Strategy Ruminations

[TinselKoala]

I'm glad you're posting real results with good equipment.

I am just guessing here, but the difficulties you are reporting MAY be due to the following.

The discrepancy between the several probes that you show on the same signal in the scopeshot above is a classic illustration of "probe skew", which is a result of slightly different time constants of the probes themselves. Active probes, as you can see, suffer from this. Many modern DSOs have a "de-skew" option whereby this error can be removed. Also important with Hall-effect and transformer-type current probes is degaussing. Generally, the current probe should be degaussed and zeroed (using the scope's built-in functions if possible) before each measurement hookup.
This skew, as you can see, will result in incorrect calculations down the line.

The order-of-magnitude discrepancy sounds like the scope may not be auto-detecting the probe attenuation at some point-- or perhaps it's somehow being set incorrectly manually. I would suggest that you attempt to verify the probe's reading by setting up the situation where the probe is indicating what you think is the wrong reading, and then check it with the calibrator or other known voltage source.

[Omnibus]

There are two issues actually. The instant values of the voltage and the phase shift.

Now, of course I'm degaussing the current probe in absence of current, setting it to zero, with a deskew value 0.00s. Then, after applying current I'm waiting until 512 traces are averaged. The voltage probes (both the passive and the active) are also first zeroed, deskewed etc. in absence of current. So, that part I think is good.

Now, mind you, the current measurements should be OK since the output power, based only on the current measurements (and the resistance), come out practically coinciding with the theoretical output power.

As for the independent checking of the instantaneous voltage values I independently measure the source I use (the HP pulse generator) with a Keithley 2000 DMM and judging from the rms it also seems to be good. That is, there doesn't seem to be a problem with the voltage values. Besides, all three different ways of voltage measurement give coinciding results in terms of values of the voltage.

One other thing. It seems to me that the phase shift problem may be due to a slight delay in the attenuated probes due to the need the voltage to be recalculated. Probably the phase shift of the 1X probe is the correct one because the value does not need to be recalculated and therefore it appears in pair with the right time label, as it were. What do you think?

[Omnibus]

Also, don't forget the theoretical analysis. It's a problem in itself, independent of any experiments. As is seen from the data I posted above OU is inherent in the very standard theory of electricity. That deserves a separate discussion.

[Omnibus]

OK, I see Tektronix has a special add-on for precise deskewing probes used in power measurements: http://www.testequipmentdepot.com/tektronix/pdf/tekdpg.pdf So, now what? Spending more and more money for things they should've provided for to begin with? There should be some simple way to precisely deskew these probes. Otherwise, how can you ever be sure that your power data are accurate?

[Omnibus]

Now, I'm playing with the deskew of the active voltage probe, trying to adjust it so that the outcome will fit the theoretical power in. And, indeed, if instead of 0.00s I set -8ns the experimental power in becomes of the same order of magnitude as the theoretical. However, that's incorrect methodology because I'm presuming that these two values (experimental and theoretical) should coincide. No wonder why Tektronix is selling a special deskew "calibrator" for over a thousand bucks. Having that, however, is crucial.A couple of ns skew along the time axis makes all the difference in the world.

EDIT: I should keep adding to this post for some time, probably.

So, here's what I did. I adjusted the deskew value to a number which yields an experimental result (of the power input) practically coinciding (of the same order) with the theoretical at 0V offset. That deskew value is -8ns. Then I studied the power balance at a positive (1 and 2V) and negative (-1 and -2V) voltage offset and and I now see a huge difference between the input and output power. Thus, while the Tek calibration gadget would give an absolute value of the necessary deskew figure this relative method also seems to indicate that the also the experimental power balance is not according to the CoE (we already know that at certain conditions the theoretical balance itself doesn't obey CoE).

Unfortunately, I'm unable to go beyond 16ppV with my active probe (which leads to very low current voltages, at the limit of detection of my current probe) but maybe I should redo the whole thing with the 1X and especially the 10X passive probe and will report the outcome.

2nd EDIT: After checking the 1X and the 10X passive probes I found that the the 10X probe requires the application of an even lower deskew (-17ns) while the 1X probe asks for no deskew -- the deskew for the 1X probe should be at 0.00s. Like I said, the amplitude and the momentary voltage values are practically the same for all probes.

Probably I should just return the active voltage probe and save one and a half grand. I don't see what use I may have for it in this study.