Very slow DMM offset settling - humidity effect?

[maxwell3e10]

I got a Fluke 8600A multimeter, and while its hardly a precision instrument, the basic topology is an integrating ADC similar to better meters. So I figured I can play around with it. I found that it works well and is mostly within specs except for one problem.
When first turned on, it has an offset of about 0.1-0.2 mV that takes very long, several days, to decay to zero. Once it decays to zero, the meter is stable. If I turn the meter off, the offset starts to come back after about a day.
I poked around inside with a voltmeter and experimented with shorting the input using different resistors. The offset is consistent with some leakage current going into the input stage. The board seems clean and I also cleaned it with IPA.

So my current theory, which maybe totally wrong, is that there is a hairline crack in one of the input stage FETs that makes it sensitive to humidity.  Humidity is the only effect I can think of that would have a time constant of several days to "dry out". Any other ideas? I'd like to identify the problem rather than just start swapping all the parts.

[Kleinstein]

It does not need a crack in a case for humidity to have an effect. A known path is humidity to slowly hide inside the FR4 board material and make is swell. The mechanical stress can than cause offset voltages. Another mechanical path is thermal expansion mismatch to cause mechanical stress at solder joints and the SnPb solder so slowly relax some of the stress. A similar effect can also happen inside chips with stress between the silicon chip and metal carrier and the glue/solder for attachment to move.

So there is a reason that some precision instruments are left turned on 24/7.

[maxwell3e10]

I got the meter on craigslist, so it certainly could have been stored in a damp basement for years. So it could be an effect in the board, although it doesn't look damaged. The leakage current is about 1 nA, a fairly large effect. I guess I could leave the meter on for a while or heat it gently to 50 C to bake out the board, see if that helps.

[HighVoltage]

The leakage current is about 1 nA, a fairly large effect.

How did you measure the leakage current?
 

[maxwell3e10]

If I short the meter with a 1 MOhm resistor, the offset is about 1 mV. If it is short-circuited at the input, the offset is 0.1 mV, but there is an internal 100 kOhm resistor. If I short it inside the meter directly at the FET switch, then the offset is zero.

[3roomlab]

I learnt to check leakage from alex nikitin
discharge a film cap with the appropriate voltage using only the input terminals
there is a kemet film cap rated at 440G 68nF, I think there are caps with even higher insulating resistance.
you need to log the curve to better see it

To do a full caracterization of input current one would have to measure the input current at different input voltages.
There are two ways to do that:

There is always a third way  . Log the voltage at 10V range, NPLC1, and discharge a charged to 10-11V polypropylene capacitor, say, 10nF, first time from +10V, second time from -10V. You need two runs as you don't know where/if  the leakage current polarity changes. The speed of the voltage change is essentially the leakage current (i=C*dU/dt, for 10nF and 10pA = 1mV/s).

Cheers

Alex

after abt 5 hours of warm up and rounds switching around capacitors, the impedance did increase ... by alot.
but at least, now the impedance vs volts are of the expected lower volts = higher impedance curve.

edit **
@34nF 15v 10s drop = 5.33mV/s, = 181pA, = 61.1G @ 11.1v
reverse @ 204nF 18v 10s drop = -1.39mV/s, = -284pA, = 39G @ -11.1v
the reverse character is slightly weird that it required alot more charge (4 to 6x) to make it past -11v.
is this what originally K2015 is in terms of impedance?
if so, it would mean that either i need to dampen the hyper sensitivity, or enhance it further with more attention to lowering of creepages?

(plots are Mohm vs Volts)

edit*** more testing
im not sure if im doing this right, but i have no other reason (or lack of reason)  to think there could be something wrong
@68nF 10s drop  -1.364mV/s, = -92pA, = 50G @ -4.70v
@68nF 10s drop  0.874mV/s, = 59.4pA, = 82G @ +4.93v
@68nF 10s drop  -0.616mV/s, = -41.9pA, = 45G @ -1.88v <---- weird
@68nF 10s drop  0.306mV/s, = 20.8pA, = 92G @ +1.92v

geeez 82G impedance

[HighVoltage]

I learnt to check leakage from alex nikitin
discharge a film cap with the appropriate voltage using only the input terminals
there is a kemet film cap rated at 440G 68nF, I think there are caps with even higher insulating resistance.
you need to log the curve to better see it

Thank you, that was very helpful.

[maxwell3e10]

Taking a hair-dryer to the circuit board brings the offset to zero very quickly. It is staying there for a while but then drifts back up again.

Also measuring carefully the shorted input leakage current and the offset voltage read by the meter gives an impedance of 140 k\$\Omega\$ for the leakage current path. But the input circuit consists of two 108k\$\Omega\$ resistors. Since it doesn't clearly agree, it makes me believe there are multiple leakage paths, which again suggests the problem with the circuit board itself. It looks clean, though. Maybe drying it in an oven at 50C would help.

[Le_Bassiste]

depending on previous storage conditions, it may not just be humidity but also an invisible film of fungus that wreaks havoc on isolation paths. i realized that during the process of bringing a FLUKE 5450A back into spec. i knew this particular specimen came out of a large dusty and humid warehouse where it must have been sitting on the shelves for ages.

back to topic: before you put the whole shebam into your pizza oven as a last resort to prove your humidity hypothesis, you could give it a shot with a warm (not hot) heat gun and a very small nozzle to examine individual components on their drift behavior. maybe, some veteran opamp or something has finally thrown the blanket not just for humidity reasons.

[maxwell3e10]

depending on previous storage conditions, it may not just be humidity but also an invisible film of fungus that wreaks havoc on isolation paths. i realized that during the process of bringing a FLUKE 5450A back into spec. i knew this particular specimen came out of a large dusty and humid warehouse where it must have been sitting on the shelves for ages.

What did you do to get rid of it?

[Kleinstein]

Surface contamination with biological film is a real possibility. This can kind of amplify the effect of humidity and make the board surface more susceptible to humidity that than causes surface leakage currents. Heating the board would remove the humidity for a while, but it would come back once RH goes higher than some 10-50% (depending on the surface).

Cleaning can be quite challenging: just water or IPA would not do much. Some modern detergents contain enzymes that may help. Where possible also some mechanical work with a wet brush can help. Chances are most of the cleaning would be water based, followed by IPA to remove residual water and some other dirt.
Not sure the relays like cleaning though.

[Le_Bassiste]

What did you do to get rid of it?

my weapons of choice. airbrush (plus oil-free compressor @ 60 psi) is really handy to flush pcb areas below components.

[SilverSolder]

Love the airbrush idea!

[3roomlab]

I got a Fluke 8600A multimeter

you got hires porn on that meter? maybe someone can spot some dirty underwear. 

[maxwell3e10]

I've measured the leakage currents more closely and there are definitely more than 2 leakage paths, as indicated in the schematic. So its likely a board leak.

Now that humidity is smaller, the leakage current is better as well. So its a good meter for the winter . I am attaching some pictures in case someone can spot anything peculiar. There are some guard traces on the board, so leakage currents were a consideration even for this 4-1/2 meter.

[Kleinstein]

The "flexible" PCB for the display looks odd. Nice as longs it is not broken.

A leakage current of some 1.6 / 0.9 nA sounds like a lot. That is way more than I would expect for a reasonable clean board.

There are odd looking orange/yellow spots at some parts of the board. Are they real or just sone reflection.This could be a sign of a contamination, that may also extend to the sensitive area, though not visible there.

[Twoflower]

Not sure if this is intentionally open or not. I haven't checked the schematics, but for my eyes it looks strange and probably worth to do some investigations.

[maxwell3e10]

I was also initially puzzled by tapering-off traces, but I think it is part of the leakage guard system. They are all connected to ground.

The I2 current its particularly puzzling. There are no semiconductors on the left-hand side of the circuit. Perhaps some contamination in the push-button switch?

The "orange spots" is just a reflection of the light. Unfortunately there is nothing so obvious.

[Kleinstein]

The small gap looks rather symmetric. More like an intentional spark gap than a production defect.

Contamination can be in different forms. Ranging from some kind of leaking fluid from a capacitor,  deposits from cigarette smoke, flux residue, mold. The humidity effect is usually at the surface, so guard traces can help a lot here. One could try a local wipe with some IPA/water (50:50) mixture.

Contamination at the switches is possible if they use some lubricant.

[dietert1]

In our lab we are still using three Fluke 8600A. Nice instruments, +/- 200mV and +/- 2 V are electrometer ranges with very high input impedance. With inputs shorted, they turn on with +/- 10 or 20 uV offset and show +00.00 and -00.00 after a minute or so.
I remember working on those multimeters like cleaning switches and tuning the power supply, which would damage batteries when continuously connected to mains. I also implemented a protection circuit for the batteries when somebody leaves the device on unintentionally. But that is normal with test equipment, which in my experience never comes as "ripe" as consumer equipment.
Is/was your 8600A equipped with batteries?

Regards, Dieter

[maxwell3e10]

Yes, the meter has NiCd batteries, which look original and amazingly still give about 6 hours on a charge (the spec. is 8 hours). My understanding is that it is actually good to occasionally let the batteries fully discharge. There is no evidence of the batteries leaking and the meter doesn't seem to have any electrolytic capacitors, as far as I can tell.

[dietert1]

As far as i remember there are some smaller vertical modules. I would try to reseat them. The same if you see parts in sockets. These DVMs must be about 40 years old. If nothing helps, you can wash the board in warm water with a little soap like dishes (after taking out the batteries). Needs some days though to dry completely. I have done this successfully to clean one of the 8600As after magic smoke inside.

Just measured the input offset currents of our 3 DVMs using a 1 GOhm resistor. After some minutes of warmup they arrive at offset voltages of -15 mV, 8 mV and 0,3 mV . This means we are within +/- 20 pA.

Regards, Dieter