Help KELTHLEY ammeter composed of LT1012 + 2N4338 cannot be reset

[daisizhou]

One of my KELTHLEY ammeters,Zero adjustment potentiometer, adjust clockwise to the limit, and adjust counterclockwise to the limit can not be zeroed.
And I connected it to the normal current mode of the multimeter, and the displayed current value is also inconsistent, and there are deviations in each gear.
I don't know where is the fault, if you know please help me, thanks

I have added the manual, and a simple circuit diagram

[daisizhou]

I checked the resistor and it doesn't look damaged

[Kleinstein]

The circuit looks relatively simple and one should be able to get a least a crude idea on how it is made. Maybe one can also find a service manual with schematics for this rather old, lower end instrument.

Not getting a zero points to an offset problem of the amplifier, not a problem with the resistors. For just the normal mA readings there should be no need for the JFET. The JFET may be needed for the integrator. For the rather large currents the current should still first flow through the shunts and integration would be separate (not linke electrometers in the pA range that directly integrate the input current).
A point to check would be the LT1078 OP-amp - as it is socketed one could test it ouside of the circuit in a bead board. Somtimes just moving it in the socket could help, as the ceap sockets may develope contact problems.

Is the pot for the zero doing anything, so is there a change in the reading by a reasonable amount ?
What is the mAs range read with the reset bottom pressed ?

Looking at the manual makes me think this is a really odd design, that deserves many  :
With a 1 V burden votlage there should be absolutely no need for an extra zero adjustment. If at all the integrator part may want a zero trim for the drift if rather long integration is wanted. The power consumption also looks excessive.  Chances are the parts are worth more than a working meter of this kind.

[daisizhou]

I tested the current data of 20ma, 200ma and 2000ma.
I re-plugged the LT1078 op-amp, and checked the 3 black sampling resistors, all within the correct error range.
In the mas function, if the current is flowing, the number will be displayed on the display, and the number will increase over time. For example, in the 2000ma gear, I pass the current number from 100, 200, 300... . . . . Up to 2000, press the RESET button again, the display will return to the initial number, indicating that the mas function is normal, similar to the power meter function

Please let me know if you have any other tests you wish to test, I will test them, thanks for your help

[Kleinstein]

There seems to be an additional factor of about 1.4 that the meter is off in addition to the offset. This could be just the trimmer for the gain set wrong, but it looks like a large error for that.

For the mAs function the question it to what does it reset with a zero current (e.g. open input) and if it stays there reasonable (not too much drift). It is a bot confusing that there is supposed to be a limit before the integration actually starts and is still looks like an analog intergrator with the rather large capacitor.

Even if repaired and working as intendend I would consider this meter to be rather poor. The only special thing is the integration fuction. Otherwise the performance is rediculous poor, maybe reaching the harbor freight free meters in some aspects. A repair would be for learing, not really to get a useful meter.

[daisizhou]

I found the definition of the motherboard P3 interface and the display module, it is in the following manual, I hope it will help with maintenance

mas test video
https://youtu.be/08LEwMKziPY

[Kleinstein]

The mAs range seems to be working analog, as the DMM modul does not support digital integration. After reset the zero error seems to be the same (same fraction fo FS)  as with the other ranges. So this could be an error in the module or an error in the offset, that effect both parts. One should be able to check the module separately and see if it reads zero.

If the error is with the rest, it would really help to get an idea on the signal flow or a crude schematics. The circuit looks way more complicated than needed for the functions.

[daisizhou]

The display header is a 3½-digit meter,I found out that P3 and P4 are differential analog inputs,So I tested the display value corresponding to the current value of each gear and the analog voltage value of P3、P4.
There is a "cal" character on the back of this 3 and a half digit meter, I don't know what it is for, it has not been adjusted for the time being

[Kleinstein]

The idea of measuring the voltage at the module input is good, but the resolution is a bit too low for a decisive result. If the 1 mV zero offset is real, it looks like the problem is with the front end.
The module seems to work with some 110 mV full scale, which is strange given the large shunts. Normally the ADC works with a large full scale range, like 200 mV or even 2 V.

The extra trimmer is likely setting the reference for the module and thus the overall display gain. Chances are the trimmer setting is a bad choice.
A better setting (higher reference at the module and more gain for the main part) may improve things quite a bit.
The calibration / scale factor seems to be way off (some 40%) anyway. So there is nothing to loose adjusting the trimmers.

[daisizhou]

I took off the chip LT1078 and CD4027BE, and then let the circuit work, and found that the "mas" function failed, and the "ma" displayed normally.
Next, I removed the tc7660 and found that the value displayed by the zero position has changed significantly.
I don't know if this can determine the direction of repair ideas

[Kleinstein]

The TC7660 (similar to ICL7660) is a charge pump chip, likely to produce some negative supply or similar. Still odd, as the DVM moule likely wants to work with a virtual ground instead.
One could also use the TC7660 for a virtual ground, though a relatively noisy one.

If the mA function still works without the LT1078, this essentially means the signal from the shunts only goes through a voltage divider to than go to the DVM module.
So it is really odd to get that much offset. The offset problem seems to be with the trimmer - normally I would not expect any need for that trimmer. The ADC chips like ICL7106 are normally accurate enough to get away without.
The extra OP-amps for the mAs function may be needed to implement the starting level / suppressing drift of the integrator.


A wrong setting with the trimmers could be the main problem. It looks like the signal is divided down quite a lot (like 1/20) and than the ADC run with too low a reference. So the idea is to crank up the ADC reference from some 55 mV to some 1 V and than also crank up the gain trimmer for the mA and mAs functions from some 5% to 95%. This should reduce the offset problem by something like a factor of 20 - maybe good enough. Changing the gain trimmer could also effect the offset trim. The change to the trim is massive (like near one end to near the other end).

[daisizhou]

thank you for your reply,In order to confirm that the "ma" function is normal after removing the LT1078, I tested it again, and the following is the picture I took.The test is the moment when 10ma current is input.
I don't quite understand which chip you mean by "trimmers", how can I adjust it?

[daisizhou]

By the way, I found the signal for the zero potentiometer, looked all the way, and found it went through a 92K resistor and ended up going to pin 15 (REF BG) of the DPM600

[Kleinstein]

Changes are the trim-pot for the ADCs reference is the one on the bottom side of the DVM moule, near the CAL lable.  There is srew near the cal lable to adjust. This should set the full scale range for the module.
One end of the zero potentiometer (= trimmer)  getting a signal from the Ref. BG pin makes absolute sense, as this is a 1.2 V reference voltage. The point is more how they get a negative side. As the ADC input is differential, this could be a fixed offset to the other side, e.g. with some kind of divider there. The gain setting may also effect the offset to some degree and with the gain setting way off the normal setting this could be the problem.

[daisizhou]

I lost.I don't know what other useful measurements, tests can be made

[Kleinstein]

It is not so much about a measurement, but changing the adjustment pot under the DVM module and the mA gain on the main PCB.
A first step could be to adjust the pot on the module. This would be bringing up the voltage between ref- and ref+ from some 55 mV (which should be the current value to get some 110 mV full scale). I instead of some 55 mV I would target some 500 mV or maybe even more (it depends on how the main board is made).
After this adjustment the meter should read the current way to low so 10 mA should read more like 1 mA.
The 2nd step is than to adjust the gain pot on the main boad (e.g. apply 15 mA in the 20 mA range to make it read 15 mA). This would be quite some change, so not just half a turn, but more like 10 turns.
With some luck the zero adjustment may than also work - at least the error should be much smaller.

[daisizhou]

First step, do you mean to adjust the potentiometer in the picture?Just let me input the actual current of 110ma on the 200ma current gear, and adjust the potentiometer to make the DPM600 display drop from 184.5ma to 110ma. Is that how it goes?
The second step is to adjust the gain potentiometer on the motherboard, do you mean the potentiometer that adjusts "ma" in the middle?

[Kleinstein]

First step, do you mean to adjust the potentiometer in the picture?Just let me input the actual current of 110ma on the 200ma current gear, and adjust the potentiometer to make the DPM600 display drop from 184.5ma to 110ma. Is that how it goes?
The second step is to adjust the gain potentiometer on the motherboard, do you mean the potentiometer that adjusts "ma" in the middle?

That is about the way, just with larger steps, more change. So more like in the first step adjust until the 110 mA read as some 10-20 mA. Only than adjust the mA gain (middle of the 3 on the main PCB) to get the correct reading.

[daisizhou]

Do I need to measure the initial resistance of this potentiometer before making adjustments?I mean be prepared to recover in case the adjustment fails。Do you think this is necessary?

[Kleinstein]

I don't think one would need the original settings. A calibration is relatively easy and it looks the settings are way off anyway. So there is absolutely no value in the current calibration (some 40% off).
There are 2 pots to set the gain (at the DVm module and on the main (separate for mA and mAs). So there is no unique correct setting. There is some freedom on how to distribute the gain over the two parts - it looks like the current setting is way to one extreme and is should be closer to the other.

It is not clear on how the DVM module is made. I highly suspect that is a kind of ICL7106 or similar. For best performance they have the integration resistor and capacitor fitted for the reference voltage range. If the capacitor and resistor is set for 200 mV operation (about the lower recommended limit), it would not work all the way to 2 V, but saturate early (e.g. 500 mV). If set up for 2 V the accuracy and noise performance would suffer if used with only 200 mV FS and even more with 110 mV FS. So one is not fully free to change the reference / pot on the DVM module.

It is hard to tell if the module is more for 200 mV full scale or near 2 V full scale, but very likely not for made for 110 mV FS. I somewhat suspect it to be a little less than 2 V, so that the signal from the shunts can be send there with only little attenuation (for trim). If the module were for 200 mV, the sensitble thing would be to use smaller shunt resistors (which helps with heat and overload protection), but not to use a 1:10 divider after the shunts.

It could be that someone tried to change the mAs range, e.g. to be more sensitive there, or may compensate for something bad (too little gain, or maybe shift the starting limit) in that branch.

[daisizhou]

I adjusted the potentiometer at the back of the DVM module.But it didn't improve.
Let me describe what happened,I put in a steady 45.5ma DC current,DVM module rear potentiometer, clockwise to the end display 120.0, counterclockwise to the end display 108.1，Then I adjusted the motherboard in the middle of the "ma" potentiometer, clockwise head display 115.9, counterclockwise head display 109.6.
I disassembled the DVM module, the following is the picture inside

[Kleinstein]

The adjustment range is surprisingly small and this would mean that the DVM module is working in a rather odd range. This makes me suspect that both pots for the gain are effecting the reference voltage in a somewhat combined way.
It gets tricky to find the problem without a schematics.  The DVM module should be essentially the standard circuit - so no need for much reverse engenniering there. Maybe get an idea how the pot is connected.
The interesting part would be connections to the ref. pins and the part around the adjustment pots.
Chances are the mA part is just passives one the main PCB and than the connection to the DVM module. The part around the OP-amps is less relevant for the start, as they seem to be only for the mAs part.
For the TC7660 one could measure the voltage to check if it created more like a center voltage or negative voltage.

[daisizhou]

I have drawn some schematics (sketches), which will be updated gradually

[daisizhou]

I made a sketch, hope it will help the analysis

[daisizhou]

After I disconnected the 22M resistor, the DPM600 display has shown 0.01 when I turn on the power.
Then I tested the DPM600, the 8-pin COM is the negative electrode and the 15-pin REF BG voltage, only 0.423V。
Is it because the ICL8069 reference inside the DPM600 module is broken? Because the ICL8069 happens to generate a 1.2V voltage reference