DC High Voltage Measurement Accuracy

Hello Shyam,

Pockels & Kerr cells are electro-optical devices used in measurements of EHV. Both operate by rotating the polarization of light through an angle which is a function of the applied electric field. Typically a beam of light (usu. from a laser) is directed through a polarizer, the active medium, and then through another polarizer (in this setup called an analyzer) polarized at right angles to the first.

In the absence of an electric field the polarization of the beam remains unchanged and very little or no light makes it past the analyzer. Glan-Thompson prisms are sometimes used for both polarizer and analyzer, but more often than not a high-quality polarizing film is used.

The output of the analyzer is sensed by a PIN photodiode and its associated electronics.

Kerr cells use liquid nitrobenzene as the active element whilst Pockels cells are based on variations on crystalline KDP (potassium di-hydrogen phosphate) and similar materials. Kerr cells are capable of sensing field variations as short as 10 ns or so whilst Pockels cells are usually a bit slower, around 40 ns or so. Pockels cells also operate at lower voltages.

Field mills are another time-tested technique. New designs have reduced error margins significantly. The downside of this approach is that field mills cannot respond to transients anywhere near as fast as you specified.

Precision resistive dividers can be used for your lower voltages, of course.

Do not be surprised that EHV measurement error can potentially increase drastically with increases in voltage. At the higher voltages you will have many problems to solve, such as measurement errors due to corona discharge, stray electric fields and stray capacitance, to name a few.

Good luck on your project!

SR

Be careful if you are designing a voltage divider. Voltage coefficient is a measure of resistance change as function of voltage drop across a resistor. I once had a special wire wound resistor designed to minimize that effect.

--- CHAS

Correct. This non-linearity must be taken into account along with a whole host of other sources of error. Humidity, for example, and temperature. Dielectrics must be kept extremely clean and hydrophobic to minimize the effects of condensation and adsorption. Mr. Shyam's work is certainly cut out for him.

Yes I know all that but what is to be done is important here. Which wire wound resistor has minimum voltage sensitivity? Any specific resistance type or make?

What should be the field gradient it air or I need to avoid using air gap all together? Right now 0.1" per kV air gap is used in all connectors. I think this is less and may opt for 0.2" per kV or even 0.5" per kV in open air.

I can keep the moisture out and can use dry atmosphere to some level.

I am aware of cosmic rays causing GM effect and short pulses in the HV Supply of low current less than 1uA so not much effect on high current version expected. Vacuum can help by reducing the space charge generated.

What else?

Dear SR

"Precision resistive dividers can be used for your lower voltages, of course.

Do not be surprised that EHV measurement error can potentially increase drastically with increases in voltage. At the higher voltages you will have many problems to solve, such as measurement errors due to corona discharge, stray electric fields and stray capacitance, to name a few."

For up to 30kV I will like to use resistive dividers. Resistive dividers in air sure will have corona current problem so placing these resistors in oil or some insulated filler medium looks more reasonable. I think Oil may degrade in time and may need replacement. Solid insulator may collect charge and it may be essential desensitized it after using it.

What is the accuracy of the polarizer to measure the electric field from 1kV and up to 30kV? Any standard high quality equipment in this direction which can measure 1ppm precision / resolution and about 0.1% absolute accuracy?

Ross Engineering manufactures precision dividers up to 1000kV.

As a matter of convenience, you may also wish to consult with resources a little closer to home. This week, the M. P. Varshney High Voltage Laboratory was dedicated at the Zakir Husain College of Engineering and Technology, Aligarh Muslim University. M. P. Varshney is an alumnus of AMU. See also the Indian Institute of Science.

Hello Sonos Research,

ROSS engineering HV probes up to 10MHz are interesting one ans also high accuracy voltage dividers for HV DC measurement.

1pF to 2pF capacitive load probes will be ideal for some of my HV pulse application for 20V/ns rate. For DC measurement, and peak to peak ripple measurement I perhaps need resistive and capacitive networks.

For Dc measurement 1G ohms 10000:1 with 1" open air isolation per 2kV step of voltage looks reasonable so resistive probe of nearly 15" for 30kV is the longest I will try and perhaps 12" the smallest to be tried. 1G Ohms will leak about 30uA from 30kV 10uA from 10kV. Assuming that we use 100pA bias current JFET amplifier across voltage divider sense resistor of 100K of 10000:1 giving 3V across 100K for 30kV, then I are still have more than 6 decades away from the maximum current level of 30uA. 1V will inject 1nA in 1G Ohms which is fairely good current for measurement, however voltage drop across 100K will be about 10uV and requires some good amplifier design consideration or change of probe to more sensitive one designed for lower voltages and the one injects greater current.

For automatic high dynamic range of DC high voltage measurement single probe looks not a very good idea. More than one probe will require probe switching circuit suitable for such high voltages. Vacuum switches are very expensive item, however may become unavoidable requirement for this setup to be more effective.

I have seen somewhere charge sampling type of high voltage measurement circuit. This circuit charges small capacitor and then discharges it into charge integrating amplifier capacitor for measurement. V=Q/C method used. Has any one tried this for 20kV and above measurement?

High voltage switch which is corona free, is something we can also add to this discussion. With movement of switching electrodes closing the switching gap, corona current bound to come into picture so what is this corona free switching any way?

For the DC measurement up to 100 kV (that was my horizon so far) the most accurate commercial probe I have come across is from CPS:

http://www.cpshv.com/products/hvp500data.htm

I have the 50 kV version of this probe but do not have it calibrated to its spec. But I have a sample of the calibration certificate that I can send to you if it is of interest. Up to 20 kV the deviation from the NIST reference probe is well below 0,01% allthough the spec allows for 0,025%, approximately doubling this value for 50 kV.

For higher voltages you might also want to check out this german high voltage calibration lab website for the employed techniques:

http://www.highvolt.de/

Good luck.

Hello Urbie

Yes, I am looking for similar product. I will consider Tektronix probe which has good response to fast pulsed high voltages also. These are in the range of US$2000 each. DC voltage divider is easier to make.

What DC accuracy do you actually require?

For instance the posting says "100V to 20kV need to be within 1 ppm precision"
1ppm=1/1,000,000 or 0.0001%.

You will probably have problems finding resistors to this accuracy.

Instead of using a voltage divider, it is probably easier to just measure the current.

Dear Carl Pugh

Normally high voltage divider resistors are manufactured with 1% to 2% ratio tracking absolute accuracy. It may be possible to manufacture at best 0.1% accurate ratio. Temperature coefficient of 10ppm/C and ageing of 10ppm per year may also be possible.

10ppm stability HV supplies are claimed by many manufacturers and some even claim to have 1ppm stability. It requires to measure these parameters close to what they are being claimed.