Grasp the nettle firmly?

Try reading Ralph Morrison's "Grounding Shielding Techniques".

My copy is the fourth edition - ISBN 0-471-24518-6 - published in 1997.

The third edition was published in 1986, and there is a fifth edition published in 2007. Apparently the second edition was published in 1977, so what I read as a graduate student in the 1960's would have been the first edition, which was pedagogic magic.

He does emphasise that stray capacitances are important, and drive currents through your notionally equipotential screens, which have finite resistances, and thus potential differences from one place to another.

Semi-rigid coaxial cable is the kind of screened structure you want, with 100% solid screening and low resistance sold copper for the induced currents to flow in. Triple screened coax, with a layer of metalised Mylar to provide a genuinely continuous screen (and braid to carry the induced currents without too much voltage drop) is almost as good, but none of them are perfect - but if you think carefully about what currents are flowing where you can often improve things no end.

Nice.

Conductors moving around in an electrical field will pick up signals. You need shielding and solid, stable construction.

Jeroen Belleman

In addition to the other excellent responses:

Beware of Teflon (PTFE) insulated cable. It is extremely microphonic in itself. In very high- impedance circuits, it may be unavoidable, so you have to shield the whole thing against sound and jolts.

The reason for standard cells' mercury chemistry, is that there is no gas evolved or absorbed in normal operation. If there were, the battery voltage would respond to partial pressure changes of its produced gas... I've heard that alkaline batteries have a sawtooth time signature, as pressure builds, then the vent 'burps'.

it sounds like your problem is soluble in several ways: (1) regulator (Zener, bandgap) and add lots of capacitors/capacitor multipliers (2) experiment with several different battery chemistries (no need to try zinc-air, those are well known to be P sensitive) (3) do something exotic, like using a reversed thermoelectric module and a (regulated) heat source (4) discover a box of old mercury batteries that has been in a researcher's hoard for a quarter century or so

Back at home now... this feels like an out of work project... and a busy day at work. I've got the 3rd ed. of R. Morrison's book. I just got it off the shelf. Most of that is way over the top for this. I think the above circuit needs to "know ground" at some point. Where and how much?

George H.

Yeah well there are lots of issues... As the cable from the bias box to probe swings around the DC level goes here and there, ~100 uV -wise.

But all that will be low frequency, as long as you don't bump it.

George H.

Yeah, There is twisted pair wrapped around 2 meters of grounded probe. FIR light pipe on the inside and vacuum can on the outside. Some thin wire in "thick" shield.. good to 4K. Teflon.

George H.

Yeah, we soldered up a four pack of AA's. And that was much better. I was thinking all about cap multiplier's and simpler RC filters. But I don't think the pickup is getting in through the power supply.

It smells like a ground issue. Where to grab and how hard?

I was thinking of how to make a dummy signal at room temperature. Maybe an LDR and LED... with the same 10 meg R's (minus the LDR)

George H.

No mystery at all, have seen this effect at much greater distances, it is thermoelectric air currents induced by the sound pressure waves, and it doesn't take much.

That takes dissimilar metals, but it's fairly easy to check. Because the box interior will make standing waves, compression/rarefaction due to the sound is the heating mechanism to expect. Using thick insulation on the internal wires (thermal insulation, really), or adding heatshrink, will be effective. Wadding up some Sorbothane might be useful, too.

If the SS probe were a problem, it'd be good to add thermal mass to the outer shield at the attachment point to copper wires.

Huh? I'm not sure what that is, but these signals are "big" ~100 uV. Battery vibration right into the local ground.. bypassing the whole detector network. Well not bypassing, but 20 Meg on one side and ~60 pF on the other.... well there's ~60 pF on both sides. So the differential is big.

Ground at the battery up top (a), the detector down the bottom (b), or something else (c)?

I'll try a and b this Friday, as long as things go according to plan.

George H.

I interpreted this as compression/rarefaction of gas, and attendant adiabatic temperature change, interacting with thermocouple effects. It's unlikely that would get to 100 uV at a few hundred hertz, but that depends on heat capacities that are unknown, and it DID seem to have some resonant character (which could build up a small signal).

Pressure effect on battery voltage was my chief suspect, but the noise persists with batteries out of circuit.

Oh no, my mistake if I said that. No acoustic pickup, with the battery out of the box. Pickup with the battery in the box, but only one lead connected, the other one just floating!?

George H.

Are the batteries metal-cased, and are the cases capacitively or conductively coupled to the box and ground?

Batteries are standard 9V and in a blue powder coated Pomona box, So I assume no DC conductivity, but plenty of capacitive coupling.

Friday was very productive, AA's are much less noisy, we grounded one side and made things better... I'm thinking that maybe the student and I can try measure the noise... see if it's mostly Johnson noise from the resistors.

I made this dummy signal with a fet shorting a resistor.. and I had a few issues. (See upcoming thread in SEB.)

George H.