A 100V battery bias supply for picoammeter measurements

[Gyro]

A while back I was trying to measure the leakage current of an electrostatic voltmeter with my low cost Picoammeter. I suddenly realised that I didn't have a sufficiently quiet low / no ripple bias supply for testing the leakage current of devices with any significant parasitic capacitance (or a capacitor), where ripple will swamp the Picoammeter input.

Not having the luxury an SMU, I don't know how other people handle this problem, but a reasonable voltage fully screened floating battery supply seemed like a simple answer. Our local Lidl also happened to have 230mAh CR2032 cells at less than £2 for a pack of six on the centre isle, which made the idea seem reasonably economical too. I settled on 100V as a suitable voltage - not that much help for a 1kV electrostatic voltmeter but probably adequate for most components that I might want to measure - 34 cells provde around 102V for the flat portion of the LiMnO² discharge curve.

Construction proved not to be much of a problem. I picked up a short length of 21mm ID acrylic tube off ebay. The stack of cells is held under light spring pressure and a combination of metal and fibre washers (Lidl again). The two washers that actually make contact with the cell stack came from my junk box. I don't know what the plating on them is but it is readily solderable and shows continuity at the lightest touch of a blunt meter probe. The whole thing is enclosed in a piece of square section Ali extrusion with fabricated sheet ends and supported by O rings and foam. I included a polarity switch, the only sensible place to put one with a screened output.

I haven't built the test chamber yet, but I intend to put a 10 x 1M resistive divider inside to provide 10V steps. With the 10uA load of the resistor chain, the load life of the battery stack will be thousands of hours and, with practical usage, I should still achieve the expected shelf life of 10 - 15 years.

[MiDi]

Another simply method is to stack 9V Batteries.
11 in series will give ~100V and cost ~15$ for ~500mAh Alkaline.
Shelf life is around half of CR2032 though, but otherwise they have better bang for the buck.

[Gyro]

Yes, I did seriously think about 9V batteries, they're certainly much easier to clip together as needed. I wanted a flatter discharge curve than Alkaline though (and something a bit more compact than 11-12 batteries). At very low load, LiMnO² give a virtually flat 3V for 80-90% of their shelf life.

[DavidKo]

Lidl batteries has short shelf life, but who cares for that price. Since you have small current loading maybe you can combine different types of batteries and this can be even cheaper solution. There are 24 CR20xx batteries . Pity that they had stopped selling 9V accumulators (at least in our country).

[Gyro]

Ha, yes. That would have made a more interesting battery stack, maybe I could have introduced different voltage taps with higher mAh rating on the lower voltages. I've never seen Lidl sell 9V rechargeables (I have looked too). They sell a NiMh charger for AA, AAA, and 9V but alas no 9V batteries to go with it. A shame because their other NiMh cells are really good.

[langwadt]

https://youtu.be/8hwLHdBTQ7s

[Gyro]

Ha yes, I've seen that video. That guy must be suicidal - I shudder to think how much cell reversal was going on there! One false move and he would have fried himself, at least he could have put insulating spacers between the batteries. It's another reason why I went for CR3032s, I don't know their short circuit current, but I'd be surprised if it is is more than a few 10's of mA (maximum recommended pulse drain is around 12mA). I haven't tried it because I'm also worried about cell reversal (and  severe loss of capacity). The closest I got to frying myself was a minor belt off the rear of the BNC when I was fitting the end plate, (predictably) no worse than I used to get when playing with old 90V radio batteries and Neons.

I'm still wondering whether to include a current limiting resistor, say, 100k in series with the output. Obviously a fuse or PTC would be no use. The more I think about it, the more it seems like a sensible idea. It would drop the voltage into a 10M voltage divider, but not excessively, and I only want a rough measurement of leakage currents anyway.

[Kleinstein]

If a current limit is needed with relatively low series restance in normal operation, one could use a mode MOSFETs with a resistor at the source, a little like the JFET current source, just higher withstand voltage. The MOSFETs turn off at some 1 V (depends on the type) and thus relatively little resistance needed (e.g. 1 K to get a 1 mA limit).

[Gyro]

Thanks Kleinstein, I must start considering active solutions sooner... and lay in a stock of depletion mode MOSFETS.

Now I actually think about it, in this specific application a 100k resistor will only drop <1V / 1% too when feeding into a 10M divider, so is effectively insignificant. A 1mA short circuit current will give me 100+ hours to discover my (hopefully one time) mistake and still retain useful capacity. I can't see me wanting to go for a lower impedance divider, 10M will allow long run times without the threat of having to replace the cells in the forseeable future, and is an insignificant source resistance when measuring pA level leakage currents.

[r6502]

Hello all,

have a look here:
veenstra (comecer) VIK 202

Here a bias supply for an ionisation chamber was also generated with standard lithium bateries.

Guido

[Gyro]

That's interesting, thanks. I hadn't seen that teardown, it's nice to see that someone has used this approach commercially.

I wish I'd had the benefit of a tab welder!

[David Hess]

Jim Williams published some designs for low noise high voltage power supplies.  Sine wave drive through a step-up transformer into a voltage multiplier is about as clean as it gets.

[langwadt]

Jim Williams published some designs for low noise high voltage power supplies.  Sine wave drive through a step-up transformer into a voltage multiplier is about as clean as it gets.

and I guess if you don't need it constantly it doesn't need run all the time, only when you need to charge a cap

[David Hess]

Jim Williams published some designs for low noise high voltage power supplies.  Sine wave drive through a step-up transformer into a voltage multiplier is about as clean as it gets.

and I guess if you don't need it constantly it doesn't need run all the time, only when you need to charge a cap

That can be done, and leave open the possibility of only making measurements when the bias supply is off.  Then a simpler but noisy supply could be used, like a boost or flyback converter driving a voltage multiplier.  For leakage measurements, the resistive divider which controls the output voltage is likely the largest load, so some effort should be made to use a high impedance divider.

[r6502]

Hello All,

I think, David was thinking of LT AN 118

It is really interesting to read - if you need a HV supply with low noise read this article.

It is also interesting, as it explains how to measure noise, so really recommended, and still avelable from the ADI webside.

Guido

[Gyro]

Yes, I've read that app note, it's one of my favorites (together with the really good one on CCFL inverters and true RMS measurements, HP3400A analysis etc.*) his apendices tend to be more intersting than the notes themselves.

I freely admit that the battery is the cop-out approach - minimum noise and short term drift for minmum design effort (100k series protection resistor now added).


EDIT: * AN61

[iMo]

Jim Williams published some designs for low noise high voltage power supplies.  Sine wave drive through a step-up transformer into a voltage multiplier is about as clean as it gets.

A colleague of mine made this design within his diploma work (mid 80ties) as we were working in a lab focused on 1/f noise measurements and we wanted a clean spectra. As I can remember he spent 98% of the time dedicated to his diploma work with the transformer rewinding exercises 

[Wolfgang]

Another simply method is to stack 9V Batteries.
11 in series will give ~100V and cost ~15$ for ~500mAh Alkaline.
Shelf life is around half of CR2032 though, but otherwise they have better bang for the buck.

I did this and it worked fine. See here:
https://electronicprojectsforfun.wordpress.com/power-supplies/battery-operated-power-supplies/a-120v-40ma-battery-operated-power-supply/