Sie sind auf Seite 1von 78


. capacitor , to give a more level DC voltage' This valuc is , however' rather

point so regulation
depcndent on the current drawn from the circuits at this
I '

]'FIIIORY AND DESIC}N pertbrmed to maintain a constant D('voltage '

Purposc of thc power suppt_v is to convert a mains suppty (240V AC 50 Hz in

Singapore) to lower DO voltage or vohages. Simpty, this invofues Four Basic 2.1 The'Iransformer
on the
steps (l ) transforming to a lower voltage (2) rectilying (3) smoothing (4) A transformer simply converls one AC voltage to anofter depending
winding' A
regulating. ratio of the number of tums on the primary and the secondary

Figure 2. I is a block diagram ofa power supply , showing how hansformer with specification 48VA 0-18 , 0-18 @2A has two secondary
these steps

relate and waveforms at relevant points- windingB cach of which can suppty 2'0A' On the other hand a 48VA

I 8-0- 18@2A specifics a transformer with a 36V center tapped socondary w'

capacity of 2A. This second example has effectively two

l8V winding whic


specilied in rms.

to suit the
Thc author has decided to use 18-0-18 V @ 2A tansformor

requirernents of the specifications' Thc choice was

abo donc basc on thc fac

that my tcam matcs Mr. Tham Kin Mrur requircs l.l85A (maximurn) of pe
has a valuc of
at various stages curreirt. The peak voltage of thc choscn Fansformer
has a vohage of 24'8 V suitable fot
Input power is from a mains source. A kanrf,ormer changes lhe 240V supply to JT} = 25.2V which after rectification
usod in the dcsign oftre powcr
direct power inpu8 to opcrational amphfiere
a lower voltage, still of an AC natrue ; the difference is shown in the warrcforms
in Fig.l. After rectifcation the waveform appcanr as a scries ofhunps , each
2.2 The Rectifier
hump corresponds to half cycle of the original AC vottagc. Rcctif^.tion is
Examplcs of rectifiers and corresponding
output waveforms arc shown in
usualty performed with diodes. Although the wavcform is now of a DC voltagc

Figrne 2.2-
, it is too irregular to be uscfirl and so must be smoothed, gcncrally using a
'l'he full wave bridgc rectifier configuration was being used in the power s

--\,, l J-- )l---o+w design. Single mit diodes was u.sed inslcad ol'iliscrete ones as it is availabl,
=l| lhc lahonlory
' t o -VE \

a) nr\Lh
2.3 Smoothins Svstem
*1,rrrf--+f-l 'fhc smoothir^g system consists of a large capacitor C just after the

SlllFr l-.'*,.
flrll wa,

Tltg I L-ft____-_,, I
bridge rectificr as shown on Figure 2.3 below. This capacitor mainly serves

filler out the input signal ripple to a smaller value.

(u) pur.r-wavr

ftp* (r) BR.IDqE

Ftp. 2.2 Rectifier confimrations

The simplest possible rectifier is a single diode , shown nFig.2.2t. This mercty
p,rsvents every altcrnative halfcycle from passing and accordingty gives the Fieure 2.3 Full wave loctification circuits

output waveform shown. It is known as half wave rectification and obiourly

Considering only the positive rect'rfied in wave form +Vi, the grapbo of the
requires a coruiderable amount of smoothing so ir not gcncrally rued. Figurr
output potential difference +V., the ouput cuflsnt, the rectifed currqrt ant

2.2b showe a frrll wrve rectificr which obviatos the problon, allowing curreirt
capacitor current for thc above circuit is drawn as follows in Fig, 2.4 in the

to pass on wery half cycle and thercby giving an ouput which ir more easily

smoothcd. Disadvantags of such a co,nfiguration is lhat it requircs a taosformcr By inspecting Fig. 2.4, the following approximation can be made:
with a center tap,ped secondary and was rcalty populr in thc days wh€n rcctifier
Vi,- V* - 1.4 ( due to diodcs vohagc &op )
diodes were expensive. Figure 2.2c , on the other har4 of
showB the typc Vh Y.*-v12 (l)
rectifer used in moet applications today.
where V,* is the mo<imum value of V- and v. is 0re peak-te'peak ripdo p
constant then we rnay assume that the capacitor discharge current i. is consi

and tlurt the charge transl-erred to the load each cycle ls given by.

Q -'ir" x t,

F'or a capacitor we have

where V is the change in potential caused by the charge Q. Then V = v,

Wemaytherelbrewrite Cx1 =i*xt,

v, - (i& x t/C (2)

Assumrng that td is much greater than t" ( See Fig. 2.4 ) thcrl

td=T/2= ll(zyl)

where f is the period and fis the frequency ofthe supply.

Substituting t. into equation (2),

1- i*/2xhC) (3)

C = id.i(2lv,)
l,tr). iltt)-.i ,tt)
For 50 Hz supply, C-- i&( l00v,) (4)


A 6800trF capacitor was used in the implementation of the actual circuit.

According to equation (3) , thcre exist a ripple,

1 = l.l18(2x50x6800lrF)
Fiq 2.{. Graphs of currents and potsntials in the b'ridgc rectificr circuit = 1.64V peak-to-peak

Note that it was assumed that i& = l I 184 the peak current drawn by the power

If we considsr a load of R" at the output, then thc timc oonstant of oapaoitor amplifi€r. The ripple voltage into the regulator would have a value of 1.64V if a

dischargc/charge is R"C. Thc capacitor is fimctioning as a chagc storage

capacitor value provided by thc laboratory of 6800uF was used. This ripplc was

device. If the capacitor dischargc timc t, is coruiderably lors than thc time corsider quite small and almost small as compared to J2 x 18 = 25.2V fsr
supply input into operational amplifiers. Thrx the operational amplitiers will limctiori By definition ,

(no load voltage) - (full load voltage)

normally. rcgulation
firll locd voltage

2..1 'lhe Reeulator Systern

2.5 (lurrent Limitine (Overload) Device
An ideal regulated power supply is an electronic circuit designed to provide a 'fhe current lirniting systen
was implemented using a technique called fold
current limiting. This is shown in the diagram bclow.
predetermined dc output voltage Vo which is independent of the load current I,
drawn from the output vohage Vo , ofthe temperahrre , and also ofany

variations in the ac line voltage. Therefore , a regulator system is require to

maintain the output voltage as constant as possible. In this project , the op-amp

series voltage regulator was used as shown below.

Fig. 2.5 Fold-back c:rrent limiting s-vstem

Fig. 2.4 op-amp series negative feadbackvottage r€gulator

Under normal operating condition V* . V", . A ruddcn dccreare in load I

causes I" to increase. Vr. becomes larger than V"" , q starb to conduct. Its
A zener diode was used as a reference voltage to fix ono input tsrminals ofthe
collector current df,aws cwrent away form the baee of pass tansistor Q'
op-amp which will be used by the opamp to ampliS thc voltagc variations
causurg \ to drop.
found in thc output. This will causc the series regulator Qr to adjust its VcE so
Since I : I + lR :I + I", both I and I" drop but keeping q conducting As
that the output voltagc to rernain as comtant as possible.
result , output voltage VL decreases. If R" deoreases firther, Q, causes I" a

The ability of a power supply to maintain its output voltage .'hen the current V, to decrease firrther.

takeir from it is known as its regulation.

A disa&antage of this fold-back is that thc regulation of ttre circuil concernbd

win be slightty poorer than normal short-circuit protection dwice . However, if

the output is suddenly short-circuited , the currort wifl be limited to a value

much smaller than the original maximum current as depicted in the diagram


o3 N)q$ + hltrink

Fig.2.6 Dual I'rackine Rsgulator

the dual racking circuit is realized by thc opamp OP2 and tansistor q. 'I

two IOk resiston are to be closely matched. thre to tho opamp, the two lC

resistors are achrally connected in series but theirjunction is at virtual grour.

The positive rcgulatcd output is used as the reference for thc negatiw supp'
-- Fig.2.5a short-circuit current using fold-back
The lower error arnplificr OP, contols the ncgative ouput by comparing th

2.6 Tracking Output average of the two output vohages with ground thercby gMng cqual positir

It was dccidcd to use dual-hachng feature to implernent the positive and and negative regulated outputs. The enor op-amp ir acnully uscd as an

negative supply voltage required by thc specificationr. This feature was obtained inverting amplifier with a pin of l. Negative feedback helpr to maintain a

from the reference "The Art ofElechonics', pp 204, and is sho$rn in the circuit constant output voltage as the load current varics. For thc negatiw output (

in Fig. 2.6. must sink the load curent and so a PNP pass transistor is callcd for.

2.7 Choice of Comnonents

The whole oower supoly systcrn was drawn with all the required featues in

shown on theAppgdlgA.

'!h. roroting m th. co@pomtu ch*n lrr.d or th.
ctv.n lqCsl@bs'tE d!darqdlrod!!-9t!
$.dlo{d: vE=vD tl Iw
vr =t*&'& 3 LR' -nq
@DgL 6.2v Ma. z€ncrdodc = 1t9mA PmRrlins- lW
OwdcrC/.frna Mo&l:Ll35l Rrg.ofsrplyirytr = +5 to Bls I=V!/&+&) Edvz=Vr rI.L,sV!+L&
+"12\/ rhqefo'! vk =L& -0"+L&)Rr'(&+Rt
rteatbr oL rplzo codrw [* ' v* = v*
ttoitcl : 5A colot ( cwa = "'
corri@.laic! dirip.&n = 65w => ve = L",R, -(v" r l-"&)R1(R, +Rt
lEEEr.iz; Modd: TIPr25 coltiNucolLcta(ctlEr= 5A d& -0 => v!=ov
Cotiddr &ric. dlipolicn = 6rw
Ik=ryu/RJ( r+&/& )
lllrbEla} rlbd.t : 2ll2n 2 cdtiN.r', rlviE di.iFri6 = 0.5w
CodiM.olLcro. curld =0.EA .. (O.f6tr.2t(t +2.11a7, =0.6?A (n rkiqr.d
Rcsisror.t current)

ChoosinsR,andR^ andR-
Heot Sinh
The short-circuit curont Lc
we note that the power raruiston can dissipate many watts. To ensrne that tl
was sct to l/3 of thc nwdmum cwrent
narsistorjunction temperature is kept below the ma,ximum speoifcd operatin,
whorc lrr^* was choeon to be 1.5 A. temp€rature , which is I 50. c for the transistors in plastic packages tike TIPI

and TIP125 , heat sinling was do4re. It is wise to be corseratiw in heat sinkin

At point 12 , Vr, = lE V design since transistor hfe drops rapidty at opqating tctnpcraturcs nc:lr
d abr

Atpointll, V'= V,r*Vr,, = l8 +0.7V = 18.7V the maximum specified. Hence a large piece of heat sink was choscn
for the

p. Fansiston Qt md Qr.
Now, 211 = lr'
nfft; "
Choosc R, = a7 kQ and & = 1.2 O

A, Vs = l8 + 1.5 x 1.2 = 19.8 V

then, for Irr r,=1.5

giving R, =2.76Kd) chooee R, = 2.7 KC)

l: IF
tll A
POWER AMPLIFIER DESIGN . . Max. output current -- 1. 12 A

ln iuldition, thc closed-loop gain rvill bc r-ct ilt -5. Tho gain of'5 is set low, so that thr)
The classification of a porver ar-nplilrcr into class A, B or AB depends on the

biasing current. The class A design is the least cfllcicnt while the class B designs power unplifiel will not tunplify th.' noisc sigJurls tratlsnrilte(l liont the pre-amplifier

suffers from cross-over distortion. The cllss Ali tlesign rvhich is the most efficient
' r:
design is chosen due to its lower heat dissipation, rvhich enablcs it to operate for a :

longer period of time without sophisticated cooling devices. 'Ihe power amplifier

circuit used in the project is shorvn in Fig. Ll iris circuit was obtain by integrating

the circuits obtained from reference chaptcr l7 (pg 6la) The power anrplitier dcsig:l:d its sll()rvll in l;ig I on the neK page consists ol'
[], and refdrence [2]
tlree tJistinct stages, namely the input . inlcrrrrudiide and output stage. -{.C. and D.C.
chapter 5 (pg 165) The specifications givcn u,crc as follow's .-
li:cdbark achievcd through t{5 and lt(r, rvlrs crrrl)l()\'ed to obtain a pre-determined gatn

the inpur stage r;onsists ol-difltrurrtinl prir Ql iurtl Q2 with biasing pedbrmed
Given Specifications :

l. To operate with +/- 15 V porver supply hv a cum:nt ;ourcc. lhc currsnt gaifl al this staeo is rclativcly small. Thc dilfcrcntial

provides a h:gdr urput resisbnce which is desirablc.

2. A lower 3-dB point of 65 Hz (pre-amplificr to set this point)
'lhe second stage consist ot'a single contnlon-emitter amplificr,
Q3 which
3. A upper 3-dB point of 1 5 kHz (power amplifier to set this point)

4. A 5W output iurplilies the input currerrt and provirlos lnuch ofthe open-loop gain. fhis transistor
inoreiLses the input resistance lillher. ,\ capacitor rs corlnect *rott tttt U*a ;J
5. A 8C) load
eollector ofQ3 to provide lrequency colnpensaflon.
'fhe output stage consists ot'cornplcmcntary parr Q5 md Q6, a V6" multiplicr
provide the
From the given specifications, the required output current and voltage were -nd a short-circuit protection. The purposo ofthe output stage is to

obtained as lollows :
anpliticr rvith a low output resistance. [n addition, the output stage r-hould be able to
supply the requirerJ load rurrent tvlthout disstpating ur unduly large rtmotnt,)l power.

The short circuit protection provents the current tiorn risrng above a predetermined
V2.r, / R: power
value, when the ()utput terminals afe shorlcd. lt aiso selnds t() prot€ct loads wttlt
impcdance lcss than lhc ratcd 8Q' iiom cxcessivc hisjr currellts'
= 612V
ll.msrsrors t]P1l0. fl1rll-i. i\( li, l ,urtl ll('l-11 ,.verc ttsctl.'lteirtechnrcal
Ma.ximum output voltage = 8.94V

I sDccrficatiuns are put ilt .\ppendix lJ.

=\/ 'm3x'/t2
'max "
+ 15V
:r) Current Source

I of
The differential pair Q I arrci Q2 was choscn to be biased with a current sourc

2mA and the V5" multiplicr stage $as choscn to be biased at 10mA' The

RC2 current source arrangement is shown in figLire 2

+ L5V



R3 C1
Fig.2 Current Source

Ia = 0'7lRa
I-Iu- 677Pu

. Ra:0.7/I
= 13 6

Fig.f Power Amplifier Circuit

.'. RS3 : 1.5 kO, RS4 : 2.7 kQ
Consider biasing at quiescent condition, i.e. input voltage
:0 V

The output voltage must also be 0V

let I.5 = 12 mA
c) DC Analvsis
'" Ic4 = Ic5 = 12 ntA

assume P: = 06 = 1000
The circuit for the dc analysis is shown in Fig. 4
Ibs : IbO= 0.012 / 1000

= 12VA

current into V6. multiplier = 10 29 mA - l2ptA

= i0.28 mA

A current of 0.5 mA was chosen to pass through Rpr with V66 = 2V

Rpz = 20.@00s

= 40000

Rp: was chosen to be a 5 k0 resistor in series with a I 2 kC) so that the multiPlier

can be more flexible in the biasing.

R8 and R9 were chosen to be I c), 3 watt resistors. The value of lc) is smal

so that it will not reduce the output voltage swing. At the same time, it is large

enough to provide negative feedback to transistors Q5 and Q6 as well as to reduce

the thermal runaway. The power rating is 3 Watts as output current reaches a

maximum of l.2d which would dissipate 0.72 Watts of power in the resistor'

The voltage swing across Q6 was limited to about l3V,

.'. voltase across R5 = 2 V

: 194 Cl

Fig. 4 DC Analysis of Power Amplifier Circuit

As can be seen from the circuit, the inputs to the bases of Q1 and Q2 have the

same value, V1 , since the voltage drop across the emitter and base of the transistor is

small. That is, the current input is negligibly small. Therefore, the gain of the

amplifier is governed by the node equation at the feedback loop of the circuit :

V1 : {Vo/(R5+R6)}xR5
Vo/V1 :(R5+R6)iR5
for a gain of 5,

choose R5 : 10 kQ, R6 :40 kQ

e) Inout and OutPut Resistances

The input resistance is approximately equai to R1 in parallei with the input

resistance of differential ampiifier. For values ofRl that are very much smaller than

the input resistance of the differential amplifier, the input resistance of the power

amplifier will thus be equal to Rl.

To obtain a large input resistance, Rl is chosen to be 60 kQ.

Since 60 kC) << 2(1 + B)(r. + 5 kQ), input resistance = 60 kQ-

Output resistance consist of the two parallel I f) resistors in parallel with the

resistance of rest of the circuit elements, which is much higher than 1 Q,

.'. output resistance of circuit = 0.5Q.


The values of the resistors and capacitors used as well as the type of

transistor is summarized in the table 1 below :


In this project, the main challengc in designing thc bridgc amplifier was thc J,.,
construction of two fixed gain power amplificrs as shown in Figurc 2(a) and 2(b). R8

Howcver, as wc can scc, the two circuits were idcntical to each other cxccpt for the
feedback network topologics applied. Hcncc, the author was requircd to dcsign onc

of thc power amplificn and thcn transform thc dcsign to obtain anothcr power Rl0
amplifio, by changing the fecdback topology. ln this casc, thc non-invcrting amplifier
was designed and would be explained in greater dctail throughout thc rcport.

Figure 3 : Basic Dcsign of A Non-invcrting Fixerl Cain Amplifier
Prellmlnary Circuit Descriotions
The preliminary design of a fixed gain non-invcrting amplifier is illusratcd in Figure 3.
Thc sccond $tagc compriscd transistor eg as a common-cmittcr amplifio with
The amplifier circuit shown above consisted of threc distinct stagcs : an input i

differential stage, an intermediate single-cndcd high gain stagc and a high-powcr , rclativcly low rcsistancc in its cmitter (R7) comparcd to its high collector resistancc

output stage. Feedback network was employed to obtain a prcdctcrmined fixed gain,
t This cnsured that the common-emittcr runplificr provides considerable high voltagr

g1\A capacitor was conncctcd across the transistor's basc and collcctor terminal ir

to providc llcqucncy compcnsation to the circuit.

The input stage employed a differential pair amplificr consistcd of transistors Q5 &Q6 ,ordcr
in order to provide largc input rcsistancc. The differential pair was biased by a
The purposc of thc output stage was to provide thc amplifier with a low outpu
constant current source made up of Ql and Q2. A 20O trimmer Tl was connected
resistancc. Sincc thc output stage would be dcaling with largc signal, it was designct.
between the emitters of differential pair and the constant current source so as to
to be ablc to supply rclative largc current to the load without dissipating an undull
provide flexibility in matching the differential pair kansistors for reducing offset
large amount of power. The output stage in this design employed a Class AE
amplifier. lt consistcd of two Darlington pair powcr transistors e9 and elO whicir
wcre biased by the Vs1, multiplier with small but non-zcro current in order tr
eliminate the crossover distortion. A constant current source was employed to [a] Constant Currcnt Sourcc

provide sufficient base current drive to the power hansistors as well as to bias the Vrt I.., was dcsigncd to providc dc biasing curront to thc diffbrential pair transistors and

multiplier. Q7 would be mounted on tle same heat sink as the power transistors as to convcntionally is small. Hcncc, l.r: 2mA.

provide thermal stabilization.

llowcvcr, in ordcr to providc sufficicnt currcnt drivc to the basc ofe9, I,.o has to bc

For the purposc of reducing noise level to minimal in the power amplifier, transistors iargc cnough. Thcreforc, under maximum output current condition with I., = lA, it

for low-noisc audio amplifier application were prefcrred. Transistors of the constant rcquircd a basc currcnt drivc of magnitudc lAlp = lmA, where to be the
B was taken
minimum cuncnt gain of Q9.
cunent source, Ql to Q4, were choscn to be from BC559 type, whereas the
differential pair employed BC179 type transistors. In order to provide thermal
compensation, Q7 used transistor of type BDl35 while the second stage transistor Q8
Following thc rulc of thumb that : 5 lBg < l(..r < l0lB,)
'fhcrctbrc, choosc I,.o = ,5mA
was determined to be of BCl09 type. The output push-pull complementary transistors
were chosen from high power TlP-series where Q9 used TIP120 and QlO used

TIPI25, These two power transistors have to be mounted on heat sinks to enhance Thc circuit ofa constant currcnt sourcc is illustratcd in Fieure 4 below :

their heat dissipation.

The specifications ofthe tansistors used above are included in Appendix B.

Clrcuit Analysis
( NOTE : All the analysis in this part is referred to Figure 3, unless otherwise stated)

From Chapter 2, it has been determined that maximum peak output currenl Io = lA.
Figure 4 : Circuit of A Constant Currcnt Source
Hence, with an effective load of 2C!, the required peak output voltage, Vo' = 2y.

For current source consisted of Q I & Q2 : [b] DC Biasinq
Biasing at quiescent condilion was considered in this analysis, i.e. input voltage w:
1., = o.6V/R2

Hence, R2=0.6V/2mA=300O set to 0V and the desired output voltage was also 0V.

.'. Choose R2 = 3300

Neglecting Ise, as at quiescent condition it was in tens of pA, almost all of the currer

supplied by the constant current source (=5nL{) would be driven into the V6
Set the biasing current for Ql :1mA,
multiplier. Hence, by allowing a current of 0.5mA to pass through the trimmer T:
Rt _ = 3.3kcl
lmA with biasing voltage Vs, = 2.4V (Note that V6s at least must be 2V),
.'. Choose R2 = 3k() *T2= =+.tto .'. Choose T2 = 5kO

By choosing T2 = 5kO fimmer, it enabled the VBE multiplier to be more flexible i

Similarly, for current source consisted ofQ3 & Q4 :

biasing thc two output power transistors (Ql & Q2). In practice, the trimmer woul
I.o = 0.6V/R4 be adjustcd until thc output transistors of thc Class AB amplifier were biasod by
Hence, R4 = 0.6V/5mA = 120O small but non-zero currcnt at ouiescent condition.
.'. Choose R2 = l20O

R8 and R9 were chosen to be two 0.250 5W power resistors due to large curent an

Set the biasing cunent for Q3 : lmA, large heat dissipation in the output stage. They were usually very small and wer

_ (4.5 - 0.6- 0.6)V = 3.3k0 includcd to compensate for possible mismatches between Q9 and QlO and to guar
lmA of thermal runawa! due to the
against the possibility temperature differenct
.'. Choose R2 = 3kO
bctwecn the input and output stage transistors. Noting that an increase in current t

Q9 causes an increase in the voltage drop across R8 and this leads to a correspondin

drop in Vrro thus stabilizing the current. Therefore, the power resistor providt

negative feedback that helps stabilizing the current at the output stage as well as th

load currcnt.

Now consider transistor Q8 and by negle cting Is ,., lc6 : 5mA at quiesccnt.
Howcver, by applying fccdback to thc circuit, the output re$istance is reduced by a
In order to cater for largcr voltage swing at collector of e8 and providc large voltage
factor of I +pA, i.c. the desensitivity factor of the feedback network, where A can be
gain' R7 has to be as small as possible and thus it was choscn to be 22ct rcsistor.
approximatcd to bc 500 and B = RBr / (RBr+Rlr) = 0.2432.
Thereforc, voltage at base ofe8 was givcn by,
R"*o o'25
+pA- =2.o4m}
Vnr=Vun+Icr.R7+Vrr, Thcrcforc,
I l +(0.2432)s00
-0.6+0. ll -4.5

= -3.79Y. Hcncc, \n1, = 2.04mO

Hence, to match the biasing voltage at the emitters of thc differential pair transistors,

by assuming that they wcre perfcctly match, Having complctcd thc dcsign proccdure, thc initial circuit design ofthe non-inverting

Vn: = Vss powcr amplificr obtaincd is shown in Figure 5 bclow.

+ C3.79) = C4.5) + (lxl0'r).R5

.. R5 = 7l0Q .'. Choose R5 = R6 = 750C)
3K 3K

Input & Output Resistances

5K BDT35
The input resistance of the power amplificr is contributed mostly by Rl} in parallel lmA
with the resistance of the whole circuit which is very large compared to Rl0.
vtiu 56K vo
Therefore, R,* = lskf!

On the other hand, the open-loop output resistance, Rnorn is givcn by a 0.25O

power resistors in parallel with the output resistance looking into the emitter of thc
power transistor. Since the output resistance looking into the emitter of the powcr

transistor is much larger than 0,250, effectivcly, Ropen = 0.25C1. Figure 5 : tnitial Dcsign of The Powcr Amplifier (With Component Valucs)

Pspice Simulation CHAPTER 4
Thc circuit illustrated in Figurc 5 was simulated using Pspice program. Thc hard copy
ofthe frequcncy responsc ofthc circuit is attached in Appcndix C.

From the frequency rcsponse characteristic, the output voltagc obtained is 1.978V, in Noise Elimination and Frequency Compensation

othcr words, thc dcsign has a gain of, Thc circuit shown in Figurc 5 was constructcd and tcstcd on the brcadboard. Thc

circuit was found to havc noisc and oscillation. and hencc certain modifications wcrc
Gain- =1978=3q50
V.u 0.5 nccdcd to solvc thc oroblcm

It is apparcnt that the dcsign is applicablc as thc gain simulatcd is vcry closc to its Thc fbllowing approachcs wcrc takcn in ordcr to climinatc thc noise in the circuit :

dcsigned value. The dcviation might bc duc to thc tolcranccs ofcircuit componcnts as (i). A 22pF capacitor (CF) was conncctcd in parallcl with thc large-value fcedback

wcll as thc assumptions made whilc calculating thc component valucs. Bcsides, it can rcsistor RF. Thc ampli{icror thc fecdback network was modificd so that a zcro

be seen that thc powcr amplificr has a fairly flat rcsponsc bctwccn 0Hz-300kHz; wirs addcd to thc transfcr function, thcrcby improvcd the phasc margin

howcveq a peaking effcct occurs at about | .26MHz. This phcnomcnon can bc solvcd Practically, this configuration will filtcr out most of thc noise produced by thc
by introducing polcs at high frequencies and the proccss will be discusscd in Chaptcr circuit at high frcqucncy. And, this mcthod is known as "lcad compensation".

4 (although it does not fall into the pass-band of the prc-amplifier). (ii). Two 0.47prF capacitor (C I & C2) wcrc conncctcd in betwccn thc power supplics

and ground in ordcr to climinatc thc noise causcd by thc power supplies. Thcir
Through the simulation progriun, thc following mcasurcmcnts can be obtaincd : valucs wcrc choscn to bc large cnough so as to have better effccts in

Input Rcsistancc, R,n = l4.92kO smoothcning thc supply voltagcs.

Output Resistancc, R,rr,,. = 2.l68mQ (iii) Aftcr applying (i) and (ii), it was found that thc output voltage of the amplificr

It is obscrved that the simulatcd results are quite closc to that wc have derived in thc oscillatcs at ccrtain high frcqucncy. Hcnce, a l50pF capacitor was connccted to

previous section. thc basc tcrminals of the input diflbrcntial pair transistors (Ql & Q2). This is

known as "polc-zcro compcnsation".

Breadboard Testing Notc that thc adding of largc capacitors will introduce polcs at high frequencics and

After eliminating thc noise, the ficqucncy responsc of thc final circuit was invcstigatcd thc cffcct ofthc compcnsations can bc obscrved tiom thc Pspice simulated frequcncy

using oscilloscope. It was obscrved that the powcr amplifier performing pcrfectly with rcsponsc curvc, which was includcd in Appcndix C. Also noting that the peaking
an input voltagc of 0.5V. Thc lollowing parametcrs wcre obtaincd in thc cxpcriment : ct'fcct occurs at l.26MHz fbr thc initial dcsign has bccn rcrnoved by introducing poles

to thc circuit at high frcqucncics.

Midband Cain = 4.0

Lower 3-dB Frcqucncy = No

Uppcr 3-dB Frcqucncy : I 55kHz ( thc bandwidth )

Thc modificd circuit of thc powcr amplificr is shown in Figurc 6.
lnput Rcsistancc = l-5.2kO

Input and Outout Rcsistancc Mcasurcmcnt

Input Resistance, R,*

Figurc 7 shows the circuit to determine input rcsistancc of an amplificr.

lmA CF

I s'''e o.2s 2.0

Figure 7 : Circuit to determine input rcsistancc of an amplifior

Procedure to determinc input resistancc.
F'igure 6 : Modific<t Dcsign of Thc Powcr Amplificr
(l) a potentiometcr R was connected in scrics with thc input source.

(2) R was set to zero, with output voltagc noted.

(3) R was then varied until output falls exactly to half its original valuc

(4) R was then removed and measurcd; R = \n

F-ouue f S,,PP!

rrV 4 co4

t\2 v
-T aado.rr.F
: I

I/v'4oc+ I

1A/z+7 :N,I(lc4
t /^F
I !i bv

y' i. + iBV
I,eo 1
:5 vA

I ail 37eb



<. -1.r'
'1 N +cct+ | .c
'l-t 220 c j^T TrPil S t,tlL
.J_ :t_
P.,c er 4r,^plif;< r^



e,r" &



2z o-S-
<t? tz5


- l\ ,
1t \l
'l' /;
+- AA.\ -+- -
Lr J

lcJ l-
1t'* |]
pL l)ov w
S a\*6'

-t-- \r^r'. la^ ^<L-c!-\*
+J( ' 't'u'{
r ay\a+c't clt"
--{/,n q-r
a- I I
r!4< rok t

t-- f
\ A
Itbro T\ +

i i rl

i ,/

_J_ - _-1+ -:, i

iiJ- \l
l-----l I

\- I



\ -/\- I

l -_ - -_z ."_ l__--




n\f - T
tJL t
Q-. Lvr
.?. ,t
-kD''"\ I

ts\- -
J q"'Y -(,e t)*r
(_*U ft-\oo oM

^{J (o"D / ,4{

trnd' 7r-h- lvt\.
, t--\- --)#-*
tfl-lV t sl -l+\-/ ,/'l
h\ld' Yv')- <ll-| I i
U \) f^J,-
>l) t) I


\ru Yl-- ( ;:/- z .^u'u ) \1Y,..-

\t p\rr

(r bp,
v-o \ . ""'^ h-
q a .i^x-<

JDD J, Ii-_ ('-
rvt -l
\ *"- dF't ,
--Oi] 6Jt'/
r ' I r^t- /L-"-1, \ \

' /47u\
q, g- 1oo\
\-r' /
)l"v -- ?'J1_r_",'n'
s",'f,- \) - lfv
, *1"'

'tPrC- I


{Yvro N
MV 'vtst
- R- ;--:1, -l\--- \
+ t+

/' ?L-
-r -)u r) J
l/ut t.'*)o r
lL n1t
<'( \- l^'UCVV) t tr . t3, ht)
I r-L49 1
r -- l- l/o

\Z J2 ) -=-
C^e-t vb
, rc T

"1vu1- *7r" \ lr^5 A Z,n )

(,;A;-- n^: L;"li ( -^^r )

u l( ( l* e)*t-t )
"-, 7o ur )
( . rt? :F
l*1.. *,*_)
SI'.^4- c<*

Y'*"b z> rP,,-e-c.

y* ("-€{. s,vl, Rq J./L


.. I

),, n^ -i
r\l A

---+-*i--f ,l
,-- -*tB
i f--
\,1 \
' ,J.1'I\
*----1,'t t 1- i tl L
-) I

i-,' #\\

fit I 1

l (/

zll t

f1^J'ft ui ran<.r l* fc.*r-, .*

\,'..* .:
I t-n
<,--{ lRe,
l-- I

\"' I
?ls 7

> I l_la
'l > |"1/

6[,'5 c, ol"V
,t wtlaf sq'.l -,V 4 ; \(2n,
c' cW p-tr.t o,f f>a,i"4tsion .-7-r,,as I' +-1" r"^ll t/lrty1 C"v, lUrl
i[ Vge a'-4.^rr' t At;>f" f
- I-p Po^ (

^4 Ror, T I *k
| --"7( Ror I V",tt $
h---{ U a;
.,r -
D ^t
V rt' p"r-,- 4+
^ "1|,
\f z, =(p*,{ a.%) .'tt
rL-Jtll----v tt

I /- --'
\,o., G + t,

tiw t
l,*y" L.( & V,,,

^1-' t'v
\-- af

t f4r--- vwr
\",r R,^ = R"// re

+fl { 7*-" v,,

\^'.2 5,tt^.t tiv

Lu<re.e^ t c*p t.f., . .r-"

,,' l n?./u, / k ,1 ,,*e
h4 lla'llor a?
, fi,htk/)ria+i n {/._e--_

C C - -0 r,\.r. ,In-.t 6n | ".^"...-- ll

,t-f l--]---k
'*6'L ?r.^ J'\
I--_ ;^ft 7
\ *-., \ --. t-
| *+ t.^"-'16; 4at,,---'v

rl I
,f -i
vi ---i L
'7'""- '

to I kHz, I to l0 kHz, and

Audio lntegrated Circuits 72
e-locked looP to demodulate
frequencY deviation of + l0
output amPlifier l0 V Peak-

gnal and convert it to a 150

op and a DS8926 frequencY OUTLINE
phase-locked looP is shown l2-l The LM387 Audio Preamplifier l2-3 Volume and Tone Control
d loop is t9 V and to the l2-2 The LM1877 Audio Power Circuits
Amplifier l2-4 AM and FM Radio Circuits
ed loop?
is 150 Hz? KEY TERMS
Audio amplifier Balance control Automatic frequency
;olution of the l0 to 100 Hz Preamplifier Bass control control (AFC)
, increase the inPut frequencY NAB equalization Treble control Automatic gain control
and a DS8926 divide-bY-100 Audio mixer Audio taper
(logarithmic) Preemphasis
Audio power amplifier
Volume control potentiometer Deemphasis
Tone control Muting (or squelching)

On completion of this chapter, the reader should be able to:
r explain the purpose of a preamplifier and describe why a preamplifier circuit
constructed using op-amps is inferior to a specially designed preamplifier
integrated circuit.
r design amplifier circuits using the LM387 dual preamplifier, including a mi-
crophone amplifier, a tape record amplifier, a tape playback amplifier, and an
audio mixer.
r design power amplifier circuits with the LMl877 dual-power amplifier' in-
cluding a simple 2 W audio power amplifi er, a2W per channel stereo amplifier,
a 4 W bridge amplifier, and a high-power amplifier.
r describe the operation of tone and volume controls in audio systems, and use
the LMl035 tone/volume/balance control integrated circuit as a stereo control.
r describe the construction of complete AM, FM, and FM stereo radio receivers
using integrated circuit components.

542 12 Audio Integrated Circuits
Introduction 'gsemicondu
Many students enter the field of electronics because of an interest in audio sys- LM387ILM387A Lo'
tems. This final chapter is included especially for those students. It introduces a General Description
variety of special-purpose integrated circuits used in audio systems such as stereo The LM387 is a dual preamplitier fo
amplifiers, PA systems, tape decks, and AM and FM radio receivers. Collectively of low level signals in apptications
noise performance. Each of the two
these devices are referred to as "consumer integrated circuits" and represent a pletely independent, with an inter
major subgroup of linear integrated circuits. decoupler-regulator, providing 1'10 d
and 60 dB channel separation. Other o
Many of the circuits discussed can be constructed from simple op-amps and include high gain (104 dB), large or
(Vss-2Vlp-p. and wide pwer bar
similar building blocks, but their performance may not be satisfactory for the 20 Vp'p). The LM387A is a rclecr
optimal reproduction of sound. Audio systems have special requirements which LM387 that has lower noase in a NAI
can operate on a larger supply voltag
are often difficult to satisfy with ordinary op-amps. In particular, audio systems erates from a single supply across the
may require higher gain, lower distortion, lower noise levels, and larger power to 30V, the LM387A operates on a su1
outputs than can be provided by simple op-amps. On the other hand, audio systems The amplifiers are internally comt
generally do not need the very high input impedance, the differential inputs, the gr€ater than 10. The LM387, LM387
8-lead dual-in-line package. The Lt\
high common-mode rejection, or the direct coupling provided by op-amps. biaed like the LM38l. Se AN-64 anc
Most manufacturers of audio equipment either produce their own audio in-
tegrated circuits or purchase specially designed integrated circuits from major Schematic and Connecl
manufacturers. These devices are usually identified only by the audio manufac-
turer's part number. Replacements must be obtained from the audio manufacturer,
and data sheets are usually not available. Even special-purpose devices produced
by the major integrated circuit manufacturers, and listed in their data books, are
often hard to obtain. We will discuss only integrated circuits that are readily
available, so that any audio enthusiast reading this book can actually build the
ln this chapter, we will develop circuits for simple audio preamplifiers, audio
power amplifiers, and volume and tone controls, introducing integrated circuits
designed specifically for these applications. Unfortunately, the field of consumer
electronics is large, and it is not possible to cover the entire audio field in one
chapter. Interesting topics and applications such as graphic equalizers, noise re-
duction systems, and video systems simply could not be included. L

The final section briefly introduces communication integrated circuits and

develops circuits for complete AM, FM, and FM stereo radio receivers. These
radio circuits illustrate the construction of complete electronic systems using in-
tegrated circuits.

12-1 The LM3B7 Audio Preamplifier

Audio systems use many types of circuits, the most basic being the audio amplifier.
Audio amplifiers are designed to operate in the normal range of audible sound,
typically from 50 Hz to 20 kHz. The main requirement of an audio amplifier is
FIGURE 1. Flrl Grin Cir'
that it accurately reproduce the input signal without adding appreciable noise or
distortion. We will see that noise and distortion requirements place quite severe
TillNational Audio/Radio Gircuits
luse of an interest in audio sYs- LM387/LM387A Low Noise Dual Preamplifier
r those students. It introduces a General Description Features
i in audio systems such as stereo The LM387 is a dual pJeamplifier for the amplification ! Low noise .0 llv total input noise

of low level signals in applications requiring optimum I High gain lo4 dB open toop
FM radio receivers. CollectivelY noise performance. Each of the two amplifiers is com-
pletely independent, with an internal porer supply ! Single supply opetation
;rated circuits" and rePresent a decoupler.regulator, providing ll0 dB supply reiection r Wide suppty ranga LM387 9 to 3OV
and 60 dB channel separation. Other outstanding features LM387A g to 4OV
include high gain (104 dB), large output voltage swing
ucted from simPle oP-amPs and r Powersupplvreiection 110d8
{v""-zvio-p, and wide powar bandwidth tis r.nrl
nay not be satisfactorY for the 20 Vp-pl. The LM387A is r sclocted version of the I LargeoutputvoltageswinglVcs -.2V)o-o
ave special requirements which LM387 that has lower noise in a NAB tap€ circuit, and . Wide bandwidth l5 MHz unity garn
can opetate on a larger supply voltage. The LM387 op'
ps. In particular, audio systems erates from a single supply across the wide rsnge of 9V '
Power bandwidth 75 kHz,20 Vp p

noise levels, and larger Power to 30V, the LM387A operates on a supply of 9V to 40V. . Internally compensated
)n the other hand, audio systems Tho amplifiers are intornslty componstcd for gains I Short circuit protecied
groatlr than 10. The LM387, LM387A is svailablc in an r perlormance similar ro LM381
rnce, the differential inPuts, the 8{ead dual-inline packsg€. The LM387, LM387A is
iing provided by oP-amPs' birsad lik€ the LM381. See AN-O4 and AN-104.

rer produce their own audio in-

integrated circuits from major Schematic and. Connoction Diagrams
ied only by the audio manufac-
ed from the audio manufacturer,
ecial-purpose devices Produced
d listed in their data books, are
grated circuits that are readilY
his book can actually build the

mple audio preamPlifiers, audio

introducing integrated circuits
rtunately, the field of consumer
er the entire audio field in one
as graphic equalizers, noise re-
t not be included.
rication integrated circuits and Typical Applications
{ stereo radio receivers. These
:te electronic systems using in-

t basic being the audio amPlifier.

rormal range of audible sound,
rement of an audio amPlifier is
FIGURE l. Fl.r circuil (Av - l(xxll FIGUnE 2. NAB ls Cicuir
lut adding appreciable noise or
equirements Place quite severe 10.1.1

' Th^ f^rm 36'rrrrlin

Absolute Maximum Ratings
s|fcly Volto Of.ataq T.mpe.ature Rang€ O'C to +70oC
Lt387 +30v Storag€Tcmrfaiur€ Rans -65"c to +l€,o"c
LM387A +40V Lead Temrrarure {Soldering. 10 €onds} 300'c
P6 Oi$artl@ lNola 1) 660 mW

Electdcal Characterislics r" 14V. unless otherwise stated.


Voltage Gain Open Loop, f = 100 Hz

Supply Current LM387, Vcc 9-30V, R1 = t0

LM387A. Vcc I 40V, R! 10 mA

Input Resiitance
Positive lnput 50 100 KQ
Negative Input 200 KQ

Input Cu,rent
Negative Input 0.5

Output Resistance Open Loop 150 f,)

Output Current Source I

Sink 2

Output Volrage Swing Peak-ro-Peak Y.c 2

Uniry Gain Bandwidth MHz

Larg€ Signal Fr€qu€ncy 20 Vp'p (Vcc > 24V), 75 kHz

Response THD < 1%

Maximum Input Voltage Linear Operation 300 mvrms

Supply Rejection Ratio f= 1 kHz 110 dB

Input Referred
Channel S€paration f= I kHz 40 60 dB

Total Harmonic Distortion 60dBGain,l=lkHz 0.1 0.5 %

Total Equivalent Input 10-10,000 Hz

Noise (Flat Gain Circuit) LM387 1 0
Figurc 1

Output Noise NAB Tap€ Unweighted

Playback Circuit LM387A 400 700

Gain of 37 dB F ioure 2
Nob 1: For wEtioh in mbicnt tempGralres above 25'C, the device musl b€ d€rar€d based on a 150'C maximud junct;on temfrature and a
themd rditbnca ol '18?pC/W junction to ambient.

Typical Applications (conrinued)

TwcPd. Fd Torn-ON NAB T.F Prmpliti.r Frqu€my Rqons ol NAB
zr' Fi$ro 2
Carcuii o{


=.0 l

s il

20 $ rs200 5e 11 21 51 1q20r


Figure 12-1 (Continued)

12-1 The 1M387 Audio Preamplifier 545

power amplifier. This section will discuss audio preamplifiers. The next section
will discuss audio power amplifiers.
The term preamplifier is used to describe low-power audio amplifiers used to
boost the audio signal level as it comes from a microphone, tape deck, or pho-
nograph pickup to a level suitable for driving a power amplifier, typically to the
standard I V peak-to-peak (0.35 V RMS) audio line level. In function, however,
they are simple amplifiers.
An inverting or noninverting amplifier using an op-amp such as the TL081C
with the circuits described in Chapter 3 (see, for example, Figures 3-5 and 3-10)
could be used for such an application. An op-amp, however, is not entirely sat-
isfactory. First, the op-amp requires a dual supply often not available in audio
equipment. Second, standard op-amps generally introduce considerable noise.
Third, audio systems may require gains as large as 1000 at frequencies in excess
of 20 kHz. (The maximum gain for a TL08l circuit at20kHz is approximately
150.) Finally, although the distortion introduced by an op-amp is very low for
small signals at low gains, it can be objectional at the gain levels encountered in
audio systems.
Admittedly, these various limitations can be minimized. As discussed in Chap-
ter 3, a single supply circuit for a standard op-amp can be used to remove the
dependency of an op-amp on a dual supply (see, for example, Figure 3-14). A
low-noise op-amp such as the TL071 can be used to reduce noise levels. An
uncompensated op-amp with external feedforu *id compensation can be used to
provide a greater gain. Several stages of amplificatiol. can be used to minimize
distortion. These solutions all, however, increase the circuit complexity. The
simplest and best solution is to use a preamplifier integrated circuit designed
specifically for audio applications.
The LM387 is typical of such audio preamplifiers. The data sheets for this
integrated circuit are shown in Figure l2-1. The LM387 is a low-noise dual (for
stereo applications) preamplifier packaged in an 8 pin DIP. It requires a single
supply from 9 to 30 V, has a unity-gain frequency of l5 MHz (about five times
higher than a standard op-amp), and a distortion of approximately 0.17o for a gain
of 1000. The LM387 is internally compensated, but must be used with a minimum
gain of 10.
ln many ways, the LM387 resembles a standard bipolar op-amp and is rep-
resented schematically by the op-amp symbol. Although it can be operated as a
noninverting preamplifier or as an inverting preamplifier, for most applications,
as we will see later, it is used in the noninverting configuration. Therefore, the
schematic symbol is normally drawn inverted to that of a standard op-amp. There
are some subtle differences between the LM387 and an op-amp. The noninverting
input is internally biased and, hence, can be AC-coupled to the source without
an external biasing circuit. The internal biasing network contains two diodes con-
nected in series, which holds the DC voltage at the noninverting input 1.4 V above
ground. The inverting input is not internally biased, and an external DC biasing
path must be provided. The DC voltage at this input will be the same as the voltage
at the noninverting input, as for any standard op-amp. This DC bias voltage is
546 12 Audio Integrated Circuits

amplified by the LM387 and appears as a DC offset voltage at the output. Because
of this 1.4 V offset, neither input can be connected directly to ground.

The tM387 as a Noninverting Preamplifier

Figure l2-2 shows the basic circuit for connecting the LM387 as a noninverting
amplifier. The circuit resembles an op-amp noninverting amplifier, but there are
several important differences. The feedback resistor Rr-is the same as for an op-
amp, but there is no R/ resistor. Instead there are two resistors, Ra and Rs,
connected to the inverting input (pin 2 or pin 7). Resistor Ra is capacitively coupled
to ground and is used to set the AC gain. Resistor Ra is connected directly to
ground and is used to establish the DC bias current. This resistor must be designed
before the feedback resistor Rr can be selected. The minimum bias current re-
quired by the LM387 is 0.5 pA, and Ra should be chosen so that at least ten times
this current can flow through it. Because the voltage at the inverting input is
1.4 V, the maximum value for Rs is

^":1# = #:28oko
Any value of Rp, can be chosen as long as it does not exceed 280 kO. Typically
we will use 100 kO.
Once resistor Rs has been selected, resistor Rr- must be designed to set the
DC gain. The DC gain must be chosen so that the bias voltage of 1.4 V is amplified
to give a DC output voltage of V6a;12 to allow for the maximum swing in the
amplified AC signal. Resistors Rr-and Rs act like the feedback and input resistors
in a noninverting amplifier, From Equation 3-5, the gain of a noninverting amplifier

Input ;

Figure 12-2 Basic noninverting amplifier circuit using the LM387 dual audio preamplifier.
12-1 The [M387 Audio Preamplifier 547

in terms of R1 and Rs is

The required voltage gain is output voltage Vccl2 divided by the voltage at the
inverting input of 1.4 V. Substitute this for the gain in Equation l2-2to get
Vcc Rr (1 2-3)
Solve Equation l2-3 for Rr:

Rr:Ra"(ff-') (12-4)

This is the design equation for Rr.

At audio frequencies above the lower cutoff frequency established by C,t, the
AC gain is determined by the parallel combination of Ra and Ra. In all practical
preamplifier circuits using the LM387, Ra will be much greater than Ra and, thus,
can be ignored in the design of RA. In a noninverting amplifier, where the gain
is described by Equation 3-5, the AC gain is

e,=fi+r (12-s)

To design R4 to set the AC gain to a specified value, solve Equation 12-5 for Ra
to obtain

R": #:1 (12-6)

Since resistor Ra is AC-coupled to ground through Ca, the - 3 dB low frequency

cutoff f ct of the amplifier is determined by Ca and occurs when the capacitive
reactance X6'o is equal to Ra. Equating these quantities and solving for Ca yields

r- I
-o - 2rf .tRo (12-7)

This is the design equation for Ct. The lower cutoff frequency f .t is usually
taken as 50 Hz for audio applications.
Because the LM387 operates from a single supply, it must be capacitively
coupled to the source and to the load. The source capacitor C1 is chosen such
that the capacitive reactance X6, is equal to the sum of the source impedance R5
and the input impedance Zin of the amplifier at the lowest operating frequency
f ct (generally 50 Hz). Setting X6:, eeual to Rs + Zin at f cr and solving for Cr,
we obtain

Ll - (12-8)
2nf ct(Rs * Zi^)
548 12 Audio Integrated Circuits

From the data sheets, the typical input impedzrrce Zin for the noninverting input
is i00 kO.
Similarly, the output capacitor Cz must be designed so that its capacitive
reactance equal to the sum of the amplifier output impedance Zo,, and the
load impedance Rz at the lower cutoff frequency /cr. The open-loop output
impedance of the LM387 is listed as 150 O. With feedback, this becomes negligible
compared with the input impedance of an audio power amplifrer (typically 100
kf,)), so we omit it. Setting X62eeual to Ra at f 62and solving for C2, we obtain

Cz: (i 2-9)
The 1M387 as an Inverting Preamplifier
Although the LM387 is normally used as a noninverting amplifier, it can be used
as an inverting amplifier, as shown in Figure 12-3. The circuit in this case is
somewhat confusing. Resistors Rr and Rs are in a different position in the circuit,
but are the same as for the noninverting amplifier. These resistors still establish
the DC gain and are calculated from the same equations as for the noninverting
amplifier. However, resistorRa and capacitor Ceare not used. Instead, we have
resistor Rr connected to the source and resistor Rp-1 connected as a second feed-
back resistor.
The AC gain of the LM387 in this configuration is determined by the ratio of
the feedback resistance to the input resistance as in a standard op-amp inverting
amplifier, but in this case the AC feedback resistance is the combination of R6,
R61, &Ild Ra. From the Thevenin equivalent circuit for these resistors. the effec-
tive feedback resistance of the amplifier is calculated to be (Rp1Re + RrrRB +
RFRB)|RB. and the gain is

Av: (1 2-r 0)
Rr(R, + Rs)


Figure 12-3 The LM387 dual audio preamplifier used as an inverting amplifier,
12-1 The 1M387 Audio Preamplifier 549

Notice the inclusion of Rs, the source resistance, in this equation. When we
discussed amplifiers using op-amps in Chapter 3, we ignored the source resistance.
In audio preamplifiers, the source resistance may be appreciable and must be
included in all calculations.
Because the DC bias current must also flow through Rr'r , there is a maximum
value of Rrr that should be used to avoid any offset problems. The voltage drop
across Ri"r due to the bias current must be much less than the 1.4 V bias voltage.
This will always be the case if we choose Rnr subject to the same condition by
which we chose Ra, or simply
RFt-.* ( RB-.* (12-11)

Normally, we will make Rnr : Rp, and typically we will use a value of 100 kO.
Capacitor Ca is a bypass capacitor connecting the noninverting input to
ground, and is chosen such that its capacitive reactance X6, is less than the
minimum noninverting input impedance Zn at lhe lowest operating frequency of
the preampllfier f ct. Solving the equation Xc,) Zinfor Ca, we obtain

,", nlu^ (12-12)

Substituting 50 kO for Zin (obtained from the data sheets in Figure l2-l as the
minim um input impedance of the LM387) and 50 Hz for f ct (a typical lower cutoff
frequency for an audio preamplifier), we find that Co should be greater than 0.064
pF.A value of 0.1 pF will be adequate for most applications.
Finally, the coupling capacitors Cr and Czare calculated in exactly the same
manner as for a noninverting amplifier by using Equations l2-8 and l2-9.

Microphone Preamplifier
We will now look at several typical applications of the LM387 audio preamplifier.
The first, and simplest, application is a microphone preamplifier. There are two
types of microphones in use: high impedance and low impedance. High-impedance
microphones have a typical output impedance of 20 kO and signal levels ap-
proaching I V. Often no preamplifier is necessary, but if one is used then a simple
op-amp buffer or low-gain amplifier, such as designed in Chapter 3, is adequate.
Because of the large-amplitude signal from the microphone, distortion and noise
introduced by using an op-amp are negligible. High-impedance microphones are
susceptible to hum pickup because of their high impedance and, consequently,
are seldom used.
Low-impedance microphones have an impedance of typically 200 O and out-
put signal levels around I mV. Usually they have a balanced three-wire output
with two signal leads and a ground return to minimize noise pickup. The micro-
phone can be connected to the preamplifier circuit in Figure l2-2 by using a
matching transformer as in Figure l2-4(a) or by direct coupling with the pream-
plifier configured for a differential input , as in Figure I 2-4(b). When the differential
amplifier in Figure l2-4(b) is used, care must be taken to ensure that the micro-
phone sees both inputs as identical for maximum common-mode noise rejection.
550 12 Audio Integrated Circuits

low impedance

(a) Microphone connected to the LM387 using a matching transformer


(b) Microphone connected directly to the LM387 configured as a

differential amolifier

Figure 12-4 The LM387 used as a low-impedance microphone PreamPlifier.

Resistors Ra and capacitors Ce are used for each input. A variable resistor Rz is
connected from the noninverting input to ground. This resistor should have a
value nominally equal to Rs., but because the two inputs have different input
impedances it should be variable to allow exact matching of the inputs to optimize
the common-mode reiection.


Design a microphone preamplifier to amplify the output of a microphone having

an impedance of 200 O and a level of 3.0 mV RMS to drive a power amplifier
12-1 The LM387 Audio Preamplifier 551

requiring an input level of 0.35 V RMS into 50 kO. Use an LM387 dual preamplifier
with a 12 V power supply. Assume a lower cutoff frequency of 50 Hz.

We will use the LM387 configured as a noninverting preamplifier and couple the
microphone through a matching transformer. The circuit is shown in Figure
l2-5 with a parts list.

Parts List
3-wire Resistors:
low impedance Ra 2.7 kdz
microphone R3 100 ka
Rr 330 kQ
C1 1.5 pF
C1 0.1 pF
C2 0.1 pF
ICr LM387

Figure 12-5

First we must design Ra and Rr to set the DC gain of the amplifier so that
the output offset will be Vcc12 for maximum output voltage swing. We calculated
from Equation l2-l that the maximum value for R6 is 280 kO. We will arbitrarily
choose a value of 100 kO for Ra. Now we can design resistor Rr from Equation

/v-- \ : /l2v
Rr : Ra
" (fr - r) loo ko ,. (r* u - r)\ = 32e ko
Use a value of 330 kC) for R6.
The AC gain is determined by Ro, which is calculated from Equation 12-6.
The AC gain is VoutlVin = 0.35 V/3.0 mV : ll7.

" : ; RF :: :*-un, : 2.84 ko
A,-l ll7-l
We will use the standard value of 2.7 kO.
552 12 Audio lntegrated Circuits

To design Ca to give a lower cutoff frequency f .t : 50H2, use Equation


,": : 1.18 pF
Choose Ca as 1.5 pF.
Finally, we must calculate the input capacitor Cr and the output capacitor C2
from Equations 12-8 and l2-9, respectively. From the data sheets shown in Figure
l2-1, the input impedance to the noninverting input Zi" is 100 kO and the load
impedance is specified as 50 kO.
Cr: : 0.0318 p.F
2trf a;/Rs * Zin) 2rr x 50 Hz x 100 kO

: 0.0637 pF
2r.f ctRt 2n x 50 Hz x 50kO
Choose both Cr and Cz as 0.1 pF.
Our preamplifier is designed.

Tape Record Preamplifier

The tape head used for magnetic recording is an inductive device. The inductive
reactance ofthe head, and hence the amplitude ofthe signal recorded on the tape,
increases with frequency at the rate of 20 dB per decade up to a frequency slightly
above 1 kHz. At this point stray capacitance and other losses cause the reactance
and signal level to decrease. The typical response of a record head is shown in
Figure l2-6. In magnetic tape recording, the high frequency response of the input
signal is usually boosted by using a preamplifier to compensate for the high fre-
quency drop in gain of the head. The resultant response curve is shown in Figure
l2-7(a). On playback, it is necessary to attenuate the high frequency signal level
back to the original level by using an amplifier with the complementary response

100 Hz 1 kHz 10 kHz 100 kHz

tigure 12-6 Typical response of a record head.

12-1 The [M387 Audio Preamplifier ss3

3l dB 3l dB
30 30
rr O
20 z0
l5 l5
10 l0
5 5

0 0
l0 Hz 100 Hz I kHz l0 kHz 100 kHz l0 Hz 100 Hz I kHz l0 kHz 100 kHz
Frequency Frequency
(a) Recording response (b) Playback response

tigure 12-7 Standard NAB equalization response curve.

curve, the curve shown in Figure l2-7(b). This process of boosting the high fre-
quency record signals and attenuating the high frequency response on playback
is called NAB (National Association of Broadcasters) equalization. The curves
shown correspond to open-reel tape speeds of l3 and 3] in./s and to cassette
recorders. Slightly different curves, used for open-reel recording at 7i and
l5 in./s, are not shown.
The curves in Figure l2-7(a), (b) are standardized curves used throughout the
tape recording industry. For recording. starting at 50 Hz, the recorded signal
amplitude increases at a rate of 20 dB per decade up to a frequency of 1770 Hz,
by which point signals see a total boost of 31 dB (a gain of 35.48). Higher fre-
quencies are recorded at a constant boost of 3l dB. For playback, the reverse is
done. Frequencies lower than l77OHz are boosted by 20 dB per decade down to
50 Hz, by which point signals see a total boost of 3l dB (a gain of 35.48).
We will design a recording preamplifier using the LM387 to boost the level
of the signal to the recording head to produce the standard NAB equalization
response. At low frequencies (less than the frequency where the recording head
gain starts to fall, approximately I kHz), the preamplifier requires a constant gain.
Starting at a frequency of 1000 Hz, the preamplifier must boost the incoming
signal at a rate of 20 dB per decade to generate the NAB response. This response
is achieved by the circuit shown in Figure l2-8, where a series R-C circuit con-
sisting of Rr and C: has been connected to the noninverting input of a standard
noninverting preamplifi er.
The design of this tape record preamplifier is similar to the basic preamplifier
described earlier. First set the DC gain by selecting a value for R6 less than the
maximum value of 280 kO given by Equation I 2- I . Then use this value to calculate
Rr from Equation l2-4. Resistor Re is designed by Equation l2-6 to set the low
frequency gain, and capacitor Ca is designed by Equation l2-l to set the lower
cutoff frequency /ga. Input capacitor Cr and output capacitor Cz are calculated
from Equations l2-8 and l2-9, respectively.
554 12 Audio Integrated Circuits


Figure 12-8 The LM38? used as a tape record preamplifier.

The difference in this preamplifier, however, is that the gain must start to
increase at a frequency of approximately 1000 Hz to overcome the drop in gain
of the recording head. In the circuit in Figure l2-8, this increasing gain is produced
b1l C3 connected to the noninverting input. This capacitor is effectively in parallel
with Ra, and at frequencies above the critical frequency /3, where X., : Ro,
causes the gain to increase by 20 dB per decade. Solving for Cg gives

Ct= 2rf (12-13)

The circuit in Figure l2-8 also includes resistor Rr in series with capacitor
Cr. If this resistor is not included in the circuit, the gain will increase at 20 dB
per decade for all frequencies above f:. This is undesirable because at very high
frequencies, the gain becomes very large and can cause the amplifier to oscillate.
Resistor Rr is included to prevent this. For high frequencies, where the capacitive
reactance of Cr is much less than Rr, the gain is determined by Rp and the parallel
combination of Ra and R:. The critical frequency, where R3 begins to limit the
gain, which we will call f a, is determined by X6. : R:. Solving this equation for
Rr, we obtain

Rs= (12-14)
2nf dCt
Normally, fa is set to approximately 20 kHz, the top end of the audio range.
Figure 12-9 shows the actual circuit that would be used for recording. Note
the inclusion of two additional elements. Resistor R+ is connected in series from
the output of the amplifier to limit the current to the recording head. The parallel
L-C filter is included to prevent the tape bias signal from feeding back to the
amplifier. It is designed with a resonant frequency equal to the bias frequency'
12-1 The 1M387 Audio PreamPlifier JJJ


Figure 12-9 LM38? tape record preamplifier connected to source and recording head'

Design a tape record preamplifier using an LM387 dual audio preamplifier. As-
sume the standard recording head response shown in Figure 12-6 and design the
preamplifier to give a standard NAB response. The input signal has a maximum
arnplitude of 20 mV RMS from a 400 O source, and the tape head requires a drive
current of 25 pA RMS. The recorder response should be from 50 Hz to 20,000
Hz. The recorder has a 9 V power supply.

The circuit is shown in Figure l2-10.

Parts List
RA 1.8 ka
R8 t00 ko
Rp 22OkA
R3 75O
Ra 100 ko
C1 0.033 PF
C2 0.033 pF
C3 0.10 PF
CA 2.2 pF
ICr LM387

Figure 12-10
556 12 Audio lntegrated Circuits

First design the Dc gain to set the output offset to *vcc. The procedure is
the same as in Example l2-1. choose a value of 100 ko for Rr and calculate Rp
from Equation 12-4:
_ Rst (;+
Rr /v-c -\ /gv _ l)\ : z2rka
I | : l00ko x {*
= -
\l.E / \z.uv /
We will use a 220 kO resistor for Rr.
The low frequency (less than 1000 Hz) AC gain is determined by Re, which
is calculated from Equation l2-6. we must first determine the desired AC gain.
The input signal has a maximum amplitude of 20 mV RMS. This is 2 x fi x
20 mV : 56.6 mV peak-to-peak. The maximum output peak-to-peak signal is
approximately 2 V less than the supply voltage, or9 V - 2V : 7 V. The required
AC gain is Vo,,/V;. : 7.0V156.6 mY : 123.7:

R^ : ..Rn
A"-t . = ,-t:qun,
123.7- I
: "'-"'ko

We will choose the standard value of 1.8 kO.

Design Ca for a lower cutoff frequency f ct of 50 Hz by using Equation l2-7:

/- |
= 2r, ,. 50H, t l^8Lo :
L:":- 2rf.rRo l'77 *F

Choose Ce = 2.2 pF as the closest larger standard value.

The frequency f3 where the gain of the preamplifier must start to increase to
compensate for recording head losses is 1000 Hz. This gain increase is achieved
by the addition of C:. Capacitor C3 is given by Equarion l2-13:

c' : tl : o'0884 P'F

znf ,R^: 2"
" looo Ht x, ls kf)
We will choose the closest standard value of 0,I pF.
Resistor R:, which limits the gain at frequencies above 20 kHz, is calculated
from Equation 12-14
R::- :?960
" 2rJ aC3 2n x 20 kHz x 0.1 pF
The closest standard value is 75 O.
Resistor Rq is used to limit the current from the amplifier to the recording
head. since the tape head requires an RMS drive current of 25 pA and the output
voltage is 7 V peak-to-peak or (7lDlf1 : 2.47 V RMS, R+ is

^ : Vo.,:
2'47 V
,5 l.A
: 98'8 ko

We will use a
100 kO resistor for Ro.
Finally, we must calculate cr and c2. lnput capacitor c1 is calculated from
Equation l2-8, using a source resistance Rs of 400 o and an input resistance Zi.
12-1 The [M387 Audio Preamplifier ,5/

of 100 kO (from the data sheet). Output capacitor Cz is calculated from Equation
l2-9 for a load resistance equal to Ra:

Cr= : 0.0317 pF
2rf ct(Rs * Zi) 2r x 50 Hz x 100.4 kO

Cz: 2rf ll x
: 0.0318 pF
ctRt 2r x 50 Hz 100 kO

Choose Cr :
Cz : 0.033 pF.
Our preamplifier is designed.

Tape Playback Preamplifier

We will now design a tape playback preamplifier. The signal recorded on the tape
has the NAB response shown in Figure l2-7(a) due to the combined effects of
the recording head response and the recording amplifier response. On playback,
we want a level output response. The playback preamplifier must boost the am-
plitude at frequencies lower than 1770 Hz at a rate of 20 dB per decade down to
50 Hz to give the required level response.
The circuit of the LM387 used as a playback preamplifier is shown in Figure
12-11. This circuit is similar to the basic preamplifier circuit in Figure l2-2,but
has a series filter circuit consisting of R+ and C+ in parallel with RF. At very low
frequencies, capacitor Ci" has a very large capacitive reactance, and, hence, the
filter branch has a very high impedance compared with Ri". The audio gain is
determined by Ro alone from Equation l2-5. At very high frequencies, capacitor
C+ has a very small capacitive reactance. and the audio gain is determined from

Figure 12-1 1 Circuit of the LM3E7 used as a NAB tape playback preamplifier.
5s8 12 Audio lntegrated Circuits

the parallel value of Rr. and R+. Because Ra will always be much less than Rp,
the parallel resistance is approximately Ra. Replace R6 in Equation l2-5 with Ra
to get the equation for high frequency gain:

e,:ft+r (r 2-1 5)

The high frequency gain required is determined from the input level required by
the power amplifier divided by the output level supplied by the tape deck (the
head sensitivity).
To design the preamplifier circuit, first design Ra and Ri- to set the DC gain.
This is done in exactly the same manner as for the previous amplifiers. Select a
value of Rr less than the maximum value of 280 kO given by Equation l2- I . U se
this value of Rr to calculate Rp from Equation l2-4.
The lower critical frequency of the NAB equalization curve, f r : 50 Hz, is
determined by setting Xco = Rr. Solve this equation to determine C+:

,r:;k, (12-16\

The upper critical frequency of the NAB equalization curve, f z : 1770 Hz, is
determined by setting Xco : Rq. Solve this equation to determine R+:

Rr: 2trf zCa


Finally, use the value of Ra to determine the value of Ra required to set the high
frequency gain of the amplifier. Solving Equation 12-15 for R,a, w€ obtain
" Av-l (12-18)

Capacitors C^, Cr, and C2 are determined as for a simple preamplifier using the


Design an NAB tape playback amplifier using an LM387 dual-audio preamplifier.

The playback head response has a sensitivity of 1200 pV, and the preamplifier
drives a power amplifier requiring 0.35 V input into 40 kO. The recorder response
should be from 50 Hz to 20 kHz. The recorder has a 9 V power supply.

The circuit is shown in Figure 12-12.
As in the previous preamplifiers, we will first set the DC gain by designing
resistors Ra and R6.. We will arbitrarily choose a value of 100 kO for Rr and
12-1 The [M387 Audio Preamplifier 559

Parts List
R1 22fJ
R3 l00ko
R6 220kd,
Ra 6.2 kO
C1 0.033 pF
C2 0.10 pF
Ca 0.015 pF
Ca 150 pF
ICt LM387

Figure 12-12

calculate Rr.from Equation l2-4:

/v-- \ /9v \
Rr = Ra. (ti - t) : rooko' (ffi - t) : 22tka
We will choose Rr : 220 kO.
NAB equalization is determined by resistor Ra and capacitor Ce. Calculate
Ca from Equation l2-16 to set the lower critical frequency of the NAB equalization
curve, /r, to 50 Hz:

c.: | :- : 0.0145 pF
2rf rRr 2tr x 50Hz x22OkQ
Choose C. : 0.015 r,rF.
Calculate R+ from Equation 12-17 to set the upper critical frequency of the
NAB equalization curve, f z, to 1770 Hz:
O, = :
#rrr: 2n x 1770 Hz x 0.015 p.F
5.99 kO

Choose Ra = 6.2 kO
Finally, we will use this value of Rn to determine the value of R4 to set the
high frequency gain of the amplifier. We will first have to determine the required
high frequency gain. Since the required output is 0.35 V for an input of 1200 pV,
the gain is 0.35 V/1200 p"Y = 292. Calculate Ra from Equation 12-18:

R, : Ro 6.2 ko
: -""" f)
Au-t - 292-t lr.3
We will choose the standard value of 22 Q for R: .
560 12 Audio lntegrated Circuits

Design Ca to give a lower cutofffrequency f 6a : 50H2, using Equation l2-7:

Ca: = 145 pF
2trf crRe 2r x 50Hz x 22Q
Choose Ce :150 pF as the closest larger standard value.
Finally, we must calculate the input capacitor C1 and the output capacitor
Cz. Calculate Cr from Equation l2-8. We will assume the source resistance Rs
of the playback head to be zero, and from the data sheets the input resistance Z.
is 100 kO.

Ct: : 0.0318 p.F

ct(Rs * Zi")
2rrf 2tr x 50Hz x 100kO
Calculate C2 from Equation l2-9 for a load resistance equal to the 40 kO input
impedance of the power amplifier:

cz : ;),o, =
2r x 50Hz x 40kO
: 0.0796 pF

Choose Cr : 0.033 pF and Cz : 0.1 pF.

Our preamplifier is designed.

Audio Mixer
The final application circuit that we will introduce for the LM387 preamplifier is
that of an audio mixer. Audio mixers are circuits for combining two or more inputs,
and are used for applications such as public address systems, tape recording, and
guitar amplifiers. We will only consider a simple two-channel mixer, although
more channels can be added with little difficulty. The mixer circuit using the
LM387 audio preamplifier is shown in Figure 12-13. Note that we are using the


Figure 12-13 Two-channel audio mixing circuit using the LM387 preamplifier.
12-1 The 1M387 Audio Preamplifier 551

inverting input of the preamplifier, taking advantage of the low input impedance
of this input to minimize crosstalk between channels (this configuration is the
same as that for summing circuits in Chapter 4).
Resistors R6 and€ designed to set the DC gain as in previous preamplifier
circuits. Select a value of Ra less than the maximum value of 280 kO given by
Equation l2-1. Use this value to calculate Rpusing Equation l2-4. Inverting pre-
amplifiers have a second feedback resistor R;'1 that must be chosen with a value
less than the maximum value of Ra. Normally, set this resistor equal toR6.
The AC gain of an inverting preamplifier using the LM387 was described by
Equation 12-10, and depends on Rr-, Rn.r. R". Rr, and R5. Equation 12-10 is
normally solved for Rr to give the design equation for this resistor:
Rr.'Rn*RrrRa+RFRB (12-19\
Rt: - Rs

The gain Ar is determined by dividing the desired output voltage level by the
individual inDut source voltase levels:
desired output voltage level
A\ _ (12-20)
source voltase level
The gain must be calculated for each input to the mixer and, in general, is different
for each input. As a consequence, separate resistors Rr must be designed for each
lnput coupling capacitors Cr (called Cr earlier) must be designed for each
input from Equation l2-8 with the Rr and Rs values for each input. The output
coupling capacitor Cz is calculated from Equation l2-9. Capacilor Co connected
to the noninverting input is simply a bypass capacitor to ensure that this input is
at effective AC ground and is calculated from Equation 12-12. As mentioned
earlier, any value greater than 0. I pF is adequate for this capacitor.
If a variable gain is required to allow for different input devices or for pro-
ducing different mixing levels, Rr on each input can be replaced by a potentiom-
eter. In this case, the capacitors Cr should be chosen much larger than the design
value to allow for small values of R, when the potentiometer is at the low end of
its ranse.


Design a two-channel mixer using an LM387 dual-audio preamplifier. One input

is from a microphone with an impedance of 600 O and an output level of l2 mV.
The other input is the signal tiom a tape deck preamplifier having negligible output
impedance and a signal level of 350 mV. The mixer must deliver a signal level of
l0 V to a l0 kO load. Use an 18 V power supply.

The circuit is shown in Fieure l2-14.
12 Audio Integrated Circuits

Parts List
Rp 560 kC)
R6 l00ko
Rfl 820()
Rp 43 kQ
Rfl l00kQ
Ctr 3.3 pF
Cn 0.1pF
C2 0.33 pF
Cs 0.1pF
ICt LM387

Figure 12-14

First design Ra and Rr. This will be the same as in previous examples. We
will arbitrarily select Rr as 100 kO, less than the maximum value of 280 kO
required by Equation l2-1, and calculate Rr from Equation 12-4:

Rr: Ra , (+ r) = rooko . f*+ - r) : s43 ko

\z.E - '/ \2.8V /
Choose the nearest standard value of 560 kO.
The voltage gain must be calculated for each input to the mixer by using
Equation l2-20:
, desired output voltage level
' :
source voltage level
For the microphone input, the voltage gain required is Av, = 10/0.012 : 833'
whereas for the tape deck input the voltage gain required is Av, = 10/0.35 :
We will select a value of Rrr : 100 kO (equal to Ra). Calculate Rr for each
input by Equation l2-19:

^ Rn,Rn + RFrRa * RrRa

K,=T-nt D

For the microphone

input, the required gain is 833 and the source resistance is
600 0:
l00kO x 560kO + l00kO x l00kO + 560kO x 100kO
l00ko x 833 - 600f,)

: 14650 - 600.f}: E64'f)

Choose the closest standard value of 820 O.
12-2 The LM1BZZ Audio power Amplifier 553

For the tape deck input, the required gain is 28.6 and the source resistance
is negligible:

: 100 kO x 560 kO + 100 kO x 100 kO + 550 kO x 100 kO

Rr. - 0f,)
100 kO x 28.6
: 42.7 kA
Choose the closest standard value of 43 kO.
Next, input coupling capacitors C7 must be designed for each input by using
Equation l2-8 and assuming that the lower cutoff frequency is 50 Hz:
2nf 62(R5 + R1)

First for the microphone input, where R.s = 600 O and Rr : 1.6 kO,

e!' 2rr x 50 Hz x (600 O+ 820 n, = 2'24 v"F

Choose the standard larger value of 3.3 pF.
Next for the tape deck input, where Rs : 0 O and Rr = 43 kO,

2r x 50 Hz x 43 kO
= 0.0740 pF
Choose the larger standard value of 0.1 pF.
We will arbitrarily select bypass capacitor Cs as 0.1 pF. This will be more
than adequate according to Equation l2-12.
Finally, the output coupling capacitor C2 is calculated from Equation l2-9,
where the load resistance is 10 kO.

: 0.318 pF
2nf ctRt 2n x 50 Hz x l0 kO
Choose the larger standard value of 0.33 rrF.
The design of the mixer circuit is complete.

12-2 The lM1877 Audio Power Amplifier

The second audio integrated circuit that we will study is the audio power amplifier.
The audio power amplifier is an amplifier designed for the audio range of fre-
quencies that has a relatively low voltage gain but a very high current gain and,
consequently, a large power gain. Audio power amplifiers are designed to drive
loads of a few ohms, typically speakers with impedances of 4 or 8 O.
There are a wide variety of audio power amplifiers, ranging in power output
from less than I W to more than 100 W. Most of these are special-application
devices, custom-manufactured for a specific piece of audio equipment, and thus
are not readily available. Some major manufacturers of integrated circuits, how-
ever, such as National Semiconductor, Texas lnstruments, and Motorola, produce
'Ja Semiconductor
LM1877 Dual Power Audio Amplifier
General Description
The LM1877 is a monolithic dual power amplilier r Wide supply range. 6-24V
designed to deliver 2Wlchannel continuous into 851 I Very low cross-over distortion
loads. The LM1877 is designed to operate with a low r Low audio band noise
numb€r of external components, and still provide . AC short circuit protected
flexibility for use in stereo phonographs, tape reorders
and AM-FM stereo receivers, etc. Each power amplifier
r Internal thermal shutdown
is biased from a common internal regulator to provide
high power supply rejection, and output O point cen-
tering. The LMl877 is internally compensated for all Applications
gains greater than 10. . lvlulti-channel audio systems
r Stereo phonographs
r Tape recorders and players
Features r AM-FM radio receivers
r 2Wlchannel r Servo amplifiers
t -65 dB ripple rejection, output reterred r Intercom systems
| -65 dB channel separation, outout referred I Automotive Oroducts

Connection Diagram Dud.ln-Line P*kag€

r0, vrtw

Ordor Number LMlE77N

56€ NS Package NltlA

Equivalent Schematic Diagram


Figure 12-15 Data sheets for the LMl877 dual power audio amplifier. (Reprinted with permission
of National Semiconductor CoDoration)

Absolute Maximum Ratings
SupplV Vollage 10.7V
Input Voltage 0'c ro +70'c
Operating TemPerarure
-65'C to +150"C
Storage TemPerature 1 50'c
Junction TemPerature 3oo'c
'10 secondsl
Lead Temperature (Solderrnq'

Vg = 20V, TA = 25"C, RL = 8A'

Av = 50 t34 dB) unless otherwise specified
Electrical Cha racteristics
Total SuPPIY Current Po=ow

Output Power THD = 1O%

LM 1 877N
VS = 2OV, BL = 8Q

ToEl Harmonic Distortion

f= 1 kHz, VS = 14V
LM1877 0.07 5
PO = 50 mw/Chann€l
0 045
PO - 500 mwlchannel
PO = 1W/Channel
vs 6
cF=5o!F CIN=01!F
Output Referted
VS=20V.VO=4Vrms -60
CF = 50!F, C1N = 0 1 sF, | = 124 42
Outpul Referred dB
VS. 7V, VRTPPLE = 0.5Vrms

Equivalent InP0t Norse

RS= 0,clN = o
j !F' BW= 20Hz 20 kHz

OutPut Norse Wldeband

RS=O.CIN=O1!F.AV=200 dB
Open LooP Gain
RS=0,f= 10OkHz.R1=8Q
lnpur Otfser Vollage
Inp0l Bies Current rilo
lnpur lmPedance Open LooP
VS = 2ov
DC Oulput Level
Slew Rate kHz
Power Bandwrd!h
Current Limll

upon device mounling technrques
,a]^g ,n".-"i *"i"ance which depends

Figure 12-1 5 (Continued)

"12-2 The 1Ml877 Audio Power Amplifier 567

Bias vcc

Output I Output 2

Gnd Gnd

Gnd Gnd

Gnd Gnd
Output 2
Input I Innnt 2

Feedback I Feedback 2

(a) Pinout of 14 pin DIP (b) Schematic symbol

Figure 12-15 Pinout and schematic symbol for the LM1877 dual power amplifier.

The 1M1877 as a Simple Power Amplifier

The circuit of a simple noninverting power amplifier using half of the LMl877 is
shown in Figure 12-17. The voltage gain is determined by feedback resistor Rn
from the output (pin 2 or 13) to the feedback (inverting) input (pin 7 or 8) and by
input resistor Rr from the feedback input through capacitor C1 to ground. Equation
3-5 is used, which is the same equation we used for a noninverting amplifier for
a standard op-amp. Normally a value is chosen for Rr, and Equation 3-5 is solved
for Rr to give the design equation for this resistor:

R,:#= (12-23)

The gain Ay required by the amplifier is determined by the ratio of output signal
level to input signal level:

. : output voltage level (12-24)

"t ;;."" uoltuee IeuA
If the output voltage is calculated from the power supply voltage by using Equation
l2-2l,the value is in peak-to-peak volts, and it is important to express the input
in the same units when using Equation 12-24.
Because the LMl877 is designed to operate from a single supply, resistor R7
must be connected to AC (or signal) ground through capacitor Cr. This capacitor
determines the lower cutoff frequency f ct of the amplifier. This is the frequency
where the capacitive reactance X6, eeuals Rr. Solving for C1 gives
tt = znf.rR, (12-2s)
568 12 Audio lntegrated Circuits

Rs Input i
= (a) Amplifier using a single supply

10,1 1, 12



= +
(b) Amplifier using a dual supply

Figure 12-17 Simple noninverting power amplifier using the LM1877.

Capacitor Cr should always be chosen larger than this value to ensure that the
capacitive reactance is negligible at all operating frequencies.
The manufacturer recommends using a compensating network consisting of
a2.7 Q resistor R6 in series with a 0.1 pF capacitor C6 connect€d from the output
(pin 2 or 13) to ground. The output is also capacitively coupled to the load (speaker)
with output capacitor Cz. This capacitor is designed such that its capacitive re-
actance X6, eeuals the load impedance Rr at the lowest operating frequency /6a.
This gives the following design equation for C2:
C^= (12-26)
2rf ctRt
12-2 The 1M1877 Audio Power Amplifier 569

Notice that we did not include Zo',. the output impedance of the amplifier. in this
equation. Although the output impedance is not given on the data sheet, it can
be assumed to be a few ohms. With feedback, it becomes even smaller and, hence,
can be neglected in calculating Cz, For an 8 O speaker as load and a cutoff fre-
quency of 50 Hz, C2 is typically 500 pF.
The signal input is connected to the noninverting input of the amplifier (pin
6 or 9) through coupling capacitor Cr. Since the input impedance of the amplifier
is essentially infinite, we can use any value of capacitor. Choose a 0'1 p.F value
for Cr.
The noninverting input must be connected to pin I through a resistor. Pin I
is labelled BIAS on the pinout in Figure 12-16. Because the LM1877 uses bipolar
technology, a bias current must flow into each input. For the inverting or feedback
input, the bias current flows from the output through Rr. For the noninverting
input, however, the source is capacitively coupled and, without pin l, there is no
path for DC bias current to flow. The bias pin must be decoupled to ground by
a large-value capacitor Ca, for which the manufacturer recommends a value of
250 pF. To prevent the input signal from being shorted to ground by this capacitor,
insert a bias resistor Ra between the input and the bias pin. The value of Ra is
not critical, and any large value will be satisfactory. We will use a value of 1.0
MO for all of the following designs. If both amplifiers are used, separate bias
paths must be established for each noninverting input (pins 7 and 8).
The LMl877 has a number of pins labelled as ground (3, 4, 5, 10, I I, and l2).
For operation from a single supply, these are all connected to supply ground. For
operation from a dual-polarity supply, they are all connected to the negative sup-
ply, All other ground connections still go to ground. Figure 12-17(a) shows the
LMl877 connecred ro a single supply, and Figure l2-17(b) shows the LMl877
connected to a dual supply. when the LMl877 is operated from a dual supply,
the bias output (pin l) is connected directly to ground. It does not need to be
capacitively coupled through Cr because the AC ground and system ground are
the same in this mode of operation.


Design a simple audio power amplifier using half of an LM1877 dual-power audio
amplifier to drive an 8 O speaker. Assume that the RMS input voltage from the
preamplifier is 0.35 V, and design the power amplifier for maximum output voltage
level. Use a single l8 V power supply.

The circuit is shown in Figure l2-18.
First we must determine the amplifier gain required. The input from the pre-
amplifier has an RMS value of 0.35 V, or a peak-to-peak value of l'0 V. The
maximum output voltage is given by Equation l2-21:
maximum output voltage : Vcc - 6V : 18 V - 6 V : l2Y
s70 12 Audio Integrated Circuits

Parts List
R7 9.1kcl
Rp 100ko
R6 1.0 MC,
R6 2.7 O
C1 1.0 pF
Cs 250 t$
C6 0.1 pF
C1 0.1 LrF
C2 500 pF
ICr LM1877

The required voltage gain, given by Equation 12-24, is

, output voltage level 12 V
* =tt
This value will be used to determine Rr. Choose Rr as 100 kO and calculate R7
from Equation 12-23:

Rr : --
Rn 100 ko : 9.09 kO
A"-l tL-r
Choose the closest standard value of 9.1 kO.
Now calculate capacitor C1 from Equation 12-25, assuming a lower critical
frequency of 50 Hz:
2rf ctRr 2r x 50 Hz x 9.1 krr
Capacitor Cr should always be chosen larger than this value to ensure the ca-
pacitive reactance is negligible at all operating frequencies, so we will choose a
value of 1.0 rrF.
Calculate the output coupling capacitor Cz from Equation 12-26, where the
load impedance Rz is 8 O and the lowest operating frequency .f.. is 50 Hz:

c'= II
2nf .rR,
= t;t-50 H;; 8 o = 398 P'F
Choose the larger standard value of 500 pF.
We do not need to calculate the input coupling capacitor C1 because the input
impedance of the amplifier is essentially infinite. We will simply select a 0.1 r-r.F
value for Cr.
12-2 The lM1877 Audio Power Amplifier 571

use the recommended values of 1.0 Mo for the bias resistor R8, 250 pF for
the bias capacitor C B, 2.'7 O for the compensating resistor Rc, and 0. I pF for the
compensating capacitor C6'.
Finally, connect all the ground pins (pins 3,4,5,10, 11, and l2) to ground'
The power amplifier design is complete.

The 1M1877 as a Stereo Amplifier

The LMl877 consists of two identical amplifiers and is designed basically to be
used as a stereo amplifier. The circuit for this application is shown in Figure l2-
19. Each amplifier is connected by using the same circuit as described for a single
amplifier. Each amplifier circuit has its own biasing, its own compensation net-
work, and its own feedback network. For biasing, both bias resistors are con-
nected to pin I and decoupled through a single decoupling capacitor Cr. The gains
of the two amplifiers can be made different if necessary.
Figure l2-19(a\ shows the stereo amplifier using a single supply, and Figure
12-19(b) shows the same circuit with a dual supply. When using a dual supply,
the ground pins (pins 3,4, 5, etc.) are connected to the negative supply, and the
bias output (pin 1) is connected directly to ground (it does not need to be capa-
citively coupled through Ca).



(a) Amplifier using a single supply

Figure 12-19 The LM1877 used as a stereo amplifier.

572 12 Audio Integrated Circuits


(b) Amplifier using a dual supply

Figure 12-19 (Continued)

Design a stereo audio power amplifier using an LM187'l dual-power audio am-
plifier. Each output will drive an 8 fl speaker. Assume that the input voltage from
the preamplifier is 0.35 V RMS and design the power amplifiers for maximum
output voltage level. Use a dual 'r12V power supply.

The circuit is shown in Figure 12-20.
The overall design is similar to the simple power amplifier designed in the
previous example. First determine the gain required from the amplifier. The input
from the preamplifier has an RMS value of 0.35 V, hence a peak-to-peak value
of 1.0 V. The maximum output voltage from Equation 12-21 is
maximum output voltage = Vcc - 6 V : 24V - 6V : 18 V

The required voltage gain, given by Equation 12-25, is

output voltage level IEV :18
source voltage level 1.0 V
12-2 The lM1877 Audio Power Amplifier 573


Parts List
Left r
fu 5.6 kO
-ljvjry-l Rp
1.0 Mo
Rs 2.7 Q
C1 1.0 pF
Cs 250 pF
C6 0.1 prF
Ct 0.1 pF
Right C2 500 pF
ICr LM1877

Figure 12-20

Select Rr as 100 kO and calculate R7 from Equation 12-23:

^' =
Rr : 100
-,- ko
---, : _5.88kO
Au-l 16-l
Choose the closest standard value of 5.6 kO from Appendix A-1.
Now calculate capacitor Cr from Equation 12-25, assuming a lower critical
frequency of 50 Hz:

v' 2rf ll'

C,- x Hz x =0.568pF
ctRt 2n 50 5.6 kO
Capacitor Cr should always be chosen larger than this value, so we will choose
a value of 1.0 uF.
Calculate output capacitor Cz from Equation 12-26 for a load impedance R1
of 8 O and a cutoff frequency fcr of 50 Hz:
ll :
Cz = 2nJ.tR.
= 2r, 50 FL '^ 8 o
398 PF
Choose the larger standard value of 500 pF.
574 12 Audio lntegrated Circuits

Select a value of 0.1 pF for coupling capacitor Cr. Use the recommended
values of 1.0 MO for the bias resistor R8,250 p.F for the bias capacitor C6,
2.7 Ofor the compensating resistor R6, and 0.I pF for the compensating capacitor
Finally, connect all the ground pins (pins 3, 4, 5, 10, I 1, and I 2) to the negative
The stereo power amplifier is now designed.

The 1M1877 Used as a Bridge Amplifier

The two amplifiers in the LM1877 can be configured as a single bridge amplifier
to supply 4 W to an 8 Cl load. This application is shown in Figure 12-21. The
outputs of the two amplifiers are connected to either side of the 8 O speaker with
the output compensating network connected in parallel across the speaker.
For this configuration to work, the output signal from one amplifier must be
a maximum when the output signal from the other amplifier is a minimum; that
is, the outputs must be 180'out of phase. This is accomplished by supplying the
input signal to only one amplifier, the left amplifier as shown in Figure 12-21. The
right amplifier input at pin 9 is effectively ground because it is connected to the
bias pin and thus to ground through the bias capacitor Ca. The input network
consisting of Rr and C1 goes from the feedback input of the left amplifier (pin 7)
to the feedback of the right amplifier (pin 8). This allows the feedback signal from
the left amplifier to serve as input for the right amplifier, introducing a phase shift
of 180' and making the amplifiers operate in a push-pull mode as they drive the
speaker. The component design is the same as for the simple power amplifier
described earlier,

lc. R6 | :.+.s.

Figure 12-21 The LM1877 used as a 4 W bridge amplifier.

12-2 The LM1B77 Audio Power Amplifier J/ J

Figure 12-22 The LM1877 used as a high-power amplifier.

High-Power Amplifier
The maximum output power available from the LMl877 is 2 W/channel (or 4 W
if the amplifiers are connected in bridge configuration), but with the addition of
external power transistors, it is possible to increase the output power well beyond
this limit. As shown in Figure 12-22, the LMl877 can be used to drive a com-
plementary pair of power transistors, with the output power limited by the current
gains of the transistors.
The resistor Rz.connected between the bases and emitters of the power tran-
sistors is used to bias the bases at 0.7 V to minimize crossover distortion. It has
a typical value of 5 O. Each transistor conducts for alternate halves of the input
cycle and is nonconducting for the other half-cycle. This is known as class B
operation, and it allows high operating efficiency for the power transistors. The
circuit in Figure 12-22 is a minimal circuit, and for optimal performance a more
elaborate discrete component power amplifier should be used. The design of dis-
crete component power amplifiers is beyond the scope of this book, and the reader
is referred to any standard text on discrete circuit design.
The design of the circuit for the LM 1877 power amplifier portion of the circuit
is similar to the basic power amplifier designed earlier. There are a few minor
differences. There is no compensating network on the output. The input imped-
ance of the external discrete component power amplifier is approximately equal
to the transistor B times the load impedance. This is much larger than the 8 O of
a speaker, so the basic compensating network would overcompensate the inte-
grated circuit amplifier. Instead, the manufacturer recommends a compensating
network consisting of an 82 pF compensating capacitor C. in series with a 27 kO
compensating resistor R6, in parallel with the feedback resistor. The feedback
J/O 12 Audio Integrated Circuits

signal is taken from the output of the discrete component amplifier. not from the
output of the integrated circuit amplifier.

12-3 Volume and Tone Control Circuits

With most audio amplifier systems, circuitry is provided for varying the loudness
of the output sound level (volume control), for varying the low frequency gain
relative to the high frequency gain (tone control), and for varying the relative gain
between the two channels in stereo systems (balance control). Of these, the tone
control is the most complex, requiring adjustment of the low frequency gain (bass
control) and of the high frequency gain (treble control). All of these controls are
easy to implement, usually as part of the preamplifier circuit.

Simple Tone and Volume Controls

We will first look at simple controls for a single-channel amplifier (monaural sys-
tem). Obviously, there is no need for a balance control. The volume control is
usually a potentiometer across the input to the power amplifier, with the input
signal picked off the wiperarm as shown in Figure 12-23(a). Since the human ear
has a logarithmic response, an audio taper (logarithmic) potentiometer should be
used so that the change in sound intensity as perceived by the ear will vary linearly
with respect to the setting of the volume control. Bass and treble controls are
shown in Figure l2-23(b) and l2-23(c), respectively. The potentiometers in these
should also be audio taper.
Consider the bass control first. This control can attenuate or boost low fre-
quency signals relative to midband frequencies, In the circuit in Figure l2-23(b),
for midrange and high frequencies, capacitors Cra and C:a effectively short out
potentiometer Rp13, and the attenuation is

. Rzu
Kta + R:u

Ir lPut


L lr^

(a) Volume control (b) Bass tone coDtrol

Figure 12-23 Volume and tone control circuits.

12-3 Volume and Tone Control Circuits 577

At low frequencies, the capacitors and potentiometer boost the signal or cut it
relative to this attenuation.
To design this bass control, first determine values for Rrs, R26, and Rp6. The
ratio of RrslR28 and RpalRrs sets the amount of "boost" and "cut." If both ratios
are 10, then the boost and cut in decibels are each 20 log l0 : 20 dB, a typical
value for a bass tone control. Choose the capacitors such that X6,, : Rrs and
Xcru : Rzs at frequency /r, where the bass attenuation is to effectively start (the
-3 dB point). Solve for C1s znd C26:

Cra:*k (12-28)

L2B - (12-29)
2rf 6R26
The treble control in Figure 12-23(c) is similar. The resistor and capacitor
dividers are reversed from the bass control, and the control attenuates or boosts
highfrequenc'y signals relative to the midband frequencies. For midrange and low
frequencies, capacitors Crr and Czt have high impedances and give an attenuation

. Crt
: -=--:- (12-30)
C17' I C27

To design this circuit, first choose a value for potentiometer Rp7. Design
capacitor Crr such that X6-,, : Ret at the frequency f r where the treble atten-
uation or boost is to effectively start (the -3 dB point). Solving for ClTgives

Crt:*k: (1 2-31 )

fiom preamplifier
Bass L Its
con t rol Treble
Audio Audio
taper taper


to power amplifier

Figure 12-24 Combinations of tone controls and volume control in a typical circuit.
578 12 Audio Integrated Circuits

The ratio of CzrlCtr sets the amount of boost and cut. If the ratio is 10, then the
boost and cut are each 20 dB. Choose resistors Rrr and Rzz such that R'r :
X6,. &rd Rr, : X6., zt frequency /r. The following equations give R17 and R:r:

Rrr = (12-32)
Znf rC rr
Rzr = (12-33)
2nf rCzr
Figure l2-24 shows how the gain, bass, and treble controls can be incorporated
into a typical audio amplifier circuit. The tone control circuit is usually placed
between the preamplifier and the power amplifier, and the overall circuit gain is
designed to compensate for the midband attenuation of the tone controls.
Tone controls are sometimes incorporated in the feedback loop of a separate
preamplifier. There is little advantage in this for most applications, and cost is
an additional amplifier stage and more complicated circuit design.


Design a single-channel tone/volume control network. The bass control should

become effective at 600 Hz with a boost/cut of 20 dB. and the treble control should
become effective at 1500 Hz with a boost/cut of 20 dB.

The circuit is shown in Figure l2-25. Use 100 kO audio taper potentiometers for
all of the controls.
First design the bass circuit. The ratio of RralRza and RpolRts sets the amount
of boost or cut. Thus, for a boost/cut of 20 dB, the ratio is simply 10. Since Rpg
has been chosen as 100 kO, RrB : Rpsll} : l0 kO, and R4 : Rral1O :
1.0 kO. These are all standard values.
Now calculate capacitors Cra and Cza from Equations l2-28 and 12-29 for a
frequency /a of 600 Hz:

CtB:*k: 2n x 600H2 x lOko

: 0.0265 pF

Czn:*k= 2" t 600

H, 1.0 kCl
: 0.265 p.F
Choose values of 0.027 pF for Cra and 0.27 p.F for Cza.
Next design the treble circuit. Since we have chosenRpTas 100 kO, we can
design capacitor Crrfrom Equation l2-31 for a frequency /t of 1500 Hz:

c,, :;*-,*, 2.rr x 1500 Hz x 100 kO

: 0.00106 pF
12-3 Volume and Tone Control Circuits 579


from preamplifier

control Treble
Audio Audio
taper taper

C:n Auditr
Rn taper
c0ntrol r.'''
to power amplifier

Parts Lisi
Resistors: Capacitors:
Rrs l0 kO Cta 0.027 pF
R:s 1.0 kQ Cy 0.27 pF
Rrr I l0 kO Ctr 0.001 pF
Rt ll kQ Ct 0.01 [F
Rps 100 kQ pot
Rpr I00 kQ pot
Rpr, 100 kO pot
R.r t0 kQ

Figure 12-25

Choose a value of 0.001 pF. The ratio of CzrlCrr sets the amount of boost/cut.
For a boost/cut of 20 dB, the ratio is 10, and a value of 0.01 pF will be used for
capacitor C2r.
Now design resistors R17 and R21from Equations 12-32 and l2-33 for a fre-
quency fr : 1500 Hz:

Rrr : :
#*_: 2n x 1500 Hz x 0.001 pF
106 kO

R.r= 2'trf rl :-=l0.6ko

'\:r 7C27 2rr x 1500 Hz x 0.01 pF
Choose values of I l0 kO for Rrr and 1l kO for Rzr.
Finally, select a value for resistor R:. This resistor isolates the bass and treble
controls to prevent one control from loading the other. We will arbitrarily choose
a value of 10 kO.
The volume/tone control circuit is designed.
s80 12 Audio Integrated Circuits

The 1M1035 Dual Tone/Volume/Balance Control

The basic controls we have discussed are widely used and, for a monaural system,
are entirely satisfactory. In a stereo system, however, two complete sets of tone/
volume controls are necessary, resulting in a relatively complicated circuit. Fur-
thermore, in a stereo system, the tone/volume controls for both channels must
be varied simultaneously. This requires the use of ganged potentiometers and
matched components, making the tone control system relatively costly. Integrated
circuit technology has produced an alternative: the LM1035 tone/volume/balance
The data sheets for the LMl035 tone/volume/balance control are shown in
Figure 12-26. The pinout of the chip is shown on the first page of the data sheets.
The LM1035 operates from any supply between 8 and 18 V, provides control for
both channels simultaneously, and only requires the connection of several external
capacitors. The volume, bass, and treble, for both channels, and the balance are
all controlled by DC voltages applied to four control pins (pin 12 for volume, pin
14 for bass, pin 4 for treble, and pin 9 for balance). This DC voltage (typically
from 0 to 5.4 V) can be supplied by an external control circuit or obtained from
potentiometers connected to an internal 5.4Y zener regulated supply (pin l7). A
linear variation in voltage produces the proper logarithmic audio response in the
volume, bass, or treble, so standard linear potentiometers can be used.
A typical stereo tone control circuit using the LM1035 is shown in Figure 12-
27. The bass and treble responses for each channel are defined by capacitors C6
and Cr, respectively. For channell, CB is connected from pin 6 to ground, and
Cr is connected from pin 3 to ground. For channel 2, Ca is connected from pin
15 to ground, and Cais connected from pin l8 to ground. For standard operation
with approximately l5 dB boost or cut, the manufacturer recommends Ca : 0.39
pF and Cr = 0.01 pF. lncreased boost and cut can be obtained by reducing C6
and proportionally increasing Ca.
Capacitors Cr and Cz are the input and output coupling capacitors, respec-
tively. The input capacitor C1 is designed such that its impedance X6, at the lowest
operating frequency .f.. is equal to the sum of the input impedance Z;n and the
source impedance Rs. The resulting design equation for Cr is

Cr: (12-34)
2nf ct(Zi" + R.s)

The input impedance Zin for each channel is typically 30 kO, and the source
resistance R5 depends on the preamplifier or other input circuit. The output ca-
pacitor C2 is designed such that its impedance X6. at the lowest operating fre-
quency .f.. is equal to the sum of the output impedance Zuu, and the load imped-
ance Rz. The resulting design equation for Cz is

Cz: + Rr)
(r 2-3s)
2nf a;1(Zo6
Notc that the loudness network is part of the electronic volume control (al- electronic
though there are cxtcrnal components). AIso note that a failure symptom for the volume control
loudness functions is usually difficult to define. This is because the loudness
function attcnuatcs thc midrangc signals so that the ear hears what appears to be

the same lcvel across the audio range. (In some amplifiers, the loudness function Audio
boosts thc treble and bass. but this is rare.) to
Unfortunately, all ears are not the same, and not all loudness networks define power
midrange at the same frequcncics. Usually, the customcr complains that "there is no amp
diffcrence when I play the tape or recording with the loudness function on or off." 220k 100 3.9 k
4.7 k 47 ft
To troubleshoot such a symptom, apply an audio signal (say between 7 and 10
kHz) with loudncss off. Then press loudness and check for drop of about 20 dB in
lcvel at pin 4 of ICooo. Repeat the test at 50 Hz and 20 kHz. There should be
substantially no changc in lcvel (at pin 4 of lCuor) at the low and high ends of the 47 tt
audirl rangc (unlcss a boost circuit is used) even though there is a change at the
No mattcr what type of loudness circuit is used; if there is no change in audio 5.6 k
Icvcl at any frcqucncy when thc loudness function is switched in and out, there is 47k
a problcm in thc loudness circuit. Start by checking the loudness switch circuit, 5600 p
ICe,,1, lCc.a, and the network connectcd at pins 2 and3 of ICuoo.
Treble 50k
50k ---'-' C-taper
0.33 p
1.5 k

f '-I-
tv u
Bcfore we gct into thc output circuits for our particular amplifier, let us review
typical tonc, balancc, and playback-cqualization functions usually associated
with prcscnt-day arnplifrcrs.
V V I"-
One clrannel of a tone-control network (IC).

loss prcsentcd by the circuit can be eliminated or minimized as dcsircd. Figure

5.6.1 Tone and Balance
5.7 shows one channel of a tone-control network using an IC amplifier. Note that
Tone (treblc and bass) conrrols are found in most hifi amplifier systems. Balance the circuit docs not provide for volume control. This is bccause an clectronic
controls arc uscd in stcreo amplificrs to balance the gain of both channels. volume-control circuit (such as shown in Fig. 5.6c) is used- The tone-control net-
Atreble control provides a mcans of adjusting the high-frequency response of work of Fig. 5.7 is connected between the electronic volume control and the out-
an audio amplifier. such adjustmcnt may be necessary because of variation in put or powcr amplifier, as described in Sec. 5.6.3.
rcsponse of thc human ear or to correct the frequency response of a particular
A bctss control provides a mcans of adjusting the low-frequency response of an 5.6.2 Playback Equalization
audio amplificr. Such adjustment may be necessary because of variition in re-
sponse of thc human ear, which does not respond as well to low-frequenry Many playback-equalization circuits are found in modern audio amplifiers. Most
sounds (at low levels) as to high-frequenry sounds at the same level. Also, cou- involve the use of frequency-selective fcedback between stages or from thc out-
pling capacitors present high reactance to low-frequenry signals. Both of these put to the input of an amplifier (typically an IC amplifier).
conditions rcquirc that the low-frequency signals be boosted (in relation to high- The feedback network usually consists of resistances and capacitances that
frequency signals). form a fecdback circuit. At any given frequency, the amount of fecdback (and
There are many circuit arrangements for tone controls. Some involve the use thus the frequency response) is set by selection of the appropriate RC combina-
of adjustable feedback (mainly in treble controls). other circuits involve bypass- tions. As frcquency increases, the capacitor reactance decreases, resulting in a
change of fccdback (and a corresponding change in frequency response).
ilg ttt" coupling capacitors with adjustable reactances (mainly in bass contiols).
However, the most common tone controls are RC filters using audio-taper poten-
tiometers as the adjustable R portion of the filter. Basic Playback-Equalization Network. Figure 5.8a shows a basic playback-
cqualization circuit whcrc thc voltagc gain of two stages (with fcedback) is about
_ In present design, tone-control networks are often used with IC amplifiers.
one advantage to the lC-amplifier and tone-control design is that any insertion cqual to the feedback-circuit impedance divided by the source impedancc. (ln

this case, the source impedance is the emitter-resiStance Rt value.) Thc fccdback
impedance is the vector sum of the R, resistance value and thc Cr rcactancc
valuc. The voltage gain of the two stages can be set to any desired lcvcl for any
givcn frequency with this simple feedback circuit.

IC AmpliJier Equalization Network. Figure 5.8b shows an equalization nctwork

that uses feedback between the input and output of an IC amplificr (thc prcamp of
a stereo system in this case). The closedloop (with feedback voltage) gain of the
proamp is set by the ratio of the feedback network to resistor Rr.
Thc fecdback for the phonograph (RIAA) equalization network is composed of
C3, Ca, R., and Ro, and the tape network (NAB) is composed of R5, C6, and Cr.
Preamp Notc that CD playcrs normally do not require playback equalization.
-l RIAA Playback Equalizalion- Figure 5.8c shows the standard RIAA equalization
curvc for phonograph use. The recording curve is the invcrsc of thc playback
R3 R4
curvc, so addition of the two curves produces a flat frequency-versus-amplitudc
51k 750k
response. In phonograph recording, the high frequencies arc cmphasized to re-
v3 v4 ducc cffccts of noise and the low inertia of the cutting stylus. The low frequencies
1500 p 5600 P
Phono are attenuated to prevent large excursions of the cutting stylus. lt is thc job of the
(HrAA) frequcncy-selective feedback network to accomplish the addition of the recording

""i,tl3t,l'i:.[$l:",:T'""" the ptayback network be the exact invcrse or tr," ,J
51k cording compcnsation since each recording system is slightly diffcrcnt. However,
Tape optional guidclincs can be applied. A typical audio range is from 2OHzto20kHz,
33ul IPS
so thcrc is a rolloff at both the low and high ends.
Fl6 910 p In thc circuit in Fig. 5.8b, the linear rolloff is produced by dividing the play-
820 k 7lz IPS back nctwork into thrce sections. The RrC, section sets thc 10-Hz point at 3 dB
down from the 20-Hz point, the RoCo section covers frequcncics up to about 1
R2 kHz, and the R.C. nctwork covers higher frequencies.
c2J- NAB Playback Equalization. Figure 5.8d shows the standard NAB equalization
15pr curves for tape usc. Again, thc recording curvc is the inversc of the playback
V (b)
curye, so addition of the two produces a flat rcsponse. Likewise, thc high fre-
quencics are emphasized and the lows are attenuated. However, unlike the
phono playback, tape playback tends to flatten out after about 3 to 4 kHz.
A different rcsponse is required for different tape specds. Figure 5.M shows
thc plas,back responsc curves for both 3% and 1Vz inches per second (ips). Up to
+20 +30 about 1 kHz, thc curvcs are almost identical. Because there is only one frequency
brcakpoint (whcrc thc curvc must start to flatten) for each tapc spccd, a simple
+20 RC corrrpcnsation nctwork is sufficient (instead of the multisecond network used
+10 33t+ tps for phono playback).
Voltage +10
Voltage Thc brcakpointfor 33/4 ips occurs at about 1.85 kHz, whercas the breakpoint
(dB) 0 \ for 7 t/z ips is at about 3 .2 kHz. (Thc reactance of Cu equals the rcsistancc of R. at
(dB) \ 1.[i5 kl'tz, whereas thc reactancc of Ct equals thc rcsistancc of R. at3.2kHz.)
-10 *
l'/2 tps
Notc that thc accuracy of both the RIAA and NAB compensation is only as
good as thc conrponcnts used. Typically, resistors and capacitors with a I perccnt
029 (or bcttcr) tolcrancc arc uscd in discrete-componcnt playback nctworks. Also, it
10 1oo 1k 10k 10 100 1k 10k rnay bc ncccssary to trim thc value to get an cxact (or near cxact) performancc
FrequencY (Hz) curvc for truly good hifi perfornrance. (Rernember this if you nrust rcplacc any
FrequencY (Hz)
(d) corrponcnt in any playback-equalization network.)
Becausc Co and C7 block the dc path for thc IC preamp fccdback input, Ro is
5.8 Amplifier playback-equalization circuits' addcd when the phono-tapc switch is in eithcr tapc pgsition. Althotrsh R. lirrlits
suspccl lL6(,6. Also cllccK tllc Dass ano treuls conrrols srnco [noy are rn the
negativc-fccdback path of ICuoo. Mrdpornt

. If.therc is audio 3J piry 1 and 7 of IC6o6, check for audio (at about the same
level) at pins 1 and 18 of ICro,. If the audio level at ICro, is substantially differeni
from. the lcvcl at lCuuo, trace the audio path through ilii subsonic filtei.
The subsonic filter switch sou, setting should have little or no effect on the
audio levcl, exccpt at vcry low frequencies (below about 20 Hz). If you notice To FLg'or
drastic_change in audio levcl at different settings of suur, from aboui l kHz and through lCes2
up, \t9k for problems in thc subsonic filter circuit (suc-h as leakage in Cu,6).
To speaker
.^ If thcrc is audio a! p]lns 1.and 1.8-of lCro., but not at pins l0 ind 13, Iispect
lcr,,,. Beforc you pull ICr,,, (heat sink and aii), make sure that the 45-v suppiy is relay RY7e1
applied to various ICr,,, terminals.
__Jh".45-V (both plus and minus),supply is appried through Ryro.. In turn,
RYr,,, is turncd on through Qru, and err,rwhen pin :+ of ICn,,,
loes low'1wnen the Sr* (normallY closed,
amplificr powcr switch is prcssed). Mosiof thc other Ics reCeiie operating power
opens at 100" c)
whcn thc powcr cord is plugged in (whether the power button is piessed-oi not).
Becausc of the hcavy'.current drain and high heat dissipation, po*". amplifier
it turned on only during play. This is ty:pical for most umptitiers wtrere thc
19r,,powcr (Mounted on lO7s1 heat sink)
final Thermal
stagc is a single rc in the +o- to-so-w rangc. If thc 45-V supply is
abscnt at ICr,,,, suspect Ry7o2, erur, erur, or ICuo, (ch-eck pin 34 of ICn.i,'for
low). Audio to soeakers
If. thcrc is audio at pins l0 and 13 of ICro, but the audio does not reach
spcakcrs^ or hcadphones, suspect Ry7o1, eror,
eroo. Also check for a low at
pin_33 of lCr,,,. Pin 33 should go low aithe iime ", time as pin 34. Overload
Remembcr that thc power-output protection circuits deicribed in Sec. 5.7
designed to cut off thc audio circuit in the event of an overload. Defective pro-
tection circuits can cut off thc audio, even without an overload. Audio

FIGURE 5.10 Amplifier output-protection circuits'

multaneously, pins 36 through 40 clf IC.r,,, arc pulscd to llash thc function-display
Figurc 5.10 shows the o_utput-protection circuits. The overload-protection
circuit portion of Fl-uur.
prcvents damagc to ICro, when a low-impedance or shorted'
speaker is con_
nected. 'fhe midpoittt-p?ten-tyt protection iircuit prevent, ouruglio
ihe speaL-
ers in case of a defective tcrof and is turned on when a oc p&entiar (usuaily
called dc ffiet) is present^at the output of ICro,. The thermarprolii'tio, 5.7.2 Midpoint'Potential Protection
prevents damagc to ICr6, from excessive heat.'
The midpoint-potential protection circuit functions by monitoring the dc output
from IC-,. ln theory, there should bc no dc output.from-ICru, to the speakers.
(excessiv6 direct cuirent can damage th_e speaker coils-) HoweYe.r' as a practical
5.7.1 Thermal Protection matter, therc may be as much as 11.7 V at the ICro, output without damage to
ttte splakers. The midpoint-potential protection circuit (called the dc offset
ThermaLprotcction switch sru. is mounted on the ICro, heat sink
and is normalry proteLtiott circuit in .ot" is iurned on if the 1'7-V value is exceeded'
closed..This keeps Dru, and gno, off."Wittr gno, noiconducting, If there is any dc output frbm ICro, to the speakers, this potential c-auses C7,3
pin 3 of ICro, is high, andIe"."fr^"^biased
pin 33 of ICno, r..irtt-r to* to t.ii"fi" rp"ul"., to chaige through Rr,rL and R- Cr, the average value of the speaker
nected to the ICro, output. "on- operation, wittr ttre dc output from lCror less than +1.7
V, th-e midpoirit-protection circuit is turned off'
. . lf the temperaturc of the ICro, heat sink rises to 100oC, S"n. opens. fbrward
biasing Drur. This turns pou2 on a.tio produces a low ai pin
i;f iEr;-u"n!". tr,"r" If the u.r".ug" charge across C7r, increases above + l'7 V, Q7o7 is-turned on'
conditions, pin 33 of ICoo, goes high to disconnect thc speakers'"tlorn
iCro,. Sl_ ro.*aiJuiasl nE Dror. itris applies'ilow to the base of Qzro, turning pzro on and
i;il uiasin! pr"". This applies a high to the base af Qsoz, turning Qnoron, and
c.auses prn 3 ot lcno, t. go low. ICnu, then produces a high at pin 33
the speakers (and to pulse the rl-oi, function display) u'r airi,u*.a.-

should be compared to those found in the
5.7.3 OverloadProtection ;ifi"t:;;;;;";;;";;;..J"i",
litcraturc tor tne,'p..iti" amplificr bcing scrviccd. Likewise, the service-
p.o"!a"i.r should be foll6wed. (When all else fails, follow
The overload-protection circuit is the same for_both channels, so only literature adjustment
the right instructions.)
channel is covered hcre. Audio_output from ICro, to the "''il;;-,.5-u1 the amplifier described in this chapter does not,have any internal
$eake;-is appli"eJ
through Rr,ro., a 0,22-o resistor. This resistance is'mich smaller than
the ,p';k;; presen!-duv 19 audio amplifiers'
load impedance (rypically, 6 to l6 O). .dj;r;";i;ontroti.'rnit is truc for many treble, and balance controls' In the
During normal opcration, the voltage across R,u* is very smalr. If As a minimum, n,,ori utpflii"is ttu,re U-asi,
a shorted rp;cific tertin!'und adjustment procedures in the service literature'
or very low-impedance speaker is connected, exciisive oltput cuirent flows
through Rr16*, and the voitage across Rrru* increases sharply.' "b;;';f
ffi;;;t"it i; consider rihen tcsting anv audio amplifier follow'

, Resistors-R71ap arrd_R715R are connecGJ as a.voltage OViOer across Rrro* to

the base o-f Ozo-sn. As the current through Rrro. increaies
lUecause of a short o,
Iorv-impcdancc. load), the voltage appli6d to
d)qr* in...ur'"r,-iurnii 0ror* on. 5.8.1 AmPlifier Tests
This forward biases Dror, turns on
Qrro, for*i.O""biur", Drou,'t;;;; b""")
applies a low to pin 3 of ICro, to disionnect ^a
the speakers"'J"a f"rr" frii;r"., operatcthcvolumecontrolforamidrangevolumclevelorasspecifiedinthe
display. service literature, uno bass, treble, ind balance controls to their midrange'
ii;;;;-"ppt a {_kHz ".iitt"
rin"-*u,r. signal to each of the various inputs (phono,
turncr, Cb, tapc play' auxiliary' 9!":)'
t'f,c Lutuni" control until the outputs across
the speakers (and/or tape
5.7.4 Output.Protection Troubleshooting rccord) are identical. li;h;;il]if*, hut o'ftont-panel level-indicator,
adjust the
the same levcl indication.
The front-panel function display FLno, should flash on and off, and ;;l;;;;i;oi untir both channcls show
get equal output
should be disconnected when lny <inb of the foilowing occurs: FLro,
the speakers If the balanc" .un,rJ rnust be sct far from thc midrange to
becomes (withanidentical,ignu'r.utuotr'inputs),.thereisaseveremismatchcondition.
overheated.(the ICru, hcat sink reich_es 100'c), the consianii;;.;tiJi'd" is not limited to the
applied to the spcakers exceeds !1.7 v, or thespeake. output tine
vortage This can bc thc rcsuu ;i;;";i";t il the balance circuits but
can bc a mismatch in lCror fig. 5.9)
is at any impedance below that of the speakers).
is'strorteo 16r balancc nctwork. n,r, .*Jrpl", the problcm
;lh;;;;il i, ui'utt'inpui, of thc amplifier. In such cases, it is neces-
.late,Except f<lr thc low-impedance output, these conditions are difficult to
simu_ sary to rcplace
tCr' as-ola"fug^' though one channel may be good'
making thc circuits difficult to check. Arso,.if you do .r"".ri in si-utating "\"n input or one o.utput of the am-
any onc of thcse conditions and the protection circuits are not functioning Of coursc, if thcre irl'*iir-ut.n at only irnc
is a mis-
crly, you can damagc the cquipmentlfor exampre, burn out th".p;;k;;
propl plificr, thc problcm .r; il';i";;d do*n casily', For cxample'
prcamp shown in Fig'
or ovcrheat ICrur).
coir and/ match at only thc ph.,;.";;;;: tutp""t lCo,,, lthc
record output'
5.4). On thc other n"nJ, lfiit tism.atch
upptutt af only thc tape
. !f you must check the c.ircuirs, try shorting the speaker rincs (either or R
," Cj"l-"j tgmnolariJy ee.ry tcmporaril"yl. Ctr'cct that the iunctionL portion or suspcct lCn,,, (tapc rccord buffcr' Fig' 5.'a)'
(withai t-q:l'l:"t signal at
ol lf both chann"rt p.oA,t"-ttt"tjtiuily the'same signal
rL,)()r tlashcs on androff and that thc speakers are disconnected. If the next step is to test the
porarily short pin 3 of lCro, to ground, and check for a nasling
not, tem- both inputs and with ,;; b"l;;;;;*i."t ot midran-gc),

spcakcrs disconnccted.
;i;ily with the .ane" of both thc bass and trcble controls'
'"'?;p;;;lly; * at some spe-
,rt. t oi. and treble controls havc a 8- to 10-dB range' in this chap-
If thc display flashes, and the speakers are cut with pin 3 of ICo^, shorted but t.,"trol of the amplificr dcscribed
cific frcqucncy. no,.
not when thc spcakcr lines arc shorted, suspect "*unipit;;;;;;; treble conirol has a t 1 0-dB range at 10
eror, drur, elror, dion, e716, aod tcr has a r g-dB rangc oi iin ifr, whcreas thc
i.H;. ;0 Hi and 20 ki; oi" urt6 common bass.and trcblc i too-uz,fienal to the
Qrur. frequencie-s'
.^lf thc display does not flash and speakers are not disconnected with pin
3 of Set both thc bass ;;;i;;;ntrols to midrange..Applv
lCo-u, shortcd, suspect ICro,. s-o thai ihl outputs of both chan-
You can also check try;,t19 anode of Dro, is-at ground (unless inputs of both channcls. Sct the balancc control
ri"[i:-rl_1_00',C or.highcr). tf it is not, suspJct thatSro, is ipen. ---- -ICro, heat nels are identical-
the bass control from
you can also chcck the bases of. Keep the treble and balance controls at midrange'-Vary
eror-and gror. E"oth bases should be 0 V other. Note that the output of Jach chinnel varies about 8 dB
(ideally) but.may be at some potentiir i6is than i-t.z one extreme to the
v, wirhout triggering rhe at thc bass midrange setting'
protection circuits. If the basei are at some varue uilo"" una below the output existing apply a 10-kHz sig-
in exclr;';i;1.;'V, pin 3 of Return both the b";;;nJ tt"ui" Jonools to midrangeland
ICno, should go low, and rhe display should n".t. so that both out-
nal to the inputs of boil ;il;t;;lt .
if through L"uu" the balance control set
Qtrc or Qsoz. ".t,-;6;'br;, puts are identical.
. ; ;;;; ; ; ; ;. ;; #.'N'
above and below the output cxisting
il; ;il: t,i ifi I :itJ,i"i'':il #';i';if" ::H: I'J:il 5.9 PRELIMINARY TROUBLESHOOTINb
at the trcble midrangc sctting.
If the amplificr passcs thc tcsts dcscribcd here, it is reisonable"to assume that Before you plunge into any amplifier circuit, here are some rather obvious but
the amplificr is functioning normally, and no troubleshooting is required. often overlobked checks that may cure mysterious problems'
If the amplifier does not tunt oz (either manually or by remote), make sure
5.8.2 LoudspeakerTests ittut tte iower cord is secure in the receptacle and that the receptacle has
In addition to ctre.cking amplificr characteristics, the loudspeakcrs should also bc If the amplifier operates manually but not by_ remote, make sure that the con-
checked. This brings up somc probrcms. Arthough it is possible to test a loud- trol cables are secure in the coriect jacks. If all control cables are good, try
speaker for such characteristics as,sound pre""rri" tevetisvr) undcr laboratory replacing the remote-unit batteries.
conditions, thc^most practical tcst is "by ear." Unfortunately, you and the cus-
tomer havc differcnt cars, so thc results are unccrtain (at Ueiti.- If there is no sound, check that the speake-r switches are set properly. Then
To further comflicatc. thc spcakcr problcm, s,mc speakers arc adjustable. For check the input selector. It is possible that the wrong audio-source comPonent
example, thc spcakcrs shown in Fig. 5.ll havc volumc controls (cailed pads, or has been seiected or the component is defective. Try playing the amplifier
potcntiomctcrs with morc one scgment) in both thc midrange and twceters. with another source. It is alsb possible that the audio cables between the
Although thcsc arc not usually.customcr adjustment controls, th"c pads are often ,ou."" and amplifier, or the tp"uket cables, are disconnected (even-though the
adjusted to some arbitrary sctting matcir the customer's ear."' control cables are good). If the front-panel output_power-lwel display is flash-
From a test or troubleshooting standpoint, make sure that the speaker adjust- ing, with no sound-, the output-protection circuit has tripped'
ments (if any) can control volume at thocorresponding speaker and that the con- If there is httm, the turntable may no.t be properly.connected to,the amplifier
shields may !e broken. It is also
trols are smooth (no abrupt changcs in volume as thc-control is adiusted). ti.ounO terminai (Fig. 5.1), or the audio-cable
iossible that the uuiio-"ible connector may not be the cor-
lespondingjack. If the cables, connectors, and ground terminals all appear to
be good, Check the installation notes in Sec' 5'1'4'
If the left or right speaker (only) is dead,.temporarily reverse tht l"jl a.nd-right
Joeakei leads.-tf tne run1" rp"ui.". remains dlad, the speaker is at fault. If
p'roblem moves to the other speaker, one channel of the amplifie-r or input au-
lio tout." is at fault. Tempoiarily reverse the left and ca-is
;i;r-; the amptifier input.'If the same speaker remains dead,thetheaudio-source
probably ut f*tt. If the other speaker goes dead, suspect
oruT L
I Woofer

Toru l=

22lt Midranl le volume pad

, 33rLtr 0.5 mH
t^ /

'uI i*1
. J ,u,.,-
0.033 p
I Midrange

4.7 lt

0033p o33mH+rF I Tweeter

Tweeter volume pad

FIGURE 5.fl Tlpical speaker circuirs.
Yr 4/SP2|I



r Object

1.1 To gain experience on the use of various software and hardware

tools in a digital signal processing development system.
r.2 To implement and investigate the performance of an adaptive
filter and a FFT spectrurn analyzer- in real-time by using the
DSP32C digital signal processor.

2 Introduction
Apart from using discrete digital components, there are in general three
approaches for implementing any digital signal processing function.
The fastest approach is to implement the processing in rrhardwirerr by
designing an Application Specific IC (ASIC). Although the algorithm
will then be difficult to be altered later, the use of ASICs will lead to
the simplest and the most cost effective hardware when the volume is

In the otlier extreme, the rnost flexible way to implernent any

processing function. .. is through the use of general .purpose
microprocessors. While such processols are highly programmable and
are supported by a variety of high level development tools such as
compilers, simulators and emulators, they are usually not optimized for
signal processing purposes. As examples, their sophisticated addressing
and interrupt handling capabilities are not needed in most algorithms,
whereas the common mulitply-and-accumulate primitive is either not
provided or only very slowly executed.

The last and increasingly popular approach to realize a signal

processing function is by using Digital Signal Processors (DSPs), which
are essentially specialized microprocessors with architectures tailored for
implementing the most common signal processing primitives. These
devices started to appear in the commercial market in the early 1970s
and have been used as the processing engines in numerous
speed-critical applications in speech, image, communication, aerospace,
control and computer systems.
Invariably, all DSPs include a hardwired multiplier-accumulator on
chip and can execute one instruction in a few clock cycles so as to
overcome the speed bottleneck in real-time signal processing
applications. Also, since most processing algorithms are dependent on
just a few fundamental primitives, these devices only have a sinple
instruction set, a few addressing modes and some very elementary
Yr 4/SP2/2

interrupt handling capabilities.

Some common DSPs are the 16-bit fixed-point TMS32010 and DSP16
processors which are rated in the order of five to a few tens of Mips.
Although they can cost as little as $10, these devices are fixed-point
in nature and the programmer has to take care of overflow, scaling
and truncation problems. Also, since high level compilers may not be
available, these processors are sometimes cumbersome to use.
To overcome these difficulties and with ease of use and large dynamic
range in mind, some 32-bit floating-point DSPs such as the
TMS32030 and DSP32 have also been introduced in recent vears.
With their accornpanying compilers and othel high level support iools,
these processors are ideal in the initial stages of developing and
implementing sophisticated algorithms.

3 Equipment
DSP Development Systern.
Function Generator.
Noise Generator.
Dual- Trace Oscilloscope.
Passive Summer.

4 DSP32C Development System

In this laboratory, you will first download some programs implementing

a FFT spectrum analyzer and an adaptive filter onto a PC based
DSP32C development system and to investigate the performance of the
algorithms under various real-time operating conditions. Subsequently,
by using the development system, you will develop, debug and test an
improved adaptive algorithm for enhancing a FM signal in broadband

Apart from the PC which functions as the terminal, the DSP32C

development system consists of the various software and hardware
components listed in Table 1. Note that the DSP32C is a 32-bit
device capable of performing 12.5 million floating-point
multiply-and-accumulate operations per second. Also, a C compiler
is available and this will ease the burden of developing any real-time
application considerable.
Yr 4lSP2l3

o Assembler, C compiler and simulator

r Maths and application library
o Program downloading, debugging and other utilities
o DSP32C based development card with
o 50 MHz 25 Mflops DSP32C propcessor
o 144 kwords of 32-bit static RAM
o 2 16-bit A/D and 2 16-bit D/A

Table 1 Components of DSPS2C Deuelopment System.

5 Rectification and Clipping

To learn how the development system can be used for software and
hardware development, first study carefully the program "M1 . Crr listed
in Appendix A.1. Basically, this is a C language program which
directs the DSP32C processor to read the inputs from the two A/Ds,
rectify and clip the input from channel A, and then output the results
to the two D/As.
Note that the program includes an interrupt service function which will
be executed whenever the processor receives an interrupt from a
hardware counter. Tlie counter is memory mapped to the DSP32C
and by calling a function to write the appropriate word to this counter
at the beginning of the rnain function, the sampling frequency can be
software controlled.
As for the hardware counter, both A/Ds and D/As in the system are
memory rnapped to the processor. The memory mapped addresses are
given in the interrupt service function whose main task is to pass the
values read from the A/Ds and write the appropriate outputs to the
D/As through four global variables declared at the beginning of the

As it stands, the C program contains only a high level description of

the processing tasks to be performed. Before this description can be
executed by the processor in real-time, some low level processol
initialization and plogram relocation information has to be provided.
As listed in Appendix B, the former information is given in the form
of a DSP32C assembly prograrn "STARTUP.S" which will actually be
executed first when the system is allowed to run. This startup
program initializes the various control registers, defines the interrupt
vector table and some other constants used by the compiler, and then
calls and passes control to the main function in the C program. Note
Yr 4/SP2/4

that the starttup program also contains a low level interrupt service
routine which becomes activated first and subsequently calls the high
level interrupt service function in the C program.
The remaining information needed for real-time execution is given in
Appendix C which lists the content of the memory map file
rrMEM0RY.MAPrr. Basically, this file shows where each program segment
will be located and whether the segment will reside in ROM or RAM.
Note that the C program will always be located in the text and data

To run the C program rrMl.Crr in real-time, the appropriate object

code has to be generated first. Assuming that the development system
has been properly installed and that all the program and memory
mapped files reside in "\8G", this can be achieved by executing the
following commands.
Firstly, execute
cd \EG
to change to the program directory. Secondly, execute the command
d3as -l STARTUP.S
to generate a relocatable object file ''STARTUP.0" for the startup
program, followed by
d3cc -S M1 . C
to generate an equivalent assembly program "Ml.S" for the C program.
Lastly, execute
d3cc -l -lap -1m -m MBII0RY.MAP -s STARTLP.0 --o M1.0lJ.I M1.S
to generate an absolute load file rrMl-.OLT" after assembling and linking
"M1.S" to "STARTIJP.0" as well as the Application and Maths Libraries
in accordance with the memory map file "MEM0RY.MAP".
Having generated the absolute load file, execute
d3nm M1.t)UT
to verify that the location for the sampling frequency variable fo is
800680 Hex. Note that you may have to use <C0NTR0L-S> to freeze
and re-start screen scrolling.
To load "M1.0LT" to the development boald to test the algorithm, first
copy Ml.t)Ur \LSI32\M1.0tlr
cd \tSI32

to copy the absolute load file to the directory

containing the Debug
Monitor program as well as invoke this software for controlling the
Yr 4/SP2/5

DSP32C hardware card.

Press <ALT-S> to get the System Menu, select the <[0AD) option and
enter "M1 .0lllil to load the absolute load file onto the board. Press
<ALT-R> to get the Run Menu, select the <RLIN> option to run the
algorithm in real-time. Apply a sinusoidal signal to channel A of the
system and observe the DAC Outputs from both channels on the

Now press <ALT-R> and select the <HALT) option to halt the execution
of the processor. When the processor has been halted, press <A[T-l{>
twice to move the pointer to the Data Window. Press <A[T-+{>, select
<DATA BASE> and enter 'r800680" to examine the content at this
memory location for the sampling frequency variable fo. To see the
displayed data in floating-point format, press <ALT-D> and toggle the
Memory Display format to <32-bit>, <FL0AT> and <SIGN>, finished by
pressing <ESCAPE>.

With the cursor pointing to 800680 Hex, press <ENTER> to change the
content at this location-to "1.6E+1rr. Run the loaded program"again
in real-time by pressing <ALT-R> and selecting <RUN>. Has the
sampling frequency been changed from 32 to 16 kHz?
Now exit the Debug Monitor by pressing <ALT-S> and selecting <QUIT>.
Does the DSP32C continue to run?

6 FFT Spectrum Analyzer

A C proglam C" for implementing a FFT based spectrum analyzer'
in real-time is listed in Appendix 4.2. Study this program and
outline the implementation methodology. Execute the command

in " 1EG" to run a batch file "M2 . BAT" containing the various
commands discussed earlier to generate the absolute load file for this
program and activate the Debug Monitor.

Load the absolute load file "M2.01.fI" onto the develooment board and
run the program. Apply a sinusoidal signal to channel A of the board
and observe the DAC Outputs from both channels on the oscilloscope.
Vary the frequency of the- sinusiod manually and note down how fne
spectrum changes. Does the spectral peak always have the same
magnitude? Why? Now vary the frequency of the sinusoid
continuously by switching on the FM modulation of the function
generator. Observe and comment on the spectrum when the
modulation is small and when it is large.
Yr 4lSP2/6

7 FIR Adaptive Filtering

-Studyfilter to enhance a
A C program "M3. C" for implementing an adaotive
narrowband signal is listed in Appendix A.3. this program and
outline the irnplementation methodology. Execute the command

in "\EGrr to run a batch file "M3.BAT" to generate the associated

absolute load file and activate the Debug Monitor. Note that the
memory location for the filter length variable is 800684 Hex and you
can change the filter length by changing the content at this location.

Run the program for a filter length of 4, 8 and 12. For each filter
length, observe and cornpare the two DAC outputs when the input is
given by a noise-free sinusoid, a sinusoid with broadband noise, and a
FM modulated sinusoid with broadband noise. Note that in the last
two situations, you have to use the passive sumlner provided to add
the outputs frorn the noise and function generators. Is the DSP32C
processor capable of implementing an adaptive filter of length 12?
Explain your observations.
Appendix A.4 shows the listing of another program, rrM4.Crr, which
performs the sarne task as "M3.C". Show that the two programs are
equivalent. Execute the command

in rr\EGrr and run thisprogram in real-time, repeating your

observations for filter lengths of 4, 8, 76, 32, 64 and 128. At which
filter leneth does the DSP32C processor become too slow to implement
the adaoiive filter? Why is there such a wide disparity between the
speed performance of the two programs? Give reasons.

8 IIR Adaptive Filtering

One problem with the FIR adaptive filter implemented is that the
SNR improvement is approximately proportional to the filter length.
Thus, a 20 dB SNR improvement would require a filter of length 100
and demands considerable computational resources to be implemented.
To overcome this problem, IIR adaptive filters can be employed.
While there are numerous difficulties with making IIR filters adaptive,
a simple and quite well behaved IIR filter can be used effectively to
enhance a narrowband signal in broadband noise.
Yr 4/SP2/7

The processing equations for such a filter are given by

y(n) = x(n) - aw(n)y(n-1) - &y(n-z) ,

e(n) = y(n) + w(n)y(n-1) + y(n-z),

o(n) = (1 - o)/4:w[n)"y(n),
py(n) = 0.99py(n-1) + 0.01y(n)2,
w(n+ 1) = w(n) - 2pe(n)y(n-1)/py(n),
| -2, w(n+r) < -2
w(n+ 1) +]2, w(n+I)>2,
I w(n+1), els ewhere
x(n) = inPut,
Y(n) = outPut after IIR filter
e(n) = prediction error,
o(n) = outPut after scaling,
w(n) = predictor weight,
py(n) = estimate for power of y(n),
a = pole magnitude,
p = misadjustment.
Essentially, the algorithm first makes useof a second order filter with
complex poles to
improve the SNR of the underlying narrowband
signal. The enhanced sinusoid is then tracked by a second order
predictor with complex zeros constrained to lie on the unit circle so
that adjustment of the weight w(n) to minimize the prediction error
power will steer these zeros to null the sinusoid. Lastly, the weight
w(n), which determine the zero positions, is used back to control the
pole positions to ensure that the poles of the IIR filter also track the
narrowband signal.

Determine and sketch the zero and pole positions of the adaptive filter
in terms of the weight w(n) and also describe the purpose of each
update equation. Develop a C program for implementing the algorithm
by copying the program "M3.Cil and the associated batch file "M3.BAT"
in "\EG" to I'M5.Cil and rrM5.BAT" in ''\W0RKSP" and editting these files.
For a : 0.95, 0.99 and p = 0.01 and 0.0002, run the program in
real-time and compare your observations with those for the FIR
enhancer. What are the advantages and disadvantages of the