Sei sulla pagina 1di 11

314 IEEE TRANSACTIONS ON NUCLEAR SCIENCE February

HIGH STABILITY NUCLEAR PULSE AMPLIFIER ANALYSIS*


Kenneth F. Hatch
Lawrence Radiation Laboratory, University of California
Livernmore, California
Summary-Developments in radiation detec - vide verygood temperature stability. This four-
tors have established aneed for maximum ampli- transistor loop will be analyzed in detail using ex-
fier stability. This paper derives detailed design pressions developed from the Z-parameters for
equations for the transistor current feedback amp- feedback loops in general. Expressions for input
lifier that are necessary for careful gain stability and output impedance, gain, and gain stability will
analyses. These expressions clearly define the be derived in terms of transistor parameters and
open loop gain limitations, and they indicate cir- passive circuit elements. Measured values of the
cuit variations which greatly improve the tem- loop's characteristics will be compared with pre-
perature stability. The parameter rd (1/hoe) of dicted values in the "Experimental Results" sec-
the grounded emitter stage is seen to require the tion.
greatest attention. It limits open loop gain, but
stability of the open loop gain can still be DEVELOPMENT OF FOUR-TRANSISTOR LOOP
achieved. The design equations are general, and
modifications to the basic loop further improve The availability of high beta transistors today
stability, reduce input and output impedances, or makes the simple two-transistor loop adequate
enable the use of high impedance input. A four- for some high stability amplifier applications.
transistor feedback loop has been tested that has However, the input and output impedances of this
0. 01 ohms input and output impedances, a gain of simple loop are relatively high, and large coup-
50, 15 ns rise time, and a temperature stability ling resistors must be used between loops to main-
of 1 0 ppm!/ 'C. tain over-all amplifier temperature stability. For
higher power gain the coupling resistors must be
INTRODUCTION reduced. However, to maintain the same tem-
perature stability, the input and output impedan-
The development of solid-state detectors ces must also be reduced. This is accomplished
capable of 0. 1% energy resolution has placed ex- by adding two more stages to the simple two-
treme stability requirements on the amplifier and transistor loop, and the resulting four-transistor
analog-to-digital converter used in pulse height loop proves to be very useful whenever the source
analysis. Also, differential counting of small and load impedances can be small. A theoretical
peaks on the steep slopes of an energy spectrum comparison of two cascaded two-transistor loops
requires maximum amplifier stability. An en-- and a single four-transistor loop is given in
vironment changing a few degrees centigrade in Figure 1. All transistors are considered ideal
ter-aperature can develop drift in the amplifier current amplifiers for this comparison, but it
that will degrade the 0. 1% resolution unless the serves to illustrate the potential improvements.
amplifier stability is very good, 0. 01%/ 'C or
better. Therefore the use of transistors, whose Other factors remaining the same, the addi-
current gains may change 1%o/ 'C each, necessi- tion of a grounded base input stage will lower the
tates the use of large amounts of negative feed- open loop input impedance beta times (see Fig. 2).
back to achieve the required stability. There is no increase in loop current gain, but Rin
becomes re instead of Pre. To lower output im-
If one chooses to use a two-transistor cur- pedance, however, one must increase the loop
rent feedback loop, he will experience difficulty gain, and this is accomplished by adding a second
in meeting the above requirements, especially un- cascade emitter follower at the output. Large
der worst-case conditions. For example, a five- power gains can thus be achieved by using small
loop amplifier with over-all temperature stability coupling resistors.
of 0. 01%/ 'C will require 20 ppm/ 'C stability per
loop. With two transistors, the open loop current Although there are four transistors in the for-
gain will change nearly 2%!/ C (if f changes 1% ward loop, the frequency response is primarily
each), and a feedback factor of 1000 is necessary dependent on the response of Q2. The common
to achieve 20 ppm/ 'C. Such feedback factors are base input stage and the additional emitter fol-
difficult to obtain with two transistors. In addi- lower do not disturb the frequency stability of the
tion, the input and output impedances of the 2- loop because they contribute little phase shift
transistor loop are troublesome and vary with compared to Q2. QZ has a very high collector
temper-tture. They should be quite small com- load impedance (Pfo 4RL'),and the high impedance
pared to the external coupling resistances to main- in turn produces a high Miller effect capacitance
tain temperature stability and linearity. which slows Q2's response well below w . Since
all transistors except Q2 have a bandpa&s equal
A feedback loop using four transistors in the at least to w the over-all open loop response
forward path has been developed that easily meets will be apprgximately the same as Q2's response,
all the requirements. The input and output im- and it is therefore stable.
pedances of this loop are very low because it has
high loop gain, and large feedback factors pro-
*
Work performed under the auspices of the U.S. Atomic Energy Co-mmission.
1965 HATCH: NUCLEAR PULSE AMPLIFIER ANALYSIS 315
A biasing problem exists because the input The true current gain of each stage depends
impedances to Q3 and Q4 are so high (typically on the current division between the bias resistors
1 meg and 10k respectively). Bias resistors less and the transistor input impedance (hie). * They
than 10 meg and 100k will shunt significant signal are designated here by prime variables, so that
currentto degrade the performance of the loop.
A system of bootstrapping has been devised that
alleviates the problem, (Fig. 3A). Figure 3B
represents the circuit when C3 and C4 are signal
A0 =
0 i 4 rd+R ' (3)
shorts, and Figure 3C shows an equivalent signal
flow circuit. Since ore is typically 1 k, 0k bias For convenience the current gain of Ql is included
resistors are now adequate. Note that R6 com- in B., and a new parameter y; is defined so that
bines with the load, while R4 and R5 combine in
parallel at the input to Q4. The complete biasing I1 = Y,; F; , (4)
is shown in Figure 4.
where
ANALYSIS OF FEEDBACK LOOPS
To aid the development of gain expressions (5)
for the four-transistor loop, Z-parameters are R, tRin
determined describing a general closed feedback The gains of Q3 and Q4 are designated S ' and
loop for which the open loop gain and input impe- 0 ', and exact expressions appear in Table II.
dance are known. See Figure 5 and Table la.
Next the voltage gain, input impedance and output
impedance of the closed loop are expressed in We will now show that an effect of rd is to
terms of the Z-parameters, RI and RL. Applying limit the attainable open loop gain,Ao. First
the formulas of Table Ia, we obtain the results note that the common base output resistance of a
given in Table Ib for closed loop voltage gain, in- transistor follows the relation:
put impedance and output impedance, in terms of
open loop current gain and other known parame- rc = 1rd- (6)
ters. It is now necessary to determine the open
loop current gain. Also note that Rxd the input impedance of Q3, is
DETERMINATION OF OPEN LOOP CURRENT Rx = $;a 4I .
(7)
GAIN, Ao
L

The forward path of the feedback loop is Now from (3), (4), (6), and (7), we obtain
shown in Figure 6A. For a first approximation,
the open loop current gain might be written as the rd y4rC I41
product of short circuit current gains of each
stage. Thus:
A0 = 34 rd+R rd+;6 1R L (8)

A0
A0 = cc1
L F 3+l)~4l
4+ (1) As 0. and @' approach infinity:
There are two factors that make equatiori (1) y rc
optimistic. First, because of the high input im- A (max) = R-I QE D. (9)
pedance of Q3 (it can be 1 megohm), the outpiut L
conductance of Q2 can shunt most of the signa
current to ground. To evaluate the gain of Q Also the feedback factor for the closed loop
equivalent circuit must therefore be valid for under this condition is limited. It is defined as:
large load resistances, and base width modul F = AO/ (closed loop current gain) = IF/Iin.
tion effects become important. Sz>cond, the I
resistors can shunt enough of the signal curr Thus
to cause a substantial reduction in gain. The
two factors also affect the temperature stabil y,rc RL y,rc R
of the loop. F(max)= R I R +RF R RF R +RF
L L I. L
To account for the shunting affect of rd, RL+RF
equivalent circuit of Figure 6B can be used.
current gain of Q2 is simply reduced by the c y, r
rent division of rd and Rx. Thus RF (10)

A0
rd
+1 ) ( 4+ I ) rd+Rx
( O; (2) Therefore maximum gain and maximum feedback
1i;5 factor are obtained when rc of Q2 is as large as
possible. Also, srmaller values of RF produce
Certain conclusions can be drawn from equation larger feedback factors.
(2) concerning the maximum attainaoble gain for
the loop, but first a few statements about bias In the typical four-transistor loop, limit (9)
resistor shunting will be given. will often be approached. For example, if
316 IEEE TRANSACTIONS ON NUCLEAR SCIENCE February
4
R - 1 k, r =10 meg, Y =1, then Ao(max) = 10, ing R_ larger, and re f4 comes into play only
but if , 6- $3 = = 100, equation (1) predicts when RL is very small. Therefore the stability of
Ao = 10 . Limiting obviously occurs. It is ex- the four-transistor loop can depend largely on rc
pedient to rewrite equation (3) in a form which whenever ;jand ¾' and/or RL is large enough to
emphasizes limit (9). make Rx >> rd. From curves of hob with tempera-
ture (2N697), we see that rc can be more stable
From (3), (4), (6); or (8), with temperature than P. Therefore increased
temperature stability can be obtained by increasirg
and 0t3' even though the gain itself does not in-
A = {I /
r

d =Y_ rc Oa crease significantly.


o Z3
brd+Rx rd+RX
DISCUSSION OF MERITS OF FOUR-
y rc k, TRANSISTOR LOOP
The use of four transistors in a feedback loop
R x (1 +r d R) (10) does not necessarily lead to instability as might
be assumed. In fact the rise time expression
Substituting Rx = '34R L/ (10) becomes represents a single R-C time constant, easily
predicted. The benefits of using four transistors
Y,; rC @s 1 !841 YU~rc appear to be well worth the extra stages, and
Ao = much higher power gain can be obtained, with
L 1(l+rd/RX) R 'L(1+rd/Rx) (11) excellent temperature stability. The open loop
gain will nearly always be limited by rd of the
Therefore the open loop gain is given by equations grounded emitter stage, but the achievable tem-
(3) and (11) where the latter emphasizes the limit perature stability will not be limited. The feed-
for large d; and 0,B. back loop is easily biased and displays excellent
saturation characteristics. As an added feature,
In a more exact analysis a term re4' is added the base circuit of Ql can be driven, enabling
to R ' to represent the series output resistance of high impedance non-inverting operation. Also
Q3 aihd Q4 (important when RL is small). * Also variations of 3- and 2-transistor loops can be em-
the detailed analysis of input impedance and cur- ployed subject to many of the same analysts and
rent gain for Q2 appears in Appendix I, where limitations expressed for the four-transistor case.
base width modulation effects are included. The
large load resistance for Q2 necessitates use of An expression for closed loop time constant
a more exact equivalent circuit, and a new para- has been developed in Appendix II. The simple
meter, n (base width modulation factor), appears result that
in the expression for R. The exact equations
are presented in Table E R.
T = cObRF(l+ R ) CobRF
TEMPERATURE STABILITY
In Table Ib, equation (4), the temperature has been found experimentally to be correct.
stability of the closed loop is expressed in terms Values of RF less than 1 k tend to produce over-
of a coefficient times the open loop gain stability shoot, but 5 to 10 ns rise time can be achieved
and a coefficient times the open loop input resis- with 1 k. Note that the effect of placing addi-
tance stability. The input resistance of Ql is tional capacity from collector to base of Q2
simply rel(or rei+ rb / 1 ) which has a stability (paralleling cob) is nearly the same mathemati-
of 1/3%/ °C at room temperature. On the other cally as placing it directly across RF. Since
hand, AA /AO may be a couple of percent and has this further slows down the response of Q2, the
a coefficient somewhat larger than ARin/Rin' frequency stability of the loop should be im-
Therefore the stability of Ao is more important. proved (see Development of Four Transistor Loop,
paragraph 3). It was found, in fact, that 5 pf
As a first guess at the temperature stability shunting cob stabilized the output loop of an amp-
of AO, assume rd of Q2 does not shunt significant lifier so well that 1000 pf capacity could be driven
signal current. Then the gain is merely the pro- directly at the output, causing only a slight change
duct of betas, and the stability is the sum of the in rise time. Other loops without this stabilizing
stabilities of each beta. However, some shunting capacity produced steady oscillations. Therefore,
usually does occur, and with the four-transistor to integrate within a loop, it is better to place the
loop the gain often depends more on rc than on capacity across the base-collector of Q2 rather
beta. The gain stability expression then becomes than across RF.
very complicated. Equations (3) and (11) are
differentiated to determine exact coefficients for The predicted output and input impedance with
each important temperature dependent parameter* four-transistor loops is usually less than 0.1 Q.
The results are tabulated in T.ble III. When limitation (9) holds, the open loop gain is
inversely proportional to R ', and we find that
It can be seen from equation (2) of Table III the output impedance, R tends to be a
that the coefficients of the terms ARx and Ard de- function of the load resistance. As R is de-
crease as Rx is increased. Also, the effect of Y, creased (thus RL'), Ro also decreases and the
on temperature stability can be reduced by mak- percentage of output impedance to load impedance
1965 HATCH: NUCLEAR PULSE AMPLIFIER ANALYSIS 317
tends to remain constant. Note that this is true by the use of Cob padding.
only while , and P. are large enough to make
Rx > rd. Also if R approaches zero, re4 of Q4 Estimation formulas for the feedback loop
will limit the open loop gain, raising Roo It has characteristics are presented in Table IV.
been possible to drive 10 ohms with less than 0. 01
ohms output impedance for excellent temperature EXPERIMENTAL RESULTS
stability. Likewise, input resistors of 100 ohms
or less can be used to derive large voltage gains. Figure 4 shows the test loop used to compare
For experimentation, the author has successfully experimental measurements with calculations.
used 2.7 ohms for RI and 5. 4k for RF to obtain a The impedances mneasured were less than ten
voltage gain of 2000 in a single loop. milliohms and ground currents became a serious
problem. It is very important to keep all bypass
A close estimate of the open loop gain can be ground currents from flowing in the signal ground
made by noting equations 1 and 2, Table II. If lead. The input impedance was determined from
Rx << rd, then AO a P, $F*. Also if Rx >> rd, the voltage developed at the summing point due to
A rC /RL '. If the product of betas is roughly a known input current. Likewise the output im-
equal to rc /RL then the actual gain will be about pedance was measured by driving a current into
half of either estimate. Also note that the re- the output, with the input grounded, and noting
quirement RX >> rd is the same as > the difference voltage developed at the output.
rC/ R L. The extremely low level voltages measured were
below the noise level, and, using an auxiliary
Concerning the temperature stability, Case I amplifier with a voltage gain of 4000, scope pic-
of Table III will amount to a summation of beta tures were taken as shown in Figure 7.
stabilities, so the closed loop gain will be EA
divided by the feedback factor. For case II, the Experimental results appear in Table V
feedback factor is limited to rc/RF when Rx >>rd, where RI and R0 are 0.01 ohms or less. The
but then the stability of Ao is improved, and we rise time of this feedback loop is approximately
don't need as much feedback. However, the sig- C bRF, (6pf x 2. 61 k max), which is about 15 ns.
nal current shunted by R; is beta dependent. Tge temperature stability was very difficult to
This effect appears in the term y, which depends measure, but a net stability of -5 ppm/ °C was
on the input impedance of Q2. Also note that the noted in a temperature run from 0°C to 50°C.
betas of Q3 and Q4 appear in the y. term because However, the resistor RF had a coefficient of
of Rx and the base width modulation effect. There- about -8 ppm/°C (measured separately), so that
fore the maximum stability of Ao occurs when @ the net active circuit stability is about 3 ppm.
and $4, and/or R ', are as large as possible, This measurement was very crude, but in any
provided that R;, is very large to minimize the case, it is safe to assume the stability is better
effect of y;,. than 1 0 ppm/ 'C.
Simple Ohm's law calculations allow the com-
plete circuit of Figure 4 to be determined. If the ACKNOWLEDGEMENTS
collector-emitter voltage of Q2 is smaller than Vce The author owes his deepest appreciation to
for Q3 and Q4, then Q2 will saturate first. Under C. A. Van DenHeuvel for his many hours of care-
this condition, Q3 and Q4 idle until Q2 comes out ful study and aid in developing this paper, and to
of saturation. Thus, the maximum output voltage R. C. Kaifer for his helpful criticism and super-
is equal to the quiescent collector-emitter voltage vision throughout the development of this project.
on Q2, and overload voltages at the input produce
very little dead time. The overload current simpiy
transfers through the grounded base stage, with REFERENCES
no gain, and charges the base of Q2. Storage time
(1) Paul E. Gray et.al, SEEC NOTES 2 -PEM:
of Q2 represents the only dead time. Physical Electronics and Circuit Models of
By returning RI in Figure 4 to ground through Transistors (John Wiley and Sons, Inc. New
,

a capacitor, the base of Ql may be driven for York,nLondon, 1963), pp 8-21, Fig. 8. 13.
high input impedance, non-inverting operation,
with a gain of Gv (R +R1)ARI). Also any two or
=

any three of the transistors can be used in a feed-


back loop, provided the grounded emitter stage is
included. These variations may be used whenever
low input or low output impedance is not needed.
The general analysis is still valid for any of these
cases, particularly the rd limitation. Various
combinations of these loops have been cascaded in
the design of a high-stability nuclear pulse ampli-
fier with a gain of 1000 and rise time of 60 ns.
The temperature stability of 0. 01 %/ °C for the
over-all amplifier is due primarily to passive
elements in the coupling networks. The output
loop and the high -gain loop are of the four-tran-
sistor variety, and excellent oscillation-free out- * The details of these derivations appear in the
put results into any length of unterminated cable Appendices of UCRL Report 7947 to be compiled.
318 IEEE TRANSACTIONS ON NUCLEAR SCIENCE February
APPENDIX I-ACCURATE EQUIVALENT OreRL g° +nOCRL + Ore
CIRCUIT
Calculation of hi, and h21 from Base Width g (RL+"IORL+Ore) + OL./rc + 1
Modulation Equivalent Circuit
6 -1-I ~~VC C The term go OR~ Lre )
(RL+ << 1 + O RL/rc
Factoring the denominator:
DEN= 1 +RL (gc + go + In0go + Orego/ RL
Derived from SEI:9 NOTES II, PEM with ele- (+,nO) go gc << Ogc; re/ RL< 1,
ments redefined. 1
SO DEN r 1 +RLObc = 1 +RLgd
LOW FREQUENCY GROUNDED EMITTER
EQUIVALENT CIRCUIT Substituting (3) for go in numerator,
k I dW
n = base width modulation factor = I
re RL/rc - RL+ OrR L +Ore
I +RL/rd
It can be calculated from:
n =
re/rd -
hl_ (2) ,[re(l+R- /rc )+nRL1
+ RL rd
The value of ro is chosen so that hjdz = rd.
ge-go
l/ro = l/rc - C/(Pre) (3)
go(6ge+'nge+G L) = ge(gc+8 GL) L
Note: the inverse of resistance r will be denoted
g. Also Multiply NUM and DEN by ,reRL. Also DEN simp-
lifies to 1 + RL/rd as above:
I / = 6. Thus go = gc- 6nbge- (35
hzl (l+-gre )
Solution Node Equations =
RL/7r-d gore << 1

V(5ge +go) - goVc = Iin


Finally:
V(ge-go) + Vc(go+lge+GL) = 0
hll = rb + :[re(l+RL/rc)
lI +RL/rd
+ IRLI
hll rb++ lin =rb
= V + 1All
1+ R LIrd
where 1A11 =go+ge+GL
Rin, as it appears in the text, is hl1.
he1 = IL/Iin
I= = GL- IA; = ge -go
APPENDIX 11-TIME CONSTANT
A = (5ge + go)(go +?ge + GL ) + go(ge -go) From Table lb, with Rin/Ao <<< RF; Rin << RF.
Sgeg0 + 6Tge+geG + go2 + lgego + goG
2
+ goge go _

in
From (3)
+ [L(RF + RL ) +R-U RF+ R)]+ RI
go = gc - n5ge * go (ge + lge + GL) t gegc -RF/ RI
- T6ge2 + 6ge' + 8geGL [(RF+RL)(+Ri) ] + 1
A = g(ge + Tgr + GL) + ge (gc + GL)
1
gO+lge+G
go(&ge+rge+G~L) + ge(gc+5GL)
L Let AO => AO/(I +S/PO).
multiply NUM and DEN by RreR L:
1965 195HATCH: NUCLEAR PULSE AMPLIFIER ANALYSIS 319

XQut _ -RF/RI TABLE Il-OPEN LOOP CURRENT GAIN


Vin [ (RF+RL)L(1+ Ri ) + 1
AOLR Case I RX < rd
-RFAQ RL/RI
1. A0 = $. P3 P4 [rd/ (RX+rd) I
fAOR L+[ ]} + SI ]/Po Case II Rx > rd
Vout RFAoRL/ RI(AoRL +[ I)
V.in
=
i+sL ]/PofAoRL+[ 11 2. A-r=
R +re4 R+r I~
L -I
Therefore,
3. 01= y
small AoR
Pi P (A R +[ IW/[ Il = PotlW+ OL1R
(RF+RL )(l+ in 4.
=R2 + Rin,.
R-;Ere2 (l+Rx/rc) + nfl ]
AR 5. Rin
PI Po 0 L
1 + /rd
_R
(R
1 ( )~~~P ...

6. 1f = base width modulation factor


kT 1 dW
Determination of PO Po = dominant Miller effect q W dVGB
time constant of Q, .

7. can be measured from n = re/rd-hre.


0
p ::P;~ obRXrd
R+
Rxrd where Rx = R
RLRF
+RF 4

8. Rx = 4 (L +re4)
0,O4, Il+RXrd1
II
AO= =@3@4@8
+Rx/ rd 9. r e4 = (re4 rh4 )

F (RX+rd) F ;4rd R
rh3 )74
;3
+fe3 tre4
o'LCobRxrdL_ Rx+rd (RF+R Rin)
lo. o" = @a'+ I ;
1
=4
- 4' + i
P1 = -IR
Cob Rx (RF+R L) (I +-in)
i1. 03 = 0aR3 / (R3 +%rea +rba) 03 R3 / (R3 +Rin3)=

$3 $4 (R +RF)
12. 04'= 4 R4 /(R4+ 4 re4 +rb4) @4R4/ (R4+Rin4)
CobO/ KRLRF(RL+RF)(1+Rl ) CobRF(l + I
=

Note: rc, rd, T1 and Cob refer to transistor


iT, Cob RF('+RIi)|
Tl=CobRF(l+
= IL)1
Q2, where rd = 1 - rc -= 1
Rin3 = :3 rez+rbJ ; Rin4 = ftjre4rb
13. Rise Time = CobRF(l+Rin/RI).
320 IEEE TRANSACTIONES ON NUCLEAR SCIENCE Feb;iruary

TABLE IIl-TEMPERATURE STABILITY TABLE IV-ESTIMATION FORMULAE

Note: The notation Ax signifies 1 ax or % Define: GI=


RF+R
R
L.
;R =
RF+R L
change in x per 'C. L L RF+R L
Case I Rx < rd AO = ;e I4 Lrd/(RX+rd)l
I
F = AO/GI = feedback factor
1. AAo APS'+AS +A$4'+[Rx/(Rx+rd)][Ard-ARx] 1. Case 1 Rx << rd AO z ,B,~,, 4
Case II Rx>rd 2. Case 2 Rx >> rd Ao P: rc/RL = AO(max)
AO = [yd rc/ (RL)+re4,)][Rx/ (Rx+rd)1 Note: if ,2 35z4 = rc/R; A0 = 1/2 0
3. F(max) = rc/RF
2. AAo =Arc+Ay2+[rd/(Rx+rd)][ARx-Ard1
4. R A, Rin/ F
_ [re4/(RL +re4 )]Are 4
5. RoszRF/AO
3. AO,' ACL,+ AO - [Rin2 / (R;e +Rin2)]ARin2 6. Case 1. AAO E Li
A5
)'[4 53 /re
4 AO, R. A Case 2. AAO = Arc minimum
)'rbL
-R~ ( -Arb +re Are)
7. Rise Time ; CobRF (Cob refers to Q.).

5. AOf' = same as Aki' with subscripts changed TABLE V-EXPERIMENTAL RESULTS


to 4.
Transistor Data re 5 rd
6. ARx AOj+AOS +[re /(RL +re')]Are
Q, 26Q - -
7 Ar e4' Ar e4
Q.ke 26 Q 500 62K
8. AyB = Aa1 - [Rin; / (R;d +Rin ) ] ARin2 Qd 25 Q 135 -
rb,;e rd rb. QZI 1, 1 Q 150 -
9. ARin =
Kin Arb~ +D
rdL in~ From CKT. Fig. 4 4/ = 150, 5' = 92, 02. = 150
lr cRx Rx = 10(150)(92) = 138K
+ +Rin Ar
Are
(rd+Rx)
A0 = (150)(92)(150) 62
= 640,000
+
re z~ r C +Rx
ARx R1 = 26 10(624)10 - . 01060
Rin rd+Rx
(1
R - 2610(1+26/51. 1)+51 1 = 00625Q
RX F B3rb +re -. 640, 000 0622
+ +rd+Rx (1-2
x
Rin )Arc From Figure 7
1= V~/ Iin~
-1/Zcm(Smv/cm) -
+' Rx
F kAlrd+re2+rb,, 1 A'rq
RI ; V1 = 4000 (Aux Amii/p) = 0. 625
6 v
V

- 2
Vo = 0. 2v into 10; =n 76. 6 pa
ESTIMATES
- ~t20ma R1 261 .008150
A02' A 0;ASARRx
At
+4; 3 ;AtA4
R;_, AtR
a _R in4i R AR
R 4
+RnAS 4 ° - Iin AV0 t(0. 3cm)(. 2v/cm)
=( 4000
_ 0 m
4
o
AVt .O05mv
.015
Rim tout = 20 ma, Ro = = . 0075%2
R;d+Rin Calculated Measured
R1 .011 Q . 008 Q
RO .006 Q . 0075 Q
1965 HATCH: NUCLEAR PULSE AMPLIFIER ANALYSIS 321

No+e :
RI) R0 o o+io n &d
RrR,l R23
II iR I
4ro, e(tAO¶ °V%s
IK F R, Filre Lb
So
1K RK

2- Casco.8e Loops

Troxvsis+t0r Do+ot
re= 1OfL
,B= 100
ibt = 1 %/°-c Four Tronsis+or Loop
A
2 -XSTR
4- XSTR
Loop I Loop 11

Vaoltie Go,n Gv 1 3.3 3.3 20


CuLrevd Ga.,i GT 3.3 2 6.6 loo
Inpu+ Ivmpedtrice R, .433 30 .001
RI 1K 3t0 50
°/, oJ E + erval R$ _,043 o.1 ,0oo
-v

°'J+Puf Im-pedance RO .30 .533 oo, 1'


T-wo Transis+or Loop RL 300 500 l
0/1 of External FL o.l .107 ._1_ _

Feedk6ix k Factor F 2300 3300 10,000


T,.peroture- Stal;1;+/ 9 ppwn 6 pmvv 15' PP- 3 rpm

Fig. 1-Comparison of two-transistor and four-transistor loops.

LIow IA ?LbAOO4CL .L
$UVAMAiO40. P*%%ar
Fe GRcbINt4

YJx (-T2)wt *Cob$2 RX cob (S., p


4::O-MLLV-% lLV*.C- TTLqvA
2 to)
~OWCTRA%15TR
~~~~+C

rrrDk.KLol

%DOITICO> Of 6QVU4 0D 5A%VS


58- NC%rg

Fig. 2-Development of four-transistor loop.


322 IEEE TRANSACTIONS ON NUCLEAR SCIENCE February

c~~~~~~~~~p~~~~~~~~~y~~c

M4Xt,
__________ F\.3- CO, 24 FIA. 3C- ECAU\VALLET C'RCxWT e
____________; Q.C. %'zzon

Fig. 3-Multiple bootstrapping.

MM5s2? M2#484~ At!?Z~ /z$f


.Lk
aL.LL. w I -2

Fig. 4-Typical biasing for four-transistor loop.


1965 HATCH: NUCLEAR PULSE AMPLIEIER ANALYSIS 323

cho.ro.cJeri za.tion of forwo.rRA R; n)S


, z
loop

VI V2
t11-r
V"1'
R Vz, Vovr R. RF
Vi. 1 ??
h I

R,= 4i,

Fig. 5-Analysis of feedback loops.

6.2 . 1-2/ii,
M.1'611 01

_1177

F~~-Xwt/31o
L

(a)

1-i1,- .F3 r aR
q
-
Q- -- - - - Fig. 7-Input and output impedance measurements.
Upper Trace-Input impedance-appears
II~~~~~~~~ as a displacement in the noise. Vert. = 0.005
v/cm; Horiz. 100 ns/cm; I = 20 ma. Lower
-

Trace-Output impedance-appears as a dis-


placement of the noise at the leading edge of
L__ ____Jb the signal. The trailing edge exhibits a large
displacement (off the scope screen) because
of the discharge of capacitors (having been
(b) charged 20 ma for 100 ns). Vert. = 0.2 v/cm;
Horiz. -100 ns/cm; I = 20 ma.
An auxiliary amplifier, gain = 4000, was
Fig. 6-a) Forward path of four-transistor loop. used in both traces. Polarity of pulses is
b) Simplified equivalent of Q2. (R2 very large). inverted from actual case.
324 IEEE TRANSACTIONS ON NUCLEAR SCIENCE February

To-b6e IC. - Z-pa.rcame+er Forrv utI e_


= R;nA _'1Sin -Ao RF R;n-+ RF
4 A R121 + A0 I + Ao
7 =

VoJT =
RL -iR
Viv% (2 , Rs Z22 RL) -
)( \>i

ouL+pv 1,pceJavlce
RI=pU Ir Zeorz c
R, i I ZZ 2+RL ions=2z-

ToxblAe 16 C I osed Loop Ecitxc. ;ons

.su 6s+ +u4%n3 Z-parcomrners if+o formuAae Fl Iasge. A,,

1. =
A. R- RLA/A R _ RF
yin R:r [I+ AO ( 1+ RI: R;in] R;n (Rp+ Rj)

Ri
RF + R, R-nRF+ RL
2. RI = L(1+ AO) + RF + Rin

R1RF* Rin RF + Rin R: RF ( I+ in) +R;n


3.' i° Rs~~(I + Ao) + R;n, -T
Ao
p,ffe,re.n+io+e 1. for 4empe-rai+ure s4aL;kl;y
A. if: = 6

G AA0AA RF (R R

A o RF ( R;nr RFL R RL11


R 1.v A, F RLR JJ

Potrebbero piacerti anche