Electrometer Amp: Darlington NPN

I soldered up the simplest possible “electrometer amplifier” at Squidwrench, based on Charles Wenzel’s writeup:

Electrometer Amp - MPSA14 NPN Darlington
Electrometer Amp – MPSA14 NPN Darlington

It’s an MPSA14 NPN Darlington transistor, with the base soldered directly to the Victoreen 710-104 ionization chamber collector pin. The flying leads connect to an ordinary digital voltmeter set to read voltage, rather than current, so that you see the voltage created by the transistor’s collector current through the meter’s input resistance.

The MPSA14 data sheet specifies DC current gain hFE > 10 k for low collector currents, with a graph suggesting it might be somewhat larger. Alas, all those are for “ordinary” currents, not the countably finite number of electrons coming from an ionization chamber, but let’s assume 10 k is close.

I used a Radio Shack 22-805 DMM, set to auto-ranging DC volts. The specs say the input “impedance” is 10 MΩ for all voltage ranges, so let’s run with that, too.

With 24 V (actually 24.6 V) applied to both the chamber (through the red wire) and the DMM (through the yellow wire), it read 250 mV: a mere 25 nA through the 10 MΩ meter resistance.

Assuming a transistor gain of 10 k, that’s a chamber current of 2.5 pA.

The ionization chamber specs say it produces 5 pA at 0.5 röentgen/hour → 100 mR/h produces 1 pA.

No, I do not believe the Squidwrench Operating Table is bathed in gamma radiation at 250 mR/h.

I should wipe down the transistor to see if that reduces the external leakage, then try a few others, but obviously the signal will remain lost in the noise.

We replaced the DMM with an oscilloscope and 10 MΩ probe, which conclusively demonstrated that unshielded high-impedance circuits make excellent 60 Hz receivers.

6 thoughts on “Electrometer Amp: Darlington NPN

  1. In my early days, I was product engineer for the National LM216 darlington superbeta opamp family. According to the old datasheet, max Ibias is 50 to 150pA, depending on grade. We were having problems failing QA after the parts had been re-marked at the (notably non-cleanroom) mark and pack operation. I couldn’t figure it out, but Bob Dobkin had a hunch and took some of the parts to the men’s room and subjected them to the hand soap. The Men’s Room Clean did the job… (Second hint–breathing near the parts caused the bias current to soar–far more than an unmarked part.)

    As it turns out, doesn’t matter what package you use–my recollection says we were doing this on can parts, but plastics should respond similarly. Prototyping the clean, we ended up with some liquid detergent, DI water and an artist’s brush. Your mileage will vary.

    1. breathing near the parts caused the bias current to soar

      Wisely is it written that any sufficiently sensitive instrument is a thermometer. The output takes quite a while to settle down after turning on the power, even for a circuit consisting of a couple of transistors…

      I’m not convinced the alcohol wipedown had any effect at all, but I did establish that a Q-tip isn’t the right hammer for the job. [sigh]

      1. In the dirty darlington opamp, it was a sensitive hydrometer. IIRC, you need an immersion clean and a decent rinse. (One of the laboratory spray bottles works wonderfully too, but a household spray bottle should do the trick.) Sounds nasty to dunk electronics, but it works–once it dries off. I’d use off-the-shelf distilled water for the wash and rinse.

        1. It looks like I can get away with AC coupling the output to ignore all the leakage, because I know the Basement Laboratory isn’t bathed in gamma radiation. Just saw some spikes that look like they should be gamma ray ionization events, so this absurd lashup looks like it’s kinda-sorta working…

  2. Having similar issues debugging an energy harvester. Not only are the currents in the nA/pA range, they’re all AC waveforms of some sort, and there are some bizarre ground loop issues due to the circuit topology.

Comments are closed.