Keithley 175A not working

[MikeK]

So I got a Keithley 175A multimeter off eBay for cheap.  It powers up, but won't properly measure anything.  I found that the 6.4V reference Zener diode only measures 4.47V.  This could be the problem.  But before ordering this part ($8 plus shipping) I'd like to do more testing.  Any reason not to do either of these?

1. Remove Zener and apply 6.4V with a floating power supply.
2. Or, remove Zener and use 5.1V Zener (1N4733) plus two small signal diodes (1N4148) to achieve the 6.4V.

The problem with the Zener solution is that the datasheet test current for the 1N4733 is 49mA, which is much more that what the 6.4V reference Zener is spec'd at (2mA).  And indeed the reference Zener appears to run at 2mA in-circuit (based on the math).  I ran the three-diode chain at 2mA, however, and it is almost dead on 6.4V.

Is the Zener solution reasonable to allow me to do further testing?  What could have caused the reference Zener to fail in the first place?...I know the multimeter is probably 30-40 years old, but there is a 5.1V Zener on the board that tests dead on 5.1V.

Attached is the part of the schematic showing the reference Zener.

[Kleinstein]

The zener combination would be OK for testing, even if not very stable (low TC).  It is very rare that a 6.4 Zener difts towards 4.5 V. Chances are there is a different fault, that causes a low voltage.
A first point to check would be the supplies, if there is excessive ripple (e.g. due to failing electrolytic capacitors).   Another possiblity would be something loading U103 too much.

[MikeK]

Yeah, I thought the failed Zener was odd, especially given it's low operating current.  All the supplies tested good, though...Except for the ones involving the 6.4V reference Zener of course.  I'm going to remove it and test it out of circuit.  And I think I found confirmation for the 5.1V Zener in the datasheet: Izt2 appears to be the minimum current, which is 1mA for the 1N4733.

[Kleinstein]

The 1N4577 ref. zener is supposed to run with 2 mA to give a low TC. It would work with a higher / lower current, but normally with a higher TC.
The 1N4733 is a more normal low cost zener and the test currents in the data-sheet are more like example values. The 1 mA are more like a common point for comparisons not a real minimum for the diode.

Testing the diode sepately is a good idea, though unsoldering is some stress to the diode. So this may effect the calibration.

The +5 V supply in the K175 is quite important. The ADC raw conversion result is relative to the +5 V supply and than a seprate measurement of the -2 V ref. value is used to measure the scale factor ( 5 V supply relative to the -2 V ref and some resistor ratios). So a noisy 5 V supply can cause quite some extra noise. It is only a 4.5 digit meter - so it may be good enough.
IF the -7 V ref. part is way off this can cause quite some trouble for the ADC, as this is used to shift the zero.

[MikeK]

IF the -7 V ref. part is way off this can cause quite some trouble for the ADC, as this is used to shift the zero.

It is indeed way off.  It measures -4.5V instead of -7V.

How about if I inject current at that op-amp output?  I could apply 5.1V across the output Zener, but allow for more current.  If the voltage across the reference Zener then picks up I'll know that something is loading that op-amp.  Good?

[Kleinstein]

-4.5 V at the OPs output and -4.47 V at the zener points to a problem with the OP, like too much loading there or maybe the negative supply to that OP not high enough.
One could apply more current to the zener (an extra resistor from the negative supply). Chances are some part is drawing too much current from the -7 V. 
So it may be worth at looking how the measurement functions behave. This may give a hint on what additional parts are not working properly.

[Dr. Frank]

It's quite unlikely that the Zener is defect.
Check first the supply voltages  of the OpAmp, then check if the OpAmp itself shows a big offset, by measuring on on its + and - inputs.
It's much more probable that this OpAmp broke down, so replace it with an OP07 or something similar, just for testing.
Check the datasheet of this OpAmp, if it has something peculiar, before replacing it.

Frank

[MikeK]

Yes, the negative supply is not high enough.  Since the output should be -7V, the 5.1V Zener indicates that the negative supply should be -12.1V, but is actually -9.73V.  I'll hunt down the negative supply, since it's not a 12V 3-term regulator.

[Dr. Frank]

also check capacitors for shorts, which might pull down supply or the OpAmp, like C116.

[MikeK]

I'm not certain about the -12V rail.  I measure -9.5V, and that's what the resistor values of the regulator indicate that it should be.  But then the op-amp can't produce -7V?

[Kleinstein]

The 12 V zener is not for regulation, but more for protection in case something goes wrong. The 9.5 V look about right with the resistors at the LM337. The manual calls for -10 V +-10%.
A  bad U103 is very real possibilty. It could also be U103 oscillating - so just a DMM may not be enough to check. If U103 does not provide any current, VR103 (5 V zener) could provide about -4.5 V. This may be intended for start-up, but should not be the normal operation current path.  This makes the op-amp U103 (and maybe a solder joint / socket) a prime suspect.

[MikeK]

Yeah, with a -9.5V negative supply U103 should be able to achieve a -7V output.  If C116 is leaky, like Dr. Frank suggests, that could bring the output down.  These are almost 40 year old electrolytics afterall.

[Kleinstein]

C116 may be bad, but a leakage in the 20 mA range with some 5 V would be an exceptional case. Tantalum caps usually fail short and Al cap are more failing open or maybe with leakage in the µA range.
C116 is loading the OPs output and could be an issue in choosing an OP to replace U103.

[MikeK]

The 6.4V reference Zener does appear to be okay.  Without pulling it I can apply 7.0V across it with 20mA limiting current and it does regulate to 6.35V which is within spec.  Anything less than 20mA, though, and the Zener voltage is less.  So something stealing current does make sense.  Still may pull it to see if it regulates at 2mA, which appears to be what it's supposed to run at in circuit.  If this had previously gone bad and someone had replaced it with a not-exact replacement the 2mA wouldn't be enough for a more generic Zener.

[MikeK]

Curiosity got the better of me and I pulled the 6.4V reference Zener.  At 2mA it's right at 6.35V.  Even at 1mA it was still in spec.  Rats.

[Dr. Frank]

As expected. Have you already checked the OpAmp if its inputs are on the same potential?
Some types (esp. low bias bipolar ones) break down on their inputs and then latter might draw current.

If the output of the OpAmp is not working, or is pulled down by a defective component, it will not provide enough current via R120A to supply the zener.. that could also be the root cause for the excessive current draw from somewhere.

Please do these simple measurements as proposed!

Frank   

[Dr. Frank]

I'm not certain about the -12V rail.  I measure -9.5V, and that's what the resistor values of the regulator indicate that it should be.  But then the op-amp can't produce -7V?

I confirm: The LM337 seems to be programmed for about -9.4V supply.
The OpAmp probably provides rail-to-rail input and output, therefore, is capable of handling -7V output.

Frank

[MikeK]

As expected. Have you already checked the OpAmp if its inputs are on the same potential?
Some types (esp. low bias bipolar ones) break down on their inputs and then latter might draw current.

If the output of the OpAmp is not working, or is pulled down by a defective component, it will not provide enough current via R120A to supply the zener.. that could also be the root cause for the excessive current draw from somewhere.

Please do these simple measurements as proposed!

Frank

Sorry, I thought I had checked this.  OpAmp output has a voltage so it must be working?  Inputs are off by too much.  I just took some measurements shown in the image.

[MikeK]

And with a 3mm blue LED at the op-amp output there is enough current to make it bright.

[Kleinstein]

Besides the OP, there is also the 5.1 V zener toward the 9.5 V supply that can provide current as long as the voltage is lower than some 4.4 V.  Given that the voltage at the OPs output is at 4.5 V and thus just at the limit the zener can drive, chances are the OP is not providing much or any current and the current from the zener it that sets the voltage.
So probably time to pull the OP or at least check it's supply (pin 4) directly at the chip.

[MikeK]

Yeah, I'll pull the op-amp and test it to see if the output is working.  If failed, I'd replace it along with the other TLC271 on the board...Might as well.

[Dr. Frank]

U103, the TLC271, supplies -7V to 3 other ICs, up to -30mA. (A very crude design, to use the A/D reference voltage for this).
U101, 102, 104 normally should draw about -2mA each.

One of these 3 ICs might as well be defect, besides C116, VR103 (less probable), and of course CMOS OpAmp U103, itself.

I would de-solder U103, then at first externally supply -7V at its place, current limited, of course, and measure the current consumption of these 3 other ICs.

If that is on the order of less than 10mA, U103 is probably defect.

Check its functionality on a breadboard in a non-inverting amplifier configuration, and also check its input bias currents, if these are on the order of pA (or lesser than ~ nA), and if the output is able to pull -10mA at least. See datasheet for details.

Frank

[MikeK]

I'm thinking this would be an easier test of U103, without desoldering it:

Desolder one leg of the 5.1V Zener and use a DMM to see how much current is going through it.  Anything significant means U103 isn't working.  Yes?

[Kleinstein]

The -7 V are not really used as a reference for the ADC. They are used to shift the input signal, so that the actual ADC only sees a negative voltage. The actual ADC reference is the +5 V supply from a normal regulator for the logic. The divider for the -2 V used to measure the scale factor is directly from the zener.  So from the noise perspective it is even worse - but this is only a 4.5 digit meter and not a 5 or 6 digit one.  The better ones with a similar ADC (e.g. K193) derive the +5 V ADC ref from the reference at least.

Lifting 1 pin of  VR103 (5.2 V zener) is an measuring the current is another option, though it is still hard to tell how much current U103 is than delivering. So one would still not be sure if U103 is bad or not.

One could try to measure the total current from the -10 V supply (e.g. turn the switch of and measuren the current there). There are not many other consumers at the -10 V. There should be no more than maybe some 10 mA for the -15 V.  If U103 is OK it should take more like 20-30 mA alone to try to force the -7 V before it reaches its limit. So a low current consumption would mean a bad U103. With a high consumption U103 may still be OK.

[Dr. Frank]

I'm thinking this would be an easier test of U103, without desoldering it:

Desolder one leg of the 5.1V Zener and use a DMM to see how much current is going through it.  Anything significant means U103 isn't working.  Yes?

No.
You will not learn anything from that.
As said, the other components might draw too much current, even if U103 is working correctly.
Frank