High Voltage Current Sensing

[DrMag]

I'm working on an application where I need to measure individual currents through a set of detectors in a vacuum chamber with limited feedthroughs. Ideally it would make sense to use a low-side current sense, but unfortunately the way the detectors are mounted links them all at the ground side so there is no way to insert a current sense between the end of the detector and ground. The high-side is kept at around 10kV, and I'm looking for a way to measure the current through the system short of an electrometer on each, requiring more feedthroughs than I have available.

My mind came up with this idea, which I'm certain is bad, but not exactly certain why. Assuming everything were well isolated from earth ground, why is this a bad idea? Does anyone with far more experience than I know a better way to do this? The currents are typically 100-200 uA.

[TimFox]

Assuming that your amplifier is a differential instrumentation amplifier (not an op amp), that would sort of work, except that the output voltage would be referenced to a useful point (not shown in your drawing) that is at high voltage.  A typical instrumentation amplifier has an output common connection, which can be tied to the common point (the junction between batteries in your drawing) for the amplifier power supply.
You then need to put a meter inside the high-voltage terminal for each detector feed.
In my grad-school days, feeding 100 kV to a cathode, we did exactly that with a large d'Arsonval meter that we read from a safe distance visually.  However, if your detectors are drawing small currents (like microamps or less), you would need a transimpedance amplifier (op amp with feedback resistor) to measure the current, with appropriate protection at the input.  (We had very small current, and used a combination of neon bulbs and semiconductor diodes before the op amp summing point.)

[T3sl4co1l]

How is it kept at 10kV?  Your answer will be there.

[DrMag]

The external power supply keeps it at 10 kV, but there's only one supply feeding all the detectors.

TimFox, thanks--that actually explains why I felt this method was wrong. Perhaps everything would work at its own reference ground, but sending the data (read through ADC, sent serially) would then be done at high voltage--not going to end well.

Does anyone know of other techniques for measuring high voltages?

[T3sl4co1l]

Measuring high voltages, just use a bigass divider, no worries.  But measuring small voltages (or currents) at high isolation?  Digital conversion followed by fiber-optic is about all you can do.

Currents can be measured in isolation, using enough turns of wire around a Hall effect sensor.  These are usually direct reading (no core, ~single turn) for 10-100A+ sensors, and use many turns (and a core) down into the single amp range.  They aren't usually made for, like, single mA or less, but the principle might be applied.

If you need to sense the current in a bunch of common anodes, consider flipping the sign.  Use negative ground.  For example, CRTs were often driven either way: first, oscilloscopes needed the deflection plates (and face) at ground, so the electron gun cathode was operated at -2kV or so.  Any control voltages (referenced to the cathode) were either generated in place (e.g., high voltage pots for intensity and focus), or coupled using capacitors (AC coupling), or more complicated methods.  (The typical Tektronix "Z axis" coupling used an entire additional HV supply, purely as a series battery to the CRT grid, so intensity could be controlled with DC rather than AC coupling.)  Magnetic deflection CRTs (e.g., almost all TVs) used significantly higher voltages to achieve brighter pictures, and had a much greater need for the cathode end to be near ground (applying high bandwidth video to the cathode or grid; the last generation of high resolution Trinitron displays delivered over 100MHz analog bandwidth!).

So, it sounds like you may have a need for the same consideration: if all your important signals are referenced to the anode, why not make the anode ground?

Of course you haven't mentioned what the cathode end is, if anything, and whether there's anything important going on down there... so you may just be screwed instead.

Tim

[wbeaty]

To get your op-amp outputs (or any analog signals) across many kilovolts, make your own high-kilovols optocouplers from LEDs and photodiodes placed at opposite ends of 4cm plastic pipes.   (If you don't have plexi rods, you'll find that white heatshrink pipes works great, especially if covered up with slightly-oversized black heatshrink, to prevent crosstalk with nearby pipes.

On these circuits you can 'linearize' the LED's optical output by using another op-amp plus two identical LEDs in series, plus a local photodiode facing one LED as feedback to control the LED.  That way both photodiodes put out a linear signal, even when separated by an inch of space and tens of kilovolts.    (Heh, even use an LED as your photodiode, just two LEDs on either end of the light-pipe.)

Have you discovered USB chamber feedthroughs?      736-SWH1S304CLLPV is Mouser pn for a long 4-pin Lemo coupler, fits in threaded hole with gasket.    Then just chop a USB cable and add appropriate mating Lemo connectors.

Is this high vacuum?  Can you find any well-sealed batteries which won't outgas all over the place?    Many batteries don't play nice with micro-Torr environments.

MIT's doctor VandeGraaff solved his floating power supply problem by connecting two DC motors together with a long insulating rod.    Power one motor, the long rod spins, and the other motor becomes a DC generator putting out roughly the same volts as the driven motor.   Of course he was trying to send kilowatts across seven megavolts, to light his desk lamps high inside the man-carrying upper spheres of his "atom-smasher," but the mini version works fine, if you can keep the motor from overheating in vacuum.

I've been to the top of this one below.  The rope ladder inside, it has missing/broken steps!   (No, that's not me looking out of the access hole.)