KV/µA source, how do you measure the output?

[Leuven]

I have a 100KV source at 100µA, and trying to measure the output with a multimeter.

I cobbled together a voltage divider, but it reads about 10 times less than what it is supposed to. Can I use an op amp as a buffer?

[Ian.M]

No. You can't do it that way.   Either the OPAMP input will rail or it will die as there's nothing to stop the current to the +in through the 100Meg resistor taking the input past its Vcc rail except whatever internal protection network it has.  You *NEED* a lower arm in the divider.   Also a 100Meg resistor with 100KV across it would pass 1mA, which is ten times what your EHT source is rated to output.  You need a minimum of a 1G upper resistor in the divider, which would then draw 100uA, and if you are interested in the EHT voltage when the source is lightly loaded a 10G resistor would be preferable.   Doing that cheaply with reasonable accuracy is going to be challenging.

[ArthurDent]

I don't know how accurate you have to measure the voltage but D.C. voltages of that level can be measured by an electrostatic voltmeter and there are videos showing how to make a DIY unit. Calibrating it could be tricky but it is a way to measure high voltage with basically no current being drawn.

https://en.wikipedia.org/wiki/Electrostatic_voltmeter

Another way would be to look up the distance a spark will jump in dry air and do the calculations on how large a gap your supply voltage will jump. Also check articles on Van De Graaff generators to see how they do it.

[Leuven]

Thanks Ian, so there's so shortcut to it. I figured the impedance is not high enough, but I clearly don't understand how op amps work. I will add more resistors. Accuracy of the measurement is not crucial, I just need to confirm the KV output is within some 5% of what it displays.

[Leuven]

I don't know how accurate you have to measure the voltage but D.C. voltages of that level can be measured by an electrostatic voltmeter and there are videos showing how to make a DIY unit. Calibrating it could be tricky but it is a way to measure high voltage with basically no current being drawn.

https://en.wikipedia.org/wiki/Electrostatic_voltmeter

Another way would be to look up the distance a spark will jump in dry air and do the calculations on how large a gap your supply voltage will jump. Also check articles on Van De Graaff generators to see how they do it.

Interesting ideas, thanks. I reckon it's more involved compared to a voltage divider. It'd be worth doing as an experiment.

[JustMeHere]

Two finger multimeter 

[joeqsmith]

I would be interested in seeing this supply you plan to measure.   Can you post a picture?

You can't drive much of a load at 100uA.   I made a video showing a vintage meter that would work assuming the output is DC.   
https://youtu.be/n7z9DjCaxD8

[Leuven]

I would be interested in seeing this supply you plan to measure.   Can you post a picture?

You can't drive much of a load at 100uA.   I made a video showing a vintage meter that would work assuming the output is DC.   
https://youtu.be/n7z9DjCaxD8

Hehe, nothing as interesting as the stuff you do I'm afraid. It's an electrostatic powder coating gun, I'll post a picture tomorrow. The control unit outputs 6-21VDC at 1 amp into a voltage multiplier (found inside the gun), and at the tip of the gun you have a thin electrode. Output is 25-100KV. Powder shoots past the electrode, gets an electrostatic charge and wraps nicely around your (previously grounded) part.

Problem is, these multipliers have a relatively high rate of failure, compared to the rest of the equipment. And the manufacturer wants something like 1500 USD for a replacement. I wanna keep on top of it, i.e. test the output from time to time.

[Dundarave]

Would it be possible to monitor the voltage & current going into the multiplier and measure the difference in power usage between when the powder coat is flowing and when it is not?

It may not give you an absolute value for the output (in terms of either voltage or current), but by comparing the results with a known good system, you could possibly detect a drop in efficiency and infer some kind of go/no-go metric to inform a decision to replace the multiplier when the metric drops too low.   Depending on the efficiency of the multiplier, you might be able to come relatively close to a realistic measurement.

[Leuven]

Would it be possible to monitor the voltage & current going into the multiplier and measure the difference in power usage between when the powder coat is flowing and when it is not?

It may not give you an absolute value for the output (in terms of either voltage or current), but by comparing the results with a known good system, you could possibly detect a drop in efficiency and infer some kind of go/no-go metric to inform a decision to replace the multiplier when the metric drops too low.   Depending on the efficiency of the multiplier, you might be able to come relatively close to a realistic measurement.

I think I could do that, figure out nominal board output voltage and current at idle and controlled load, and keep an eye on it with an extra voltage/amp display. It's reasonable to assume there will be a drift, if the multiplier is malfunctioning. It's a catch 22 though, I don't have access to another system. I'll try the voltage divider first, but would sure be handy if I implemented your idea..

[Dundarave]

It's a catch 22 though, I don't have access to another system.

If you measure a solid difference in power flow between when the powder coat is flowing and when it is not, you really wouldn’t need a comparison system: what else could account for the power usage except charge flow via the powder?

 Then you just use your values measured today as your baseline for the future.

[Ian.M]

That seems troublesome to build.   Only the 10x 1G resistors need to be in oil or potted.    It would probably be a lot more convenient to get axial resistors and assemble them into a thick walled Perspex tube that can be used as the EHT probe and fill it with melted petroleum jelly. Perspex is OK up to 90°C for short-term use, so you can use a hair-drier or hot air gun on low to keep the jelly melted while you tap out all air bubbles.  Why petroleum jelly, not oil?  Well its a decent dielectric, and if the seals aren't perfect it wont leak so readily as oil if you can keep it below its melting point, yet its still very easy to clean off the resistors if you ever need to do any maintenance.  The tube ends could be plugged with neutral cure silicone sealant, but my preference would be brass plugs with double O ring seals.   

Although screening the 10G resistor assembly is desirable it severely increases the dielectric stress on the perspex tube, and you would need 100KV insulation over the screen at the 'hot' end. Also the effect of the distributed capacitance to ground from the resistor chain to the screen would be to over-voltage the first few resistors during fast transients e.g if the EHT source sparks to the probe or to ground.

Direct readout on a DMM on a 2V range should be possible - its input impedance will be  high compared to the 100K lower resistor in the divider. Put a good quality 0.1uF capacitor across the 100K resistor to swamp the stray capacitance across the resistors in the divider chain.  Use a NE2 neon bulb in parallel to protect the DMM if the 10G resistor chain breaks down or flashes over

Caution: if the ground lead comes off, it will blow the DMM.

[Leuven]

I think the easiest and cheapest way is still the voltage divider. How does this sound: 50 x 36MΩ resistors + one 470KΩ, all rated for 3.5kV. I can get them cheap - about £15 all in. Current flow 50µA across the resistors, accuracy well within 5%.

[Ian.M]

It depends on the length of the resistors as 50 in a row is going to be unreasonably long.  If you need 20mm each including joints that's a 1m long EHT probe!   Folding or zig-zagging the resistor chain makes it a *LOT* harder to manage the stray capacitances and stress on the insulation.  Personally I'd expect anything over 60cm long to be pretty unwieldy to use.  Fewer, higher value, higher voltage resistors would be preferable.   Of course you could put the resistor chain in an oil bath separate from the probe, but then you'll need >>100KV rated flexible EHT cable and there's a much greater risk to operator safety.

[joeqsmith]

Dang, that's an expensive repair bill.   I assume the end goal is to save money.   Do you think that the output degrades and you could have it serviced before a major failure happens?   Or maybe know when to order a new part before it totally fails?   IMO, it seems like you would just keep a spare part on hand (assuming it's for a business) to avoid down time or maybe find a more reliable gun.   

[Leuven]

It depends on the length of the resistors as 50 in a row is going to be unreasonably long.  If you need 20mm each including joints that's a 1m long EHT probe!   Folding or zig-zagging the resistor chain makes it a *LOT* harder to manage the stray capacitances and stress on the insulation.  Personally I'd expect anything over 60cm long to be pretty unwieldy to use.  Fewer, higher value, higher voltage resistors would be preferable.   Of course you could put the resistor chain in an oil bath separate from the probe, but then you'll need >>100KV rated flexible EHT cable and there's a much greater risk to operator safety.

You're right, I can bring the number down some.

[Leuven]

Dang, that's an expensive repair bill.   I assume the end goal is to save money.   Do you think that the output degrades and you could have it serviced before a major failure happens?   Or maybe know when to order a new part before it totally fails?   IMO, it seems like you would just keep a spare part on hand (assuming it's for a business) to avoid down time or maybe find a more reliable gun.   

I'm pretty sure a replacement can be bought from a Chinese manufacturer at a much lower cost, that's one option to go about if it breaks. Some prefer to buy the brand name at full price, I suppose it all depends if you're making money with the equipment or not.

For my part, this is a home workshop and I bought the unit second hand. Before anything else, I want to confirm it's working ok. The manufacturer also recommends periodic checks, a problem (like a grounding issue) can creep up and lower your output voltage, and you wonder why you're not quite getting a good result with your wrapping. Maybe it's in the multiplier, maybe it's your setup. It's just good to have eyes on that output, for any eventuality.

[joeqsmith]

Did the manufacture specify how to test it, what equipment was needed and such?  If not you could ask them what they recommend.  At least this way you are following their guidelines so if something is suspect, you can at least avoid having to explain your using something home made.  If I were a supplier and heard this, I would question the results no matter what.   

It may not be all that expensive.  Maybe something like a fencer (tester for electric fences).   

[Leuven]

Did the manufacture specify how to test it, what equipment was needed and such?  If not you could ask them what they recommend.  At least this way you are following their guidelines so if something is suspect, you can at least avoid having to explain your using something home made.  If I were a supplier and heard this, I would question the results no matter what.   

It may not be all that expensive.  Maybe something like a fencer (tester for electric fences).   

They do sell a test kit, which is basically a DMM and a fancy 100G high voltage probe. For megabucks, I'm sure, but even if it were cheap - I really don't see any reason why their HV probe would be any better than a good home-made HV probe.

They also mention about checking the multiplier's resistance with a megger. I couldn't find any more info about that, but you can't beat an actual voltage test.

http://emanuals.nordson.com/finishing/files/powder/107004e02.pdf