Cooling the electrometer front-end

Piotr Wyderski · 2022-05-24T18:37:34+00:00

Gentlemen,Since there is a discussion about ovenized oscillators, here's something for restoring the broken symmetry.I am planning to build an electrometer based on ADA4530-1. Due to the insanely low leakage currents required, a regular PCB seems to be a no-go. This instrument will be a one-off, so I can safely ignore the usual DFM rules. Namely, I am considering a dead-bug arrangement of the critical components and using air as insulation. For mechanical stiffness and reasonable thermal conductance to avoid gradients, a ceramic base plate would be one idea. But the ceramic can be a part of a TEC module. Question: would it be worth the trouble to cool the ADA4530-1 (and maybe the ADC and the reference) to ~10 degrees C, in terms of noise/thermal input offset etc.? Best regards, Piotr

The highest-impedance thing I ever did for a production application was a 1 G / 50 G TIA for a scanning surface voltage tool used in semiconductor manufacturing.

Here's a post about it from 2013. (Ignore the stuff about reverse-polarity diodes--there was a package outline mixup, long fixed.)

Cheers

Phil Hobbs

========= On Mon, 09 Sep 2013 17:15:01 -0400, Phil Hobbs snipped-for-privacy@electrooptical.net wrote in <msmdnROcIY7Lp7PPnZ2dnUVZ snipped-for-privacy@supernews.com:

On 09/09/2013 04:27 PM, Jan Panteltje wrote: > On a sunny day (Mon, 09 Sep 2013 16:02:25 -0400) it happened Phil Hobbs > snipped-for-privacy@electrooptical.net wrote in > snipped-for-privacy@electrooptical.net: >

Might be possible, but the package is a lot taller above the leads than below.

There are reverse-polarity Schottky pairs, but the ones I've found are all 4 volt RF mixer/detector things.

The application is a super high-Z front end, where I need to bootstrap the contact-to-coil capacitance of a relay. The circuit is like this. K2 is single-coil latching, and switches the actual feedback resistors, while K1 engages disengages K2 for about 30 ms after a transition of either polarity. (The coils are both about 1.3k, so 650 ohms * 47 uF ~=

30 ms.) K1 is the wrong way up, so I need to be able to invert the polarity of D1 and D2 without a board spin. *------------*------------* | | | 0 *---* D1A --- | --- | | A | D2A A -12 --* *--- | . C + | *-----* . O K1 *------*---------0 From gate | | . ..I | | driver IC | --- . L - --- | | D1B A . | . D2B A | | | . | . | | | *----.-------*----.-------* | | . . | | . . | | . K1A .K1B | | /0 . 0\ | *----0/ + K2 - . \0----* | 0---*---COIL-------*---0 | | . | | | . | CCC *--.-RRRR--||--* CCC 47uF | . | | CCC . CCC GND CCC . CCC | . | *--.-----------*---< Bootstrap . . . 50 G total *------RRRRRRR-.--RRRRRRR------* | . | | . K2A | *--------------0\ 1G | | \0----RRRRRR--* | 0 | | |\ | | | | \ *---------------*----0 Vout 0----*-------| -\ | I_in | \ | | >-----------------*

... etc....

It's interesting that the contact capacitance of K2 is large enough (0.2 pF or so) that I have to short out the 1G resistor. If I'd let it float, its Johnson noise current would have dominated at high frequency, i.e. above a few hundred hertz. The capacitance itself is noiseless, so shorting out the 1G resistor allows the frequency response to be fixed in the second stage without hurting the noise.

Cheers

Phil Hobbs