k increments, and some may require minimum order quantities.


For free samples or technical support, visit our website.

Innovation Delivered is a trademark and Maxim is a registered trademark of Maxim Integrated Products, Inc. © 2009 Maxim Integrated Products, Inc. All rights reserved.
the manufacturers’ data sheets. Place
Serial port tests digital circuits bypass capacitors as close as possible to
the IC’s power and ground. You can re-
Yury Magda, Cherkassy, Ukraine
place the MAX232 with a MAX225 or
A PC’s serial port provides sig- will run on the 2005 version, as well. MAX233.EDN
 nal lines that you can use to To create the application, select the
read voltage levels of digital circuits. “Windows Form Application” from the R E FE R E NCE S
You can use the port to test digital templates in the project wizard. Place 1 “MAX220-MAX249 ⫹5V-Powered,

TTL (transistor-to-transistor-logic)- the text-box, label, and button compo- Multichannel RS-232 Drivers/Receiv-
level circuits. You just need to convert nents on the project’s main form and ers,” Maxim, January 2006, http://
the TTL levels to RS-232 voltages, and assign titles for them. You should place
you can add a multiplexer to increase the serial-port component on the de- MAX220-MAX249.pdf.
the number of signals that the serial sign area of the project. Then, set the 2 “CD54/74HC4051, CD54/

port can sense. appropriate parameters for the serial- 74HCT4051, CD54/74HC4052,
The circuit in Figure 1 uses a port component, including the port CD74HCT4052, CD54/74HC4053,
MAX232 IC from Maxim (www. number, baud rate, data bits, parity, CD74HCT4053 High-Speed CMOS to convert RS-232 and stop bits. Logic Analog Multiplexers/Demulti-
voltage levels to TTL levels (Reference When you build the circuit, follow all plexers,” Texas Instruments, 2004,
1). A 74HC4051 from Texas Instru- precautions concerning the MAX232
ments ( lets you select any and 74HC4051 wiring according to cd74hct4053.pdf.
of four digital inputs and route them to
the serial port (Reference 2). Listing
5V A3 A2 A1 A0 5V
1, which is available with the online
version of this Design Idea at www.edn.
com/090625dia, lets you control the 16
RTS (ready-to-send) and DTR (data- 16
1 X0
terminal-ready) pins in the serial port 14
that selects the signal under test. The C1 ⫹ C3 15
CTS (clear-to-send) pin then reads the 1 ␮F ⫹1 ␮F 12
3 2 X3
signal under test into the PC. 1
4 6 5
The four digital-input signals, A0 PC COM PORT C4
C2 ⫹ MAX232 2
through A3, from your device under DB9 1 ␮F X6
1 ␮F ⫹ 4
test connect to the first four inputs, X0 1 5
6 74HC4051
through X3, of the multiplexer. Only 3
7 RTS 13 12
one of those signals can pass through to 4
8 CTS 14 11 S0
9 DTR 8 9 10 S1
the X output, Pin 3, at a time. By set- 5
9 S2
ting the appropriate binary code on the 15 6
serial port’s RTS and DTR lines, you 7
can select the signal to pass through
the multiplexer (Table 1).
The PC software, running on Win-
dows XP, sequentially sets those binary
combinations on the port’s RTS and Figure 1 This circuit lets you pass up to four TTL-level signals to an RS-232
DTR lines and reads the digital signal port to read their status.
on the CTS line. The software then
reads the status of the selected bit and
displays it when you press the “check-
status” button (Figure 2). The code is
written in Microsoft C# 2008, but it


Signal to X pin RTS bit DTR bit
A0 0 0
A1 1 0
A2 0 1 Figure 2 A main window of the running application shows that input lines A0—
A3 1 1 A2 have high logic levels and A3 has a low logic level.

48 EDN | JUNE 25, 2009

15VIN, 4MHz Monolithic Synchronous Buck Regulator Delivers
5A in 4mm × 4mm QFN – Design Note 467
Tom Gross

Introduction produces an out-of-phase clock signal so that parallel

The LTC®3605 is a high efficiency, monolithic synchronous devices can be interleaved to reduce input and output
step-down switching regulator that is capable of delivering current ripple. A multiphase, or PolyPhase®, design
5A of continuous output current from input voltages of also generates lower high frequency EMI noise than a
4V to 15V. Its compact 4mm × 4mm QFN package has single-phase design, due to the lower switching currents
very low thermal impedance from the IC junction to the of each phase. This configuration also helps with the
PCB, such that the regulator can deliver maximum power thermal design issues normally associated with a single
without the need of a heat sink. A single LTC3605 circuit high output current device.
can power a 1.2V microprocessor directly from a 12V 1.8VOUT , 2.25MHz Buck Regulator
rail—no need for an intermediate voltage rail.
The LTC3605 is specifically designed for high efficiency
The LTC3605 employs a unique controlled on-time/con- at low duty cycles such as 12VIN-to-1.8VOUT at 5A, as
stant frequency current mode architecture, making it ideal shown in Figure 1. High efficiency is achieved with a low
for low duty cycle applications and high frequency opera- RDS(ON) bottom synchronous MOSFET switch (35mΩ) and
tion. There are two phase-lock loops inside the LTC3605: a 70mΩ RDS(ON) top synchronous MOSFET switch.
one servos the regulator on-time to track the internal This circuit runs at 2.25MHz, which reduces the value and
oscillator frequency, which is determined by an external size of the output capacitors and inductor. Even with the
timing resistor, and the other servos the internal oscillator high switching frequency, the efficiency of this circuit is
to an external clock signal if the part is synchronized. Due about 80% at full load.
to the controlled on-time design, the LTC3605 can achieve
very fast load transient response while minimizing the Figure 2 shows the fast load transient response of the
number and value of external output capacitors. application circuit shown in Figure 1. It takes only 10μs
to recover from a 4A load step with less than 100mV
The LTC3605’s switching frequency is programmable from of output voltage deviation and only two 47μF ceramic
800kHz to 4MHz, or the regulator can be synchronized to output capacitors. Note that compensation is internal, set
an external clock for noise-sensitive applications. up by tying the compensation pin (ITH) to the internal
Furthermore, multiple LTC3605s can be used in parallel L, LT, LTC, LTM, PolyPhase, Linear Technology and the Linear logo are registered
trademarks of Linear Technology Corporation. All other trademarks are the property of
to increase the available output current. The LTC3605 their respective owners.

R5 C7
100k 2.2μF D1
12V C1 C2 0.1μF
22μF 22μF 25V
16V 16V SW 1.8V
SW C3 C4 5A
SW 47μF 47μF
SVIN LTC3605 6.3V 6.3V
10Ω 0.1μF SW
R4 C9 R1
71.5k 0.1μF 10.0k
DN467 F01

Figure 1. 12V to 1.8V at 5A Buck Converter Operating at 2.25MHz



DN467 F03
DN467 F02 500ns/DIV

Figure 2. Load Step Response of the Circuit in Figure 1 Figure 3. Multiphase Operation Waveforms of the
Circuit in Figure 4. The Switch Voltage and Inductor
3.3V regulator rail (INTVCC). This connects an internal Ripple Currents Operate 180° Out of Phase with
series RC to the compensation point of the loop, while Respect to Each Other
introducing active voltage positioning to the output
voltage: 1.5% at no load and –1.5% at full load. The where each slave’s CLKIN pin takes the CLKOUT signal
hassle of using external components for compensation of its respective master. To produce the required phase
is eliminated. If one wants to further optimize the loop, offsets, simply set the voltage level on the PHMODE pin
and remove voltage positioning, an external RC filter can of each device to INTVCC, SGND or INTVCC/2 for 180°,
be applied to the ITH pin. 120° or 90° out-of-phase signals, respectively, at the
1.2VOUT , 10A, Dual-Phase Supply CLKOUT pin.
Several LTC3605 circuits can run in parallel and out of Conclusion
phase to deliver high total output current with a minimal The LTC3605 offers a compact, monolithic, regula-
amount of input and output capacitance—useful for tor solution for high current applications. Due to its
distributed power systems. PolyPhase capability, up to 12 LTC3605s can run in parallel
The 1.2VOUT dual-phase LTC3605 regulator shown in to produce 60A of output current. PolyPhase operation
Figure 4 can support 10A of output current. Figure 3 shows can also be used in multiple output applications to lower
the 180° out-of-phase operation of the two LTC3605s. The the amount of input ripple current, reducing the necessary
LTC3605 requires no external clock device to operate up input capacitance. This feature, plus its ability to operate
to 12 devices synchronized out of phase—the CLKOUT at input voltages as high as 15V, make the LTC3605 an
and CLKIN pins of the devices are simply cascaded, ideal part for distributed power systems.

100k 2.2μF
22μF 10Ω 0.1μF RUN SW 1.2V
VON 47μF
0.1μF RT1
CC2 162k
22μF 10Ω 0.1μF RUN 0.1μF 0.33μH
DN467 F04

Figure 4. 12V to 1.2V at 10A 2-Phase Buck Converter

Data Sheet Download For applications help, call (408) 432-1900, Ext. 3229

dn467 LT/TP 0609 155K • PRINTED IN THE USA

Linear Technology Corporation
1630 McCarthy Blvd., Milpitas, CA 95035-7417
(408) 432-1900 O FAX: (408) 434-0507 O © LINEAR TECHNOLOGY CORPORATION 2009
DAC calibrates the negative input equal to that on its
positive input. The output current de-
4- to 20-mA output current pends on the value of the sense resis-
tor, the gain of the current-sense am-
Ronald Moradkhan and Steven Lau, plifier, and the control voltage.
Maxim Integrated Products, Sunnyvale, CA The DAC provides the control volt-
Industrial controls make heavy where I OUT is the output current, age that lets you automate the calibra-
 use of 4- to 20-mA current loops VCONTROL is the control voltage, RSENSE tion procedure. By selecting the right
to transmit process measurements be- is the sense resistance, and KCSA is the value for the sense resistor and by using
cause current loops retain information gain of the current-sense amplifier— a suitable resistor divider for R1 and R2
in the presence of noise and changes in 20 in this case. The circuit comprises at the output of the DAC, you can ad-
loop voltage. The loop circuit requires IC2, a Maxim ( just the circuit’s output to 4 mA when
proper calibration to ensure accurate MAX5304 DAC; IC3, a MAX4376T the DAC’s digital input is zero-scale
readings. The circuit in Figure 1 cali- current-sense amplifier; IC 4 , a and 20 mA when the digital input is
brates the loop by generating a current MAX420 op amp; and Q1, an N-chan- full-scale. Figure 1 shows the compo-
in response to a control voltage: nel IRFL4105 MOSFET. The op amp nent values you need to achieve that
VCONTROL lets the control voltage set the output condition.
I OUT = , current because it forces the voltage on With a zero-scale digital input, the
DAC output is 0V and the resistor di-
10V vider produces 0.6V at the op amp’s
R3 positive input, forcing the output cur-
C4 rent to 4 mA. With a full-scale digital
R4 0.1 ␮F input, both the DAC output and the
15 midpoint of the resistor divider are at
Q1 the 3V reference voltage, forcing the
C3 IRFL4105
0.1 ␮F RS⫹ RS⫺ R5 output current to 20 mA. A transfer
VCC IC3 curve relates the output current to the
0.1 ␮F control voltage (Figure 2).EDN

C1 GND ⫹
10.2k CCXTD
0.1 ␮F
V⫺ 0.1 ␮F

R2 0.1 ␮F
C2 255k
0.1 ␮F



Figure 2 The circuit in Figure 1

Figure 1 This DAC-controlled 4- to 20-mA transmitter allows digital control produces a linear output current
of the loop current. versus digital control voltage.

circuit that this Design Idea describes,

Alarm tells you and the gadget still operates with the
original 9V battery. The circuit oper-
to close the refrigerator door ates by sensing a decrease in resistance
Boris Khaykin, TRW Automotive, Livonia, MI of photocell PC1 that results from
light in the refrigerator when its door
The circuit in Figure 1 is a sim- ilar one I recently read about (Refer- is open. A counter is in a reset state
 pler and safer device than a sim- ence 1). A few years ago, I built the when PC1 is in the dark, and its resis-

JUNE 25, 2009 | EDN 51

tance is greater than 30 k⍀. Usually, resistance is lower than 15 k⍀—that is, state is approximately 2.5 mA.
the dark resistance is greater than 200 when the light bulb in the refrigerator You can use almost any photocell,
k⍀, and current consumption at this is on. Then, in 20 seconds, it turns on such as the Jameco (www.jameco.
state is less than 40 ␮A. Oscillator- a buzzer for 20 seconds or until some- com) 202403 CDS0018001 with 200-
counter IC1 starts counting when PC1’s one closes the door. The current at this k⍀ dark and 3-k⍀ light resistance.
This circuit uses a Ra-
dioShack (www.radioshack.
com) 273-074 buzzer. You
can use any similar piezo-
VDD D1 electric buzzer with an op-
9 1N914
OUT2 2 erating dc voltage of 1.5 to
C1 R2 ⫹ 15V. V1 can be as low as 3V.
⫹ 270k BUZZER
R1 10 nF 10
The trade-off is that using a
V1 20k IC1
voltage this low gives you
ⳮ 4060
11 longer battery life but lower
volume of sound.EDN
12 Q5 R E FE R E NCE
VSS 1 Babu, TA, “Alarm
PC1 8 Sounds When Refrigerator
Door Remains Open Too
Long,” Electronic Design,
March 26, 2009, pg 46,
Figure 1 This gadget, placed inside a refrigerator, sounds an alarm when the refrigerator http://electronicdesign.
door is open for more than 20 seconds. com/Articles/ArticleID/

Battery Management at its best

Battery Sensor Interface

Ultimate precision for small signals

2 independent acquisition
channels in one device

Configurable to fit a wide range

of present and future battery
management applications

Raleigh (919) 676 5292 • Detroit (248) 202 7100

52 EDN | JUNE 25, 2009

No need to duplicate yourself
BuyerZone saves you time and money
Get free advice, pricing articles, ratings and quotes on Business Purchases from digital copiers,
to phone systems, to payroll services and more. Visit today to make
fast and efficient business purchase decisions. At BuyerZone, you’re in control. We’re just here
to make your work easier.

Join the millions who’ve already saved time and money on many of their Office Equipment
purchases by taking advantage of BuyerZone’s FREE, no obligation services like:
• Quotes from multiple suppliers
• Pricing articles
• Buyer’s guides
• Supplier ratings
• Supplier comparisons


Call (866) 623-5564 or visit Where Smart Businesses Buy and Sell

A division of

Product Info
1 Photos - Detailed product
photographs and expanded
views help identify the exact
part you need
1 2 Ratings - View product reviews
and ratings from other users
3 Availability - Check stock
availability and stock on order,
so you know exactly when your
product will ship
4 Recommended Accessories -
Compatible accessories are
prominently displayed next to a
5 Datasheets - Download
2 manufacturer datasheets for
in-depth product data
6 Overview - Brief description of
5 the product and its capabilities
7 Specs - Quick view of the
product’s technical specifications
8 Similar Products - Easily find
similar products when the search
6 7 8
result doesn’t quite meet your



Browse over 1.2 million electrocomponent product solutions and all the
information you need to place your next order.

© Allied Electronics, Inc 2009. ‘Allied Electronics’ and the Allied Electronics logo are trademarks of Allied Electronics, Inc. An Electrocomponents Company.

Innovation: genius, Application engineers: Improving on

practice, or luck? serving the customer PCB design
The Newest
New Products
EDN 2008 Innovation Award
Winners and Finalists:

Cortex™ ARM®
8051 PSoC® Encore!® PIC32 Microcontrollers
Z8 Flexis™ AVR® MSP430 HCS08

Coldfire® Piccolo™ Microcontrollers:

PIC® Z80® ST7 TMS320F2802x

HC12 ST6
HCS12 ST9 Microcontrollers

The ONLY New Catalog Every 90 Days

Experience Mouser’s time-to-market

advantage with no minimums and same-day
shipping of the newest products from more
than 390 leading suppliers.

The Newest Products

For Your Newest Designs (800) 346-6873

Over A Million Products Online
BY R I C K N E LS O N , E D ITO R -I N - C H I E F

genius, practice,
Three columnists recently
touched on this issue, and
reached three different conclusions.
David Brooks, writing in The New York
Times, contends that practice makes
or luck?
be synonymous with intelligence: “Intelligence is a matter
of output, not scores in a test. Einstein was unsuccessful
perfect (Reference 1). He begins by at school and no great shakes as a mathematician, but he
positing “certain paragons of great- was creative and insightful. ... A vivid interest in things, and
ness—Dante, Mozart, Einstein—whose an active desire to understand more about them, is a major
talents,” certain romantics would characteristic of intelligence.”
claim, “far exceeded normal compre- Where does this writing leave us? I’m inclined to agree
hension, who had an otherworldly access to transcendent with Frank, but he provides no recipe for success. We can’t
truth, and who are best approached with reverential awe.” revise our genes or command luck to smile on us. All we
Today we know better, Brooks claims, noting, “In the view can do is follow Brooks’ advice and practice. But we can
that is now dominant, even Mozart’s early abilities were not infer one suggestion from Grayling: Organizations wishing
the product of some innate spiritual gift. His early composi- to foster innovation must create an environment that fosters
tions were nothing special.” Mozart, according to Brooks, in its employees a vivid interest in and active desire to inno-
owed his talent to a father who made him practice. vate. That’s not an easy task in this day of budget cuts and
Is practice, whether with or without innate talent, layoffs—which can distract even the most innovative engi-
enough? Not according to Robert H Frank, a Cornell Uni- neers, wherever their innovative spark originates.
versity economics professor, writing in the The Huffington One way to learn to focus is to study how successful
Post (Reference 2). “There’s no question that hard work innovators operate. The stories of three of them appear
and talent make someone more likely to achieve economic in this special section. Jim Williams of Linear Technology
success,” he writes. “But for every successful person … warns of the dangers of the rigid setting of goals, which
there are hundreds of others who are just as talented and individuals might meet at the expense of companywide
work just as hard, yet earn only modest incomes.” He con- innovation. Chuck Grant of Cadence attributes his innova-
cludes, “Even talent and the inclination to work hard are tive career in part not to a narrow focus on one area but
themselves heavily dependent on chance.” rather to the chance to gain experience in test, marketing,
In engineering, given sufficient talent and—if Frank is training, sales, and customer support. And when asked
right—luck, what might practice do for us? It might allow how he became an innovative engineer, Cadence’s Ken
us to memorize equations and programming languages, Wadland says simply, “I don’t follow rules.”
for example. Is rote memorization helpful? Not according Read the profiles in this issue, and review our previous
to AC Grayling, writing in The Guardian (Reference 3): “EDN Innovators” profiles at Let
“It is a common presumption that if people know a lot, they us know what you think.
must be intelligent. Anyone who can reel off capital cities
or count to 10 in several languages ... is counted a bright REFERENCES
spark.” But, he continues, “There are plenty of very bright 1. Brooks, David, “Genius: The Modern View,” The New York
people who do not know the world’s capitals and cannot Times, May 1, 2009,
count in other languages, because they have never had a 01brooks.html.
chance to learn them. ... By the same token plenty of people 2. Frank, Robert H, “Success and Luck,” The Huffington Post,
know lots of facts without being creative, thoughtful, quick- May 2, 2009,
witted, humorous, and perceptive—the marks of true intel- success-and-luck_b_195162.html.
ligence.” 3. Grayling, AC, “Knowledge and genius,” The Guardian, May
How do these questions relate to innovation? Grayling 1, 2009,
doesn’t use the term, but for him innovation would seem to genius-knowledge-iq-tests.


Breaking Through Performance Barriers
Speedster is the ™

World’s Fastest FPGA

New Speedster 1.5 GHz FPGAs

Breaking through the speed barriers of traditional FPGAs, High-Performance Applications
fully reprogrammable Speedster devices from Achronix UÊ iÌܜÀŽˆ˜}
Semiconductor are capable of operating at 1.5 GHz UÊ/iiVœ““Õ˜ˆV>̈œ˜Ã
peak performance. UÊ ˜VÀޫ̈œ˜
Speedster’s unique 40 lanes of 10.3 Gbps embedded SerDes
UÊ ˆ}ˆÌ>ÊÈ}˜>Ê«ÀœViÃȘ}
and four independent 1066 Mbps DDR2/DDR3 controllers
enable extremely high I/O throughput to match the device’s
outstanding internal performance. At 3X faster, Speedster
opens up new worlds of application design previously
See what’s possible:
unavailable to engineers using traditional FPGAs.
Familiar Silicon. Familiar Tools. Fast Time-to-Market.
Speedster uses familiar LUT-based fabric and standard
synthesis and simulation tools, so designers can use
standard RTL.
Jim Williams is a staff scientist at Linear Technology Corp. Longtime EDN
readers recognize Williams as a vital contributor of analog-themed articles over
the last 30 years. Williams has worked for 27 years at Linear Tech and has held
previous roles at National Semiconductor and the Massachusetts Institute of
Technology. EDN Technical Editor Paul Rako asked Williams about the chang-
ing role of application engineers—from essentially a support function of sales-
people to full-blown system designers. Williams explains how a modern view of
application engineering can ensure that your company will remain innovative.

Application engineers:
Application engineers have a differ- that the typical analog customer is not
ent job today from in the old days. looking at your parts catalog. They’re
How have things changed in the coming to you to solve a problem.
years you’ve been in the business? They’re ignorant of analog technology,
I’ll direct my comments at analog- but that’s no crime. They’ve got 500
Jim Williams application engineers. The big change line items on their board to check off.
is that the customer who calls you up They’re not analog designers. They just
on the or otherwise contacts you is generally need to access analog technology, and
changing role not an analog designer. They knew a lot that’s what you’re going to do for them.
more 20 years ago about what they were They’re not coming to you to ask you
of application doing than you ever would. They had questions about components. They’re
engineering specific questions on specific line items coming to you to solve a problem.
in a data sheet.
Those people are still out there, but So 20 years ago, what was an appli-
they’re in the minority now. The cus- cation engineer doing?
tomer who’s calling you up today, as The classic application-engineering
a rule, is not an analog designer job 20 years ago was [working on] data-
but needs to access analog sheet support and writing app notes
technology. So rather than about existing parts. … [They provided]
a specific question about a high-level support for existing parts—
specific part or a specific how to use existing parts—in the appli-
characteristic of a specific cations they [felt they were] likely to be
part, they’ll call you up used in. So it was largely a reactive job.
and they’ll tell you, “I have
so much space, so much And now what do you see?
time, so much power, and so That reactive component is still there,
much money. What do I do?” although in this company, the designer
They’re not asking you detailed writes the data sheet; we insist on
technical questions. They’re that. He may seek assistance with the
asking you what to do. application engineer, but the designer
That means you’re in the ser- is responsible for getting the data sheet
vice business now; you’re provid- out the door. Those components that
ing design services. That also means application people were engaged in 20


“The fundamental responsibility of an analog-
application engineer is to understand his
customer’s problem and provide a solution.”

years ago are still there. But in a well- a field-application engineer? Are We’re not saying he has to be at the
run analog company, they’re second- or those two different jobs, or are headquarters. … He needs a lab, and
third-tier parts of the job. The fun- those roles converging? he needs time. The greatest leverage
damental responsibility of an analog- They’re two different jobs, but there’s any engineer has in doing his job is
application engineer is to understand overlap. The biggest single difference time. And, nominally, the field guy is
his customer’s problem and provide a between the two is that the factory- spending the bulk of his time working
solution. application guy is much more future-ori- with customers on existing problems
For me, that solution may involve ented, and he’s much more laboratory- with existing products. The factory guy
this company’s products. I’ve also based because he needs a laboratory. should be spending the bulk of his time
provided solutions to customers that The field guy, unless he’s got a labora- thinking about what needs doing in the
use this company’s products and com- tory at home, hasn’t got access to a labo- general sense.
petitors’ products because those are the ratory, or, if he does, it probably isn’t as
realistic solutions. I’m the customer well-facilitated as [that of] the factory Do the application groups for field
advocate, and I’m seeking a long-term guy. So the factory guy can spend a lot and factory have a responsibility to
relationship with a customer. But what more time on research and development show customers how to use soft-
I’m doing and what I hope the other for future products. The field guy is busy ware tools?
application engineers are doing around servicing existing customers with exist- Well, let’s talk about tools. There’s LT
this company is … servicing those ing problems, which is important, obvi- Spice. There’s Webench. There are all
customers who want to access analog ously, … but the factory guy’s not doing the various software tools. But there are
technology but don’t have the exper- that. If he’s focused on servicing existing also screwdrivers, shears, curve tracers,
tise to do it and don’t have the time to customers with existing products and and X-acto knives. The application
develop that expertise. getting existing sockets, he’s mortgaging engineer’s job is to emphasize and show
That’s a large part of the job. Part his future. He’s eating his seed corn and the customer what tool is appropriate
and parcel of it is that you’re garner- developing nothing. for what task. For some tasks, LT Spice
ing a feeling for what needs to be built That’s the fundamental difference is the appropriate tool. For other tasks,
next. You’re defining new products. between the two positions: The fac- a cut-down X-acto knife is the appro-
And an application note today that’s tory guy has the luxury of being able priate tool, and that should be reflected
written by an application engineer may to spend more time on futures and on in applications. What’s the real down-
involve support of an existing product, issues. If factory engineers are pushed in-the-dirt way to get from A to Z? Is
but it’s much more likely to be written toward devoting 90% of their effort it Spice? Is it an X-acto knife? Is it a
around current technical issues. to getting sockets, you’ve success- breadboard? Is it cutting copper clad?
fully mortgaged the company’s future. Is it some fusion of all of those? Tool
Do you distinguish between an Whereas the field guy is with custom- development, tool use, writing it up,
application note written by an ers all the time. They’ve got today’s how to measure, how to simulate, when
application guy versus the applica- problems surrounded by today’s parts to measure, when to simulate, where to
tion section of a data sheet written that they can get their hands on now. simulate, and when to cut copper clad
by a design engineer? So they’re the heroes of today. But are all part of the application engineer’s
There’s a complete difference. An in a well-run application effort, the job in educating the customers on how
application note, if it’s properly written, heroes of tomorrow are in the factory. to solve their problems.
stands a good chance of having a 10- or That doesn’t mean the factory doesn’t Customers like data sheets. Custom-
15-year lifetime because it’s issue-cen- provide backup for the field guy when ers like app notes. Customers like pub-
tered; it’s not product-centered. There’s things get sticky. You’ve got to do that, lications that educate them. Customers
product in it that is illustrating various but that plays into tomorrow, because like advice over the phone. Customers
types of solutions, but, if the application you look at what comes over the wall like software programs that help them
note is really well-written and well- that the field people can’t handle, and design. But what customers love, what
thought-out, it’s applicable even when you see trends, which suggest products. sells a product like nothing else, is that
the parts are long gone from the scene. simple little cardboard box arriving in
You’re writing about issues. You’re writ- And it doesn’t physically mean that the mail with a breadboard that works
ing about approaches. the guy works at a factory. It could when they plug it into their system. No
be that he works at a local office sales pitch, no sales routine, no software
Do you make a distinction between but is acting as a factory-applica- program, no phone conversation, no
a factory-application engineer and tion guy, right? e-mail sells products like a working


Focusing on a green supply chain?






5 7

see clearly
IHS knows green Design, manufacture and trade anywhere
Whether it's compliance with regulations, Providing expertise and information designed
ensuring sustainability and high reliability, to manage product lifecycles and enable green
or managing supply chain volatility, IHS performance, IHS critical insights offer decision-
understands that green means different things making agility across the supply chain.
to different people. • Eco-Friendly and Lead-free Design
• REACH Chemical Performance
Supply chain, sourcing, and engineering profes- • Greenhouse Gas footprint
sionals rely on IHS information to comply with • High Reliability and Sustainability
regulations, demonstrate social responsibility, • Obsolescence and PCN/EOL Management
and maintain uninterrupted product flow. • Global Standards, Trade, and Security

Contact IHS and go green

Worldwide: +1 888 752 0334

US/Canada: +1 303 397 2892
“There’s no patent on intelligence. Anybody
can learn to do anything. It’s just people
haven’t got the time.”

breadboard mailed to a customer. I can’t are topic-centered. The thesis there is Don’t accept responsibility for the
say that loud enough or long enough. that an educated customer is a better design; there are potential legal prob-
Nothing sells like a board that works customer, and an educated customer lems.” Many companies went out of
in the customer’s system. That is the will come back for more, I hope, to your their way to tell their application guys,
ultimate analog-application support. company. “Don’t get knee-deep in a design and
Nothing beats that. The noble way says you educate the accept responsibility.” Today, accept-
customer through publication. The less ing responsibility is a sales tool. That’s
What about compensation of fac- noble but effective way is the refer- turned 180⬚. There is a range of solu-
tory- and field-application engi- ence design where the customer says, “I tions available from a number of com-
neers? The field people get bonuses haven’t got time to come up to speed. petitors, and, if you want to distance
for filling sockets. Are you against You’re providing me with these publica- yourself from the competition, you’re
having factory-app people compen- tions, but I haven’t got time to come going to have to accept responsibil-
sated by filling a socket? up to speed. I need that little cardboard ity. It’s based on the complexity of the
This might get me into trouble in some box with the breadboard.” product, and it’s based on its availability
sectors around here, but I’m against all That is a reference design. It’s essen- in one form or another.
forms of goal-setting for factory-applica- tial because customers are out of time.
tion engineers because they’ll pursue They haven’t got the time to come up So this way is the new way. That’s
those goals and let other things slide to speed to execute the architectural how you see application engineer-
by the wayside that could be the future issues in a circuit themselves. And ing developing?
success of the company. there’s also a real marketing issue. The analog business is a service busi-
If you sit down and you agree with Through-hole is dead. We can bread- ness. I can walk into a customer’s facil-
somebody that you’re going to be board here at the factory, but, by and ity and hand them data sheets and app
rewarded if you do this, this, and this, large, customers can’t try stuff anymore. notes and parts. That may have been
then that’s what a lot of people are The parts are too damned small. a sale 25 years ago. It isn’t anymore.
going to be tempted to do. Then they’re You’re going to have to take the part They haven’t got the time to read the
going to see this, this, and this, which you’re trying to sell for the problem data sheet. They haven’t got the time
look interesting and, potentially, fruit- you’re trying to solve, and you’re going to develop analog expertise. They
ful, but they know if they do this, this, to have to incarnate it on a board-level may not have the inclination; they’ve
and this, they’ll get rewarded. reference design, because the custom- got other things to do, and they can’t
It may be that goal-setting in a field ers haven’t got the time to develop the breadboard with the parts. They say,
engineer’s job seems unavoidable, expertise they need to execute a design “So your parts are interesting, your app
because they’re servicing existing cus- and because they have a hard time notes are pretty, your data sheets are
tomers with existing parts, and that’s breadboarding and playing. pretty, but I need something I can clip
the only metric that management Reference designs and demo boards into my system that’ll work.” That’s
can use. But in a factory job, which are important. I would say of the demo a complete shift from 25 years ago,
is nominally R&D-based, if you line boards that get adopted and the refer- when management was telling applica-
up a bunch of goals for somebody and ence designs that get adopted, a third tion engineers, “Be very careful about
tell them that’s the way to success in get used pretty much the way they are, getting knee-deep in your customers’
this company, you’re destroying the and two-thirds get used as a place to designs. We don’t want to get sued for
company’s future. You’ll have a bunch start. The customer comes back to you a field recall.” That’s 180⬚ out of phase
of people who are successful according and says, “I fired up your demo board with what’s going on today. You’re ...
to their goals and a company that stalls and looked at your reference design. It over your knees, over your head in your
in innovation. seems to work pretty well, but I needed customers’ designs.
twists here, there, and the other place.”
I’m curious how you see application But the point is, the reference design With many companies, there’s no
engineering helping with reference provides an advanced place to start analog team. There’s no analog
designs in general. talking from. engineer.
There are two ways applications can There’s something else that comes They’ve got other things to do. There’s
export design expertise—which, osten- to mind with reference designs. In nothing magic about the analog field.
sibly, I hope they have—into the world. most analog companies 20 years ago, It’s just people have other things to do.
One way is what I call “the noble way.” the word from on high was “Don’t get There’s no patent on intelligence. Any-
That’s through application notes that involved in your customer’s design. body can learn to do anything. It’s just



Avnet lights the way to reliable,

durable, and sustainable LED Solutions.
Choosing LED technology for your design is only the first step on the path.
Avnet Electronics Marketing helps light your way to selecting LEDs that
meet your reliability, visibility, and availability requirements. At each stage
of the design cycle, our team of illumination-focused engineers gives you
access to the latest information on LED products, ensuring you find the
right solution to fit your specific design needs. When tackling the challenges
of thermal management, power driver stage and secondary optics, our
experts are your source for leveraging the benefits of LED technology.

As a unit of Avnet Electronics Marketing, LightSpeed brings together the

> SECURITY world’s foremost LED, high-performance analog and optical/electrome-
> MEDICAL chanical manufacturers along with best-in-class technical expertise and
supply chain management services – affording you quicker time to market.
Working together, we can help you bring your ideas to light.
> ARCHITECTURE For more information and to view the latest issue of Light Matters

©Avnet, Inc. 2009. All rights reserved. AVNET is a registered trademark of Avnet, Inc.
“If you can’t explain how something you did
works to a general audience in a general kind of
way, you don’t know how it works.”

people haven’t got the time. You walk will appear on the other side of the something can be done, the only ques-
into a customer’s facility, and you look fence. No one knows quite how. It’s a tion is how. There’s no advantage to
at their blackboard, if they have one; tunneling process. secrecy. Whatever advantage you’ll get
there’s 50 items to be checked off there, by protecting some body of knowledge
and four of them are analog. Do you think application engineers that you have is dwarfed by the good-
They don’t need to develop in-house should be writing magazine articles will and the good orders and the sense
analog expertise. They’ve got analog as you do? that you’re a problem-solving ally when
companies who will do it for them. That Application engineers should definitely you disseminate that knowledge.
doesn’t mean there aren’t companies write articles, for a couple of reasons.
out there that are well-steeped in analog The most obvious reason is that it’s So customers at system companies
expertise. It just means that most of good for the company’s image, but, appreciate the openness of a mod-
the products being built today that use more important, if you can’t explain ern application engineer?
analog technology are being built by how something you did works to a They’re looking for somebody they
companies that are not steeped in ana- general audience in a general kind of can trust technically as a partner. Most
log-design techniques and don’t have to way, you don’t know how it works. It’s technological trade secrets are short-
be. That’s what they look to us for. a great way to test your own level of lived at best. You’ll do your company,
understanding of what you just finished your professional reputation, your pro-
Where do we find this new breed doing. fession, and everybody else a lot more
of application engineers? Writing is essential. It’s an essential good by saying, “Here’s this problem I
The far side of Alpha Centauri. I don’t part of an application engineer’s job had, here’s how I solved it, and here are
know where the hell you find them. and communication skills. It’s harder the results.”
You like to think you can breed them, to find people with really good com- There’s really no advantage to hold-
but it takes a long, long time and a lot munications skills than with technical ing back. I don’t recall ever holding
of burning of fingertips. To some extent, skills. Writing is important because it back a measurement or design tech-
you can breed them. To some extent, tests your knowledge of your ability to nique from publication in an app note.
they find themselves. You still find kids understand what you really did. It puts Remember a number of years ago,
coming out of college who’ve been your company in a good light, and, when there were soft errors in memo-
playing with electronics since they were career-wise, it puts you in a good light. ries? Intel figured it out. It was [caused
in grade school. Those people still exist. I can’t imagine working on a difficult by] alpha particles. Intel released it
They’re aberrant. They’re weirdoes, but problem over a protracted period of and gave it away, and a lot of people
they’re wonderful weirdoes. time and then not writing it up. Make said, “Intel is crazy. Intel shouldn’t do
You still find those people who got your contribution. If you’ve got some- that.” So you’re making friends, you’re
addicted early and found a way, despite thing you think is worth talking about, generating credibility for yourself and
the surface-mount revolution, to hack make your contribution. your company, and [disseminating it
electronics in grade school and high obviates any] short-term benefit that
school. You do find people who’ve been I see a difference between an artist you can gain by keeping things close to
doing board-level design, sometimes and a tradesman. An artist will say, the chest.
your own customer, who wants to come “Here’s the palette, here’s the paint There are certain manufacturing
over—not often, but it happens. Those I used, and there’s my canvas. Have processes, trade secrets, and the fab,
are the two major sources: lifelong cir- at it.” They share all that important stuff like that, that you’re going to play
cuit freaks and system guys who’ve been stuff, whereas a tradesman would close to the vest. But you’re not going
working for board-level-product houses say, “Oh, this is my secret little to have to play them close to the vest
who want to come over and do this thing, and I keep it to myself.” You for long because they’re going to be
kind of work in a semiconductor com- take kind of a higher-plane view of obsolete in six months or a year any-
pany. Also, to some extent, inbreeding things and act like an artist. Why way. But in application engineering,
within the company [and] mentoring not keep everything secret? measurement technique, circuit-design
[help]. But there is no official, if you I think [the “father of the atomic technique, whatever, you’ll generate
will, path toward finding these people. bomb,” Robert] Oppenheimer was more new customers by printing it all
It’s quantum mechanics. If you line up right. There are no secrets. The only in an app note than you’ll lose because
enough people on one side of the fence, secret to the atomic bomb is that it some competitor took your scheme and
experienced system designers who want works. ran with it.—interview conducted and
to work in a semiconductor company Once a skilled technocrat sees that edited by Paul Rako








www.cUI .com | 800.275.4899

SERDES-Capable FPGA Solutions for Wireless and Wireline Applications


Built for high-speed communications protocols, the ECP3 helps you keep your power budget
down, and applications features high. At half the power and half the price of competitive devices,
now you can do more with less.

Learn more about the ECP3 family at:

©2009 Lattice Semiconductor Corporation. All rights reserved. Lattice Semiconductor Corporation, L (& design), Lattice (& design), LatticeECP3, ECP3 and specific product designations are either registered trademarks
or trademarks of Lattice Semiconductor Corporation or its subsidiaries, in the United States and/or other countries. Other marks are used for identification purposes only, and may be trademarks of other parties.
In 1995, OrCAD had the most popular PC-based PCB (printed-circuit board)-schematic
tool. Another popular tool was available from PADS, which Mentor Graphics now owns.
Engineers designed PCB schematics in OrCAD and laid out the board in PADS, a clumsy
approach that still persists to this day. Rather than redesign its PCB-layout tool, OrCAD pur-
chased Massteck, a small company offering a good layout tool and, more important, an auto-
router that achieved 100% routing on complex boards. PCB designers Al Akermann and Gene
Dancause, Massteck’s founders, contracted with Ken Wadland, a computer-science guru, to
do the autorouting algorithm and database architecture. Just as vital, they hired board designer
Chuck Grant, who first did product testing and then eventually wrote software. Cadence
in 1999 bought OrCAD. Sadly, OrCAD no longer supports the original Massteck-based
OrCAD Layout tool, instead favoring OrCAD Editor, a stripped-down version of Allegro,
Cadence’s enterprise-class layout software. Wadland and Grant now work for Cadence in a
Massachusetts design center. EDN recently had the chance to interview them.

Improving on
PCB design
Where did you get your education? the design and completely finish that
CHUCK GRANT: I got an economics area. Once you clear out that most
Ken Wadland and degree from the University of Rochester. difficult area, you just move on to the
Chuck Grant: the Other than using some computers in next one, and you keep stringing them
those courses, I hadn’t been interested out until you get to the end. When you
innovators behind in them. But the PCB-design business get to a clear area, you could pop in
was growing by leaps and bounds, so a via and use the other layer to finish
OrCAD Layout they were looking for PCB designers. I it. We knew that the value added for
software just went in and I liked what I saw. PCB design is in autorouting. So we
KEN WADLAND: In seventh grade, I said, “Let’s take the technique we use
started Wadland TV and Radio Repair. ... to divide and conquer as a manual
In junior high and high school, I started designer and apply it to software.”
learning assembly-language program- KW: I couldn’t resist the allure of
ming. I went to Worcester Polytechnic money. [Autorouting] was the only rea-
Institute and got my degree in math- son I was attracted. At that point, there
ematics, but I took every computer- were only two hard problems in the
science course they offered. I got my world: AI [artificial intelligence] and
master’s degree and my doctorate autorouting. Autorouting is a mathe-
at the University of New Hamp- matically complicated problem. Beyond
shire. I went to Pittsburgh State being mathematically complicated,
College, and I created the [com- there are the aesthetics and the electri-
puter-science] curriculum there. I cal-design rules and everything else that
was the first chairman of the com- made it just a challenging product.
puter-science department.
What made your innovations so
What gave you guys the entre- effective?
preneurial spirit? CG: In a start-up company, you wear
Ken CG: There was a lot of evange- a lot of different hats. The hats I wore
lism in that we had a unique went from testing to marketing to train-
Wadland way of designing PCBs. The ing, sales, and customer support. One
way we divided and conquered was of the nice things about a small com-
to start from the most difficult part of pany is that you really can try all those


things out and see what you like and start a project unless at least one other to get into, and a lot of that happened
gain a lot of experience. We had some person failed at it. Why waste your time after the purchase by OrCAD.
good PCB designers in that company on something easy? The downside is:
who were proficient, as well. I decided You fail a lot. To do innovation right, Users love the spreadsheet func-
that I would like to learn how to be a you have to ignore the details until you tions in OrCAD Layout. How did you
programmer, and, of course, I learned have already established the concept, come up with incorporating spread-
from the best, which is Ken, so that was and then you see if the details work. If sheets in the product?
a great opportunity for me. [Innova- you’re too detail-oriented, you reject it CG: We were unique in that we lever-
tion] was driven by the PCB designers too soon. You think of really outlandish aged existing technology before a lot of
because we were used to the time pres- things, most of which won’t work. You other people ever thought of doing that.
sures and deadlines and the need to get get a bunch of outlandish things, and We knew about Lotus 1-2-3, which we
it right the first time. [We felt] “Hey, we then you start pruning them down to used on a PC for our own accounting,
get to invent our own, so let’s do what one that can actually be usable. If you so we leveraged the idea of spreadsheets,
we want to do, and this is what it’s going can’t suspend disbelief for a long enough not really realizing that they would take
to look like.” time, you can’t get these other ideas. over the world later. It was just some-
KW: Well, we did not have a marketing One of the things I teach in my classes thing that looked like it would work well
group separate from our PCB designers. is: Never write code until you’ve found for us. It allowed us a lot of flexibility.
All the people in sales and marketing three ways of doing it. A lot of people We could lock and unlock nets, establish
were PCB designers. So when they gave [take] the first [approach] they think library components or library symbols,
us the requirements, they understood of. Chances are, it’s the wrong way or and attach them to the components
what they were talking about. a less-than-perfect way. So you want to and all the things we needed to do. It
find three approaches to the problem, all worked well with that spreadsheet
What drove your product innova- analyze all three of them, and then pick interface, so we went to town with it and
tion? the best. People just don’t do that. They used it because it worked so well.
CG: We really had to be evangelists in just don’t take the extra time to do that KW: Making a system that’s simple
interactivity. The marketing machines extra step. is very hard. One of the early win-
of all the CAD companies were touting ners was the Lotus 1-2-3 spreadsheet.
“push the button.” That was their idea, Why was the program so intuitive There was some brainstorming between
“push the button, and we’ll route your and easy to use? Gene Dancause and me, but it was my
board for you.” Well, real PCB design- CG: [It was not] that some marketing attempt to write a minimum number of
ers know that [approach] never works. spec had been translated through an lines of software to get maximum capa-
You get as much automation as you can, engineer. It was a very direct experi- bility. We wanted the ability to make
and if you can guide it, all the better. ence. Those of us all the way up to
The interactivity we built in was that the president of the company would
you could choose where on the board be using the software and sending the
you wanted to start routing. You could feedback to Ken and to the engineers.
start in two different places and have We were very much customer-focused.
them join up in the middle. So there I think one of the early innovations
was interactivity, and then there was was to use interactive forums on the
the push and shove, which everybody Internet to get feedback from custom-
copied. The idea that you could pick a ers. Everybody could read the answer,
single line, route it between two other and you’d end up with a nice, long
lines, click, and [have] everything push FAQ [frequently asked-question] list.
out of the way was something that no The OrCAD influence, what it really
one else had. We would let you see brought to the table in Layout, was an
what it accomplished, and, if it didn’t emphasis on ease of use. Back when
finish, you might say, “Well, I can fix up we began, PCB designers were career
that mistake over there, or I’ll take care people, who would spend hours and
of it.” But we wouldn’t break any design hours on a single machine, on a single
rules. We would leave unroutes where piece of software. It didn’t matter if it
we couldn’t get them to fit. If you didn’t was complicated because they would
like it, you could go back to what you learn it eventually. Layout tools were
had before. We had an undo right from plenty complicated, but, as time
the beginning. went by, engineers started to use
the software. So we would Chuck
How do you become an innovative take into account their
engineer? wishes and requests and Grant
KW: I don’t follow rules. I don’t like to tried to make it easier


“click” “click” “click”

Kellysearch clicks. Every time.

Kellysearch finds your next great supplier in three clicks or less. Talk about an elegantly
engineered solution. Log on now at
“Every time we got a board to 100%, we jumped up and
down. Getting 99.97% was cool, but if you hit 100%?
Wow, we were excited.”

changes to more than one object at isolated the database from the interac- an autoplace algorithm, and we put
once. So we came up with this spread- tivity by using an object-oriented type enormous resources into it. It’s a sim-
sheet because our database resembled of programming so that we prevented pler problem mathematically in the IC
SQL [Structured Query Language]. It database corruption, which was a com- world, where everything’s symmetric.
was kind of a relational database under mon problem. In most CAD systems, It’s a problem in the PCB world, where
the hood, so producing a spreadsheet the programming was done so that you every component is a different shape.
was a piece of cake. And then editing could manipulate an object directly. [We Things are not interchangeable. If
it through the spreadsheet was just a lot made] a copy of the object, let the user one’s a quad pack, you can’t switch that
less software to write than writing dia- manipulate it, and then let the user com- with a two-pin surface mount. They’re
logues for everything you wanted to do. mit that [copy] to the database. It would incompatible. So we failed in placing.
It was basically being lazy. all happen under the hood. What that As far as I know, no one has ever had a
allowed us to do is to recover from errors successful place tool in the PCB field.
How did you conceptualize the without the user’s being aware of it.
color-rule interface? KW: We knew from the beginning Was there a personal victory you
CG: The idea was that the program that a 64-[kbit] database was not going had?
was WYSIWYG [what you see is what to be big enough. Back then, we had CG: I will never forget one of those
you get], so you could say, “OK, I want expanded memory and extended mem- sessions when I had one important
to turn on these objects.” This might ory. One of them came in banks, and customer who absolutely needed to
be the etch and the pads on the top the other one worked only in 32-bit have the ability to move components
layer, and that’s my Gerber [file], and mode. We took this “get-put” method- and have minimization take place as
then you just print it out. So then you ology: ... Under the hood, when you say they were moving. [We had] the first
have a set of these [files], which are on “get,” we’ll switch in the right bank, product I know of that had interactive
a spreadsheet, and each one of them or we’ll switch to 32-bit mode and get component moving and then [PCB]-
establishes a set of rules [of] what’s the data for you. When you say “put,” trace minimization as you were moving
going to be visible for that artwork we’ll do the same game back again. So it. Initially, it was just a rat’s nest, and
layer. If you have the assembly layer, the application lived in the 16-bit Win- then we had a customer who said, “I
the silk screen, and all of those things dows environment, but it could access have to see the real traces.” I definitely
that also come with a PCB, ... that’s unlimited memory. We had a complete remember that one.
what will print out on your Gerber. separation between the UI [user inter-
KW: The problem they posed to me is: face], the algorithms, and the database. Was there a company victory you
“We want to see the colors in different I had used that concept when I worked remember?
ways at different times.” Some engineers with relational databases, so I under- CG: In 1994, Jeff Hawes, [a former ap-
wanted to see the top layer red; some stood the concept, and I applied that to plication engineer at Cadence, OrCAD,
wanted it green; and some wanted to an object-oriented database. and Massteck], and I went to the PCB
see this [color] differ some of the time. Design Conference and we entered
I told them “I can’t do all that. How What was the biggest problem you MAX EDA into the competition. They
about if I just give you a color set and felt you had to take on? were assigning points to PCB-design
you can change the color set when you CG: The biggest problem in marketing systems, and what they did was pretty
want?” The other thing that came into was that the PCB designers don’t make comprehensive. You had to design a
it is my campaign to protect people who buying decisions. They can influence board, sight unseen. They gave you a
are colorblind. My son is colorblind. them to some extent, but upper manage- net list and the parts. You had to design
So I refused to have any products that ment and engineering make most of the the board, create the netlist, bring it in,
had fixed colors. Everything had to be decisions. It was difficult to convince autoroute it, and print out Gerber art-
adjustable. We came up with this idea. people who’d never done PCB design work—all within two days. ... We won
We call it “color set.” You set up all the that interactivity was critical because that contest in 1994 and 1995, and that
colors the way you like, you save it, and someone else would come in and say, was the year OrCAD bought us. When
then you pick from those color sets. “We don’t have interactivity; you just we found out that we had won that com-
push the button.” Well, you can push petition, we were elated.
Was there any innovation due to the the button on ours, too, but that’s not KW: I’m not sure there’s any one [vic-
limitations of a 286 PC, and did you really what you want to do. I think we tory], but basically every time we got
program in C? did eventually win that war, where inter- a board to 100%, we jumped up and
CG: It was all C. The other innova- activity became important to people. down. Getting 99.97% was cool, but if
tion that we had was in the way we did KW: I’ll tell you the biggest problem you hit 100%? Wow, we were excited.
software best practices. This [idea] was that didn’t get solved. We tried to get a —interview conducted and edited by
Ken’s innovation. We ran C, but we good place algorithm. We tried to have Paul Rako


Looking to buy?
With BuyerZone, every big deal
is no big deal at all.
BuyerZone is the trusted, unbiased site that brings buyers and sellers of business purchases
together to make the purchasing process easier while saving both parties time and money.
Credit Card Processing, Building Services, Home Security, Trade Show Displays, Forklifts and
Construction Equipment, Business Purchases and more... Let us do the work so you don’t have to.

Join the millions who’ve already saved time and money on many of their purchases by taking
advantage of BuyerZone’s FREE, no obligation services like:
• Quotes from multiple suppliers • Pricing articles
• Buyer’s guides • Supplier ratings
• Supplier comparisons


Call (866) 623-5525 or visit Where Smart Businesses Buy and Sell
A division of
than 10 Hz. The accelerometer mea-
sures dynamic acceleration with user-se-
lectable measurement ranges of 2, 4,
8, and 16g and employs built-in orienta-
tion sensing using simple register reads.
Available in a 330.95-mm, 16-lead
LGA package, the ADXL346 costs
$3.04 (1000).
Analog Devices,

Voice-coil-motor driver
uses I2C interface
 Suiting autofocus and zoom ap-
plications in camera-phone lens-
es, the A3907ECGTR voice-coil-motor
driver has an I2C serial interface and a 2.3
to 5.5V operating range. The driver has a
102-mA maximum output current and
Motion sensor provides acceleration values operates at 40 to 85C. Programmed
using the I2C in-
in absolute-analog outputs terface in 100-A
 Providing acceleration values in the form of absolute-analog outputs, the
LIS352AX three-axis-accelerometer motion sensor operates with supply volt-
increments, the
I2C inputs set
ages of 2.16 to 3.6V and provides stability over temperature and time. Internal regu- the internal 10-
lation voltage makes the measurements insensitive to variations in power-supply bit DAC output-
voltage that typically occur in battery-operated devices, such as mobile phones and voltage reference
other portable devices. This method enables space saving and reduced cost by con- for linear-current
necting the sensor directly to the battery without a separate voltage regulator and control using an output-sink MOSFET
by providing compatibility with the power-supply voltage that the application uses. and sense-resistor feedback. Addition-
The LIS352AX has high stability over a temperature range for zero-gravity offset al features include a 10-nA input-sup-
and sensitivity, a 0.3-mg/C offset drift, and an accurate output over a full-scale ply current in sleep mode and I2C-com-
range of 2g. Available in a 351-mm plastic package, the LIS352AX motion patible logic-input levels. Available in a
sensor costs $1.30 (10,000). 1.4650.965-mm, six-bump WLCSP-
STMicroelectronics, CG package, the A3907ECGTR costs 34
cents (1000).
Allegro Microsystems,
MEMS motion sensor shock and static acceleration, allowing
measures tilt, shock, and use as a tilt sensor. The device includes
an on-chip FIFO-memory block storing
acceleration as much as 32 samples of X-, Y-, and Z-
 Measuring tilt, shock, and ac-
celeration, the ultra-low-power,
axis data and offloads the FIFO function
from the host processor. This process al- CIRCUITS
high-resolution ADXL346 iMEMS ac- lows the host processor and other pe-
celerometer operates at a 1.8V primary ripherals to enter sleep mode when not Drivers suit white LEDs
supply voltage. The digital three-axis in use. Features include a 0.1- to 1600-
smart-motion sensor provides select- Hz selectable bandwidth and 150-A with flash-mode capability
able measurement ranges and measures
dynamic acceleration from motion or
power-consumption ranges at 1600-Hz
bandwidth and as low as 25 A at less  Claiming an 80% efficiency in
flash mode, the TB62730WLG

72 EDN | JUNE 25, 2009

and TB62750FTG step-up dc/dc con-
verters for white LEDs integrate an
NMOS transistor. Requiring one exter-
nal inductor, the devices drive 800 mA
circuit outputs for each meter and
provides a real-time-angle-tracking
scheme to avoid display errors. The
device enables communication with
using lithium-ion batteries. Features
include overvoltage protection, a 2.8
to 5.5V supply-voltage range, and a
temperature-shutdown function. The
the microprocessor using a three-wire
serial link and provides an error-status
output pin. Available in an SSOP20
package, the MLX10420 costs 98 cents
TB62750FTG also uses an integrated (10,000). This advertising
constant-current driver for a red LED. Melexis Microelectronic Integrated
The TB62750 comes in a VQON24- Systems, is for new and
24 package, and the TB62730 comes
in a 2⫻2-mm WLCSP-16 package. current products.
Each device costs $1 (10,000). Analog Class D
Toshiba America Electronic Com- amplifier has closed-
loop architecture
Gauge driver runs  The 10W CS3511 stereo-ana-
log-input Class D amplifier uses
as many as three true spread-spectrum-switching-con-
troller technology and a delta-sigma
air-core meters modulator with a patented closed-loop
 Aiming at automotive and in-
dustrial applications, the triple-
architecture. Features include 0.025%
THD⫹N at 5W and 0.019% at 1W,
air-core-gauge MLX10420 driver has support for single-ended or fully dif-
a serial interface and drives two 360⬚ ferential inputs, and operation from
meters and one 90⬚ meter. The de- 9 or 12V. The amplifier offers built-
vice uses sin/cos PWM drivers and se- in protection and error reporting with
rial commands to control two sets of automatic recovery from overcurrent,
CMOS power bridges. An on-chip os- thermal-overload, and undervoltage
cillator sets 9-bit-per-quadrant-resolu- conditions. Measuring 6⫻6 mm in a
tion PWM and a 10-bit angle display. thermally enhanced QFN-32 package,
Other features include a power-on the CS3511 costs $1.39 (10,000).
self-test, which detects open- or short- Cirrus Logic,

Company Page Company Page
Achronix Semiconductor 58 LS Research 26
Advanced Linear Devices 20 MathWorks Inc 19
Agilent Technologies 2, 29 Maxim Integrated Products 45, 47
Allied Electronics 54 Mentor Graphics 17
Analog Devices Inc 15 Microsoft Corp 8
austriamicrosystems AG 52 Mill Max Manufacturing Corp 9
Avnet Electronics Marketing 63 Mouser Electronics 4, 56
BuyerZone 53,71 MP Associates Inc 38
Coilcraft 7 Murata Power Solutions Inc
CUI Inc 65 (formerly C&D Technologies Inc) 28
Digi-Key Corp 1 Pico Electronics C-3
Express PCB 26 5, 41
IHS International 61 Trilogy Design 73
Ironwood Electronics 73 Vicor Corp 27
Jameco Electronics C-4 Xilinx Inc C-2
Keil Software 37
Lattice Semicondcutor 66 EDN provides this index as an additional service.
Linear Technology Corp 42 The publisher assumes no liability for errors or
49, 50 omissions.

JUNE 25, 2009 | EDN 73


UART and transmit display data with

Weather or not: the other. To save time and effort, we
All signs point to “no” ordered an OEM board from a vendor
we had used before, and we recycled
a number of embedded C routines to
handle the serial-data streams. The
only remaining task was to write some
simple C code to transfer weather data
from one stream to another.
The initial work went perfectly. I
had hard-coded some test messages in-
to the interface, and the sign displayed
them perfectly. I set up the serial-input
buffer and state machine on the receiv-
ing port, and I could see the extracted
weather data in the debugging output.
I had only to include the weather data
in the output stream. I expanded the
output buffer to hold the larger data,
and I then reset the system.
Panic quickly set in. Instead of my
orderly flow of temperatures and pre-
cipitation counts, I got only garbled
letters and symbols. Something was
seriously wrong. Even though we had
tested the serial code in a number of
other applications, I began to comb
through it, looking for any mistake I
could have made. I checked and dou-
ble-checked every inch of the code.
usiness had been going well at my former place of

After banging my head against the wall
employment, a start-up company selling micro- for a few hours, I shut off my computer
processor-controlled weather stations to schools and went home.
The next morning I realized what
and broadcast entities. The start-up was taking the problem was. Prying up the main
off, orders were picking up, and our booth was RAM chip on the OEM board from its
popular at trade shows. Always on the lookout socket, I found that one of the data pins
for things new and different, our president had become enam- was bent under the chip, causing all the
ored of scrolling LED signs, and he insisted we needed one in ASCII characters to get scrambled be-
fore I sent them to the sign, which duti-
the booth at the next show. “Just imagine: Up-to-the-second fully displayed the gibberish anyway.
weather readings scrolling and flashing in bright colors will be After we straightened the pin, it
right behind the presenter! Who a look at the interface specification. I was not long before we had a working
wouldn’t stop to take a look at that?” was pleasantly surprised to find that sign-interface module, just in time for
he said. Although I had reservations it supported a rich serial protocol for the next trade show. Even our president
about the aesthetic appeal of baromet- transmitting and updating messages was impressed. “Looks great!” he said.
ric-pressure readings blinking in rain- for display. I needed only to convert “Now, I have this idea for a weather
bow hues, I had to admit that it would our serial-data stream from the weather billboard.”EDN
be a nice challenge that would show off station to a format that the sign under-
some of our company’s technical exper- stood. The job was starting to seem a JT Klopcic is a technical specialist with

tise. So I agreed, but I regretted it when lot easier. United Parcel Service Information
I found out that we had limited time to We wanted a stand-alone unit that Services (Timonium, MD). You can
get this show on the road. we could later sell, so we went with reach him at
The boss had already selected a an 8051-family microcontroller, which
vendor for the scrolling signs, so I took would receive weather data on one

74 EDN | JUNE 25, 2009

te grity
nen t In
itical C
r t e r s
Cr ve C o n u t
O u t p
D 0 0 0 V D C
t o 1 0 ,
2V 2 to 5000 VDC Outputs
Ultra Miniature
Surface Mount and Thru Hole
Single and Dual Isolated Outputs
Military Upgrades Available

Military Applications
100 to 10,000 VDC Output -400 and -550 to +850C
Proportional Control Up to 10 Watts Operating Temperatures
Military Components VV Series
Wide Input Range
-550 to +850C Operating Temp
3.3 to 350 VDC outputs
Military Environmental
in 1/2 Brick and Ful Brick
Screening Available
Military Upgrades Available
3.3 to 500 VDC Outputs
LF/LM/FM Series
M/MV Series
Isolated - Regulated
Programmable to HIGH POWER-Wide Input Range
6000 VDC Output 2 to 350 VDC Output
Wide Input Range HVP Series Isolated to 300 Watts
8 to 60 VDC Input in 1/2 and Full Brick
2 to 100 VDC Output LP/HP/XP Series
OR/IR/JR/KR Series
2-20 Watts

36 to 170 VDC Inputs

Terminal Strips - Thru Hole
3.3 to 48 VDC Outputs
Single and Dual Output
LV/HV Series

Also Ac-DC single and 3 Phase.

Power Factor Corrected. 3 Watts to
5 to 500 VDC Output
to 50 Watts
2000 Watts Models.
Wide Input Range
Electronics. Inc.
143 Sparks Ave., Pelham, NY 10803
Call Toll Free: Regulated
QP Series
Complete Listing of Entire Product Line
Fax: DC-DC Converters • AC-DC Inductors • Transformers
914-738-8225 and Inductors. E Mail:
What is
the missing

Industry guru Forrest M. Mims III has created a stumper. Can you
figure out what's missing? Go to to see
if you are correct and while you are there, sign up for our free full
color catalog. It's packed with components at prices below what
you are used to paying.