another small breakthrough on our NERD technology.

[TinselKoala]

Then why use an FG at all? As humburger showed a year ago, all you have to do is poke the circuit once and it will oscillate until you "unpoke" it.  Just set the thing up, and give the Q2 gates a little pulse of positive. The thing will start ringing and won't stop until you turn those gates off again with a corresponding ground or "negative" pulse to drain the charge away.
Without seeing what the FG is set to, UNLOADED, I don't think it's possible to actually say or determine from scope traces what the circuit is actually getting from the FG the way it's hooked up here.


But what about my optocoupler idea? Would the addition of the optocouplers preserve the oscillations and mosfet switching? I know it would take the FG out of the loop as current source.
I think this: The optocouplers might work as I anticipate, but Rosemary won't accept the circuit built that way even though it oscillates and makes all the other effects... because it won't charge the batteries.

[TinselKoala]

(snip)
You may want to double-check with Rosemary, I think the only analysis that she is interested in is when the function generator outputs -5 volts.
(snip)
MileHigh

What does this mean? If the FG is "set" to "output -5 volts" at, say, one cycle per ten seconds, and I check the terminals with a voltmeter, what am I going to see? And then if I reverse the terminals at the voltmeter, what am I going to see?

Or even simpler: make an LED pair, parallel but anode of one to cathode of the other one and vice versa. Let's call this a "backtoback pair". Now hook your pair to the output of the FG set to make an "output of - 5 volts". What do you see? Turn the LED pair around, what do you see? You will see that one LED lights in each case.... the one with the ANODE towards the "negative" output lead of the FG. What this means is that there is a 5 volt potential difference between the terminals of the FG.... only the reference level has changed. The FG will still put out "positive" wrt the other terminal of itself. How could the LED light at all if the FG was putting out "- 5 volts" ?

So the one lead of the FG's output swings from -5 volts to zero volts WRT the other lead. The second lead of the FG's output must therefore be at a POSITIVE 5 volts ABOVE the - 5 volt setting when it is at "zero" or ground potential. I know this is confusing and I'm not explaining it well, nor do I yet know what it does in the circuit. But try the back-to-back LED pair in your favorite FG or sim and you will see what I mean. All it really means IF THE FG IS TRULY FLOATING is that the red and black wires exchange sign. When the FG is NOT floating, as in the Ainslie case, I'm not certain YET what the effect is but I doubt if it has any effect on the oscillations EXCEPT to confine them to one or the other set of mosfets. I can put the oscs in either q1 or q2 or both, make them large or small, whatever, depending on the FG's amplitude and offset.

In the Ainslie circuit, the FG's "negative" lead is connected to the common grounds of everything else, either directly or thru the shunt resistor.

[MileHigh]

TK:

Some more thoughts and some history for your consideration.

Poynt simulated the circuit and removed the function generator completely and substituted a voltage source for the function generator like I mentioned.  He added some inductance to account for the wire lengths, he may have added some stray capacitances, and the circuit oscillated just fine with a comparable waveform to Rose's setup.  Naturally the virtual power probes showed the real power consumption of the simulated circuit but I think Rosie refused to acknowledge this part of the simulation.  She was in a tizzy to see the oscillation though.

You mentioned the function generator unloaded vs. loaded in the circuit.  I have been simplifying that in my head and didn't really consider that which was a mistake.  Part of the reason is that Rosie's unlabelled 'waveforms' in the DSO captures almost give me a headache to look at.  However, if we suppose that Rosie and the RATs were looking at the CSR and the voltage across the CSR was just a junk inductive/capacitive coupling waveform, and not the true current waveform, then that would be a *BOMB* for the RATs.

But then why not use the function generator as the CSR itself?   When you look at the unloaded function generator voltage you see the true voltage being output by the high-quality negative-feedback amplifier on the far side of the 50-ohm output resistor.  That voltage is always rock steady no matter what.  So if you scope the voltage at the function generator output when it is outputting "-5 volts" under the circuit load, then you have the current waveform!  You also have to assume that the 50-ohm resistor + interconnects inside the function generator form a near-zero inductance path.

Of course you might have to export a data dump (not sure if you have that level of equipment) and then just process it in a spreadsheet to compensate for the offset and resistor value to get the true current waveform when the Q2 MOSFET array is a quivering and in delicious spastic feedback mode.  Then you have the elusive true current waveform without having to "change the setup."  Because we know that if you change anything "all bets are off."

MileHigh

[TinselKoala]

Can you remind me again why " I " need the exact CVR waveform?



(I have not yet even _begun_ to roll out the heavy artillery..... )

(And yes, fwiw, included in the arsenal is a top of the line TEK DPSO with math and export capability, but how many shovels do you need to rebury a zombie anyway?)

[MileHigh]

TK:

I'll try respond to the issues that you raised.

Agreed that there is no need for the function generator at all.  Rosemary and the RATs simply did not realize that.  Agreed on unloaded vs. loaded and that prompted me to do my previous posting.

With respect to the optocoupler, don't you think it's a moot point since we know the function generator serves no real purpose?  The real way to take the function generator out of the loop is to have the proper voltage source and a 50-ohm resistor, i.e.; exactly what Poynt did when he simulated the circuit.  That way you preserve the path for the current to flow.

However, the other side of that argument is that Rosie Posie might go bonkers if you removed the function generator.

When I say that the function generator is outputting "-5 volts" that's the center-conductor signal terminal relative to the outer ground terminal on the BNC connector.  So the ground shield connection is "0 volts" and the signal terminal is "-5 volts" while the function generator is outputting the low component of a sloooow negatively-biased square wave.  Note relative to Poynt's reference schematic in post #1505, the function generator is not floating.  It's tied to a common ground point along with the battery set and the scopes.

I hope that is clear to you.  For the two back-two-back LEDs, let me shift gears on you.  Since the LED pair cause a relatively minor voltage drop relative to the battery set + function generator total voltage supplied to the circuit, it would be fun to insert them in series into the main current loop to get the "LEDs of Doom" effect where they clearly show that the batteries are draining as they power the RAT circuit.  To do a quick and dirty test to show which is the dominant direction for the current flow.

One last comment; everything I am saying is based on the Q2 array quivering only because of the negative voltage output by the function generator.  So I confining myself to "RAT oscillation mode."  If you want to explore what happens when Q1 switches on, that's your show!

MileHigh