How to monitor high voltage capacitor bank?

Question

I am working with a large ring launcher and a capacitor bank that has the potential to charge up to 3000 volts in a bank of capacitors totaling 12,000 μF. I need to provide a sort of tap to measure the current capacitor bank voltage outside the circuit with a meter. First thought was a 10:1 voltage divider but I'm worried that might leak current and drain the caps too fast. A secondary concern, I'm not sure I could limit the current enough to bring that down to a safer level as well.

Is there a more efficient way to monitor the voltage with minimal leakage or would a voltage divider be the more correct solution? Is there way to give a safe voltage/current on a tap in either regard?

Answer 1

Sounds like you need a 1000:1 potential divider with a very high impedance.

A simple resistive divider chain using resistors of a very large value plus an opamp with a high impedance input will work. e.g. Use 1G ohm resistor made of 5 x 200 Meg 1206 resistors, and a single 1 Meg 1206 resistor at the bottom of the chain.

Feed the voltage across the 1Meg into decent JFET opamp, configured as a buffer.

As described above, for 3kV you will need to use a resistor chain to prevent breakdown of the resistors and conformally coat the result to prevent air and surface leakage.

This design would create a 3uA discharge of the cap. If you want lower discharge go for larger resistors and spend more time thinking about other forms of leakage. You should be able to get to about 10GOhm before running into issues with other losses.

Answer 2

A 1000:1 voltage divider night be better. Make sure your series resistor has an adequate voltage specification. If you don't want to buy high voltage resistors, then often what folks do is put several ordinary ones in series. You are unlikely to run your cap bank down too quickly with (say) 10 off 10Meg resistors in series.

With any cap bank, it's good practice to have a resistor permanently attached to bleed the charge off when unused.

Answer 3

In addition to Neil_UK and Jason Mason:

Have a look to differential probe projects on EEvblog and others, with the exception than you may not need a differential input signal (actually you need just a half of the differential probe) and probably you also don't need a compensation RC circuitry, also you don't need high bandwidth.

An high input inpedance and low bias current opamp is needed, for example ADA4817-1, ADA4530-1. Then you buffer this signal with cable driver opamp, for example THS3202.

This is a Micsig differential probe - new approach and low cost. See resistor string with parallel capacitor string for compensation. The input amplifier is ADA4817-1 Source EEVBLOG . Older variants of Tektronix, Lecroy,.. probes used a matched differential pair JFET transitor as input pre-amp, but this is very expensive part, newest opamp seems to do the job quite well.