STEORN DEMO LIVE & STREAM in Dublin, December 15th, 10 AM

[Omnibus]

@teslaalset,

Now, I've been thinking about this and it occurs to me it would be interesting to analyze what levels of experimental errors must there be in order to get the effects I'm observing. Remember, I'm reporting times of OU, not some several percent. I'm pretty sure in the current value as well as in the resistance value. All errors, then, should be emerging from the measurement of the voltage. I just measured the capacitance of the 1X and the 10X probe with my RadioShack capacitance meter and found that the measured values for the 1X probe are on the order of 10uF while that of the 10X probe are on the order of 50pF. Significant difference, indeed (as expected according to the manual). The shunt theory may be in order here meaning that the 1X probe will not only show lower voltage but will shift the phase more than the actual phase of the actually applied voltage. Now, I plugged in into the voltage column of the spreadsheet a voltage value of about 0.700V less than that in the column of the current (the current supposedly is all right) leaving the shift of the phase angle unchanged. The result was nearly 1% OU. To be ignored -- the voltage amplitude isn't the culprit. Much greater effect seems to be that of the phase shift, only if the phase shift is greater than that corresponding to the actual phase shift of the RC circuit. This can be achieved by lowering the capacitance. However, I thought we said the capacitance was increased compared to the real one. Something doesn't jive here. That's on the experimental side.

[Omnibus]

@gyulasun,

Turns out the Keithley 2000 multimeter I have is of the level of the Fluke ones you suggested. However, here's what I got for the same signal:

1X probe - 32.6ppV
Keithley - 11.1896ppV
10X probe - 33.2ppV

The Keythley has 1MOhm input resistance and about 100pF capacitance, just as the 1X probe. It is obviously no good, though. Now, both the 1X and 10X probes have +-2% accuracy which, combined with the 8 bit resolution of the scope (or stemming from that resolution) makes this system almost hobby-like one, doesn't it? Well, DPO 7000 used by Steorn are 11 bit at high res and this makes them also not exactly the perfect instruments if we're really concerned with high accuracy (theirs being way better than mine). Now, I didn't go on with modeling the 1X probe while measuring with the 10X probe because it doesn't make sense at this point -- both probes have accuracy on the order of +-0.650V.

[Omnibus]

Now, speaking of 1X vs. 10X probe accuracy, one accessible way is to find out experimentally (recover) what would the value of a resistor be by measuring I and V across it and then compare it with the known value of its resistance. As a bonus we also calculate the OU -- both probes appear to show some slight OU (if you'd agree that the integration is over a full period). See first the setup. Notice the trace of parasitic inductance in the 10X traces which isn't seen in the 1X traces. That may be contributing in killing the OU effect when measuring with the 10X probe. The two probes seem to yield a similar result regarding the average R which is lower than the actual R value.

[gyulasun]

@Omnibus

You wrote:

"1X probe - 32.6ppV
Keithley - 11.1896ppV
10X probe - 33.2ppV "

The Keithley 11.1896V cannot be peak to peak but RMS, right?
Converting RMS to pp it gives about 31.6489ppV, well within accuracy spec.

You did not include what frequency you used for these voltage amplitude tests.  Your Keithley meter has about a  .6% accuracy between 50kHz to 100kHz AC and 4% from 100kHz to 300kHz. Above 300kHz it is not specified.

I cannot recall at the moment if your "OU"  occurs at much lower frequencies than 700-800kHz?  If it does, then I think your Keithley meter would be a good choice on measuring AC voltages at say 20-30kHz or lower where it has a  .1% or better accuracy.  In the some kHz range you could still use your current probe too.
And if your earlier RS transformer (you put aside in favor of the RC circuit measurements) still has "OU" in the some kHz range (use the audio range from 1-2kHz to 10-15kHz max) then the Keithley could also be used with its  .1% or better accuracy.  By the way the Keithley has AC current measurement feature too, it has a .1% accuracy from 10Hz to 5kHz in the 1A range, try to consider these.

With the Keithley meter you have a more precise measuring instrument for AC voltages up to about 50-100kHz and for AC current measurements up to about 5-10kHz than your DSO scope.


On your probe input capacitance measurement you wrote:

"I just measured the capacitance of the 1X and the 10X probe with my RadioShack capacitance meter and found that the measured values for the 1X probe are on the order of 10uF while that of the 10X probe are on the order of 50pF."

How can the 1X probe have 10uF, it sounds extremely high, even if your cap meter is from RadioShack.  A 10uF capacitor could be considered a short circuit (as a piece of wire) from 30-40kHz and up in frequencies and surely would have shunted the 20-30ppV of the FG output into the some hundred millivolt range. You can test this at say 100kHz: measure the output from the FG by the Keithley (say you output 5V RMS = 14.14ppV) and then connect the 1X probe (which is connected to the scope input) also to the FG output and see how much change (i.e. amplitude decrease) the Keithley shows the moment you connect the 1X probe.

You mean on the "capacitance of the 1X and 10X probes" their input capacitances between the "hot" pin and the crocodyle clip, right?

rgds,  Gyula

[Omnibus]

@gyulasun,

Thanks again for the good analysis. Of course the Keithley voltage is RMS. My bad. As for the frequency range where I'm observing OU, it is typically above 200kHz so the Keithley won't do any good (the probes have 2% accuracy in that range). For the time being I'd like to stay at the 700-800kHz range to be systematic. The voltage data I presented is for 800kHz. What makes an impression is that even if we consider the 4% accuracy of the Keithley multimeter (probably it's lower but let's take that for the sake of argument) then that value is closer to what the 1X probe measures. Could it be that the 10X introduces some parasitic inductance as seen also from my latest data? All in all, aside from the 8 bit resolution of my scope (which probably is the greatest concern), I'm getting more confident that the 1X probe gives more reliable results. That's why I was saying this has to be tested independently with better equipment. Like I said in @Omega_0's string, however, my attempts to talk some prestigious labs into replication it are meeting with dead end. They don't want to hear about it. What else is new? Will keep trying, though, because we need a clear answer one way or another.