High voltage probe and high voltage resistors

[xzswq21]

Hello, I want to make a high voltage probe (20kV), there are several different types of high voltage resistors with different Materials: Metal film, ceramic and etc
The working frequency is 30KHz to 100KHz. Which material should I choose?
Thanks

[bob91343]

Primarily, the material choice is concerned with temperature coefficient of resistance.

Having said that, the majority of resistors are helically built and thus have parasitic inductance.  There are many techniques for constructing resistors to minimize parasitics.  Your best bet is to contact one or two manufacturers and ask for recommendations depending on your requirements.

[David Hess]

At that frequency anything can be used except wirewound, but pay particular attention to the maximum voltage rating which is what distinguishes high voltage resistors from common resistors.

[2N3055]

Define high voltage: 5kV, 20kV 50kV?

[xzswq21]

Define high voltage: 5kV, 20kV 50kV?

20kV to 30kV

[tggzzz]

http://hvstuff.com/high-voltage-resistors

[TimFox]

And what resistance do you need?
As stated above, almost all high-voltage resistors will be helices, and hence inductive.  The non-inductive construction techniques will reduce the voltage capability by reducing spacing between nodes at different voltages.
However, the resistance is usually much higher than the inductive reactance (and hard to measure).
For high resistance values, the best temperature co-efficients are probably metal-oxide (cermet).  Metal film can have better tempco, but the resistivity is lower and the resulting resistor will be physically larger.
A good manufacturer of precision high-voltage resistors (in the 20 to 30 kV range) is Caddock  http://www.caddock.com/Online_catalog/high_voltage/high_voltage.html  and they are available through distribution.  They have a large range, some of them are good for high frequencies, as well. 
If you really need high voltage at high frequency, you may be better off with a capacitor divider, so long as you don't need to measure DC.  Choose the resistive load on the capacitor tap to obtain the required low-frequency cutoff.

[xzswq21]

If you really need high voltage at high frequency, you may be better off with a capacitor divider, so long as you don't need to measure DC.  Choose the resistive load on the capacitor tap to obtain the required low-frequency cutoff.

Do you have any information about Thick Film resistors? Ohmite manufacturer produces high voltage thick film resistors such as maxi-mox series:
https://www.mouser.com/datasheet/2/303/res_maximox-253273.pdf
I have chosen two 500Mega Ohm MOX-3 resistors but the freight charge is 150$ !

One of my problem is the capacitor type, as you know high voltage capacitors are Y5P and Y6P, can I use these or not?

[TimFox]

Thick-film is another term for metal oxide or cermet.  A thick film is applied like through a silk screen, being an ink comprising resistive stuff and insulating glass.  When fired, it binds to the ceramic substrate and the the binder in the ink burns off.  The antonym is "thin film", which means a metal alloy deposited from vacuum.
I would never use X, Y (such as Y5P), or Z dielectric ceramic capacitors in a measurement system, since they lose capacitance drastically with applied voltage.  For ceramic, you need NP0 or C0G (basically synonymous).  However, there are special capacitors for high voltage, such as reconstituted mica or high-voltage polypropylene film.
I have used the Ohmite maxi-mox resistors, which have reasonable performance, but I prefer the Caddocks.  I have used them up to 40 kV, with a more complicated structure (with corona guards) up to 100 kV.
Note that all of these non-metallic resistors have a voltage co-efficient of resistance that is substantial at high voltage.  I was able to get matched co-efficients from Caddock to make a voltage divider, and they sell pre-assembled voltage dividers that are specified.  (I keep reminding people that Ohm's Law is an approximation that is often useful, but not guaranteed for a given resistor-shaped object.)
Your Ohmite data sheet does not specify the voltage co-efficient, some Caddock parts are so specified.
Ohmite makes a "Supermox" series with typical specifications on VCR.  https://www.ohmite.com/assets/docs/res_supermox.pdf
A 30 kV rated Supermox can have a "typical" VCR of 0.1 to 0.15 ppm/V which sounds good, but when you multiply by 30 kV that is a shift of 0.3 to 0.45% in the resistance.
You didn't give a country flag:  to where would the freight charge get that high from Texas?

[David Hess]

Define high voltage: 5kV, 20kV 50kV?

1.6 to 10kV for metal film resistors and higher for thick film resistors.

As stated above, almost all high-voltage resistors will be helices, and hence inductive.

The high voltage thick film resistors I have seen all use non-inductive serpentine construction.

[TimFox]

I believe that some Caddock resistors are serpentine.

[NiHaoMike]

Given that a high voltage probe is going to use resistors well into the megaohms, inductance is unlikely to cause any problems. Capacitance is going to be significant and that will have to be compensated for.

Here's a video on DIY high voltage probes:

[xzswq21]

I would never use X, Y (such as Y5P), or Z dielectric ceramic capacitors in a measurement system, since they lose capacitance drastically with applied voltage.  For ceramic, you need NP0 or C0G (basically synonymous).  However, there are special capacitors for high voltage, such as reconstituted mica or high-voltage polypropylene film.

You didn't give a country flag:  to where would the freight charge get that high from Texas?

Before I come here, I had googled about high voltage capacitors, for low voltage filters I always use C0G capacitors but for high voltage capacitor I couldn't find anything in Digikey or Mouser except Y5P, N4700, Z5T, . I found several large size and unusual and very expensive capacitors. Can I use 1kV C0G capacitors in series? In theory there is a problem, for example, the voltage drops by 1kV per capacitor but there is 20kV and 19kV on the first capacitor!

20kV -||- 19kV  -||- 18kV -||- 17kV  -||- 16kV  -||- ....

I think I can't use 1kV capacitors

I wanted to buy from Digikey.com, the shipping cost for other components is 30$ but for the Ohmite resistors is 150$. It's better for me to find the the components from other distributors.

[PartialDischarge]

The inductance in hv helix resistors is not a problem due to the extremely low currents handled. The real reason for serpentine construction is to avoid magnetic field coupling from external sources into the resistor, this magnetic field is not filtered by most shield cans, specially at low freqs.

Capacitors used for hv are tricky, as their rating should be set according to partial discharges, not Ac or dc rating.
Capacitors in series do not distribute the voltage linearly unless compensated by resistors. This is a well studied problem in hv ceramic/glass insulators for distribution power lines

[Gyro]

Here's a 10kV peak R-C attenuator I built for my Gay MTA Memory (peak capture) Voltmeter. I included as much detail as possible in the post, including how I chose parts and calculated the capacitances for the input impedance of the Voltmeter and connecting cable strays...

https://www.eevblog.com/forum/testgear/high-voltage-x10-attenuator-for-my-gay-mta-memory-voltmeter/

An important note that this is not intended or suitable for high energy circuits, or to handle 10KV DC continuously (resistor wattage ratings). But it may help you in the construction of a higher voltage probe (more stages).

Due to the nature of the termination and the input characteristics of the meter, I wasn't able to directly look at the square wave response on a scope - I was more interested in the overall frequency response of the probe and meter combination, so I had to check the roll-off with a signal generator. It made it to 5MHz though.

It might be of some help anyway.

[xzswq21]

Here's a 10kV peak R-C attenuator I built for my Gay MTA Memory (peak capture) Voltmeter.

https://www.eevblog.com/forum/testgear/high-voltage-x10-attenuator-for-my-gay-mta-memory-voltmeter/
I was more interested in the overall frequency response of the probe and meter combination, so I had to check the roll-off with a signal generator. It made it to 5MHz though.

You have used SL temperature coefficient capacitors, Do you recommend me SL capacitors? Even 1MHz passband flatness is good

[Gyro]

You have used SL temperature coefficient capacitors, Do you recommend me SL capacitors? Even 1MHz passband flatness is good

It's not so much a recommendation, as what is available. Ideally you would use C0G or NP0, but these are not commonly available, that I could find, in 6kV rating (possibly overkill but I wanted a good safety margin with 3 capacitors in series on 10kV - the MTA original probe stated 15kV pk).

Murata information indicates that SL is a class 1 dielectric and 5% tolerance, even though probably not as good as [Edit: higher Dissipation Factor than],  NP0 and C0G (also class 1). So seemed good enough for the purpose - certainly much better than the Yxx and Zxx dielectrics found in other high voltage ceramics.

Here's a link to the Murata High voltage capacitor and dielectric data... https://www.murata.com/-/media/webrenewal/support/library/catalog/products/capacitor/ceramiccapacitor/c85e.ashx?la=en-us&cvid=20210203042629000000

EDIT: and detailed spec sheet...  https://search.murata.co.jp/Ceramy/image/img/A01X/G101/ENG/DEC_E.pdf

[Gyro]

The TE Connectivity HB series thick film resistors are also a good choice. They're low inductance and have good voltage coefficient, I don't know what overall resistance you are aiming for in your divider (and split across how many resistors) but they achieve around 1ppm/V up to 10Meg, and 2ppm/V up to 100Meg. Not cheap, but not ridiculously expensive either.

https://uk.farnell.com/w/search/prl/results?st=hb%20resistor

[joeqsmith]

I designed and constructed a few different high voltage probes.   Normally I use wire or copper foil to form the compensation network.    This video goes over the design of my 40kV 20MHz wideband probe.  I broke the ceramic insulator a few years back and ended up changing the design a bit to improve the performance.   

[TimFox]

I would never use X, Y (such as Y5P), or Z dielectric ceramic capacitors in a measurement system, since they lose capacitance drastically with applied voltage.  For ceramic, you need NP0 or C0G (basically synonymous).  However, there are special capacitors for high voltage, such as reconstituted mica or high-voltage polypropylene film.

You didn't give a country flag:  to where would the freight charge get that high from Texas?

Before I come here, I had googled about high voltage capacitors, for low voltage filters I always use C0G capacitors but for high voltage capacitor I couldn't find anything in Digikey or Mouser except Y5P, N4700, Z5T, . I found several large size and unusual and very expensive capacitors. Can I use 1kV C0G capacitors in series? In theory there is a problem, for example, the voltage drops by 1kV per capacitor but there is 20kV and 19kV on the first capacitor!

20kV -||- 19kV  -||- 18kV -||- 17kV  -||- 16kV  -||- ....

I think I can't use 1kV capacitors

I wanted to buy from Digikey.com, the shipping cost for other components is 30$ but for the Ohmite resistors is 150$. It's better for me to find the the components from other distributors.

Of the three dielectrics you mentioned, "N4700" has a high temperature co-efficient (-4700 ppm/K), but its voltage dependence should be reasonable.  See others' comments about DC distribution across series capacitors:  you must use resistors to equalize the DC voltages.

[xzswq21]

Of the three dielectrics you mentioned, "N4700" has a high temperature co-efficient (-4700 ppm/K), but its voltage dependence should be reasonable.  See others' comments about DC distribution across series capacitors:  you must use resistors to equalize the DC voltages.

Yes , I want to use two 500MegOhms or four 250MegOhms resistors.

[Gyro]

It sounds as if you should build it out of a series - parallel combination of 500Meg resistors in that case. You don't have your country flag set, but Farnell have a 25% price break at 10pcs. I would think that other distributors would have something similar.


Edit: Sorry, I missed the "or".

[TimFox]

The simplest approach is to make a series string of several resistors (more is better) and put equal appropriate capacitors across each one.  Again, resistors have a noticeable voltage co-efficient at high voltage, but ceramic capacitors can have huge voltage dependence—read the data sheet for individual types.  When last I worked with 100 kV, good high-voltage capacitors were easier to find (used in CRT supplies, etc.).

[xzswq21]

The simplest approach is to make a series string of several resistors (more is better) and put equal appropriate capacitors across each one.  Again, resistors have a noticeable voltage co-efficient at high voltage, but ceramic capacitors can have huge voltage dependence—read the data sheet for individual types.  When last I worked with 100 kV, good high-voltage capacitors were easier to find (used in CRT supplies, etc.).

I found several resistors as following:

MOX94021007FTE: (MOX940)



MOX96021007FTE: (MOX960)







SM108035006FE:

[TimFox]

Choosing resistors with a specification (even “typ”) on VCR is a good idea.  Note that the spec is “ppm/V”, so the ppm (10^-6) must be multiplied by the voltage across the resistor to get the fractional deviation (this can be derived from a quadratic function of current vs. voltage, rather than linear).