Digilent OpenScope MZ for very high common mode rejection measurements ?

[chris_leyson]

I'm currently faced with a measurement problem that I think CERN would have difficulty with, I'm running electrical fast transient tests at 7kV 5kHz burst and would like to measure a few signals at say 3V3 and 5V logic levels just to get an idea as to what the hell is going on.
Differential measurements with, say, x100 scope probes are out of the question. For a start they're not rated for 7kV and the probe capacitance to ground would inject enough current into the signal line under test to cause some serious damage. You've got rise times of the order of 1000V/ns and that is 1A into a 1pF load.

The solution is to break the common mode path to ground and a small 'scope' with a wifi connection is one obvious choice. OpenScope MZ, 12bit 6Msps is good enough to see what's happening during a 15ms long burst. Integrate the 'scope' and it's 5V power bank into the DUT and you're good to go.
If your DUT has 5V available then you don't need the power bank.

Not an easy measurement problem by any means but a wireless connected digitaizer will give you a lot of leverage because you've broken a major common mode connection when trying to make a measurement. I will give it a shot next week and see how well it works or not. Funny how none of the interweb EMC gurus don't bother to address this sort of measurement problem. Any thoughts on the matter would be appreciated.

[wilhe_jo]

To be honest, I don't really see your problem.
If you really have 7kV common-mode on the lines, then your design is probably very bad.

I'd say everything above 10-100V of CM voltage would be a real problem.
The coupling-clamp for the burst test has a defined coupling capacitance (I can't remember the value from my head) which forms a voltage devider with your filter (and parasitic) capacitances (and CM-chokes if present) to ground in your EUT.
In addition to this, your generator should have a source impendace of 50ohm IIRC.
So you will never have the full 7kV on the lines.

Whenever I have such problems to trace down, I usually put the probes at the input pins of the IC inside the EUT.
So I use the filtering/protection circuity inside the EUT to protect my scope-probes.
At this location you should stay well within +-20V (CM disturbance+signal) for common communication applications...

The worst case would be to use an isolation transformer to get the impact of the coupling to ground from the scope further down.
Usually you need to do this if you find your device working whenever you connect the scope

BTW there's the cleverscope:
https://cleverscope.com/products/CS448

To be honst, I just got a marketing presentation on the features.
I have absolutely no experience with the real product but the numbers sound quite interesting:

Code: [Select]

Isolation Capacitance< 14pf Channel ground to chassis
Isolation Capacitance < 6pf Channel to channel
But your idea has another very interesting aspect.
Whenever you get the opportunity to trace down problems in some high-side-gate-driving circuit, that isolated scope would become quite handy.
There you really have signals "jumping" some 500V-1kV wrt ground...

73

[JohnG]

I always hate to suggest very expensive solutions, but there is one, and your haven't mentioned anything about a budget.

This will most likely do what you need, and very well: https://www.tek.com/isolated-measurement-systems

It hurts me to recommend this, because you also need a Tek scope to go with it, and frankly I have been very disappointed in Tek scopes for many years now. But, the probe works.

Cheers,
John

[chris_leyson]

@wilhe_jo thanks for the Cleverscope suggestion, nice scope 100dB CMRR at 50MHz.
@JohnG IsoVue wow 160dB CMRR and a price tag to match.

Thanks for the suggestions guys it's very much appreciated. I've had a busy week but with improved grounding and a few other tweaks we're good at 7kV 5kHz burst and 6kV 100kHz burst. Will try to measure the EUT chassis voltage with 7kV burst applied to protective earth but best guess is 2kV. I had a stab at using the Openscope to make a few measurements but didn't get very far, it only triggered once in RUN mode and locked up, turned out to be the browser, seems Waveformslive didn't work with Microsoft Edge but I've now got it running OK with Firefox. Haven't tried IE yet. Would be nice to use a local copy of Waveformslive rather than rely on an internet connection, haven't tried the micro SD card option yet to download Waveformslive from the Openscope. It just seems crazy to use a browser to interprate JSON script, what can possibly go wrong.

Unfortunately my Red Pitaya is broken, busted SDI mux I think, so it's not going to get fixed any time soon. Shame really, would have been nice to get the extra bandwidth and no network needed.

Hopefully I will grab a few measurements soon, just out of curiosity 

[Doctorandus_P]

Wireless has it's own problems.
7kV stuff is likely to generate lots of RF noise, which may completely make communications impossible.

Have you considered optical solutions?
With optical S/PDIF  you have cheap optical cable, drivers and transmitters, the cables (cheap plastic) can be several meters which would make the capacitive coupling completely negligible.

Normal mains powered SMPS supplies almost always have a high voltage capacitor over the optocoupler (through the isolation barrier).

Another way may be to look into the ADUM series of IC's (National?) I'm not sure if they can get to 7kV though. A meter of optical cable seems a much safer way to go, even if there is no "official" specification for it

[PartialDischarge]

How do you generate a risetime of 1000V/ns with such high voltage?
I’ve measured signals of a few volts riding in 1500V common mode voltages with 30kV/us of slew rate with no problem using an optical probe, and would like to try it with something faster.

Be aware also that cmrr of optical probes lowers as input attenuator are added, input range without them is only 0.5 to 1V. Besides that the cmrr measure is difficult to use in practicality since no one is measuring sinewaves, but pulses, so a rejection measurure based on slew rate should be more apt.