Claimed OU circuit of Rosemary Ainslie

[Rosemary Ainslie]

Poynt - not sure what you're doing there.  Take the shunt voltage / 0.25 * vbat.  That will give you vi.  Golly.  Strange me telling anyone how to do this?  Hope that's clear - else skype me.  Maybe I can talk you through this.  But you need to open the file.  I have difficulty there.

[poynt99]

OK, I've spoken with Rose and been informed of the calculation method they used. It is as follows:

Take each sample and multiply the instantaneous battery voltage with the instantaneous shunt current (instantaneous shunt voltage/0.25 Ohms). Then take the average of all these instantaneous power calculations.

This does not equate to the same number when calculated as I did above. In fact it comes to about -0.26W average.

The assumption I made is that the average shunt current is an accurate representation of the average battery current. In my tests they are quite close.

If I am correct, then the average numbers should be computed together (for the battery power only) as I did above to arrive at the average power consumption.

Install a power meter as Luc has built for himself, and I am confident that the average power will be quite different. This power meter consists of two large capacitors and a current sense resistor in-between. Accurate average power is easily obtained by this method, and is the method I intend to use for my tests.

.99

[Rosemary Ainslie]

Poynt.  Thanks for this.  Just remember that you need to eliminate some of the samples to accommodate the full cycle.  Else the average will be coloured by some imbalance with the inclusion of some waveforms that are not representative of a full cycle.

I can't comment on your proposed method of measurement.  The only one that is accepted by academics is the differential equations related to the data as detailed.  But I would be very interested to see if the waveform can be replicated on Spice.

[poynt99]

Poynt.  Thanks for this.  Just remember that you need to eliminate some of the samples to accommodate the full cycle.  Else the average will be coloured by some imbalance with the inclusion of some waveforms that are not representative of a full cycle.

I can't comment on your proposed method of measurement.  The only one that is accepted by academics is the differential equations related to the data as detailed.  But I would be very interested to see if the waveform can be replicated on Spice.

Rose, you should know me better by now. I did the same calculation earlier after we spoke using the 400ns data sample Glen posted. I trimmed it to exactly one cycle. The results were within a few percent. The 40us sample range gave about 17 cycles, with 57 samples per cycle. A half cycle out of 17 would not throw the results off substantially, but I wanted to do the single cycle computation anyway. So I did.

The wave form can be created in SPICE. I have already done this. It is not an exact match, but it shows the negative spike, just slightly after the big spike on the drain. With some tweaking and time, it could be improved I'm sure. The question is, what is to be gained by doing this?

.99

[Rosemary Ainslie]

  Hi Poynty.  Dare I tell you this?  One cycle is not representative unless it's periodic.  You need a range?  But it's all good.  Very little difference.

And you know why I want that waveform!!!!  Why do you ask?