HP 4140B Picoammeter

[Swainster]

I thought someone here might be interested in this HP 4140B repair.

I fairly recently bought a couple of broken 34401As with the idea of making one good one out of them (another thread). While playing with my new toys I noticed that the DC current range of these fairly high spec (for their day) DMMs had no more resolution at the low end than my el-cheapo Uni-T bench meter. This could not stand, and I started looking out for something a bit more specialised. Googling mostly turned up stuff that was out of my price range, until I got a hit on a local classifieds website. This HP 4140B was a real boat anchor but it had picoammeter in it's name, and was a piece of rather specialist kit which had been superseded by much more modern stuff, leaving it somewhat unwanted, and close enough that I could go and pick it up in person. In other words, it was perfect!

As it turned out, it was being sold by an ex-EEVBlog forum member, who will hopefully return to the forum at some point. He explained that it passed self test but that he hadn't really tested anything else out, and didn't have any cables for it (other than the IEC lead). I hauled it back and started to put it throught its paces, as best I could.

" alt="" class="bbc_img" />

[Swainster]

The first problem was that the ammeter input is triax. Actually no, scratch that. The first problem was a lack of documentation, but I solved that with a visit to the Artek manuals website, even before getting the 4140B home.

Next, a quick google for triax cable assemblies brought me out into a cold sweat. Seems like a cable could cost almost as much as I paid for the instrument, or more once I added a few receptacles. Luckily I managed to score a brand new pomona triax patch cable for under USD50. There was only one in stock and I actually wonder if it had been accidentally under priced as it was about USD100 less than anywhere else online. While that was in transit I got on with testing the voltage channels.

Did I mention the voltage channels? This is one of the cool things about this old box. As it was originally intended for semiconductor characterisation, it has 2 programmable +- 100V sources (up to 10mA each). This actually makes it the widest range "supply" in my toolkit (not counting the 10KV insulation tester). Potentially (if you'll pardon the expression) it could output 200V between the 2 channels. And here is where I hit the first problem. Turns out that the fancy channel A (with programmable ramps and suchlike) was non functioning. I guess that it probably had the most use during its working life. Channel B (only programmable for a fixed voltage) was working fine, and still within spec according to my 34401A. At this point I took a break to get back to proper work.

[Swainster]

A few days later the triax cable arrived so I could give the ammeter a go... well, sort of. I needed something to connect to the end of the triax. This turned out to be a Trompeter cinch panel mount receptable - close up, a thing of beauty, as well it should be given how much it cost. This was fitted to a metal enclosure bought from a local electronics store, and because it's a picoammeter, a challenging resistor was mounted. The manual requires a range of resistors up to 100GOhm for performance testing, but just getting a 100Gohm resistor would cost me USD45 in postage alone, so I settled for a more reasonable 10GOhm, in the form of 2 x 5 GOhm 1% parts from Ohmite. These were mounted in the little metal box, being careful not to get any mucky finger prints on it.
" alt="" class="bbc_img" />

This was connected up between the ammeter input and voltage channel B, and the output button was pressed...

Yes! - the ammeter was working fine and the readings were once again pretty good, reading 9.94pA for 0.1V/10GOhm. Take that Uni-T! Interestingly, if the resistor enclosure was left open, just waving a hand near it was enough to send the reading all over the place.

" alt="" class="bbc_img" />

Of course, one of the next jobs will be to complete the set of test resistors...

[Swainster]

What's that about the repair?

Well, poking around inside the machine (very easily accessible via a single screw which holds the lid on) revealed that the voltage channel A problem was due to a faulty relay with an open circuit coil. I should probably note here that there are test pins near the top of the machine which have + and - 120V on them, so poke around carefully. A new old-stock coil was sourced from sphere (in canada), the only one they had available. It is currently still making its leisurely way to Singapore however further poking around revealed that actually 2 relay coils had gone open circuit. No luck was had sourcing another coil, and the pitch was unusual (thanks Coto/HP), so I couldn't find a pin compatible modern alternative. The service manual indicates that it is a 12V coil with a 300V reed. In addition, by then I had managed to break one of the original reeds extracting it from the coil so things were getting desperate.

Enter well known far east budget PCB house and their $2 for 5 boards offer. A quick adapter board was drawn up to fit a modern Hamlin/littlefuse 300V relay, and because I felt like splurging ,I went the whole hog and spent $5 to get 30 boards. As it's well into the wee hours here I shall stop waffling and suffice to say that the replacement relay and adapter boards fitted nicely and worked as intended, and channel A was restored to normal operation, fancing ramping action and all.

So... If anyone needs an adapter for an alternative relay (Hamlin HE3351A1200) for the HP 4140B then send me a message or something. I can send out a spare board or 2 for postage costs. Also, I dismantled one of the failed coils to see if there was any prospect of repairing it. Turns out that the winding had developed multiple breaks - presumably due to corrosion. These all seemed to be near a strip of yellow tape that Coto used, so I'm wondering if something in the tape adhesive went to work on the copper or the lacquer.

[Keith956]

Nice work!

I also have a HP4140B (well actually I have 2), one needed a careful clean of the very high ranging resistor network with IPA  (be careful here, the most sensitive range uses a 99Gohm resistor that is unobtanium, although I guess a 100G would do at an expensive pinch). Also to get the leakage down I replaced the triax socket with a new one - it probably had a little dirt in it. With the cleanup and the input open circuit the meter will settle at 0.005e-12A after an hour's warmup. I also replaced the fan with a new quieter Papst one as the original made a bit of a racket!

Measuring anything sub-pA is a challenge. A good quality triax cable and some sort of shielded box is essential. I have a 16058A test fixture (for a HP4145B originally) that works quite well. Reverse leakage on diodes is easily measured and tends to be many pA for most, well within the 4140's capability.

Next job is to write some code to talk to it from a PC so I can plot IV curves.

[Swainster]

Plot IV curves? Isn't that a curve tracer lurking in the background or your pic?

I have to admit that I haven't had time to "play" with this instrument much yet. So far the only practical thing I've used it for was as a high voltage supply to refresh the patchy digits in an old nixie tube.

[Hydron]

Next, a quick google for triax cable assemblies brought me out into a cold sweat. Seems like a cable could cost almost as much as I paid for the instrument, or more once I added a few receptacles. Luckily I managed to score a brand new pomona triax patch cable for under USD50. There was only one in stock and I actually wonder if it had been accidentally under priced as it was about USD100 less than anywhere else online. While that was in transit I got on with testing the voltage channels.

Wouldn't have happened to be Arrow by any chance? They had a pomona triax assembly very cheap, would have grabbed it myself if it wasn't ~impossible to buy from them since brexit (no DDP incoterms to the UK any more, and it seems they now require a VAT ID too).

If you're looking to make up more cables, you might be able to get by using connectors for MIL-STD-1553 - they are designed for twinax but the concentric versions can be used with triax in a pinch too. I have managed to salvage triax panel connectors from 1553 stub bus couplers too (sometimes available very cheaply) but it's hit and miss - you need to find ones with rubbery potting rather than a hard compound.

If you can get the connectors then triax can be had from a few of the usual suspects by the meter (I think Farnell/element14 had the Belden cable used by pomona, Mouser has some Huber&Suhner options including a low-noise cable)

[Swainster]

Indeed, it was Arrow.

On the triax front I was OK with that short pomona cable for a while as the 4140B only has one triax output, but I just pulled the trigger on a nice brown box with 3 of the buggers on the back. I say "nice", but the box in question was advertised as having various ailments. That's a project for another day though, as I'll be travelling back home for Christmas.