Keithley 610C repair

You finally said something about electronics! It's wrong, of course.

Because DA could explain the voltage sag that Phil observed. He explained that to you.

DA doesn't just happen at zero volts; look up the circuit model for a cap with DA. Phil might be seeing leakage and he might be seeing DA; that was all I said. Not having his cap, I can't tell from here.

He could observe the voltage vs time and resolve which is happening, if he cares to. DA would have a different curve than leakage.

Why bother explaining such fundamental stuff?

More to the point, I should just ignore Fields. He doesn't care about electronics.

Follow-up: On 6/2/14, discussing replacing the input stage to reduce drift, you said

Which is it - drift or no drift?

Stable on the 1 pA range. Unstable on the 10 fA range.

Well, the full scale range is switchable from 10 femtoamps to 300 milliamps, so the correct answer is "yes".

Super nice meter though.

Cheers

Phil Hobbs

The discussion was improving the drift on the 10 fA range.

You knew that.

The discussion was to find ways to improve the drift on the 10 fA range.

Would your customers let you get away with such an answer?

Actually, I didn't. I'm seeing only a small, recent piece of the thread.

I don't think the mosfets are the problem on the lowest ranges. The problem seems to be random charge transfers and maybe leakages.

At heart, the mosfets are the front-end of a voltmeter. On voltage-input ranges, even with the multiplier cranked all the way up, it's stable. So the fets aren't drifting. They might be leaking, but I doubt that. A little pressure on the side of the box, bending the metal a tiny bit, makes the 10fA range go wild. Mosfets aren't doing that.

My home-made pA meter was more stable, maybe because there wasn't so much stuff inside to make e-fields.

And, no range-changing stuff, right?

Sure it changes ranges. By plugging different resistors, on Pomona plugs, into one set of jacks.

I have Z1 plugin resistors from 1K to 1T.

The lexan was added after I discovered that those binding posts are super leaky.

Hang on to that Wall wart. THey don't make transformer warts anymore, just noisy swithers. ;)

Cheers

I don't recall ever having been asked such a stupid and hostile question by a customer.

Cheers

Phil Hobbs

I think it's mostly mechanical instability combined with fairly large internal E fields. The power supplies are +- 55V or something, and all the resistors are hanging in midair. If you've got 100V gradients running around, with a node capacitance of 5 pF and a typical separation of an inch, you can get 10-fA currents from a velocity of

~ 25 mm * 10 fA / (100V*5pF) = 0.5 microns per second.

That's an average, of course, so in real life the motions would have to be much bigger. Still, it makes sense that flexing the front panel (where the switch is mounted) by a millimetre or so might make it go nuts.

A bit of strategically placed foil inside would probably help a lot.

The combination of a 100-fA FS homebrew box and a 610C is a pretty good solution. I like your pluggable thing, which is nice and flexible. For current measurements, making the loops and nodes smaller would make it faster and more stable, so I'd lean towards a fixed 1T resistor and a metal box meself.

Cheers

Phil Hobbs

We have a big box that must have 100 old warts in it. We really should throw some of that junk away, but you never know when you'll need a 17 volt AC wart.

Well, when you get round to selling your building, you can probably afford to store a couple more warts, even. ;)

Cheers

Phil

I built that before I got the Keithley, and still use it now and then. I have an aluminum chassis box that drops over the top to shield Z1 and Z2 when needed.

I was looking for a low-leakage diode. The PAD1 is sold as such, but it's actually a jfet with s and d shorted. It has lots of series r, lots of shunt c, guarantees 1 pA, and it's expensive.

Turns out that small transistors can be excellent low-leakage diodes. The c-b junction of BFT25 is amazing: sub-pF capacitance, good forward conduction, leakage hard to measure, 10 fA maybe.

One could charge a film capacitor from diode leakage, for hours or days, and probably resolve attoamps.