Page 2 of 2

Re: HV Divider

Posted: Sun Feb 03, 2019 6:43 pm
by Rich Feldman
Cristiano, please excuse me for using this thread to carry on about HV dividers.
I got a surprise last night, about behavior of plain resistors at high voltage.

This was while load-testing my detector tube bias supply, using an assortment of resistors from the junk box.
The parts for charge amplifier has arrived earlier than expected, so it was time to stop playing with the HV bias supply design. Get off the pot, as they say. HV supply will be properly reported in its own thread, after I find the camera, or finally get me a smart phone.

DC voltage and current were measured directly with the home-made instruments shown in NST Power thread.
But in round 1, only the voltages were measured. Then the currents were computed using the I=V/R formula (I=U/R in Europe, which nicely avoids using a variable letter that's the same as its unit of measurement symbol. Who else was ever taught I=E/R? )

Here's a chart that summarizes the results:
kolo6.JPG
.
The red dot measurements are repeatable, and they are based on R values measured with a 6-digit benchtop multimeter.
The load resistors responsible for the out-of-line points are NOS 2-watt carbon-composition resistors in single digit megohm values.
I bet we're actually seeing a large voltage-dependent resistance change, at the levels applied for a few seconds in my experiment.

p.s. The supply is based on small variac, tiny NST, 30-kV rectifier diode, and a 0.01-uF output capacitor. It was "softened" by putting a 1.5K power resistor in series with NST primary. The capacitor is to reduce HV ripple when load resistance is many tens of megohms.

Re: HV Divider

Posted: Sun Feb 03, 2019 7:30 pm
by Cristiano_Machado
Hello Mr. Feldman,

Thank you for the information, and fell free to keep posting contribution to this thread. It helps to better understand this issue.

I studied a bit and decided to use two Omite 200 MOhms 1% thick film resistors, rated to 20kV, but the voltage will be limited to 17kV maximum in each resistor.

Regards,

Cristiano