Need your input on this motor generator

[MileHigh]

Yes indeed but note how unusual this setup is.  You actually see it quite often on the forums.  You have the big coil discharging a train of current pulses into a circuit inside the LED light bulb that is expecting to see a 120 VAC sine wave.  The current pulses whack their way through the circuit inside the LED light bulb anyways and manage to light the LEDs.  The circuit is already trying to convert high-voltage AC into some kind of DC or pulsed DC for the LEDs and it manages to "partially digest" the current pulses which need very little prodding to produce high voltage.  It's a "super whackadoo kludge" that actually does a decent job of lighting up LED light bulbs all things considered.

How the circuit copes with the current pulses is unknown, but it could be investigated if someone wanted to go there.  One has to assume that the circuit looks like a very nonlinear load hence the need for the DSO.

A reasonable assumption is that the circuit inside the LED bulb is initially not "friendly" towards the current pulses and looks like a high impedance.  So the voltage shoots up and gets up to levels that are quasi "120 VAC look-alike" and the circuit decides it can digest these high voltage current pulses.  Once the voltage hits some unknown threshhold level something kicks in and you get a lit LED light bulb.  It kind of a "bananas in the transmission" kind of deal.

[TinselKoala]

Sure, I see all that, but I still maintain that you can get good reasonable average power measurements using two DMMs if they are used properly. And you could probably even use a photographer's light meter calibrated in f-stops to get a reasonable, if coarse, measure of luminance or irradiance from the lightbulbs, if you were rigorous about it and did some calibrations against standards.  It's a myth that you need DSOs with fancy math to do this kind of work! They are nice, like having a riding lawnmower to use instead of a push-reel type. But they don't necessarily do a different or better job of mowing the lawn. What do you think people did before there were digital scopes, just wave their hands around? Was the Watt not invented before 1975? I'll bet the engineers that designed the first digital scopes didn't have digital scopes to use, but they did just fine. The Apollo program for example ... how many digital oscilloscopes were around in those days?

[MileHigh]

TK:

If you light up an LED with a periodic voltage waveform like a sawtooth, then you have to have a DSO to measure the power dissipation in the LED (or do it thermally).  I assume that you agree with me.

How did they do it before DSOs?  I am not sure but driving home tonight a way to do it occurred to me.  It would be a variation of your trick with weighing a piece of paper on an analytical balance.  Perhaps they photographed the waveforms and then created a set of cams for the voltage and the current.  Then they ran that through some kind of cam-based multiply-accumulate mechanical analog computer.

There is a great series of clips on YouTube showing the cam-based analog computer that is used to aim the main guns on a battleship or destroyer.  It dates from the early 1950s if I recall.

MileHigh

[TinselKoala]

@MH : I'm not so sure I do agree with you. Like I said, it all depends on how you use your meters. If you do it right, I think you will get the right answer. I've done what I show in the video below about a dozen times now with various waveforms and frequencies and I consistently get within 5 percent of the scope's measurement, and at higher currents and longer duty cycles, even less error than that. Just did one with an Ainslie waveform that gave 20 mA error out of 1.810 Amps.

http://www.youtube.com/watch?v=1ST0xgf3xsw

Now you can't always multiply "average current" by "average voltage" and get the right answer in power. But if you can legitimately "assume" that the voltage is constant, while the current varies, then you can get the right answer by using Vconst x Iavg found in this way. I think. Maybe.

[MileHigh]

TK:

I forgot to mention the obvious about using tube circuits and a capacitor to make your multiplier-accumulator.  Ironically enough the transfer function in a diode can be used as the basis for making a multiplier.  Converting multiplication to adding logarithmic signals and all that stuff.

For your clip, I think I am on Poynt's side about a multimeter being more accurate than a DSO for an average current measurement.  The sampling of that big band of noise (actually oscillation waveform) may be throwing off the DSO average current measurement by a few percent but not the multimeter.  I think it's because the main waveform also causes the trigger so the sampling and the noise (oscillations) are not random relative to each other so there is the possibility of aliasing errors in the sampling.  Numerical analysis is heavy stuff.  lol  (Bits of sampling resolution also come into play.  I never liked 8 bits.)

The problem with a diode or LED of course is that the voltage is "clamped" to whatever the current is and it's not a straight line.

Now that would be a good op-amp project, making a triggerable multiply-accumulator with an adjustable event capture window.  That would give a vanilla analog oscilloscope the ability to measure the power being dissipated in an LED or whatever else you wanted.  I am sure all of the pieces to the puzzle are there in that nostalgic National Semiconductor op-amp application notes book.

MileHigh