Capacitive Voltage divider, for O-scopes (10,000:1) / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Capacitive Voltage divider, for O-scopes (10,000:1)

first previous 5 6 7 8 next last

Move Thread

LAN_403

Patrick

Fri Jul 02 2010, 07:03PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

thats my point i need 5 sig figs but i only get 2, also if you show me an instrument that includes 1.0000 pF with that many zeros and accurate, ill show you a leprechaun and a magic pixie fairy.

ill give you an example:

10.000 nF / 1.0000 pF = 10,000:1
10.000 nF / 1.0001 pF = 10,001:1 <- see!
10.022 nF / 1.0000 pF = 10,022:1 <- here too!

also,
10.000 nF / 0.7000 pF = 14,300:1
10,000 nF / 1.2000 pF = 8,300:1

0.7 to 1.2 pF was the measured amount that was accurate and repeatable on many different conditions and days, with the distance varying from 0.75 inch @ 1.2 pF through 1.25 inches @ 0.7 pF, through olive oil. digit rollovers seemed credible.

as you can see no known instrument can measure the divivder accurately thats why i want to calibrate it with a known signal, modifying the gain of the op-amp. to get very close to the ideal 10,000:1 ratio.

Mattski

Fri Jul 02 2010, 07:25PM

Registered Member #1792 Joined: Fri Oct 31 2008, 08:12PM
Location: University of California
Posts: 527

The shielding concept is good because it means that outside objects will not affect the capacitance. It also adds a constant parasitic capacitance (the two red caps in series) which is in parallel with the original capacitance (green cap). (Thanks to Fabio for the original diagram).

It's not a bad thing, just something you must keep in mind as it will increase the total HV capacitance.

A network analyzer should also be able to measure a capacitance that small, as long as you can measure below the self-resonant frequency of the capacitor where the parastic inductive impedance cancels the capacitive impedance. You could also put several of your capacitors in parallel, that would make it easier to measure with a typical capacitance meter.

wrote ...
i included R4 50 ohms, as the Sam Goldwasser article shows, is this to simulate the impedence of the BNC RG6 line? or to do impedence mathcing to the scope?

A typical 100kHz scope will have a 1Mohm input impedance, which you can increase with a 10X probe. Usually only the high speed scopes have 50ohm inputs (that I have seen). You don't need to take the impedance of the BNC cable into account because at 100kHz a wavelength is on the order of 1km in the cable, so it will act like an ideal wire easily up to 10meter length, even 100m it will be close to ideal.

klugesmith

Fri Jul 02 2010, 07:42PM

Registered Member #2099 Joined: Wed Apr 29 2009, 12:22AM
Location: Los Altos, California
Posts: 1716

Regarding the question about how to calibrate the gain of Patrick's HV probe, which has no response at DC.
The same question applies to fancy probes such as Dugg pointed out -- how does the designer or user adjust and verify that the HF attenuation is the same as the DC attenuation?

We can independently measure high voltage with an electrostatic voltmeter like:

(which looks big enough to handle many tens of kV if filled with oil).

The force on capacitor plates is proportional to square of electric field strength.
So the instrument responds equally to DC of either polarity, or to the true RMS value of an AC waveform.
A homebrew instrument could be calibrated by measuring the change of capacitance as the plates move.
Or by applying high voltage DC, independently measured with a trusted HV DC probe.

Patrick

Fri Jul 02 2010, 07:50PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

Mattski wrote ...

wrote ...
i included R4 50 ohms, as the Sam Goldwasser article shows, is this to simulate the impedence of the BNC RG6 line? or to do impedence mathcing to the scope?

thats what i thought, my o-scope is the standard 1Mohm input. so when i use this device i will set the scope for 1X @ 1Mohm input.

IntraWinding

Fri Jul 02 2010, 07:55PM

Registered Member #2261 Joined: Mon Aug 03 2009, 01:19AM
Location: London, UK
Posts: 581

Thanks to Fabio for the clear explanation of shielding. Now I understand

Do you think in practice the shielding should extend a little further upwards to more thoroughly shield the capacitor?

I think the 0.7pF input capacitor might be ok with this shielded set up. One of the problems with oscilloscope probes that I don't feel happy about is that an input capacitance of even just a few picofarads results in the probe input impedance falling to very low levels at higher frequencies. Consequently, the smaller the input capacitance you use, the better, but bear in mind that the probe input capacitance is (approximately) the sum of the 0.7pF and the capacitance to the shield, and it is this load that will affect the signal you are trying to measure!

As you say Patrick, I think the best thing is to construct one and then calibrate it. If I understand correctly, you don't need this probe to be accurate to 5 digits, you are just making the point that you'd have to measure the capacitors to that accuracy if you wanted to get their values spot on at the construction stage, and that is virtually impossible.

Patrick

Fri Jul 02 2010, 08:39PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

IntraWinding, yes on extending the shield up as high as is practicable.
and yes i meant that i cant depend on the accuracy of the caps to determine the accuracy of the division ratio.
also the Op-amp is in the capacitor dividing probe case and inch from the caps,

so the signal path is: [HV signal in] ->(HVLV caps+OPamp) -> (BNC+1 meter RG-59+BNC) -> Oscope[Display]

hopefully [HV signal in] looks like Oscope[display] and all the other crap is transparent.
so the op-amp isolates the 1 meter cable and oscope from the ball, mylar caps.

klugesmith

Fri Jul 02 2010, 08:44PM

Registered Member #2099 Joined: Wed Apr 29 2009, 12:22AM
Location: Los Altos, California
Posts: 1716

Mattski wrote ...

The shielding concept is good because it means that outside objects will not affect the capacitance. It also adds a constant parasitic capacitance (the two red caps in series) which is in parallel with the original capacitance (green cap). (Thanks to Fabio for the original diagram).

I think not, because the shield is not floating-- it's grounded. So the second red cap will have almost no voltage across it, and will not significantly perturb the attenuation ratio.

>>It's not a bad thing, just something you must keep in mind as it will increase the total HV capacitance.

On that point I agree.
Even a 0.7 pF capacitor, at 50 kV and 50 kHz, conducts 11 mA. If that current were in phase with the voltage, the associated power would be 550 watts. I wonder what power will be dissipated in the 0.7-pF dielectric due to material losses?

Fabio

Fri Jul 02 2010, 08:49PM

Registered Member #122 Joined: Fri Feb 10 2006, 12:55PM
Location: Milano Italy
Posts: 148

Do you think in practice the shielding should extend a little further upwards to more thoroughly shield the capacitor?

the efficiency of the shield will softly increase if you extend it further above the "hot" ball, but unfortunately even the input stray capacitance will quickly increase too.
shielding cannot be raised too much to preserve a decent input impedance even on higher frequencies.

Please note that the setup i proposed will raise the input capacitance to some picofarads, if a high frequency and high impedence probe is required, the input capacitance cannot be raised too much (no more than one pF) hence the screen cannot be raised higher than the "cold" ball and cannot act as a safety gap

s you say Patrick, I think the best thing is to construct one and then calibrate it

I have few potential transformers, if wired in the right way with no more than 20W of load are guaranteed for 0,2% tollerance, someone know how much can be precise if used in reverse?
I think that two 20000/100 PTs connected in backwards with both LV and HV sides wired in series and powered directly to the mains can be really useful for calibrate this probe, assuming 230V on power grid the output should be 46kV RMS sinusoidal (65,054KV peak!) of course at 50Hz only!

Ciao!
Fabio.

Patrick

Fri Jul 02 2010, 08:49PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

yes klugesmith thats why i chose to keep the HV capacity down, as i was warned in a previous thread from a year ago, also as per your 11 ma, 550 watts numbers, yes your right, and thus i chose 1/4-20 thread bolts, for low resistence and therefore the heat loss, and the reactive current can slosh in and out so long as it doesnt generate much heat. so 22.0pF might have some unwanted effects.

Patrick

Fri Jul 02 2010, 09:03PM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

Fabio wrote ...

s you say Patrick, I think the best thing is to construct one and then calibrate it

PT's like all transformers are predictable driven in reverse so long as the driving conditions are exactly the same, forward and reverse, but it must be carefully done load wise due too iron + copper losses.if i undertand you right?

ie. sine wave, 50hz normalized current, and volts per turn , same mmf, and phi.

then yes what you insinuate is possible, Fabio, you buy the plane tickets and ill do the math! yeah lets go for it!but Fabio remember that 30-50kv is all thats really needed at 50-60 hz to calibrate a 100kv probe.(i dont need100kv, as it the ratio not the absolute value which must be found for calibration

OOps! i didnt realize the quote button would cause a double post.

first previous 5 6 7 8 next last

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob