Spectrum distortion from amplifier overloads in proton-recoil proportional counting

NUCLEAR INSTRUMENTS AND METHODS 71 (I969) I 5 7 - I 6 2 ;

© NORTH-HOLLAND PUBLISHING CO.

S P E C T R U M D I S T O R T I O N F R O M AMPLIFIER O V E R L O A D S IN P R O T O N - R E C O I L P R O P O R T I O N A L C O U N T I N G * J. M. LARSON and J. E. POWELL Reactor Physics Division, Argonne National Laboratory, Idaho Falls, ldaho 83401, U.S.A.

Received 30 December 1968 The problem of spectrum distortion from amplifier overloads in proton-recoil proportional counting has been studied. These ,;tudies show that the distortion is caused by amplifier paralysis which results from severe overloading of the ionization amplifiers. An amplifier system is reported that uses pole-zero compen,;ation techniques and active baseline restoring to extend the overload tolerance of the ionization amplifiers, thereby greatly

increasing the count rate capability of the spectrometer. The compensated system is tested by measuring proton spectra at various counting rates in a plutonium fuelled fast reactor (ZPR-3). Spectra taken with the compensated amplifier system and with a conventional double differentiated system are compared. An approximate three times improvement in overload capabilities is obtained with the compensated system.

I. Introduction

overload effects since they amplify only the amount of charge out of the detector collected in the first 50 to 100 nsec. However, the overload distortion can occur in the specific ionization spectrum since that spectrum is determined by a division of the rate-of-rise by the ionization. Count rate restrictions can severely limit the usefulness and versatility of the spectrometer for in-core measurements. For example, in a plutonium fueled reactor such as Zero Power Reactor-3 (ZPR-3) 6) spectrum measurements must be scheduled prior to experiments requiring higher power operation to minimize residual gamma-ray background. Special reactor loadings are required because the amount of fuel in the reactor must be reduced due to the relatively high neutron and gamma-ray background from the spontaneous fissioning of z4°Pu and 241pu. These reductions typically leave the reactor 10% to 15% subcritical, and the resulting neutron spectrum is that of a subcritical assembly with 50% to 70% of the fuel of the reactor of interest. Selection of a maximum counting rate is also a problem. The maximum count rate capability of the analyzer is not a good guide because, depending upon the hardness of the neutron spectrum, overload distortion can occur at a relatively low count rate.

Proton-recoil proportional counting has proven to lye a very useful method for measuring fast neutron spectra in zero power reactorsl-5). Inherent in the method is the concept of pulse shape discrimination tbr distinguishing between events caused by neutrons and background gamma rays. In the typical measurement, pulses from the proportional counter are sent to a low-noise, wide band preamplifier and then to separate sets of amplifiers which determine respectively ionization (pulse height) and rate-of-rise information. The rate-of-rise data is divided by the corresponding ionization to form the specific ionization 1) which provides the necessary discrimination between neutron and gamma ray events. The ionization and specific ionization data are then stored in a two-parameter analyzer or computer. With the pulse shape discrimination system the spectrometer can be used for measurements over the neutron energy range of 1 keV to 2 or 3 MeV. One of the problems that is encountered in making spectrum measurements in a fast critical assembly is distortion of the measured proton-recoil spectrum by amplifier overloads. The distortion occurs during measurements in the energy region of 1 keV to 10 keV where the counter multiplication is high. Protons with energies in the MeV range create overload pulses in the ionization amplifiers that results in severe amplifier paralysis. Consequently, count rate restrictions must be placed upon the spectrometer since pulses that occur during the time that the amplifiers are paralyzed are either lost or distorted. The rate-of-rise amplifiers are essentially free of * Work produced under the auspicies of the U.S. Atomic Energy Commission.

2. Improved amplifier system The spectrum distortion discussed above can be reduced, if the overload capabilities of the ionization amplifiers are improved. An amplifier design that has resulted in a significant improvement in overload capability is shown in the block diagram of fig. 1. A low-noise charge-sensitive preamplifier, a pole-zero compensated linear amplifier, and an active baseline restorer are used in this design.

157

158

J. M. L A R S O N A N D J. E. P O W E L L

C~ IP

I t

LINEAR A ~ R E R

I

DETECTOR /

/

~"

'

,

:"

COR.ECT~ON"

F~{.J

~

k

J ~

OUTPUT

q

.j

Fig. 1. Block diagram for compensated electronic system.

The preamplifier is direct coupled and has a single differentiating time constant r of 500/~sec (r = RfCf). The long differentiating time constant of the preamplifier is reduced to 1.5 ~sec by the pole-zero compensation network at the input of the linear amplifier. The pulses out of the pole-zero compensation network are amplified and integrated (single RC integration time constant of 1.5 psec) by the three operational amplifier gain stages of the linear amplifier. All the gain stages within the linear amplifier are coupled by very large capacitors (coupling time constants between stages are greater than 0.1 sec) so that baseline undershoot, introduced by the coupling networks is reduced to negligible levels. The output of the linear amplifier drives an active baseline restorer. This restorer operates on the amplified diode principle 8) and is used to remove the slow component of the collected charge from the proportional counter that appears at the output of the shaping amplifiers when large overloads occur. The need for the restoring circuit can be better explained through the use of fig. 2. Fig. 2A illustrates a typical profile of the charge collected from the proportional counter after an ionizing event. The collected charge from the proportional counter is made up of a slow rising component ,6, superimposed on the fast rising step c~. The fast rising step is shaped by the shaping networks of the linear amplifier resulting in the unipolar pulse shown in fig. 2B. The slow rising component is greatly attenuated by the shaping networks of the linear amplifier and thus contributes negligibly to the pulse height and profile of the primary pulse of fig. 2B. Under conditions of overload, the primary pulse saturates, however, the slow component grows as the overload increases, thus giving rise to a pedestal following the overload, as shown in fig. 2C. (The dotted portion of the profile of fig. 2C and 2D repre-

sents the pulse shape that would be obtained on an overload if there was no slow component in the collected charge). The pedestal lasts for the duration of the collection time of the slow component and may have typical lengths of 0.2 to 1 msec depending upon

~JTION (A)

O~.._I'e( O

.50 /IE

L

SHAPED PULSE PROFI LE

(B)

0

(C)

50

I o

"-~L io

--! (Ol

0

(El

5OO usec

I0

50o

OVERLOAD PROFILE 5'O

,~,,--------~.

500

RZ. CORRECTI ON ADJUSTED FOR UNDERSHOOT ..........

4;-

. . . . . ~'g-------

RESTORED OVERLOAD 5'0

"~

Fig. 2. Pulse profiles.

5bo

S P E C T R U M D I S T O R T I O N FROM A M P L I F I E R O V E R L O A D S

the characteristics of the proportional counter used. This pedestal may be a source of severe distortion, when measuring in spectra that generate numerous overloads, as those pulses occurring in coincidence with the pedestal will be distorted. The pedestal is removed by adjusting the pole-zero compensation for a negative undershoot that just exceeds the amplitude of the pedestal. This adjustment results in the pulse of fig. 2D. This pulse may be restored by the baseline restorer and after restoration appears as shown in fig. 2E. The restorer must have a fast restoring rate to minimize the duration of the negative undershoot of the restored pulse (refer to fig. 2E). In this application, the fast restoring rate was obtained by using unbalanced currents in the restoring diodes as shown in fig. 1. One of the disadvantages of restoring is that it increases the effective noise of the systemt°). However, in this application, this increase was small, and amounted to only 0.25% of the overall spectrometer resolution. In order to take full advantage of the overload characteristics of the pole-zero compensated linear amplifier it is necessary to use a preamplifier having a large output signal capability. This capability is required to prevent the preamplifier from saturating on large overload pulses, as saturation results in increased dead time in the linear amplifiers. The preamplifier design described by Larson ~1) has a 20-V output swing, in addition to low noise and a 13 nsec ri,;e time, and for that reason was selected for use in this application. 3. Tests in ZPR-3

The pole-zero compensated amplifiers were tested by measuring ionization and specific ionization spectra in the core of ZPR-3 for count rates varying from

4000 to 15 000 counts/sec. For comparison, data were taken under identical conditions with an uncompensated amplifier system which consisted of an accoupled voltage sensitive preamplifier 9) and a commercial double differentiated linear amplifier (Sturrup Model 1415). These amplifiers have been previously used for neutron spectrum measurements 5) in ZPR-3. The data were taken with the electronics system shown in fig. 3. Signals from the detector - a hydrogen filled proportional c o u n t e r - w e r e sent to either a voltage or charge sensitive preamplifier. From here the signal was split and sent to a fast channel which determined the rate-of-rise of the pulse and a slow channel which determined the ionization or pulse height of the event. The fast channel consisted of a differentiator, fast amplifier, stretcher and inverter. In the slow channel the Sturrup amplifier was used with the voltage sensitive preamplifier to form the uncompensated system and the pole-zero compensated amplifier was used with the charge-sensitive preamplifier and baseline restorer to form the compensated system. Signals out of the fast and slow channels were sent to a dual 1024 channel ADC for digitizing and to an on-line computer for processing and storing. The division of the rate-of-rise by the ionization to form the specific ionization was done in the computer. 4. Results

Specific ionization spectra taken with the two systems over the proton energy range of 0.5 keV to 3.0keV for counting rates of 4000, 10000 and 15 000 counts/sec are shown in figs. 4 and 5. The reactor was respectively 2%, 0.6% and 0.2% subcritical for these count rates. With the compensated system the gamma-ray and proton peaks are clearly resolved at each count rate and the gamma-ray

~--~DIFFERENTIATOR~

I

,

STRETCHER

DOUBLE DIFFERENTIATED

AMPLIFIER

DUAL ADC CHARGE

POLE ZERO

SENSITIVE PREAMP

ZOMPENSATE[ AMPLIFIER

159

BASE

LINE RESTORER

Fig. 3. Comparative electronic systems for proton-recoil spectrometer.

160

J. M. L A R S O N

A N D J. E. P O W E L L

background can be easily subtracted out of the proton data. However, the uncompensated system showed noticeable distortion at 5000counts/see, and at 10000 and 15000counts/see there was no longer discrimination between proton and gamma-ray events. This smearing of the specific ionization spectrum was caused by overload effects in the ionization amplifiers since the same rate-of-rise electronics were used with the two ionization amplifier systems. To study the effects of overload distortion on the proton-recoil spectrum two-parameter data were taken at 4000, 10 000 and 15 000 counts/see with both amplifier systems. The computer was used to subtract

4,000

6000-

COUNTS/SECOND

out the gamma-ray background. A comparison of the 4000 counts/see data is shown in fig. 6 for the proton energy range of 0.66 keV to 3.0 keV, the counting time for the two sets of data being the same. In the energy region between 0.66 keV and 1.5 keV the spectra taken with the two systems differ by several percent indicating some distortion is present in the uncompensated amplifier data. We wanted to compare the two systems at count rates lower than 4000 counts/see to show that they produced comparable data but this could not be done in the 0.66 keV to 3.0 keV energy region without substantially increasing gamma to proton ratios and thereby introducing

6000.

5000-

5000-

4000COUNTS 3000-

COUNTS 3000-

2000-

2000-

I000-

I000-

4000-

-// 0

,

,

,

50

40

50

, -

60

,

,

i

70

80

90

,

I00

%~-7"--° II0 120

130

0

3o ~o ~

6OOO"

6OO0j

5000-

5000-

4000COUNTS 3000-

4000 COUNTS 3000.

2000-

2000-

1000-

I000-

:50 410 510

0

6% 7b B~ 9'o ,~o .'o ,5o 4o

15DO0

COUNTS/SECOND

0

130

30 go so 6o 7o Bb ;o ,oo ,,o ,~o ,~o CHANNEL NUMBER 15,000 COUNTS/SECOND

6OOO"

5000-

5000-

4000COUNTS 3000-

COUNTS 3000-

20 00-

2000

4000"

1000!

I000-

o

120

I0.000 COUNTS / SECOND

CHANNEL NUMBER

6000-

6% 75 ~o ~o ,6o,,o CHANNEL NUMBER

CHANNEL NUMBER

#

~o 4'o ~

do 7'o ~o ~o ,6o.'o ,~o 40 CHANNEL NUMBER

Fig. 4. Specific ionization spectra for compensated amplifier system at 4000, 10 000 a n d 15 000 counts/sec.

---//"

o

I

i

i

~

3o 40 5o 60 70 BO ~o ,60,,'0 ,~0 ,~o CHANNEL

NUMBER

Fig. 5. Specific ionization spectra for uncompensated amplifier system at 4000, 10 000 a n d 15 000 counts/see.

SPECTRUM

DISTORTION

FROM AMPLIFIER

uncertainties in the data. However, measurements were made at 4000 counts/sec in a higher energy region where a minimum amount of overloads were present and in this region the two amplifier systems produced comparable data. Comparisons of proton-recoil spectra also could not be made at 10 000 or 15 000 counts/sec because the specific ionization spectra, taken with the uncompensated amplifiers, were smeared and the gamma ray background could not be subtracted out of the two parameter data (fig. 5). In order to study the performance of the pole-zero

161

OVERLOADS

compensated amplifiers at high count rates, the proton recoil data taken at 4000 counts/see were compared with the 10000 and 15 000 counts per second data. The 10 000 counts/sec data compared within statistics but at 15 000 counts/sec the two differed by a few per cent. This effect is shown in fig. 7 where normalization is over the energy range of the measurement. These differences could be explained by a change in the reactor spectrum in going from 0.6 % (10 000 counts/sec) to 0.2% (15 000 counts/sec) subcritical. However, it is more probable that these differences are due to pulse

7000-

o

+

UNCOMPENSATED AMPLtFIER SYSTEM A T 4 0 0 0

o

COMPENSATED AMPLIFIER SYSTEM AT 4 0 0 0

C/S C/S

°oo GO00-

°o oOOo ODeo

5000 -

o

+4-

oOooo

Jr4-.p{.+4-+++ COUNTS

°°°oO 4-+ °°%00^ + ' t-++ + V°o° o ++ +4.~, +~_+ ? % o 1-+ -~r++ 4- +°?_o.b o

4000-

oo

=

3000-

0

636

•8T2

,,08

,344 .'

l.SSO

8;6

2

.b~2

. --.

' 2.288

- , . - D o T t y , _,,p~,,,..~

2.524

2i60

2 9 9' 6

ENERGY, keY

Fig. 6. Comparison of proton-recoilspectra taken with compensated and uncompensated amplifier systems.

+ °o ° 6000-

+oo + oo

+'l";e~°o 5000-

+

COMPENSATED AMPLIFIER SYSTEM AT 4 0 0 0

o

COMPENSATED AMPLIFIER

C/S

SYSTEM AT 1 5 0 0 0 C / S

o~,

4000 COUNTS

3000 -

o Do

2000 -

.636

.872

1.108

1.344

1.580

.816 2.052 ENERGY, keY

2.288

2.524

2,760

2.996

Fig. 7. Comparison of proton-recoil spectra taken with the compensated amplifier system at 4000 and 15 000 counts/sec.

162

J. M. LAR, SON AND J. E. POWELL

pile-up since no change in the p r o t o n spectrum was observed in going f r o m 2 % to 0.6% subcritical.

5. Conclusion The m e a s u r e m e n t s in Z P R - 3 show that d i s t o r t i o n o f the p r o t o n - r e c o i l spectrum can occur at relatively low counting rates. This d i s t o r t i o n occurs p r i m a r i l y in the energy region below 10 keV where o v e r l o a d s are large a n d is a result o f amplifier paralysis. Both the specific i o n i z a t i o n d a t a and the i o n i z a t i o n d a t a are d i s t o r t e d ; consequently the final p r o t o n s p e c t r u m ( g a m m a - r a y b a c k g r o u n d subtracted) is d o u b l y distorted, since the l o c a t i o n o f the event in the i o n i z a t i o n axis o f the t w o - p a r a m e t e r matrix, as well as the n u m b e r o f events at any given i o n i z a t i o n (specific ionization axis), is uncertain. A significant i m p r o v e m e n t o f the o v e r l o a d capabilities o f the ionization amplifiers has resulted f r o m a design that utilizes single R C differentiation a n d pole-zero c o m p e n s a t i o n in conjunction with active baseline restoring. This technique i m p r o v e s the overload capabilities o f the ionization amplifiers by eliminating lengthy amplifier paralysis due to baseline u n d e r s h o o t and pulse pedestal, following overloads, due to the slow c o m p o n e n t o f collected charge f r o m the p r o p o r t i o n a l counter. A t present the r e p o r t e d s p e c t r o m e t e r will p r o d u c e

u n d i s t o r t e d spectra up to 10 000 counts/sec where p e a k and tail pile up effects begin to be noticeable in the spectrum. By the inclusion o f pile-up rejection, the system should be c a p a b l e o f c o u n t rates a p p r o a c h ing 30 000 counts/sec. The a u t h o r s would like to t h a n k Mr. J. M a t t h e w s for help in o b t a i n i n g the d a t a a n d drafting the figures.

References 1) E. F. Bennett, Nucl. Sci. Eng. 27 (1967) 16.. 2) E. F. Bennett, Nucl. Sci. Eng. 27 (1967) 28. a) H. Bluhm, G. Fieg, M. Lalovic, D. Stegemann, E. Wattecamps and H. Werle, Trans. Am. Nucl. Soc. 10 (1967) 576. 4) p. W. Benjamin, W. J. Patterson and J. W. Weale, Trans. Am. Nucl. Soc. 10 (1967) 576. ~) J. E. Powell, Trans. Am. Nucl. Soc. 11 (1968) 217. 6) R. O. Brittan, B. Cerutti, H. V. Lichtenberger, J. K. Long, R.L. McVean, M. Novick, R. Rice and F.W. Thalgott, Hazard evaluation report on the fast reactor Zero Power experiment ZPR-3, ANL-6408. 7) C.A. Nowlin and J.L. Blankenship, Rev. Sci. Instr. 36 (1965) 1830. 8) R. L. Chase and L. R. Poulo, IEEE Trans. Am. Nucl. Soc. NS-14, no. 1 (1967) 83. 9) E. F. Bennett, Nucl. Instr. and Meth. 48 (1967) 170. 10) V. Radeka, Rev. Sci. Instr. 38 (1967) 1397. it) j. M. Larson, A wide band charge sensitive preamplifier for proton recoil proportional counting, AN L-7517.