HP4140B picoammeter

[Keith956]

I wonder if anyone's familiar with the HP4140B picoammeter? I have one which works, but has a rather large offset current in the lower readings.

After about an hour's warmup, with nothing connected to the triaxial input connector, the reading on the 1e-12 range is approx -0.350. The readings on the 1e-11 and 1e-10 ranges are -0.017 and -0.002 respectively. This seems odd, as it seems the 1e-12 reading is double what might be expected - although this may just be a coincidence.

I have tried the adjustments - checking the +/- 15V power is within spec, adjusting the pA meter amplifier offset and filter offset as per the manual, but while this gives nice 0.0 readings on ranges 1e-9 and above, the small offset on the lower ones remain.

Looking at the circuit of the A1 amplifier board, relay K1 is off on the most sensitive range. So the only leakage path would be R52 (the 99G resistor) and its parallel capacitor C22, R94 and Q26.

I have removed the covers on the A1 board and as far as I can see the covers have never been removed, R52 is clean and dust/grime free.

Any ideas what to check next?

[coppercone2]

might want to spray it down with isopropyl alcohol from a can, and the front connector too. its sensitive enough that your eye might not be good enough.

what I would recommend though is to get some 1 gigaohm + resistors and hook them up to the meter to see what it reads out when you try to measure a resistor. they are cheap

floating electrometer inputs are going to be bizarre, you need to test it with a source or load

https://www.jensentools.com/mg-chemicals-824-450g-99-9-isopropyl-alcohol-16-oz-aerosol/p/484-433

that device is ancient and you just can't be sure something that old is clean. maybe get 2 cans because that is a whole lot of unit

[wn1fju]

For what it's worth, on my 4140B, after a one hour warm-up, I got the following readings:

-0.546 on the 1e-12 range
0.008 on the 1e-11 range
0.001 on the 1e-10 range

With normal cabling out the front triax connector, and without a special test fixture, I think accurate readings on the 1e-12 range is kind of wishful thinking anyway. 

[coppercone2]

also keep in mind the contamination could come from within, i.e. a capacitor that sprayed a little and clogged itself and the finest of mist got on something

[wn1fju]

Some further observations on my HP 4140B for "Keith956".  I pulled the A1 card, opened up the covered input stage, and gently cleaned the board and the range resistors with alcohol.  After a one hour dry time, I ran through the offset and gain adjustments.  The (open input) reading on the 1e-12 range was now about 0.050, not the -0.546 I obtained previously.

Moreover, according to the op manual, once the reading is below 100 fA, or 0.100 on the 1e-12 range, the zero button on the front panel will work.  After I pressed it, I took the readings on all the ranges and got:

0.000 on the -12 range  (drifts to about 0.005 after a few minutes)
0.003 on the -11 range
0.001 on the -10 range
0.001 on the -9 range
0.000 on the -8 range
0.001 on the -7 range
0.000 on the -6 range
0.001 on the -5 range
0.000 on the -4 range
0.001 on the -3 range
0.000 on the -2 range

I think I will quit while I'm ahead!

[coppercone2]

Invisible dirt? Also possibly a card edge problem

[Keith956]

Thanks, I'll give it a try.

I did notice that sitting it unused for a couple of weeks let the reading increase to something like -0.800 on the e-12 range. When I powered it up it settled down to the -0.350 reading after an hour or so.

I will give it a try opening it up and cleaning the inputs and 99G resistor/capacitor with isopropyl and see if that helps.

Is it possible to tell from the sign of the reported current if it's leakage across the feedback resistor or on the input to ground (e.g. the triax connector)?

[Gyro]

I haven't seen it referenced for quite a while - You probably want to download a copy of the excellent Keithley -> Tektronix Low Level Measurements Handbook, if you haven't already got it.

https://www.tek.com/document/handbook/low-level-measurements-handbook

[Kleinstein]

Thanks, I'll give it a try.

I did notice that sitting it unused for a couple of weeks let the reading increase to something like -0.800 on the e-12 range. When I powered it up it settled down to the -0.350 reading after an hour or so.

I will give it a try opening it up and cleaning the inputs and 99G resistor/capacitor with isopropyl and see if that helps.

Is it possible to tell from the sign of the reported current if it's leakage across the feedback resistor or on the input to ground (e.g. the triax connector)?

Leakage across the feedback resistor should not result in much offset, but more like a change in the scale factor. It would need quite a lot of leakage there to get a significant offset (e.g. from digitially compensated residual offset).

Similar the Triax connector would likely not be a significant point for an offset, as there is little voltage in this area.

There are 2 possible ways to get an offset / bias current reading: one is leakage (e.g. towards the supply) to actually cause leakage current. The other parts are effects causing an input offset voltage change that only causes an offset to the readings. Here humidity in the PCB and plastic IC cases can be an issue (I don't know the details on this meter - may use less sensitiv metal case parts). One could tell the difference from the effect on the different ranges. The offset voltage effect would effect the ranges that use a smaller FB resistor more.
After longer storage it may take more than 1 hour to get back to a stable (reduced) humidity level.

[coppercone2]

I think a 450 milivolt change is saying something, that is alot

[Keith956]

I opened up the A1 board and gave the area around the input a clean with IPA. It looked pretty clean to start with though. Put it all back together and swabbed the input triax too. Then left it overnight to dry out.

On switchon it was reading -0.350 on the e-12 range, after about an hour this has gone down to -0.230. The e-11 range reads -0.012 and the e-10 range reads -0.001, all other ranges read exactly 0.000.

Could it be a small leakage/offset of the varactors C1/C2? I'm still puzzled why the reading on the e-12 range does not progress from the e-10/e-11 ones, i.e. why is it -0.230 not -0.12x ?

[Keith956]

Well after being on from 8am till 6pm the leakage reading on the 1e-12 range has gone down to -0.135 and has pretty much stayed at that the last couple of hours.

wn1fju, how did you clean your A1 board? Did you swab the whole thing including inside the shielded areas (tricky with the range resistors and reed switches mounted in pillars)?

[coppercone2]

you want to get it wet and probobly brush it, q-tip it, swab it.. hard clean up job with out the high powered spray alcohol. The spray one blasts it hard (the can has a huge amount of hydraulic force), I have seen it blast hairs out of connectors and things like that.

Maybe get the short painters brush (a tight wooster brand), soak it with alcohol, then kinda scrape/poke/brush it alot, hold it on a right angle and get alcohol in a spray bottle and spray it so everything drips off a corner.. you will go nuts trying to clean it with swabs

https://www.amazon.com/Wooster-Brush-Q3211-2-Shortcut-Paintbrush/dp/B002YC06T2/ref=asc_df_B002YC06T2/?tag=hyprod-20&linkCode=df0&hvadid=198071503086&hvpos=&hvnetw=g&hvrand=5786762561291827527&hvpone=&hvptwo=&hvqmt=&hvdev=c&hvdvcmdl=&hvlocint=&hvlocphy=9004230&hvtargid=pla-351656325596&psc=1

[coppercone2]

honestly though the one I linked is like a PCB pressure washer in a can, I have seen nasty dsub connectors that I hit look like I sprayed a real pressure washer into a flower pot.. had to wipe down the entire table after

Now, if you get contact cleaner, like QC brand from home depot, thats also pretty powerful.. much more then say a can of deoxit.

The idea is you want a 'flush' not a cleaner. Flush = pushing power. The alcohol I linked is a good flush. The caig recommended contact flush is strong, I have used it before ( I don't believe its deoxit brand, but its sold through the caig website). I am just not sure contact cleaner is aggressive enough to clean up the stuff causing your problems (it might be benefit from the slight polarity of alcohol dissolving say capacitor juice), I think its pretty non polar.

You can also try washing it with a clean detergent and a kitchen faucet with some power (the spray nozzle), flush with distilled deoinized water, preferably dip the board in a tub, shake it off, then pour clean water from the bottle, dry, spray with alcohol from a normal spray bottle, dry, bake carefully (don't melt it in a oven like my recent post).. the best would be ultrasonic aqueous detergent followed by a rinse and alcohol flush I think. But keep in mind also someone mentioned some types of capacitors might be susceptible to alcohol, I think the can does a good job because its high power, but not ultra high volume, so it gets it out quickly without too much exposure if you dry it off fast with a strong fan.

Also after alcohol if you get some distilled deionized water and really flush that out, it might reduce any possible alcohol related absorbation damage but not dirty it further (it will tend to dissolve into the water and drip off I think).. then you can hit it with a clean contact cleaner when its dry, I think it does have some action against microcorrosion that developed from the water, so long you determine the contact cleaner is clean enough.

I spend alot of time pressure washing shit because my property is setup kind of fancy and I notice alot with dirt. sometimes you need volume of water more then high pressure to get rid of something for some reason

[Kleinstein]

With old low leakage circuits one should be careful with alcohol. PS type caps may be sensitive to alcohol and in the old days these were quite common if low leakage is important.

[coppercone2]

i have been trying to think about how to seal them up, maybe if you press fun-tack around them to make a dome barrier?

do you think they would still get damage from subsequent precision swabbing to get the residual grease off from the putty?

and is there a safer solvent? I only say alcohol because its the only thing a EE knows for some reason. there is THF, acetone, xylene, etc.. most I imagine would be worse but for some reason I think there is probobly some good solvent, maybe esoteric

or you can get heat sink clips and desolder the capacitors (something I should have done if my judgement was not clouded by anger)

[Kleinstein]

PS caps are sensitive to most organc solvents. Alcohol may still be one of the less critical ones.  Water is very good solvent. Alkohol still does not dissolve grease much better than water - but it can help with flux residue.

[wn1fju]

"wn1fju, how did you clean your A1 board?"

I only noticed a little debris/flux on the A1 board right at the input connector.  I removed that with alcohol and a Q-tip.  While I was there, I gently wiped down the 99G
resistor also.  That's all the cleaning I did, followed by the offset/gain adjustment.

But alas, this morning I was getting a reading of about -0.056 on the 1e-12 range after a one-hour warmup.  My guess is that I didn't wait long enough for the alcohol to dry and the thing to come up to temperature and stabilize before I did the adjustment procedure a couple of days ago. 

Also, I know that humidity plays a big role in this stuff.  In my basement lab, the humidity gets pretty high in the summer months.  In fact, I have a Fluke 5440B DC calibrator that simply works terrible (a lot of drift on startup and won't hold calibration for long) over the summer - in the winter it works flawlessly!

Since I had probably not turned on my HP 4140B for a year before a few days ago, my feeling is that I should probably leave it on for a week, then run through the cal procedure again. 

But as I said before, I'm not too concerned about the small offset on the 1e-12 range.  Without proper test fixtures and extreme attention to cabling, there's no way I would ever expect to meet spec on that range.

[coppercone2]

what are those high z resistors jacketed with?

 part of the problem might be when alcohol evaporates its cooling the part and then the part is condensing water on it. cleaning might work better in certain humidity ranges

I bought a dehumidifier to put near my two racks in the basement last year

[Keith956]

But as I said before, I'm not too concerned about the small offset on the 1e-12 range.  Without proper test fixtures and extreme attention to cabling, there's no way I would ever expect to meet spec on that range.

You're probably right. After seeing it go down to -0.135e-12 last night, I powered it up again this morning and it was reading -0.250e-12. 7 hours later it's reading -0.115e-12. So it might just need a week powered up to drift down.

[Keith956]

what are those high z resistors jacketed with?

The seem to be glazed. There are better resistors available - correct me if my thinking is wrong, but leakage on the feedback resistor would just reduce the gain of the amplifier, not affect it's offset with the input  open.

All the immediate input components (R51, Q8, R52/C22/K1) are on PTFE standoffs, maybe there is a surface leakage path somewhere...

[coppercone2]

leakage across the resistor but leakage can also lead a path to the resistor from another junction, I think guard traces and elevation and spacers help with this

[Keith956]

Well after leaving it for a while I returned to why the HP4140B has a high O/C leakage current reading. To recap, after being left for days, when powered up it would read maybe -0.700e-12A. After a few hours this would go down by maybe half. But it didn't want to go down to < 100fA allowing the offset null to work.

Today I removed the board and opened the shield on the top of the board. I noticed the triax connector was not making a good connection with the shield metal plate on the board. As I happened to have a spare triax connector, I removed the original and fitted the new one and soldered it back in. In the process the metal shield plate it's soldered to got quite warm.

When I refitted the board, the reading was positive - about +0.100e-12. However after about 30 mins it had gone back to reading ~-0.350e-12.

I then removed the board again, removed the metal shield on the top of the board that covers the high value resistors, plugged the board back in. It was back reading positive again. But... when I covered the instrument with the top cover, the leakage suddenly dropped and read negative.

After a bit of experimentation it became clear that it was light sensitive! Pulling the curtains made it read negative, opening them and it read positive.

In an attempt to find what was responsible for the light sensitivity, I got my trusty old green laser pointer out and tried shining it on the board. The biggest impact was when it illuminated the 99G resistor or the associated 20pF capacitor. Hard to tell exactly which was responsible, as they are close and illuminating one obviously affects the other.

So - if looks like the next step is to try replacing the capacitor (it'll be easier to source than a 99G resistor, which is probably unobtainium).


-

[Gyro]

Light sensitive leakage is normally down to low leakage glass encapsulated diodes - any input protection diodes in the area? It doesn't take much incident reflected light, even if they're sleeved.

[Keith956]

Light sensitive leakage is normally down to low leakage glass encapsulated diodes - any input protection diodes in the area? It doesn't take much incident reflected light, even if they're sleeved.

No there are no glass diodes nearby.  But I did a bit more investigation. Wrapping a bit of paper round the range resistor/reed switches showed that they were the light sensitive bit.

I then carefully wiped the reed switches glass by looping a Kimwipe under and round it and pulling it gently back and forwards (I'd already tried cleaning the range resistors, to no effect).

Guess what, that appears to have done the trick! The o/c reading on the 1pA range is < +10fA now and appears stable, at least for the half hour its been powered up.

I can only assume there was some sort of 'invisible dirt', as coppercone calls it, on the reed switches. It absorbed moisture when left powered off, and lost it on drying for a while. Why it's light dependent I don't know, maybe the dirt acts like a feeble photovoltaic cell.