436-GC/456-GC Service Manual Revision B February 2019
Amplifier AR1 holds the collector of Q1 (pin 1) at ground potential and forces the collector current to flow
through R3. The final output voltage at P1-1 is then
v
out
= 2.05 x 10
6
x sqrt((l
in
+ I
sat
)(I
ref
)) + v
os
,
where v
os
is the sum of the input offset voltage of AR1 and all of the current-related voltage errors. Putting
in the nominal value of 25 microamps for Iref gives the following approximate formula for the output
voltage:
v
out
= 10
4
x sqrt(I
in
+ I
sat
) v
os
.
This formula will not be accurate if the temperatures of all of the transistors are not equal. The two halves
of each dual transistor are always at the same temperature, since they are fabricated on a single silicon
die. Q1 and Q2 track each other in temperature only by being in close proximity in an isolated environment,
however, and any thermal disturbance will cause the output of the electrometer to drift. Simply touching
one of the transistors can cause a minute or two of drift, and the output may not stabilize for 5 or 10
minutes after soldering in the vicinity of the transistors.
The time constant of the amplifier is set to 45ms by C1, and C2 eliminates high-frequency noise from digital
sources. R7 and C12 prevent glitches from the ADC multiplexer (on the main system board), which reads
the output voltage of the electrometer, from disturbing the amplifier output. Such glitches can couple
through C1 and the capacitance between the collectors of the two halves of Q1 to disturb the sensitive
input node, resulting in a much longer settling time than would be expected.
Electrometer diagnostic features.
In order to determine the input current from the output voltage, the value of v
os
in the equation in the
preceding paragraph must be determined. This is done during operation by closing the switch section of Ul
(pins 1-3) which is connected to the output of the reference log amplifier. Current through the anti-log
transistors is thus cut off, leaving the output voltage at P1-1 equal to v
os
.
Though it is not immediately obvious from the equations presented here, closing the switch section of U1
(pins 14-16) which is connected to the output of the input log
amplifier results in an output voltage at P1-1 which corresponds to an input current equal to the saturation
current of Q1. This measurement is used for diagnostic purposes only. Both of the diagnostic switch
sections of U1 are left open in normal operation.
Because the input stage of the electrometer has a logarithmic response, its behavior becomes
unpredictable as the input current approaches zero. All of the voltages applied to the circuit are above
ground potential, so leakage currents should never cause the input to reverse in polarity and saturate the
output positively. However, the response time of the circuit is inversely proportional to the current level,
becoming very slow at currents below a few hundred femtoamps. For this reason, it is impractical to test
the noise and drift of the electrometer by "capping off" the input with no input current.
To test the electrometer when no external picoamp source is available, an input current source in the
picoampere range has been provided. This source relies upon the fact that the current through a capacitor
is equal to the product of the capacitance and the rate of change of the voltage across the capacitor. C4
(10pF) is connected to the electrometer input, which remains near ground potential. Applying a negative
sawtooth waveform to the other end of C4 should thus produce a steady input current, with positive spikes
on the positive transitions of the sawtooth voltage. This waveform (modified somewhat to optimize the
recovery of the electrometer from the positive spikes) is generated by an external DAC and applied to C4
through a filter consisting of R9 and C8. The filter removes noise from the external input, while slowing the
transitions somewhat. In normal operation, transistor Q3 is turned on by writing a logic 1 to U3-12, shorting
C4 to ground. Voltage variations at P1-10 will then not create an undesired current at the electrometer
input.
To Next Page
Loading...
