High Voltage measurment with an O-scope.... / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

High Voltage measurment with an O-scope....

1 2 next

Move Thread

LAN_403

Patrick

Wed Dec 23 2009, 05:55AM

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

I have been considering building a good robust high voltage and current measurment device for an o-scope, for the past 10 years or so.
And a recent situation with Herr Zapp has riled me up. So ill deffinatley do it now. Ok so the purpose is to rectify this common
occurance....

Q: Hey how many kv do you think that is?
A: Uh...2kv or maybe 25kv.
Q: And how many watts output?
A: Uh maybe 25 watts or maybe 400 or maybe 10,000 watts.

Yeah, and where do these fancifull numbers come from?
Well often it is a baseless guess or an educated estimate if you are lucky.

Thats the problem, here is the obvious solution, we need to know voltage-time, and current-time relations. so I believe DMM's will
be useless (except for low freq, ac sine waves), leaving us the traditional O-scope traces for most applications like Iggy's and flybacks. Lets focus first on voltage, current can be sorted out later if I am succesfull at voltage measurment. With voltage division I can see
several options, (Jim Lux's website and sam's repair faq) but only two seem practical, resistive division and capacitive division.
I will first try the resistive divider, I have used several types and I have a 2G ohm one here at home. I would like to see if 100M
ohms/per 10kV would be enough for good scale, (ignoring shot noise) made out of 10-10M ohm resistors potted in epoxy. But they
will be overvoltaged for sure. This way I can keep stray capacitence down for 10-20khz wave forms. My biggest concern is harmonics, with square wave switching I wonder if all we will see is noisey bogus waveforms?

However, what is not so obvious is the ground path. I have seen some disable the o-scope ground and leave the whole thing floating,
which is dangerous and inculdes parasitic effects. (Differential inputs won't do either) To solve this I plan to use either the ISO-124 or the HP7800 optical A/D converters, once the low voltage is divided the power ground of the oscope and high voltage ground path will be independant. But both grounds will need to be kept close to 0 volts +- a few kv. Then the output of the isolation amplifier will be put to an op-amp then the o-scope. Both voltage and current signals will be isolated this way. Though with semiconductors the isolation bandwidth is a concern ususally less than 100kHZ.

This is my plan, please contribute better ideas or useful info if you have any. Especially if you think Im totally crazy

I will post anything I have forgoten later, without a double post.

Oh one last thing I would hope some of you could also verify/duplicate any results I get so we have a higher confidence factor in our measurements. I will post methods, specs, conditions, and sources later, when known.

Useful websites:
jim lux ....

sam's repair faq HV ....

Devices:
Texas instruments / Burr-brown .... ISO-124P
HP / Avago ..... HCPL-7840

-Patrick

Sulaiman

Wed Dec 23 2009, 09:19AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3141

I have a P6015 HV probe, the manual may be interresting to you

Have you looked at this

thread and earlier?

Patrick

Wed Dec 23 2009, 07:43PM

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

Yeah Suliaman that thread is useful thanks, I always do a cursory search before posting a new thread, but that old thread didnt come up. Anyway yeah I will read it again, if frequency becomes a critcal limitation then I will switch from resistive to capacitive dividers, as that is what is used at CERN and lawerence livermore labs for partical research, impulse, high frequency 300 kv stuff, and I have some PDF's describing them.

That P6015 is super dooper good, way better than that dam fluke HV one, and that scematic on the last page is useful to me.
But maybe capacitive dividers are the way to go.

-Patrick

Sulaiman

Wed Dec 23 2009, 08:57PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3141

I would say use BOTH capacitive AND resistive, e.g. 45series x (2M2 // 47pF)
Below 1.5kHz it will be resistive and above it will be capacitive - best of both worlds

ALSO the resistors ensure voltage sharing across the capacitors at dc to kHz
and the capacitors ensure voltage sharing across the resistors at high frequency

AND the resistors will rapidly discharge the capacitors - e.g. after measuring dc
AND you can use cheap low-voltage ceramic capacitors.

Patrick

Wed Dec 23 2009, 10:05PM

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

Oh ok, so both methods integrated in the same device. Well i plan to make a 100kV picofarad cap (with pipe+epoxy+acrylic), so I then just need a higher nanofarad cap at like 1-2kv for the lower end right? I guess resistors at the 300 Mohm level I can fit in my pipe too. Why so many (45) at such a low value (2.2Mega) when I have 6,000-10MegaOhm ones?

Sulaiman

Wed Dec 23 2009, 11:14PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3141

Just to highlight; e.g. 20 kV rms @ 100 kHz with 3pF probe, (TC primary)

I rms = 2 x PI x F x C x V = 37.7 mA rms !!!!! for just 3 pF

Insulation of high-voltage at high-frequency is non-trivial,
since significant current can pass through dielectrics/insulators.

Patrick

Wed Dec 23 2009, 11:42PM

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

Yikes !!! is it the dissipation factor for PVC, you mean? or just reactive current sloshing around? Is that 37ma Is conserved or dissipated?
I was worried it would be this complicated, Ill start up multiSIM v10.0

And 3-5pf was what I was shooting for too.

EDIT: Ive been able to reduce the reactive current from one giga-amp, too 40mA, to 2.3mA (at 12,000 vac RMS 10khz ) through a 3pf/3nf divider by using a 1kohm in sereis with the probe tip. I dont know if that trick will cause a lag or droop though on the o-scope?

EDIT:I get 22.9 mA at 100 ohms in series (at 12,000 vac RMS 100khz )
I get 22.9 mA at 1K ohms in series
I get 22.9 mA at 10K ohms in series, though a tiny translation was seen in the sim.
dam it, dam it, dam it, I get the same 22.9mA with only one ohm too, maybe the spice just needs a resistence, any resistence to make the RC math work? (infinti X anything = sucks)
So,I believe suliaman's 37.7mA (exact) and my 22.9 mA rms (simulated) are close approximations of the real circuit conditions.

EDIt: changed some variables on the meters and oscpe and now Iget 38.2mA,quite close to what faraday and suliaman predict.
I feel better now

Sulaiman

Thu Dec 24 2009, 09:02AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3141

The example I gave was for reactive current, 90 degrees out of phase with voltage,
"just reactive current sloshing around" i.e. an effect on the circuit being measured.

However, if you use pvc as your dielectric you're in for a shock ;)

The thermal dissipation in a dielectric is VAR x DF
VAR is Volt.Amps Reactive, in my example 20 kV x 37.7 mA = 754 VAR
DF is dissipation factor, 'good' dielectrics DF < 0.001, so Heat < 0.754 W
pvc DF can easily be >0.1 (depending on frequency) so Heat > 75.4 W in the pvc!

i.e. DO NOT use pvc as a dielectric for high frequencies.
whatever material you choose look up it's DF first.

Here are some more points to ponder;
- if your single high voltage capacitor fails what will protect your 'scope - and YOU
- commonly available opto-isolators will not stop 20kV @ hf
- unless your divider is shielded (like the P6015) 'stray' capacitance will cause errors
- 'Surface tracking', Google it, think about it, test it.

I'm not trying to discourage you, just helping to avoid some 'oversights' I've made
I have learned much more from my failures than from my successes!
As you learn more you realise that many of your successes were due to luck !!!
Also, the older I get the less immortal I feel.

Patrick

Tue Jan 12 2010, 02:47AM

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

Yeh, Sulaiman how old of a dog are you?!

Yes I know about the dissipation factor, I constructed a Pico-Farad cap that tested to 2.2pf, approx 100Kv what I want.
as for the opot's there good for 3750Vrms isolation, without flames. (or 1000Vrms w/o breakdown and still funtional.) Anyway the opto is for level conversion and ground plane seperation, like 300-500 volts on the low voltage high capacitience leg of the divider.

let me up load a pic.

100 Cap

I dont want to double post but the pic thingy button doesnt seen to work on edit tab!?
I only get the [img][img] part, with no file path in between, and this pic is super important!
In acordance with the site rules, I have decided to post a more specific thread [Hv Hf Capacitive Dividers] .

abc (unless you can explain to me how to post pics with the edit button?!)
(Much thanks too suliaman though)

-Patrick

Steve Conner

Tue Jan 12 2010, 10:45AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

The Edit function won't let you go back and add more pictures. To add pictures, you have to post them in the Attachments forum, then edit your post in that forum, copy the code/image link out of it, and paste it into the post you want to add the picture to. Search the forum for more details, I'm sure there's a thread on it somewhere.

1 2 next

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