another small breakthrough on our NERD technology.

[picowatt]

TK,

Thanks for the confirmation, but the LeCroy manual is very ambiguous as to the operation of the offset control.  It said use offset to adjust the zero volt level, and in another section, to use offset to position the trace.  I hoped the two would track!

After studying the screen shots, I thought I had it figured correctly.

Oh how I miss cursors!  All scopes here without cursors are under the bench for emergency use only!

Fortunately, I have plenty with cursors on top of the benches.

Regarding the bias current versus thru the CSR or not:

Even if the bias current was not in the CSR calculations, and one were to add +50mv to the CSR mean, that still does not appear to provide the required current to produce the observed heating.

Assuming Rosemary's heat profile is correct to within 10% or better, as well as her Rload temps, I am still having a hard time finding about 4 to 5 watts.  In fact, heat profile and temps could be a bit further off than 10% and I would still be a bit puzzled by about 3 watts.

PW

[TinselKoala]

Here's a little blast from the past using a differential voltage probe and a non-contact current probe for the data, and also showing how to use the real math capability of a LeCroy digital oscilloscope -- even a basic, bottomline one -- to make a point.
http://www.youtube.com/watch?v=IgvFHejoQEk
This isn't off topic at all, because it demonstrates that the scopes DO have the ability to extract the information we would like to see... if only used properly.

[picowatt]

TK,

How much heating of the load do you think your circuit can accomplish if you just bias on Q2 so that it oscillates?

Possibly consider connecting a bench supply set to -14V or so thru a 50R resistor to the source of Q2 so that only Q2 is oscillating and Q1 remains off.  It would be interesting to see what the oscillator section can do by itself regarding heat at the load.  If I am correct about the Q1 issue, all the March tests that indicated heating were able to do so without Q1 functioning.

If you have time of course.

PW

[Rosemary Ainslie]

As I said, I am not familiar with your LeCroy.  It is, however, my understanding that the "ofs" numbers, that is, the indicated offset, basically describes how far the zero volts line is above or below the center of the screen.

Again, referring to FIG3 in paper 1, the channel one trace zero volts line at the left of the screen is about 2.25 full divisions above the center of the screen.  That channel is set to 2 volts per division.  2.25 divisions times 2V per division is 4.5 volts.  The offset reading for channel one says 4.56V, which is in good agreement with my "optical" read of the zero volt line position.

The channel 2 zero volt line to the left is about 1.75 divisions below the screen center.  That channel is set to 100 volts per division.  1.75 times 100 equals 175.  As the zero volt indicator to the left for channel 2 is below the screen center, it will be a negative number.  The indicated offset for channel 2 is -176V, again in good agreement with my optical read.

The channel 3 zero volts indicator is about 2.3 divisions below the center line.  Channel 3 is set to 10volts per division, 2.3 times 10V equals 23.  As the zero volt line is below the center of the screen, it will also be a negative value.  The indicated offset is -23V, again in good agreement with my read.

I believe I am reading the scope correctly.

picowatt

Read the dc voltage across the shunt as per the display = -28.1mV.                  peak to peak  4.56V 
Then read the dc voltage across the battery as per the display =  73.8V            peak to peak  -172V

I think you'll agree that this conforms to the 'explanation'.  Those ofs voltages relate to ac coupling with the offset determined as AC.  The fact is that we didn't even use the display to record the applied voltage signal but rather its frequency.  My own computation as you now ask for this is roughly 4.5V or thereby - applied at the Gate of Q1.

I'll get back to the rest of your points later on today.  Meanwhile for those of you who celebrate the day for any reason at all - as I do - may you all have a truly happy day. I'll be back here, God willing - later on this afternoon.  Meanwhile I intend doing my own bit of partying.

Kindest regards,
Rosemary

ADDED
Here - finally is that download.

[TinselKoala]

TK,

How much heating of the load do you think your circuit can accomplish if you just bias on Q2 so that it oscillates?

Possibly consider connecting a bench supply set to -14V or so thru a 50R resistor to the source of Q2 so that only Q2 is oscillating and Q1 remains off.  It would be interesting to see what the oscillator section can do by itself regarding heat at the load.  If I am correct about the Q1 issue, all the March tests that indicated heating were able to do so without Q1 functioning.

If you have time of course.

PW

I have time of course.
I  think there is some heating of the load in this mode, consistent with your determinations of the current to be expected. What I believe is the following "uncharitable" interpretation, considering the casual attitude towards raw data and accurate reporting shown by the NERD team.
I believe the "five watt" figure comes from the math trace "numbers box" on the LeCroy. She is simply multiplying the raw CVR voltage drop by the battery voltage to obtain the math trace, thinking this gives her an instantaneous power curve. The data box reports the parameters selected for this trace in units of "VV", since that's what whoever set the math up told the scope the probes were used for. With her demonstrated deficiencies in observation and eyesight, it's plausible that she interpreted this VV as W and again took the boxed numbers at face value. In the LeCroys I use, the math user can tell the scope if a probe represents a current value and also THE VALUE OF THE CVR SHUNT, so then the scope's parameters box will display the correct units of "VA" or even "W". If a true current probe is in use, like in my JT demo, the scope autodetects it and displays its values in Amps instead of volts, and the calculation of the power curve will be using the right units. Garbage values in this case, but at least the right units.
So... she's reading "VV" as "W" off the math trace parameters box, and not accounting at all for the 0.25 ohm value of the CVR. I think this is the case, but she denies it.

So what I think is that the resistor heats during the "tuning" mode, in other words, while setting the circuit up and perhaps turning some mosfets on for a while... then when the oscillations are established and data recording begins, the load is already warm or even hot from the tuning. It has happened here, and I've documented it in a cleverly hidden video. And then the 200 mA during the oscs is enough to keep it from cooling off back to ambient for a while, or even warm it up a bit if it's cold.

And I can just use the F43 set to DC to make the required gate signal for continuous oscillations. This will also preserve the current path through the FG.... for better or worse, but there it is.

OK... data. I've used the F43 to make a DC output, and used its offset to move the DC down to the level where oscillation begins. I then move it further down until my inline cheapo DMM indicates around 200 mA. I measure the frequency of the oscs, and since I still have the 5 IRF830as in there it is around 4 MHz. (Sorry, somebody seems to have cornered the local market in PG50s, I don't have a complete set to use). I started with the load at 80 F (it is 10.3 ohms made of 5, 50-ohm Claristat wirewound ceramic tube 12-watt resistors in parallel, immersed in an insulated container with 250 mL mineral oil, specific heat 1.67 Joules per gram per degree C.)
After maybe 20 minutes the load is indicating around 86 degrees (analog thermometer; the RF messes with the digital ones). It's probably still warming up.

If the meter is averaging correctly and there's 200 mA in the circuit the load should be dissipating less than half a Watt. I also took the meter and put it in strict series with the FG's positive lead to the circuit. It indicates that same current... showing, I think, the current path through the FG itself. The open circuit voltage of the FG at this setting is -12 V.

Right now the load oil is at 90 degrees. Sorry, I approached this one casually and haven't been recording times at all, but it's been about an hour since I started at 80 degrees F.

I am fairly sure that the load used by the NERDs was not immersed in anything during these low-power tests, it just hung out there and had a thermocouple attached directly to it. My own digital thermometer cheapos go crazy from the RF in the circuit during oscs, but I do have an ExoTherm adapter and TC for my old Fluke 83 that might behave differently.... I suppose I'll have to dig it out.

So let's do a plausibility check. If the load is dissipating (.2)(.2)(10.3) = 0.41 Watts into the oil, and a Watt is a Joule per second, and I've run for about an hour, that's 60x60 seconds or 3600 seconds, so that makes 3600 seconds x 0.41 J/sec = 1480 Joules. If my 250 mL oil, weighing about 0.83g/mL x 250 mL = 208 g, has risen 4  degrees C, how much did that take? 4 x 208 x 1.67 = 1390 Joules.
So it's plausible to get significant load heating even at a current of 200 mA as indicated by the inline ammeter, DC drive at -12 V with the oscillations.
But I'm using the 830a mosfets. And my battery is still over 36 volts, even under the load.

(Edited to correct the temp rise in C conversion)