A:
i
1
.
.
*** Wwwww
.::..
bomo
..
:
..
I OF I
ORNL P
2639
K
.
.
:.
-
-
--
I
11
.
c
:
plus
.
;
.
✓
$
1:1
TIEFEEEE
*.
....,
.
..
.
le
.....
:
*********...-- -
les
1.25 1.1.4 1.16
#
MICROCOPY RESOLUTION TEST CHART
NATIONAL BUREAU OF STANDARDS - 1963
.
.
NOV 2 9 1966
ORNLP - 2639
CONE-661020
RELEASED FOR ANNOUNCEMENT
IN NUCLEAR SCIENCE ASSIRACTS
MC $ 7.00: 3:5
A NEW APPROACH TO DIRECT CURRENT INTEGRATION*
F. M. Class, C. C. Courtney, E. J. Kennedy, H. N. Wilson
Oak Ridge National Laboratory
Oak Ridge, Tennessee

MASTER
24
H.
TE
7
.
.
Abstract
W
:::
...
operating principle common to all four circuits
is that the voltage across a storage capacitor
is monitored by an electiometer type operational
amplifier, the output of which triggers a dis.
criminator at a given voltage level. The discrim
inator generates a reset pulse which removes a
fixed charge from the integrating capacitor and
supplies a signal to the scaler.
Y
A new all-solid state direct-current inte.
grator integraces currents as loly as 10u5damp.
ere with an accuracy of 1%. The operating
principle is similar to that of more conventional
current integrators in that the oltage across a
storage capacitor in the input circuit is moni-
tored by an electrometer type operational ampli.
fier, the output of which triggers a discrimina-
tor at a given voltage level. The discriminator
generates a reset pulse which removes a fixed
charge from the integrating capacitor. This is.
where the similarity ends. The diode pump cir.
cuit, which has long been recognized as ar ideal
circuit for removing precise quantities of charge
from the integrating capacitor, has been re-
placed by complementary silicon planar transis-
tors that operate in the inverted-mode and serve
as current switches. These fiansistors normally
have both their base-emitter and base-collector
. The technique for removing charge from
the storage capacitor is different for each of the
four integrators illustrated. The integrator in
Fig. 1A employs the time-proven diode pump.
The charge removed is Q = CV, where C is the
coupling capacitor Cc, and V is the amplitude of
the reset pulse less the forward voltage drop
across the diode. Accuracy is therefore inde-
pendent of the amplifier gain, che trigger level
of the discriminator, and the temperature coef-
ficient of the integrating capacitor Ci. Two good
examples of instruments of this type are the
ORNL vacuum-tube iniegrator, model Q-1259,
which has given excellent service since 1950, .
and the mose recent solid-state version by
Landis and Goulding. The most serious limita..
tion of these instruments is diode leakage which
limits their sensitivity to about 10-9 ampere.
V
:
-Y Y
YA
!
.
.
.
.
..
switches with leakage currents of about 10-13
ampere. Reset pulses switch these transistors
from the reverse-biased state to the active state
for a precise time during which a constant cur.
reri flows to or from the integrating capacitor.
Therefore, each reset pulse removes from the
integrating capacitor a precise charge that is
the product of the current and the period of the
active state. By the cmployment of sufficient
gail. in the operational ariplifier to 'naintain
small voltage excursions at the input, the total
effective leakage may be kept as low as 10-14
ampere. A scaler output provides an indication
of the integrated charge, and a built-in duty cycle
mr.eter calibrated in counts per second provides
an indication of the instantaneous current.
.
Si
A
.
.
.
The principle illustrated in Fig. 13 is
employed in an integrator designed by Brenner.2
This technique involves charging a capacitor Cs
to a given reference voltage of opposite polarity
to the charge on the integrating capacitor and
then transferring this charge to the integrating
capacitor Ci by means of a higir-quality relay. :
The charge removed from Ci is dependent only
on the value of Cs and the reference voltage, pro-
vided. Ci >> Cs as it normally is. The leakage
can be made quite small in this instrument; how.
ever, the reset frequency is limited by the relay
to a few hundred hertz.
PA
.
.
.
..
Introduction
NE
.
."
.
.
The ever-increasing number of accelera.
tors in this country and abroad has brought about
a widespread interest in the development of
direct-current integrators. As experiments in.
volving current measurements have become more
sophisticated, the demand has become more
pressing for integrators that have better accuracy,
more sensitivity, and a greater range than exist.
ing instruments. In general, instrument develop.
ment has kept stride with the development of the
critical circuit components used in such instru.
ments.
.
"
.
.
,:...'
Nec
.
NE
Figure I illustrates four operating prin-
ciples that have been successfully employed in
current integrator circuits. A distinguishing
The oldest of the four techniques illus.
trated in Fig. I is shown in Fig. 1C. This tech.
nique removes the charge from the integrating
capacitor by shorting it with a relay or electronic
switch. Since the charge is completely removed
from the integrating capacitor with each relay
closure, a good stable integrating capacitor Ci.
a high-gain operational amplifier, and good
trigger stability in the discriminator are essen-
tial for good accuracy. The linearity of an instru
ment incorporating this technique is poor, because
the input is shorted while the charge is being re.
moved from the storage capacitor. The only sig.
nificant feature of this integrator is that the
leakage current can be made quite low if a high-
quality relay is used.
LEGAL NOTICE
This report was prepared as an account of Government sponsored work. Neither the United
States, nor the Commission, nor any person acting on behalf of the Commission:
A. Makes any warranty or representation, expressed or implied, with respect to the accu-
racy, completeness, or usefulness of the information contained in this report, or inat the use
of any information, apparatus, method, or process disclosed in this report may not infringo
privately owned righto; or
B. Assumes any ilabilities with respect to the use of, or for damages resulting from the
use of any Information, apparatus, method, or process disclosed in this report.
As used in tho above, "person acting on behalf of the Commission" Includes any am-
ployee or contractor of the Commission, or employee of such contractor, to the extent that
such employee or contractor of the Commission, or employeo of such contractor prepares,
disseminates, or provides access to, any information pursuant to his employment or contract
with the Commission, or his employment with such contractor,
.
.
-
.
POUR
!
!
.
.
.
.
.
"
Research sponsored by the U.S. Atomic
Energy Commission under contract with the
Union Carbide Corporation.
.
.
.
AS
The principle of operation for the circuit
shown in Fig. iD is that whan the integrating
capacitor Ci charges sufficiently to trigger the
upper-level discriminator, the current gener.
ator is triggered on, which quickly discharges
Ci to the voltage level at which the lower-level
discriminator triggers. The lower level dis.
criminator then turns the current generator off.
The charge removed with each reset cycle is
Q = CIEZ - Ei), where E2 and El are the voltage
levels at which the two discriminators trigger.
10-11 ampere for silicon diodes reversed biased
at l volt is about the best that can be achieved
by the state of the art. Although a Icakage cur
rent of this magnitude is good enough for beam
current measurements, it precludes the use of
the dinde pump circuit in instruments intended
for use with ion chambers. Therefore, the possi.:
bility of using silicon planar bipolar transistors
as current switches was investigated.
The accuracy of this circuit, for currents
of sufficient magnitude to be negligibly affected
by diode leakage, is limited only by the accuracy
with which each discriminator can sense a vol.
tage level. This statement requires qualification:
the current being integrated should also be
smaller by at least two orders of magnitude than
a typicai reversed-biased collector-to-base
leakage current ICBO as low a3 1 x 10-10 ampere
at a collector-to-base voltage VCB of 20 to 30
volts. It was determined experimentally that
four types of bipolar transistors had ICBO values
ranging from 1 x 10-13 to 4 x 10-12 ampere,
when biased at a VCB of:0.1 volt. However, the
leakage current can be made much lower than the
ICBO ii the transistor is operated in the "inverted
mode," i.e., with the collector and emitter leads
interchanged. In the inverted-mode operation the
leakage current through the emitter terminal is,
from the equations of Ebers and Moll, 3
be very short for the highest current measured
if the instrument is to be linear. In addition,
the operational amplifier must have an extremely
high gain-bandwidth product.
CEFEI
IE MICRO 5
LIT
where hFEN and hFEI are the normal and inverted.
current gains respectively, q is the charge of the
electror, V is the junction bias, kis the Boltz.
mann constant, and T is the absolute terrperature
in 'K. In this relationship it is assumed that.
(1) the only voltage drops are at the junction, and
the injected current de: sities are low; (2) the
"Ea:ly" space-charge layer widening effect is
negligible; (3) he emitter and collector junctions,
individually, have a V. I characterisiic of the
form
• The advantages and disadvantages of the
different schemes illustrated in Fig. I can be
summed up as follows. The diode pump system
is fast, and excellent accuracy and long-term
stability can be achieved with minimum critical
circuitry. The range easily achieved with good
accuracy and perfect linearity is from 10-9 to
5 % 10•4 ampere. However, the diode pump
circuit is limited in sensitivity by the leakage of
the diodes. The voltage-sensing and current-
switching approach shown in Fig. ID is also
fast and has the same range limitations as the
diode funip circuit, but the circuitry required
to achieve the same degree of accuracy that is
casily obtainable with the pump circuit is com.
paratively complicated and sophisticated. Tha
Brenner circuit is very slow, but it can have.
excellent leakage characteristics if constructed
from quality components. Linearity is perfect
up to the cycling limitations of the reed relay.
The shorting method illustrated in Fig. IC is
slow and nonlinear. No one instrument described
above can fill all the needs of an experimenter
whose experiments include the integration of
very small currents from ion chambers as well
as beam currents. An ideal instrument would
have the speed and accuracy obtainable with the
diode pump circuit, current leakage as low as
the best reed relay, and be capable of dumping
charges as small as 10-12 coulomb.
I=IS
and (4) no drift fields exist in the base except ut
the junctions. Even though the equations by
Ebers and Moll were based on the be javior of
junction transistors, Verster4 verified that the
agreement between theory and experiment is very
good for silicon planar transistors. Although the
experimental data in Table I for twelve transis-
tors picked at random does not agree very well
with theory, it does point out the advantage of.
is.verted-mode switching. It is possible that
some of the disagreement between theory and
experiment is due to header and encapsulation
leakages. The experimental data does indicate,
however, an average ratio of ICBO/IE equal to .
68 for the twelve samples tested; and that most
of these transistors, if properly matched in
complementary pairs, are usable as current
switches.
Design Considerations
Reset or Charge-Dumping Circuit
Using the Transistor as a Current Switch
Solid-state operational amplifiers with
input current leakages of less than 10-14 ampere
became a reality with the development of insu-
lated gate field. effect transistors. It became
apparent that it might be possible to design a
wide-range solid-state current integrator having
a lower leakage current than existing designs if
a fast, low-leakage solid-state switch could be
founi. At the preser.t time, leakage currents of
· Resetting is accomplished in the ORNL
model Q-2895 current integrator in the following
Amplifier
manner. Complementary silicon planar transis-
tors of the type described in the preceding sec-
tion serve as current switches (see Qi and 22
in Fig. 2). In their normal quiescent state they
are in the inverted-mode condition with both
junctions reverse biased by 100 millivolts. Under
these condition, the efíective leakage contributed
by the switches is the algebraic sum of the two
emitter currents. From Tabie I, complementary
· pairs can be picked that will give an effective
leakage of less than 10-14 ampere. Reset pulses
switch these transistors from the reversed.
biased state to the active state for a precise
time during which a constant current flows to or
from the integrating capacitor. Using the tran-
sistors in this manner has the advantages of
both normal and inverted modes of operation.
The current that flows for the duration oi the
reset pulse is
VR · VBE + VD,
The calibration of the integrator is inde.
pendent of amplifier drift and gain, provided that
the sum of the error voltage due to the signal and
the drift voltage as referred to the input remains
less than the reverse bias on QI and Q2. Hence,
little effor: was made to design a superior oper.
ational amplifier. The amplifiez is a differential
type employing both common-mode feedback to
the current source for the input pai: and voltage
feedback to the unused input. It has both inverted
and noninverted outputs, either of which can be
selected by the polarity switch depending on the
sign of the current being integrated. A matched
pair of 2N3608 insulated-gate field-effect transis-
tors in the input stage provides a high input im-,
pedance with a leakage current less than 1 x 10 14
ampere. The overall gain is designed for 100
with a yain-bandwidth product of 1.3 megahertz:
ISBE
where VR is the reference voltage, Vp is the
forward drop across the diode clamp, and VBE
is the base-to-emitter voltage for the transistor
in the active,or on, state. Since VBE and VD
have opposite signs, almost perfect ternperature
compensation can be achieved by matching the
diode current of the clamp with the emitter cur.
rent of the switching transistor.. Since the base
driving impedance is the low dynamic impedance
of the diode and the input always remains at
essentially ground potential, the switching tran-
sistors and RE provide a current source that
has essentially the same stability as RE. The
charge removed from the storage capacitor by
each reset pulse is
Discriminator.
The discriminator actually consists of
two trigger pairs. The function of the first pair,
which is capable of free running when its dc
threshold voltage is exceeded, is to prevent
"Jock-out" and to provide a reasonably constant
trigger pulse for the second pair.. The function
of the second trigger pair is to generate a reset
pulse I that is extremely constant in width. . .
Methods for generating accurate timing pulses
have been covered in the literature and will not .
be discussed here.
Q = IT + CIVÁ - VB + VD),
.
Rate Meter
The rate meter covers the range of I to .
2000 counts! second in four ranges. It is a duty-
cycle type having excellent stability and linearity.
Although its intended use is only as an indicator
to aid the operator in adjusting the beam, its i
accuracy is limited only by the panel meter lin.
earity and the accuracy with which the individual
ranges have been calibrated.
Packaging and Controls
where I is the width of the reset pulse in seconds
and C is the capacitance from the emitter termi.
aal to ground (note: the capacitance from the
emitter to the base and collector is degenerated
by unity gain). If C is small (about 1 picofarad)
and the product of I and I is large (greater than
10-8 coulomb), the effect of C can be neglected.
Unfortunately this condition cannot always be
obtained, and the capacitance C therefore deter-
mines the lower limit of the charge that can be
reliably and accurately dumped. It is extremely
difficult to remove charges sanaller than 10-10
coulonıb when range switching is employed unless
a compromise is made in the reference voltage
VR that determines the reset-pulse amplitude.
By using a reed relay in a very unorthodox
manner one can have two ranges, with the most
sensitive range being 10-10 coulomb (see Figs.
3 and 4). This technique keeps the total capa.
citance in the emitter circuit under 2 picofarads.
The high range is 10-7 coulomb; it is intended
for integrating beam currents.
The model Q-2895 current integrator is
packaged in a standard NIM module having a
four-module width. It can be powered by either
the standard NIM, bin power supply or the standard
NIM 50-milliampere modular supply. As shown
in Fig. 5, the integrator has seven controls on
the front panel. The four used in normal opera. .
tion are (i) the CAL-OPERATE selector switch
that permits the operator to ground the input. ..
Select calibration test currents of 10-7 or 10-5
ampere, or switch to the operate position; (2) a
RATE METER range switch; (3) a CURRENT
POLARITY switch; and (4) a CHARGE RANGE
switch. The other three controls, used only
occasionally, are a BALANCE control for balance
ing the amplifier, a TRIGGER control which sets
the dc voltage level of the amplifier outputs with
:
3
respect to the trigger level of the discriminator,
and a meter switch. The meter switch allows
the operator to monitor either the balance of the
amplifier outputs or the voltage level of the out-
put being used with respect to ground.
Conclusions
Many of the features of this instrument
have been omitted because we have attempted to
present an operating principle rather than a de.
iailed description of an instrument. The present
model has good long-term stability with a repeat-
ability of l part in 2000. The lowest range on
this instrument rernoves 10-10 coulomb with each
reset pulse. This can be reduced to 10-11
coulomb by reducing the amplitude of the reset
pulse the same order of niagnitude. Since the
emitter leakage current on many of the silicon
planar transistors tested for this application was
low enough to produce oniy negligible error in
current measurements as low as 10-12 ampere,
they make the best, fast low-level current
switches we have evaluated.
References
1. D. A. Landis and F. S. Goulding, UCRL. .
11828, 1964, pp. 222-223.
2. Roul Brenner, "A Precision Current inte. '
grator," Instituto de Energia Atomica, Pub-
lication No. 67, March 1964.
J. J. Ebers and J. L. Moll, "Large-Signal
Behavior of Junction Transistors, 1.Pro..
ceedings of the I.R.E., December 1954,
pp. 1761-1772.
T.C. Verster, "Silicon Planar Epitaxial
Transistors as Fast and Reliable Low-Level
Switches," IEEE Trans. Elec. Dev, ED-ll,
May 1964, pp. 228-237.
ORNLUWG 66-10495
CHARGE DUMPING TECHNIQUE EMPLOYED IN CURRENT INTEGRATOR
10
INPUT
SCALER
. INPUT
: t
-
.
.
.
DISCRIMINATOR
TO
INPUT
SCALER
DISCRIMINATOR
(c) SHORTING METHOD
(a) DIODE PUMP
TO
SCALER


TO
UPPER LEVEL 1
DISCRIMINATOR
SCALER
DISCRIMINATOR
LOWER LEVEL
DISCRIMINATOR
+Cs
CURRENT
GENERATOR
REF
VOLTAGE
(b) CAPACITOR CHARGE
AND DUMP
(d) VOLTAGE SENSING AND
CURRENT SWITCHING
1
ir
;
-
L'
.
i
...
,
v...
.
.
..
..
.
.
..
.
:



1
!1'
1
.
.
2

.
.
1
1
އެލްރޭނެތާތެޖޭތތްބިންތަ
CURRENT INTEGRATOR
Q-2895-1
5
poli
25
20
நம்பழ
10 0 10
BALANTË
10-5,
IN
CAL-OPERATE
IZERO
10-7,
OP 10
CTS/SEC
1001K
2 K
POST
NEG
10-7
TRIGGER
CHARGE RANGE
CUKRENT
.



܂
ܪܐܪܘ ܙܬ ܝ܀ ܕ ܀܀ ܝܗ
ܢܐܐܪ
܀ ܀ ܀:
܀
ܕ ܙ
.
܂ ܘ
ܪܝ ܟ .
- .
ܫܙܝܫܚܝܙܝܫ ܐܥܺܫܶܕܪܶܬ݁ *
"
܆ !
ܬܪ
*
s.fi
ܕ
ܢ
ܝ
!
.
: :
;
1"ܟܙ
. ܚܐܚ ܝܘ. ܀ܬܗ . .
i;
ܕ
ܙ
f.
:
ܫܪܪ:"*
; | .;
%1
1
0
2 ܂
.:
'
܀
-r
.
'
: ' -
,
'
ܘ
܂
.ܕܤܚܝܝܪܚܗܫܡܳܫܫܪܳܪܽܓܰ x X
Y
£
ܛ ܚܲܕ݂ܝܚܢ܂ ܂
14
5
4 ܂
.
.
ܗ ܢܕ
1
܂
· ...• • •
ܙܙ ܝ
܂ ܐ
11
ܕܙܢ ܀
ܙ
ܐܠܕ
ܙ
ܝܙ܀
ܫ
܂
.
܂
.
܀
. . '
ܢ
' . .
܂ 1
ܢ
܂
.
.
ܙ
'
*
' .
-
'
..
ܪ
܃ ܝܼ
: ܀ ܂
܂.
ܪܚܐ
܂
4.: ܃ ܂ ܃
, ܐ .
f, .
ܝ *
1 :: . -,!
1
:
; ܝ
re
ܠܙ
ܐܐ
, :
1
:
ܐ
ܚܐ 00
HV
'
I
. . - ;
ܕ
- ܙ
.
1;
ܕ
15 M 1
: .- : ;
'
ܪܢܕܠܙܶܕ݂ܪ
.܂
.
.ܪ
ܚܕܕܢܢܕܗܐܝܕܕ
ܙ
.ܝܟ
ܝ
.
.
܂
ܐ
ܙ
܀
ܝܝ. .ܝ.ܝ ܐܢܝ: ܐܘܪܝܝܣܝܫܙܫܫܳܓܕܒܗܫ
ܚ
ܐ܀
ܐ
.܂
--;": *
,
«- …
..
܂
ܙܐ 7
ܕ݂ܺܝ܇
• ܐ
• •
܀
ܪ
ܪܐܟ
ܪ
܀
܀
ܘ
܀
ORNL-DWG 66-8904A
RESET
PULSE
-
0.
V
+12v (REFERENCE)
30 ML + 12v (REFERENCE)
CLAMP
- 12v
ERE -
INPUT
INVERTED
MOSFET
DIFFERENTIAL
AMPLIFIER
A=100
AND
DISCRIMINATOR!
SHAPER
RATE
METER)
n
ERE
+12v
INVERTER
TO
1 -12 (REF-
CLAMP ERENCE) ;
CALER
SCALER

O V
..
Q-2895 DC Cuţrent Integrator.
ORNL-DWG 66-10496
CURRENT SWITCH WITH RANGE SWITCHING

RESET PULSE
INPUT
1N914
+12V
412 v
-0.1 v
-12 v
War
. MOSFET
| DIFFERENTIAL
AMPLIFIER
MU
INPUT
2
Disku
HALL
.
1
.
..
NY
mi
20:*
NAS
ORNL-DWG 66-10494A

COMPARISON OF ICBO & IE
IE
IE
50
TEST CONDITIONS FOR
CURRENT GAIN - VCE = 0.6v IB = 1pa.
Ісво — -vc3 =0.1v. .
IE - VB and Vc = 0.1v Reverse bias
TYPE
SAMPLE ICBO
amps x 10-12
-13 HFEN HFEI - ICBO
camps x 10-13
2N4338 NPN
1.7
-1.00 150 12.0 17
SILICON
-1.30 280 2.4 384
0.47
-0.32 285 5.0
PLANAR
2.1
- 1.70 470 1 1.5 12.3
EPITAXIAL
0.68
-0.10 330 14.0 68
-0.17 11100 3.4
2N4250 PNP
1 -4.6
1.6
315 1.8
29
DIFFUSED
-2.5
0.17 300 10.2 147
-0.063
0.085 380 3.0 17.4.
SILICON
-0.102.
0.28 350 3.01 3.6
PLANAR
-0.21
0.20 430 3.
5 10
1 -0.215
0.085 300 3.0 1 25
_1.87
-
__
.
>
9.
.
ZUIS. MAE
ce
+
Wit 1W 3W
-
1
'.
END
DATE FILMED
12/ 29 / 66
.
.
..
..
-
--
"
M
EN
'
TS