Claimed OU circuit of Rosemary Ainslie

[poynt99]

And of course you can add a 3rd stage filter. Or you can get wild & crazy with more advance filters. No big deal here. Just enough filtering to get a good shot at this on the scope to determine the DC voltage across the battery shunt to calculate the DC current. Personally, I seriously doubt that we need any filters, as the scope could probably do it without the filters. Scopes with probes are typically 10Mohms, with 10 or so pF. If the gang thinks that's taking away the "free energy" effect, then use the ultra high impedance op-amp. I have one that has 2e+15 ohms with not that much capacitance.

Paul

The filtered shunt does a good job as is shown in the test plan. The only thing I added and would recommend, is some good quality film capacitors across the big electrolytics, from 0.22u to 2.2u or so.

You can measure the voltage across each capacitor as I show in the test plan, or measure the voltage across the shunt directly, either works well, as I have verified this with my own measurements.

.99

[poynt99]

Hi MileHigh,

Thanks for the outline. If they're concerned about the large caps taking away the effect, then maybe .99 can use the method that I suggested from the start, a simple op-amp that has a filter stage-- very simple, quick, and easy to build, maybe 5 minutes top. The op-amp has ultra high impedance and will not intefer with the battery or Ainslie circuit.

What do you think?


Paul

I see what you are getting at Paul. Yes this would work with some active filtering etc. There is no reason however that this could not be done on the low-side 0.25 Ohm shunt--it shouldn't matter, and it is already there. In fact it would be easier here as the common mode voltage is very low.

But, both have a common problem, and that is the shunt inductance. The filtered high-side shunt effectively eliminates this problem, and is one reason it was chosen.

.99

[MileHigh]

Paul:

Interestingly enough, .99 pointed out to us how even a cheap multimeter can make very accurate DC voltage measurements of a (DC + AC) voltage signal because of the dual-slope integration measurement technique employed.  Very briefly, the voltage signal is integrated by using a capacitor as part of the A/D circuit, which in essence is a low-pass filter.  You can also be very accurate with the DSO if you are doing enough samples per cycle, which is the case with the Tektronix DSO's being used.  So I doubt that an op-amp filter will get you any better measurements than a decent multimeter or the DSO.

Rosemary:

I think that Poynt is holding back on uploading data because he finds it suspect.  If I recall correctly he did upload some runs of DSO data for his last round of measurements where at least the battery output power (POS) was in accord with the analog measurements.  He is not being evasive, he is just avoiding "garbage in - garbage out."  The only partiality I see in Poynt is his desire to generate valid data that he can stand by.

In contrast, look at all of the excitement over measuring 100 watts of power "across" the load resistor when the thermal data does not bear that out.  Those data runs done by Glen and possibly Aaron are invalid and should have been retracted as soon at is was apparent that the thermal profiling was not in accord with the DSO data.

As far as the "returning energy" goes, the moment that was shown in one of Aaron's early "oscillation" videos then it was acknowledged.  I think that we have a pretty decent handle on explaining it but it may not be perfect at this point.  To be more precise, it's the energy being returned to the supply source per pulse and not the voltage that is significant.  The pulse is 100 billionths of a second wide, and there is very little power being returned to the supply source.  I'll just repeat two points that I have made several times before:  If the supply source is a battery, my gut feel is that more than 99% of the energy in each return spike becomes heat, and does not charge the battery.  And who really cares about the return spike?  It's all about the thermal power dissipated in the load resistor compared to the net power outputted by the supply source.

Also, there is no logical reason for different supply sources making any significant differences in the thermal power dissipated in the load resistor whether it be a battery, a bench power supply, or either of the two passing through the low pass capacitor and sensing resistor setup to make a DC current measurement.  I am really just going in circles here and that's why I have quit the thread.  Perhaps I will comment after all of the new data is digested at the end of November.

You are very welcome Poynt!  Really big sheww!  lol

MileHigh


P.S.:  For everybody:  If all else fails there is no reason why you can't thermally profile the MOSFET and the shunt resistor.   That will definitively prove that there is no excess energy anywhere.  There are just a few minor setup issues for doing a proper thermal profile for the MOSFET and if somebody is not sure how to do it then just post some questions.

The mystery of the missing heat pies.  lol

[poynt99]

I'm going to state the bottom line here as things seem to be shaping up now that I've taken a few measurements with my new LCR meter.

Calculating current from a voltage probe measurement across a shunt is futile. The 4" probe ground lead itself has 0.15uH of inductance, and combined with the 0.12uH inductance of the shunt resistor, this is more than enough to skew the real current measurement in the circuit as I have shown in the simulations.

The only reliable method I can see at this moment to obtaining a valid current measurement, is by using a current probe.

I might add that my "cheap" inductance meter isn't worth much at all for low inductance measurements. The load resistor I've been using all along was measured to be about 29uH, whereas now with the new Kelvin meter, the inductance is actually 10.9uH.

Glen, send me one of your old and new shunt resistors, and I'll measure them for you if you wish, however, it's a futile proposition it would seem to expect valid results, even with a shunt of 0.001uH, because the wiring and probe ground lead itself carries enough inductance to throw off the measurements a significant amount.

My suggestion in order to get to the bottom of the power accounting (without spending $k), is to do essentially what Paul has suggested, and that is to measure the temperature of not only the load resistor, but the MOSFET and shunt resistor as well. Then treat all three as an unknown and compare them to a CONTROL just as has been done for the load resistor itself.

Then it is a simple matter of doing the power accounting to ascertain how much power each component is dissipating. That leaves only the challenge of obtaining an accurate POS measurement, and a solid methodology for that has been clearly outlined in the test plan using the DC voltage meters and the filtered shunt.

.99

[PaulLowrance]

The filtered shunt does a good job as is shown in the test plan. The only thing I added and would recommend, is some good quality film capacitors across the big electrolytics, from 0.22u to 2.2u or so.

You can measure the voltage across each capacitor as I show in the test plan, or measure the voltage across the shunt directly, either works well, as I have verified this with my own measurements.

They're complaining that the 10000 uF caps might be shorting out current spikes, and they have good reason to believe so. A good testing method minimizes the influence on the DUT, and those caps fail at that. No offense, but placing 2 10000 uF caps is just not that professional. That's why from the start I recommended using an op-amp with low pass filter.

My 2 cents. Sorry if I'm getting a bit frustrated because this is a one day test job at most. More like 1 hour. 

Paul