Sie sind auf Seite 1von 5

Proceedings of 34th International Electronic Manufacturing Technology Conference (IEMT2010)

RF shields that can be integrated with IC test handlers

Chin-leong Lim
Avago Technologies
Penang, Malaysia
Email: chin-leong.lim@avagotech.com

Abstract — This paper describes several radio PCBs utilized microstrip traces because their ‘semi-open’
frequency interference shields that have been developed structure enabled electrical contact with the DUT leads.
for integration with high-speed bulk-input turret-test IC Additionally, biasing and matching components could be
handlers. The shields were developed to mitigate soldered to the signal-side of the PCB. Unfortunately, the
interference to noise figure measurements of Low Noise
Amplifier components. Two categories of shielded
absence of shielding on the component-side of the
enclosures were evaluated for shielding effectiveness and microstrip testboard leads to a quasi-TEM mode which is
ease of incorporation into the existing machines and susceptible to radiated interferences [iii].
manufacturing processes. The first category enclosed the
handler’s working area in its entirety, while the second Commercial shielding solutions were shield rooms [iv]
one enclosed the testboard only. Variation in the and portable test enclosures [v]. However, commissioning
testboard shield design was required to suit different a shield-room can disrupt production for ~2 months for
collet trajectories between handler models. The shielding the construction of the room, the disconnection and
effectiveness (SE) was measured according to the MIL- reconnection of mains wiring, network connections,
STD-285 standard. The different designs exhibited SE in compressed air, lighting, ventilation and finally, to
the 16-49 dB range.
Index Terms — Shield, interference, noise figure, test perform an in-situ shielding effectiveness (SE)
handler. verification. Additionally, management frowned on
shield rooms because they hid disciplinary problems such
as sleeping and eating, and there was no flexibility to
I. INTRODUCTION reconfigure floor space.
Packaged semiconductors for Low Noise Amplifier
(LNA) applications have noise figure (NF)
specifications that must be guaranteed by production
testing. NF measurement in a production test facility
encountered radio frequency interferences (RFI)
which were traced to cellular phone use. The false
high NF readings at the device-under-test’s (DUT)
900 MHz test frequency caused good parts to be
rejected. Because customers demanded testing
replicate the end application (LNA for cellular
service), the test frequency could not be changed to
an interference-free channel. The yield loss was
estimated to cost USD 44k/month [i] for one package Fig. 1 A small copper-mesh shield room manufactured by
family (SOT-23/143). This paper describes handler- Lindgren, USA. Costing >USD30k, it could accommodate two
scale & testboard-scale shields which were IC handlers and their associated test racks.
developed for integration with “bulk-input, turret-
handling” [ii] type of IC test handlers. This type of A portable test enclosure like fig 2 is useful for shielding
handler suits the tiny package style typical of LNA the DUT and testboard from RFI during hand-testing.
components. But even the smallest model cannot fit inside the handler
small turret and its lid cannot be integrated with the
handler’s operation.
II. BACKGROUND
On a test handler, Printed Circuit Boards (PCB) or
“testboards” are used as interfaces between the test
equipment and the DUT. The PCBs are designed into
the isosceles trapezium shape so that 4-8 of them
could be arranged circularly inside the turret. Using
vacuum, collets carry the DUTs to different test sites
around the turret. Various Direct Current (DC) & RF
measurements could be performed by moving a DUT
sequentially through the different testboards. The

Xplore Compliant ©2010 IEEE


FINE MESH
SQUARE TUBE FRAME
PANELS

TWIN-DOOR FOR
SERVICING

249 mm
183 mm CONTROL
PANEL
(OUTSIDE)
TEST HANDLER
Fig. 2 Another commercially available solution is the
ENCLOSURE
portable test enclosure. Shown here is the Ramsey
Fig.
STE2200 with its lid partially open. This model which is
3 A handler-scale shield fabricated from fine Al mesh
the smallest in the series costs ~ USD400.
retrofitted over the SRM handler
Although testboard-scale shields have been described B. Testboard-Scale Shielding
by others [vi,vii], they are not commercially offered.
They are also not sized to be retrofitted to existing Due to the tight space inside the turret, this second
turret-style handlers. Additionally, their SE has not category of shields was designed to conform to the
been reported. testboard PCB shape. The enclosure height was dictated
by the need for sufficient separation from the microstrip
and other sensitive components such as air-core
III. RF SHIELD AND IC TEST HANDLER INTEGRATION inductors. The types A-C enclosures have lids on top to
allow servicing access to the enclosed testboards.
A. Handler-Scale Shielding However, the seam between the top lid and the bottom
cover will compromise shielding integrity. RF leakage
The working area above the handler’s main platform around the seams was minimized with a 4mm lid overlap
was enclosed with aluminum (Al) mesh that has been and gasketing with adhesive-backed copper (Cu) cloth-
traditionally used to fabricate Malaysian-style food tape (RS P/N 240-315). The same Cu cloth-tape was
cabinets. A square-tube steel frame provided used to form a 360o seal at where SMA jacks exited the
structure for the mesh and attachment points for the enclosure. Feedthrough capacitors of 3.3 nF value
door hinges. The open-bottom frame was attached by (Murata P/N DFT 304-803SS332) filtered all DC entry
screws to the handler’s metal platform. Twin doors at points.
the front facilitated handler servicing. BeCu finger-
stock gaskets (Instrument Specialties P/N 0078-0014- The testboard shields require an opening on the top
02) were affixed to the door-frame for RF-sealing of surface or the lid for the collet and DUT to enter.
the gaps. The door surfaces were left unpainted for Unfortunately the opening formed an aperture in the
low resistance when compressing the gaskets. Fig.3 enclosure which compromises SE.
shows the handler after being retrofitted with the
shield. Three shielded enclosure variants (henceforth, to be
referred to as types A-C) were designed to suit the
The main disadvantage of the handler-scale shielding different collet trajectories of two IC handler models.
is the profligate use of materials as a large volume Handlers manufactured by Ismeca (www.ismeca-
has to be enclosed. The second disadvantage relates semiconductor.com) first moved the collet horizontally
to many unfiltered points of entry into the shield as to the landing site before lowering it. So, a small opening
wires have to pass through holes on the platform for on the enclosure’s top surface was required for the collet to
control, power supplies and for connecting the test- bring the DUT to the testboard inside. The result is the
boards to external test equipment. type-A design shown in fig. 4 with an Ø5mm access hole
in the lid.

Conversely, handlers made by SRM


(www.simeca.com.my) simultaneously moved and
lowered the collet. As shown in fig. 5, the SRM collet
traced a curved path over the testboard as opposed to the
Ismeca’s up-down trajectory. Therefore, the part of the
lid that lay in the collet path has to be removed. This
resulted in the type-B shield shown in fig. 6 which
has 10mm wide “corridor” on its top surface. 96 mm
Unfortunately, this large aperture has a serious
repercussion on the enclosure’s SE.

AREA WITHOUT LID


FEEDTHRU
CAPACITORS ACCESS HOLE FOR
COLLET LID LID

SMA DUT LANDING


67 JACK AREA

LIDS (2 HALVES ) FEEDTHRU


TESTBOARD
Fig. 4 Testboard enclosure type-A96with an Ø5mm opening
CAPS

suitable for Ismeca handler’s straight up-down collet


trajectory. The enclosure closely followed the PCB COLLET CU CLOTH
outline. TRAVEL GASKET
Fig.
6 Testboard shield type-B shown with opened lids

COLLET A subsequent iteration, the type-C enclosure was


designed to solve the impaired shielding due to the large
DUT aperture on the type-B lid. The key difference from the
previous A-B types is that the type-C is an “active”
TESTBOARD enclosure. As shown in fig. 6, the type-C lids consist of
Fig. 5 Depiction of the curved collet trajectory of the SRM static and moveable portions. The sliding lid is attached
handler due to the simultaneous horizontal and vertical to a pneumatically-driven linear actuator. Initially, the
motions actuator fully retracts the sliding lid to create an opening
identical to the type-B’s corridor for the collet and DUT
to enter the enclosure. Once the collet and DUT have
landed on the testboard, the actuator shaft fully extends
to position the sliding lid over the opening. This
"trapdoor" mechanism of the type-C significantly
reduces the aperture size over the type-B.

LINEAR ACTUATOR
ACTUATOR SHAFT
SLIDING LID

FIXED
LIDS SHIELD
BOTTOM
PEDESTAL

HALF
AIR IN-
OUT
HOSES
Fi
g. 7 Testboard shield type-C with pneumatically-actuated
sliding lid to minimize the large opening on the enclosure Testboard shields’ SE, which varied significantly
lid for SRM-style collet entry. The sliding lid is shown in between types A-C, was primarily dictated by the
the closed position. aperture required for collet and DUT entry. Predictably,
type-A demonstrated the best SE due to the small collet
IV. TEST METHOD opening and type-B the worst because of the large
opening required for the SRM collet. The type B’s poor
The MIL-STD-285 plane-wave SE test [viii] specified SE was significantly improved (~20 dB) by incorporating
that the receiving aerial and detector (measuring a sliding lid in the subsequent iteration, the type C.
receiver) be contained inside the evaluated
enclosure. As the space inside the described shield The type-A testboard shield has the best cost-benefit
enclosures was insufficient to accommodate a ratio among the 4 shield types evaluated. Its comparative
receiving aerial, a modified setup was used where a cost to commercial products is about 1/100 of a shield
PCB containing a microstrip trace, TL (L = ~7cm, Z0 room and 1/3 of a small sized enclosure. However, its
= ~50Ω,) served as the aerial equivalence. One end advantages are partially offset by its being compatible
of TL was terminated with a 0805-size 50Ω resistor with only one make of IC handler.
and the other end soldered to an end-launch SMA
jack receptacle (Johnson P/N: 142-0701-831). TABLE I
Knowledge of TL’s equivalent antenna factor (AF) is Shield type SE* (dB) Cost
not required as SE is a relative measurement. The (USD)
aerial replacement was connected directly to the Handler-scale (fig. 2) 32.6 500
measuring receiver via SMA adapters without any Testboard-scale type-A 49.2 150
coax cable to minimize unintended coupling. A Testboard-scale type-B 16.0 150
single 900 MHz test frequency was used to replicate Testboard-scale type-C 36.9 250
the end application. * 900 MHz.

ATTENUATOR
ANT VI. CONCLUSION
V
One handler-scale and three testboard-scale RF shields
SOURCE DETECTOR
ANT
1' 1'
have been investigated for reducing interference to LNA
SHIELD
NF measurement in volume manufacturing. They were
developed in response to the lack of commercially-
available solutions that could be retrofitted to existing
ATTENUATOR handlers of the bulk-input, turret-test family. They also
TL V
have cost and space-saving advantages over commercial
SOURCE DETECTOR
enclosures. Although their SE performances are modest
ANT 50Ω
SMA
compared to commercial enclosures, they may be
SHIELD JACK adequate for low severity interferences.
Fig. 8 (top) MIL-STD-285 SE measurement setup and All three testboard-scale shields significantly obscured
(bottom) modified setup to test very small enclosures view of the enclosed testboard. The reduced visibility
hampered maintenance work, especially the alignment
between collet and testboard. So, the type-A testboard
V. RESULTS
shield, despite its high SE, was unpopular with the
The measured SE for all evaluated shields fell short maintenance crew. In the end, only the handler-scale
of the theoretical prediction. Designing these shields shield was grudgingly accepted.
around the IC handlers imposed many difficulties in
term of unavoidable apertures required for DUT
entry. Additionally, design of the handler-scale VII. ACKNOWLEDGEMENT
shield is hampered by unfiltered wires entering the The author thanks S. M. Tan for fabricating the
enclosure through the handler’s platform. As a result, testboard-scale shields.
both handler and testboard-scale shield fell short of
the +90dB performance that commercial enclosures
are capable of.

For the handler-scale shield, reflection loss from the


mesh surfaces was not optimal due to the poor
conductivity of oxidized aluminum, Al2O3. The
unfiltered control wirings entering the enclosure
from the bottom were probably degrading the SE as
well.
I REFERENCES
[] K. O. Yap, "Electromagnetic Disturbance Analysis in End Of Line", Hewlett Packard (M) Sdn. Bhd., Penang,
Malaysia, internal report, 1999.
ii[] J-P Lanteri, C. Jones, and J. R. Mahon, “High Volume Microwave Test” in The RF and Microwave Handbook, M.
Golio, Ed.
Boca Raton: CRC Press LLC, 2001, ch. 4.8
iii[] L. G. Maloratsky, “Reviewing The Basics Of Microstrip Lines,” Microwave & RF, Mar. 2000
iv[] Lindgren product specification, “EMC Industrial Shielding,” [Online] Available: http://www.ets-
lindgren.com/page/?i=EMCIndustrialShielding
v[] Ramsey product specification, “STE series RF Test Enclosures,” [Online] Available: http://www.ramseytest.com/
vi[] G. S. Hill, “Method and apparatus for
electromagnetic interference shielding in an
automated test system,” US Patent 7218095, May 15, 2007.
vii[] C. P. Weinraub, “Shield for tester load board,” US Patent 7259549, August 21, 2007
viii[] Military Standard Attenuation Measurements for Enclosures, Electromagnetic Shielding for Electronic Test
Purposes, Method Of, United States MIL-STD-285, 1956.

Das könnte Ihnen auch gefallen