Measuring floating voltages with oscilloscope

[jitter]

Originally Posted by DF96

My money is on 'operator error', which includes misinterpreting results, ignoring parasitic components, and using unofficial 'grounds' (including the operator's body). We have tried to explain what is happening, but you seem to desire a more esoteric explanation.

Agreed, I have seen all sorts of unexpected things with probes held on traces.

Even though I don't deal that much with scopes, I do know that measuring with it (1:1, 10:1) in high impedance circuits is like asking for trouble. Not all oscillators like the load of the probe (incl. its capacitance) and stop working the moment you touch one of its leads. Tracing an error on a board that's not working you're falsely led into believing the oscillator doesn't work...

Measuring for oscillating circuits, the extra capacitance might actually help in stopping it (or the reverse) temporarily...

Quite often boards that refuse to be programmed can be "helped" by putting a probe on the relevant trace. Putting the probe on the trace creates a situation in which programming is possible (after which the boards keep working without the probe because the programme is in it).

Then there's blowing up components by connecting scope ground to a ground in the circuit that is actually 300 V above earth potential...

[AndrewT]

Akis,
my question is safety related.
It has nothing to do with your measurement errors/techniques.

[akis]

Yes the device under test was floated by using a battery to power it, so no relationship to earth at all.

I have done more thinking and taken measurements.

As you have said, trying to measure at high source resistances is asking for trouble.

I calibrated the probes in the 10X setting using a square pulse. I found that the calibration is different at 10KHz than at 1KHz. I settled with 10KHz calibration.

Then the channels A and B must be adjusted so that in differential mode (ADD) the sum is 0 for the same source signal at the highest sensitivity (eg 5mV per division). If you do not do that then you will read different values when you reverse the probes. This answers one of my questions.

Still there are readings I do not easily understand. I have taken them all down and will have to study more in case I discover some pattern or explanation.

On another board someone mentioned "self capacitance" but I did not really get it.

The outcome of all this is the confidence in one's instruments and measurements.

I have so far learned how to adjust the scope, how to be very suspicious of its capacitances and how 200KHz is not a trivial frequency.

I think paying 300 pounds for a differential probe which is guaranteed to work is probably worth it

[cliffforrest]

Some progress, AKIS! Good!

BUT, since the source is fully floating why not simply ground (scope gnd) one output terminal and remove all need for differential measurements?

Please explain why you think this is not appropriate.!

Cliff

[DF96]

Originally Posted by akis

Yes the device under test was floated by using a battery to power it, so no relationship to earth at all.

No. All RF circuits have a relationship to earth via capacitance.

Self capacitance can seem a bit strange to people who think of a capacitor as a two-terminal component. Self capacitance is like a capacitor with only one terminal - the rest of the entire universe forms the other terminal. It can be a significant issue in high impedance RF circuits. Every node in the circuit will have some self capacitance.

[akis]

Originally Posted by cliffforrest

Some progress, AKIS! Good!

BUT, since the source is fully floating why not simply ground (scope gnd) one output terminal and remove all need for differential measurements?

Please explain why you think this is not appropriate.!

Cliff

Hi, yes just that. Since the oscillator is truly floating, to the best of my abilities, I figured it would not make any/much difference if I earth its common, or its output pin. However this is far from true, and not intuitive at all. It is part of the "mystery". I did not have time to think about it more, and that's why I did not reply earlier, nevertheless, if one side of the circuit has just 2pF more earth capacitance than another, then at 200KHz 2pF is 400K Ohm, and that would account for the "weird" results. Part of that was observed, I think, when I powered the circuit with two 9V batteries, and then realised that in some measurement modes, where I did not ground the "common", touching the battery cases surely made a huge difference.

If I ever manage to put all this into the simulator and receive similar readings I will post here. Basically a model of the two batteries, oscillator and probe - so that the simulator reads more or less what I read in practice.

[DF96]

Do simulators include stray capacitance to the rest of the universe? I suspect not. A full EM wave simulator will do this but not most circuit simulators.

We did keep telling you what was happening, and why. I'm glad you now believe us.

[KSTR]

Originally Posted by DF96

Do simulators include stray capacitance to the rest of the universe? I suspect not.

LTspice does have a stray capacitance option from every node to its reference node (0), and I always use it. Stray conductance is there as well. Of course it's only lumped element stuff, no wave propagation involved.

[DF96]

That's good. Can you set a different value for each node?