High voltage sensing without any current drain

[DanioIO]

Hi,
How can i sense a 1000V without using a resistor voltage divider? I cannot drain any current from the input - I need to get bellow 100nA.

[Psi]

ac or dc
If ac, what frequency?

A voltage divider with a resistance of 10meg is doable. you just need a opamp to buffer the output before you can use it.

EDIT: sorry, misread, i see you need less than 100 nano amps. That is getting a bit tricky.

[Moseven]

They do exist, but I don't know what they are called.
Perhaps Electrometer : http://en.wikipedia.org/wiki/Electrometer
There is a method, using a capacitance : they are called non-contact surface voltage meter, or electrostatic fieldmeter http://en.wikipedia.org/wiki/Electrostatic_fieldmeter.
The most common name is Electrostatic voltmeter : http://en.wikipedia.org/wiki/Electrostatic_voltmeter

Dave was using a detection pen a few days ago in a video. But I read the using capacitance in a smart way it is possible to actually measure the voltage.

[HighVoltage]

Use a capacitive divider instead of a resistive divider.

Depending on what you have to measure, you probably can use an existing conductor of your DUT as one of the electrodes.
1 pF capacitance is more than enough
And this works well for AC and DC.
Usually I am building my own capacitive dividers for this kind of measurements

 

[dannyf]

I cannot drain any current from the input - I need to get bellow 100nA.

Isn't that self-contradictory? Either you allow current drain (less than 100na?) or you don't allow current drain. You cannot have both.

[suicidaleggroll]

Is that 100nA limit instantaneous or time-averaged?  How often do you want to sample?

[Richard Crowley]

There very well may be several practical solutions if we understood the big picture here.
Asking this as a theoretical, abstract question severely limits the kinds of responses you will get.
For example, there could be non-contact, electrostatic possibilities if the conditions are right, etc. etc.

And what does "sense" mean? Does it mean actually MEASURE?  To what accuracy?
Does it mean DETECT presence/absence?  To what tolerance?
And, as others have asked, at what "sample rate"? 
What circuit constraints?  Power available?  Signal output?
Your question only raises more questions.

[Ian.M]

Short of using a thermionic valve with an isolated heater supply and an even higher voltage for the anode supply as a cathode follower, I don't think there's any way of getting zero current except by fairly complex electrostatic measurement techniques.

[Marco]

And this works well for AC and DC.

How do you balance the leakage currents for DC?

[Marco]

Short of using a thermionic valve with an isolated heater supply and an even higher voltage for the anode supply as a cathode follower, I don't think there's any way of getting zero current except by fairly complex electrostatic measurement techniques.

Why go for Valves at 1000V? Why not a high voltage MOSFET source follower? (Followed by a normal divider.)

[T3sl4co1l]

Presumably, a cold tube would leak less than room-temperature (let alone elevated-temperature) silicon?

You can do a sampling ADC sort of thing where the capacitor divider is grounded, zeroed, then switched to the voltage node and sampled.  When the sampling is quick, leakage current of the sampling / measuring circuit has little effect.  Note that, since this is a charge pump, it still draws current proportional to sample rate and capacitance, so you can't sample very quickly.

1kV and 1pF draws 100nA at a sample rate of 100Hz.  Assuming the switching devices have much less than 1pF of capacitance themselves, which is a very tall order (1mA 1500V MOSFETs *could* exist, but I doubt you'll happen to find one available anywhere).

This is why esoteric switching methods seem to be suggested: it's slow enough to use high voltage reed relays; high voltage photodiodes (or phototubes); thermionic tubes (specifically, medium to high voltage, filament cathode types where the thermal time constant is short, the capacitance is small, and the voltage standoff is high); even CdS (or other) photocells, given the appropriate dimensions (it would have to be a very large "bright" resistance indeed, however).

A 1ms time constant (or less) would be appropriate, so the switch can have an on-resistance of under 1Gohm.  It should have an off-resistance much greater than this, the difference giving the gain error.  I.e., a 1G on, 1T off gives a 1000:1 on-off ratio, which will lose 1/1000th of the signal, or 99.9% gain -- not too bad, so mediocre parts like CdS are suitable here, as long as they are appropriately chosen.

Tim

[Marco]

Presumably, a cold tube would leak less than room-temperature (let alone elevated-temperature) silicon?

AFAICS a source follower with a cascode-bootstrapped drain is going to leak about as much as a small signal MOSFET, which is to say fAs in normal operation (ie. as long as it can follow fast enough). Almost all of it through the package rather than the silicon oxide at that.

How about two of these? One for a cascode current source, the other for input and cascode. Obsolete parts, but that's why they're cheap. Don't have to worry about cosmic rays either.

[T3sl4co1l]

I don't see how a power device could ever possibly have lower leakage than a truly small geometry device (of any construction -- namely, from my list above).  There's simply more cross sectional area to leak through.

I don't get what you mean about SiO2, because current's not flowing (or not not flowing..) through the gate, it's the channel (solid Si) that can't turn off all the way.  Topology can't change that fact; leakage just flows right on through the cascode.

Do you mean to use a small device at the bottom, and cascode it for voltage?  Ohh... that might work.  But you still have to mind geometry, because an 8A device will have... I don't know, ballpark 10^5 more gate oxide area than a small RF transistor, and that *will* end up leaking more!

You also completely blow the capacitance requirement by using a power device, because Cdg will be in the 10s to 100s of pF (and much, much more in saturation).   At least, if the sample rate should be more than fractional Hz.

Tim

[Marco]

The MOSFET in the device I linked is a low voltage part, but you could use a 2N7002 or JFET instead (with the JFET you won't really need the gate protection diodes either). The bootstrapped source follower also minimizes capacitance (source and drain voltage follow input voltage) but high frequency none of this is ever going to be. I mean something like this :

[T3sl4co1l]

Oh right!  For the sensor... my mind was still on switching 

[dannyf]

Assuming that you do allow some current drain, some cmos opamps have bias current on the fa range.

To further reduce it, you could use a follower + a resistor network to easily boost the input current over 10Gohm territory.

[macboy]

You could use a null technique. You need an adjustable DC source as high as your source being measured. Adjust it to approximately the value of the unknown voltage and measure the difference between them. Adjust the adjustable source up or down to achieve a null. Now no current flows through the meter that is detecting the null, and you can measure the voltage of the adjustable source using another suitable meter.

Assuming that you do allow some current drain, some cmos opamps have bias current on the fa range.

To further reduce it, you could use a follower + a resistor network to easily boost the input current over 10Gohm territory.

You must have missed the 1000 V part.

[Marco]

With a TOhm of input resistance on an opamp you can make quite a high impedance divider, of course you're going to have some problem with RC times ... the main problem is that OP didn't really say what kind of AC components he has and what kind of AC currents are acceptable.

You can also boot strap the power connections for an opamp in much the same way as the source follower ... still need the 9V battery though (because of the lack of PNP/P-MOSFETs at these voltages, there's other ways to solve it but a battery is easiest).

[Dr. Frank]

Macboy got the correct answer..

FLUKE manufactured several Null Differential Voltmeter..with internal 1000V generation.

I do not remember, which one it was.. 882, 883, 884, 887, 889...

They had Kelvin Varley divider, mikro Volt differential null chopper amplifier which is able to withstand 1000V overload, and precision reference , in this case up to 1000V..

In case of Null, the effective resistance is indefinitely high, only a few pA would flow.
 These are still available on the bay.. But you have to chose the correct model..

Frank

[Marco]

Dunno about the Fluke ones, but Keithly 6517A/B's are a bit dear.