another small breakthrough on our NERD technology.

[TinselKoala]

@MH... you hope that it's clear to me.

It's perfectly clear to me. Is it clear to you?


Check the latest video, Electric OU: Supplement: Function Generator Positive and Negative Outputs.

http://www.youtube.com/watch?v=hz4o37g8XmI

If the results from the INSTEK are different, I'd like to see it. Consider the implications for the CVR current trace. I'm not sure if I grasp them totally yet.... but if the performance of the circuit can be derived from electrical parameters AT ALL.... well, see the last photo I posted.

And what happens when the Q1 switches on instead of the Q2 array is that the device can't carry as much current, so at the same gate drive settings you get less power to the load and more dissipated in the single mosfet. You can still have oscillations on this drain signal too, if you like, it's not a problem, even though that transistor might not have long leads, as in my case. There's no other major difference that I can see.

[MileHigh]

You have amazing toys that go Boom!

Yes, you don't need to crunch DSO captures to show that there is no RAT magic.  "That old RAT magic la daa la daa...."

I am still contemplating this notion that the whole time the RAT team was looking at the wrong waveform for the current because the CSR was out of the main current loop.

Looking back at Poynt's diagram you can see the entire wire that includes the CSR, the Q1 source, and Q2 array gate inputs would all be super high impedance except for the tie to ground on one side.  So it's possible considering the high frequencies involved that they were picking up capacitive-inductively coupled junk the whole time.

Hmmmmmmm....

MileHigh

P.S.:

There's no other major difference that I can see.

You are forgetting that there is no longer a 50-ohm resistor in the loop, so with Q1 on you will likely dissipate more power in the inductive resistor assuming the single MOSFET can handle it.

[MileHigh]

TK:

I groked the function generator video.  It's a really nice function generator by the way.  It has that late-70s look to me, when they made "real stuff."   

Anyway, not sure if you will replicate all the way, but what the RAT team did of course was crunch the battery voltage times battery current calculation and got "COP infinity."  Of course the battery voltage measurements were suspect the whole time because of the cable lengths and Poynt showed Rosemay simulations of how varying inductance/wire length changed the power measurements but Rosie Posie refused to accept this.

I am still not sure about this potential issue I am raising about the fact that the CSR on the schematic is not actually showing the actual current in the circuit when the Q2 array is in defib.  I would not be surprised if I am missing something and am wrong and Poynt will correct me.

But, to go out on a limb, here is a thought experiment:

Rosie's setup now shows "COP infinity" when they crunch the DSO data dump. (Or the DSO crunches it directly, I can't remember.)  Let's assume the primary problem is the junk voltage data across the CSR and the battery voltage measurement is also an issue.

So, with the same setup, they instead use the function generator trick described a few posts back to get the correct current waveform.  They also follow Poynt's recommendations for proper probe positioning for measuring the battery voltage.  They then crunch the new DSO data and get under unity with the batteries discharging into the load.

Ya never know.....   Then the fat lady sings....

That old RAT magic had me in its spell....
That old RAT magic that they weaved so well...
Icy fingers up and down my spine....
The same old witchcraft no longer sublime...



MileHigh

[TinselKoala]

@MH: I know what they did with the spreadsheet data dumps... they "think" they generated an instantaneous power data set by having the spreadsheet multiply the voltage data by the current data (really the voltage drop across the CVR as registered by the scope... and using a value of "0.25 ohms" exactly for the value of the CVR. I would have expected a true value within 5 % of that, but exact.. from sand resistors... OK, fine, maybe they spent some time matching, never mind). So they did VxA at every timeslice of their sampling interval. Then presumably they again multiplied this by the length of the timeslices and added these up to get an approximate numerical integration of the power data-- for the energy flow.
Why oh WHY did they not simply use the scope's math functions to do this and display it in real time? Shall I tell you? First, because they don't understand how to do it or what it  means, and second... because it probably won't support their claims, if the raw data is measured right, that is.

A couple more videos are being processed right now: dependence of oscillations on load, aluminium heat sinks, RF emissions from the circuit.




ETA: How do you do any arithmetic operations on real numbers in a spread sheet and wind up with an "infinity" in your result? Think about it.....

[MileHigh]

We had better take our compact reel-to-reel audio tapes now and do that hissy white smoke thing with them.

Cue that Mission Impossible Black Edition music....   

lol