Fluke Scopemeter - overcautious or safety violation? 2

[Skogsgurra]

I am having an ongoing conversation with Fluke about certain aspects of the Series II scopemeters.

One of the aspects is what voltage limits that apply to the isolated inputs. Especially between what Fluke calls 'Probe reference' which is the same thing as the ground clip (or 'cold' side) of the probe and real (earth) ground. There is a diagram in the manual saying that voltage between probe references and and between probe references and real ground must not exceed 30 V when the signal frequency is more than 25 kHz.

I can understand that specification. That is why I always use an isolation amplifier or differential probe when measuring high-power VFDs with PWM inverters.

Fluke top technicians say that I can use the inputs directly (with the 10x probes), without any extra isolation "because the voltage above 1 kHz will be very low - a few volts maximum".

I do not agree. The derating curve and actual measured data can be seen here:

http://gke.org/pub/files/Fluke Scopemeter Series II maximum voltage between probes 1.pdf

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[Marke]

Yes I sure did. They are looking at the RMS values of each harmonic, but the RMS value is low because of the square waveshape. The insulation is damaged by the peak "transient" value that initiates the breakdown. The energy is in the lower frequencies present, but once the breakdown has begun, it does not stop!!

Best regards,
Mark.

Mark Empson
Advanced Motor Control Ltd

 

[Skogsgurra]

That is also true, Mark.

But it is not as simple as that (or complicated, for some). I haven't even started talking about the transients (they are reaching 900 V and not 750 V, I misread the graph). All I am worried about so far is the waveform with 'body' or energy - the square waves. Transients are probably not a problem because this is a 1000 V CatIV instrument and that means that it can take single 8 kV transients without exploding - which is quite an achievment for a scope.

Back to the issue at hand: If you look at the carrier fundamental, you have close to 270 V at 31.5 kHz. And around 450 V if working with a 690 V motor. That is more than ten times the limit - and around hundred times the voltage Fluke says there is. Remember that they say "only a few volts at 1 kHz"?


Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[waross]

Pete;
Re the 3:1 limit with 10 to one probes. 1000 volts is a safety limit, not a reflection of the probes ability or the ratio of the 1:1 probe to the 10:1 probe.



Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[Skogsgurra]

Yes Bill. But, still: it has nothing to do with the probes. It is all about voltage between probe references and between probe references to ground. Please try to grasp that.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[VEBill]

... The spec is voltage between probe grounds (inputs are isolated) and between probe grounds and 'real' ground. That is where the problem is. ...

Sorry if this point was already discussed: Doesn't it run on batteries?

 

[Skogsgurra]

It can run on batteries. But there is also a mains adapter.

The specs do not mention if mains adapter is connected or not. But 'private correspondance' says that it is impossible to use a small, light, SMPS mains adapter with a wide input voltage range because it would not fulfil the CatIV requirements.

I think that is a non-issue. SMPS can be built as sturdy as any scope. The PSU used is a very old design. With a heavy transformer and a switch for 110-ish and 220-ish voltages. That alone is a problem, if I come back from the US and forget to switch back to 230 V, I will fry the mains adapter.

I opened the adapter to see what's so special with it. Nothing, I can safely say. The PCB is single sided. No through-hole plating. The mains lead was squeezed when i put the cover back. Then the adapter didn't work. The cable had pressed the soldering pad off the board and separated it from the copper trace going to the transformer. Would not have happened with a double-sided plated-through PCB.

Fluke quality?


Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[electricpete]

Pete;
Re the 3:1 limit with 10 to one probes. 1000 volts is a safety limit, not a reflection of the probes ability or the ratio of the 1:1 probe to the 10:1 probe[/waross]
At the time I wrote that, I was thinking about the voltage limit between probe and probe reference (page 111) and not the voltage limit between probe references (page 112). As you can see, the curves look very similar.

The probe is some kind of voltage divider which brings the non-reference lead voltage closer to the reference lead voltage.

ASSUMING that the basis for this limit lives inside the meter and not in the probe (this is open to discussion, no-one has ever defined the basis for this limt), then we should be able to improve the measured probe-input limit between non-reference and reference inputs by close to the probe scaling factor. However since the common is not scaled, the probe does nothing to address any voltage difference between commons. That is why I suggested to measure from voltage to ground which can be measured using waveform subtraction. When I suggested this, Gunnar pointed out a specification in the back of the book some specifications that accuracy decreases with frequency. When I pointed out that those specifications do not mention subtraction so presumably have little to do with phase-to-phase or phase-to-ground hookup, Gunnar said we can discuss that in another thread. There may be a perfectly good answer. If there is I certainly would like to hear it. Imo, the important issue should not be what some anonmymous guy told Gunnar on the phone, but what are the capabilities of the Fluke scopemeter to safely measure PWM waveforms (without add-on's like isolation transformer).


=====================================
(2B)+(2B)' ?

 

[electricpete]

That is why I suggested to measure from voltage to ground which can be measured using waveform subtraction.
should have been:
That is why I suggested to measure from voltage to ground and compute phase to phase voltage using waveform subtraction.

=====================================
(2B)+(2B)' ?

 

[Skogsgurra]

Dear Pete.

I always measure between motor phases. Period. If you want to do it otherwise, please do so. But you will be very disappointed with the result.

The transient response is different for different probes. There are no separate DC, LF, MF1, MF2 and HF adjustments on these probes (like some high end Tek probes have) so it is not possible to have identical transient response up into the 100 ns region.

Why can you not let this go? Or at least start an own thread about it? There we can go into that problem in deepest detail. But, please, not here.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[electricpete]

"common" was a bad choice. Correction as follows:
However since the common is not scaled, the probe does nothing to address any voltage difference between commons.
should've been:
However since the reference input is not scaled, the probe does nothing to address any voltage difference between reference input.

=====================================
(2B)+(2B)' ?

 

[Skogsgurra]

Right! Pete. That's where the problem is. Let's forget about that arithmetic. It doesn't work. I can do some measurements to illustrate that. Stand by.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[electricpete]

Thanks. Phase to phase measurement vs phase to ground with arithmetic... it will be interesting to see the difference.

=====================================
(2B)+(2B)' ?

 

[Skogsgurra]

Yes. But don't have acess to that motor this Sunday evening. Tomorrow, I hope.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[electricpete]

Whenever you get to it would be great. I have a LPS reserved for you fwiw.

We have a fluke scopemeter. We also have some VFD's. The weird thing is that we only have vfd's on our small non-critical machines, while my responsibility is the large critical DOL fixed-speed machines...so I have never gotten any opportunity (excuse?) to do measurements on our vfd's. But I'd like to be ready if the opportunity presents itself.

=====================================
(2B)+(2B)' ?

 

[Skogsgurra]

I did a little excersise on the bench. Wanted to do A-B instead of B-A, which I had. That is NOT possible. I cannot select A as operand 1, because I already have that as operand 2. And, I cannot change operand 2 either, because I already have chosen B as operand 1. A catch22 situation that is absolutely unnecessary. Screen dump attached.

The scope is full of such things. They haven't had any software manager, it seems. And no SW QA either. That is what I have been trying to communicate for more than two weeks. They cannot understand what say. They hence do not believe me. It is a sad situation.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[Skogsgurra]

Sorry. Screen dump also.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[ScottyUK]

Gunnar - is a work-around to select C or D as one measurand while you reconfigure it the way you want? Seems like a pain in the ass, but is it possible?


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[Skogsgurra]

Yes. That is what I did. But, to make it real PITA, when you have 'parked' the last channel - you are kicked out and have to find that math menu again and set (the now not grey) channels. Lots of such things. Did I tell you about the inverse movement of channel C when channel B is inverted? There's more. Get a 190-204 and be amazed at what errors a bunch of hired SW patchers with no SW manager or QA can achieve.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[electricpete]

The transient response is different for different probes. There are no separate DC, LF, MF1, MF2 and HF adjustments on these probes (like some high end Tek probes have) so it is not possible to have identical transient response up into the 100 ns region.

For typical PWM output, isn't it the case that during the time when one phase to ground voltage is switching, the other phase to ground voltages will generally be constant? (i.e. only one phase switches at a time in space vector control). In that case, there should be no need to compare the transient response of two probes since just one phase to ground probe sees switching at any given time. (right?)


=====================================
(2B)+(2B)' ?

 

[electricpete]

I know you are already familiar with this, but for the sake of clarity, the PWM that I am picturing is as shown in Figures 11 and 12 here:

http://www.google.com/url?sa=t&source=web&cd=1&ved=0CBgQFjAA&url=http://

http://www.infineon.com/dgdl/AP1609...z63CAw&usg=AFQjCNE1b1RcAabrwxu2NBCzMKKUTceXKQ

The point again is that only one phase switches at a time.

=====================================
(2B)+(2B)' ?