Reliable Measurements and Simulations: Power In and Out and Efficiency n

[PhysicsProf]

  We have discussed the critically-important matter of painstaking measurements before; it is worth it's own thread here as reliable measurements are a principal goal with me and many others.  And I have some particular questions based on recent observations.
Let's start with quotes from a related thread: http://www.overunityresearch.com/index.php?topic=764.msg12282;topicseen#msg  --

A few days ago, I was able to spend some time with the Tektronix 3032 and several variations on the basic JT circuit.  Measuring Pout, Pin and n as discussed above, I might summarize n values to show the spread in values obtained and recorded:

n = .81, .52,.59, .63, .52, .35, .45, .80, .13, .77, .61, .79, .88, .62, .94,  .99, .92, 1.0, .69, 1.06

So yes, there were some "interesting values", though I would say the last value is still consistent with unity (given the +-3% errors as previously stated, 2 sigma).
 
In the process of taking measurements, I realized the importance of reviewing HOW the measurements are taken, and the importance of simulations AS A WAY OF CHECKING THE MEASUREMENT METHOD, for example where the probes are attached.  Also, if there is an input of power in the circuit somewhere (and somehow) -- just how will this show up? as an increase in current at some point, or an increase in voltage along a path?  I think the latter, from observations, but all this has to be checked -- and then re-checked.

I believe simulations will help in this effort of re-checking.

This issue is so important that I propose to come back to this tomorrow on another thread I'll start -- "Measuring Pin, Pout, n -- Simulations etc."

To which .99 kindly replied:

Sounds like a very good plan Professor.   I'll help out where I can.

One thing I will suggest, is that you try the RC filter/two-DVM method for obtaining a Pin figure. I will draw up an amendment to the schematic to show exactly how to do it. This will not only provide you with another method to obtain Pin, but a way of checking your scope Pin measurements. If you do not have two DMM's (digital multi-meters), you can buy them quite cheaply. As a bonus to this method, not only is it inexpensive, accurate, and accessible (not requiring a scope channel), it allows more freedom as to where and how the Pout can be measured, because there is no longer a common ground between the two measurements, i.e. the scope grounds always limit this flexibility. The Pin measurement is essentially "floating" wrt the scope ground.

.99

I do have two DMM's and look forward to your alternative method, .99 -- very much in fact.

Meanwhile, I would like to understand results I obtained with the Tek 3032 last Friday, as attached along with the schematic for the straightforward transistor-resonator circuit (TRC).  The TRC circuit discussed here is shown on the right; a standard JT circuit is shown on the left for comparison.  There is one change -- a 1N4148 diode has been added at point 5 before (in series with) the LED.  Here are details of the circuit (on the right):

1.46V  AA battery (measured by DMM)
CSR1 = 1.1 ohms
Rb = 979 ohms
MPS2222A transistor
Capacitor (C2) = 151 pF  (note:  corrected to pF; measured)
Diode at point 5 = 1N4148
LED = red diode

Ferrite toroid (2 cm OD, 0.9 cm ID, 1.1 cm tall) was would bifilar with eight windings of 23 gauge enameled copper wire.  Here are measured values for the two windings, using an MCP meter BR2822:

L1 (between Points 5 and 7); L2 between point 5 and capacitor C2 (feedback loop).

L1 at 120 Hz:  R=0.0336 ohms, Z = 0.0614 ohms,  L = 68.0 uH, C = 0.

L1 at 1 KHz:  R=0.0359 ohms, Z = 0.4257 ohms,  L = 67.4 uH, C = 376 uF.

L1 at 10 KHz:  R=0.0631 ohms, Z = 4.27 ohms,  L = 66.9 uH, C = 3.8 uF.


L2 at 120 Hz:  R=0.0322 ohms, Z = 0.0699 ohms,  L = 82.0 uH, C = 0.

L2 at 1 KHz:  R=0.0351 ohms, Z = 0.5124 ohms,  L = 81.3 uH, C = 311 uF.

L2 at 10 KHz:  R=0.0666 ohms, Z = 5.08 ohms,  L = 80.8 uH, C = 3.14 uF.

I invite (and would much appreciate) .99 and/or Humbugger and/or others to do a SIMULATION of this circuit, to see whether one can:

1.  Replicate the observed Power In (Pin) and Power Out (Pout) waveforms, even approximately.  
I used:
Mean Pin = Mean (V1*V2), given that I2 = V2 to a close approximation since CSR1 = 1.1 ohms.  (which I measured afterwards; for future runs, I will look for a CSR1 closer to 1.0 ohms)
Mean Pout = Mean (V3*V2), where V3 is measured between points 5 and 3 on the labeled schematic attached.

2.  Pin appears basically as "negative" because of the way V2 is measured -- I will however set the +y axis downward and refer to this as "positive power", hoping this will not be too confusing.  Pout is more interesting in that it shows both "positive" and "negative" power excursions, repeatedly.

3.  Spikes in the data are also seen and should be replicated in the simulation.

4.  Note that I acquired Mean power values over a number of cycles, to improve the accuracy, so details in the waveform are not easily seen.  But important features are seen and should be reproduced by a successful Simulation.

5.  MEAN Pout comes out to be a very small value, 367 micro-V*V = 0.367 mVV, owing to the fact that the power curve Pout goes both positive and negative.  Clearly, the voltage and current in this part of the circuit are frequently of OPPOSITE PHASE (OP, previously called OOP).
We can calculate efficiency n:

n = Pin/Pout = 0.367/21.57 = 0.017 = 1.7%.  

Viewing the Tek taking Mean values for Pout, I see that Pout is fluctuating around ZERO, and when I hit the stop button, the value came out to 0.367 mVV (very small compared to input power) -- but it could have been zero or even of opposite sign from the input power if I had hit the stop button at a different instant.
Trouble is, the LED was glowing very brightly and I flat don't believe this result of n = 1.7%.


I would like to see what a decent simulation says about the efficiency n, and if the method used is correct (which we discussed rather thoroughly before, Mean Power instead of RMS Power, etc.)   Clearly the RMS output power is much greater than the Mean output power, which latter is close to zero, and I am left wondering about the method we have developed.

There seems to be something wrong, and I hope the simulation and discussion will elucidate what is going on and how to get an accurate assessment of n for this straightforward circuit.

[EMdevices]

4.  Note that I acquired Mean power values over a number of cycles, to improve the accuracy, so details in the waveform are not easily seen.  But important features are seen and should be reproduced by a successful Simulation.

Professor, I believe the above approach is counterintuitive.  On the one hand it's self evident that larger data samples improve the mean,  on the other, the scope could be introducing larger error, at a faster rate, when the time base is made wider, and the resolution reduced.   We are dealing here with sharp pulses and quantization error plays a big part.

Here's what I suggest.

Zoom in the time base to see only one period of the pulse, maybe two.   Then freeze the scope shot and write down the average number.  Then unfreeze the scope and repeat, manually writing down the values and averaging.   Don't let the scope do all the "fun" work for you!  

EM

[PhysicsProf]

Professor, I believe the above approach is counterintuitive.  On the one hand it's self evident that larger data samples improve the mean,  on the other, the scope could be introducing larger error, at a faster rate, when the time base is made wider, and the resolution reduced.   We are dealing here with sharp pulses and quantization error plays a big part.

Here's what I suggest.

Zoom in the time base to see only one period of the pulse, maybe two.   Then freeze the scope shot and write down the average number.  Then unfreeze the scope and repeat, manually writing down the values and averaging.   Don't let the scope do all the "fun" work for you!  

EM

 Sure -- IF you could actually freeze the time base at EXACTLY one or two cycles...    IF you can actually do this, pls tell me how!

Also, this will NOT solve the "problem" noted above, since the frequently-out-of-phase pattern of Pout will still "cancel out" to about zero for the MEAN value of output power...  with the LED glowing very brightly all the while!

PS -- a successful simulation will also reproduce the frequency of the waveform oscillation, about 1.8 MHz -- that's MEGAhertz. 
I'm challenging the ability of a simulation to do this AND give the correct Pin and (especially) Pout waveforms.
If it does work, I want to be able to run the simulation myself (for sure) and learn with it.

I look at this as a type of experiment now to test the abilities of simulation packages...  a challenge, if you will.

[EMdevices]

Profesor,  are you having trouble zooming in on only one cycle of the waveform because of scope limitations?  or is it a trigger problem, where the waveform jumps around?

Play with the trigger level, I'm sure those spikes will trigger it just fine.  If it's a scope limitations,  meaning not enough time base resolution,  then oh well, you're out of luck.  

EM

P.S.   When I played with these circuits in the past I tried to filter and then I measured Voltage and Amperage.  I never trust digital scopes with their quantization errors especially when it comes to spikey waveforms.

Most important thing to keep in mind with spikey waveforms is that they have a very high frequency content, (proportional to the voltage rate of change, or dV/dt) and most oscilloscopes, will have a limited bandwidth, so only part of the signal energy gets through.

[PhysicsProf]

Profesor,  are you having trouble zooming in on only one cycle of the waveform because of scope limitations?  or is it a trigger problem, where the waveform jumps around?

Play with the trigger level, I'm sure those spikes will trigger it just fine.  If it's a scope limitations,  meaning not enough time base resolution,  then oh well, you're out of luck.  

EM

P.S.   When I played with these circuits in the past I tried to filter and then I measured Voltage and Amperage.  I never trust digital scopes with their quantization errors especially when it comes to spikey waveforms.

Most important thing to keep in mind with spikey waveforms is that they have a very high frequency content, (proportional to the voltage rate of change, or dV/dt) and most oscilloscopes, will have a limited bandwidth, so only part of the signal energy gets through.

@EMD:
Look, the Tek 3032 is a 300 MHz device and I'm not having any trouble getting down to one cycle, approximately.  I've done this, just did not capture the waveform to show (could do this next trip to the University) because I was looking mainly at evaluating the MEAN power, which suggests that one use a goodly number of cycles to avoid "boundary" issues, say for 1.4 cycles (where one cannot get exactly one cycle so will have trouble getting accuracy).

Why don't you trust this scope for "spikey" waveforms, for calculating mean Pin and Pout? 
Again, are you seeing the problem that Mean Pout ~ zero and yet the LED is glowing brightly?

[EMdevices]

very good, so you can zoom in, I was confused my appologies.  

I always advocate close up analysis of a waveform then massive averaging over many waveforms and use of averaging functions on a scope.

EM


P.S.

Why don't you trust this scope for "spikey" waveforms, for calculating mean Pin and Pout?  

I don't trust digital scopes in general when it comes to spikey waveforms like I explained earlier, and specificaly in the pictures you posted, because I don't see enough SAMPLE POINTS in that sharp transition, and lots of error is introduced by this.

Again, are you seeing the problem that Mean Pout ~ zero and yet the LED is glowing brightly?

yes, noise is swamping out the current signal through the 1 ohm resistor, what a poor setup to measure Pin/Pout. !

[Allcanadian]

@EMdevices

P.S.   When I played with these circuits in the past I tried to filter and then I measured Voltage and Amperage.  I never trust digital scopes with their quantization errors especially when it comes to spikey waveforms.

I would agree, I have been down this road before and I learned a valuable lesson. I once spent 4 hours setting up a pretty simple CFD model in Solidworks CAD then another 30 minutes setting up the solidworks flow simulator then the CFD solver ran 3 hours crunching numbers. In the end it did not tell me anymore than I had learned with a simple fan and a piece of yarn on the end of a stick, lol. All these gadgets and "fluff" as I call it are fine but they should never replace common sense.
In most cases simply filtering the input and output is a far better solution than $20,000 oscilloscopes and complex simulators.
Regards
AC

[PhysicsProf]

very good, so you can zoom in, I was confused my appologies. [snip]

yes, noise is swamping out the current signal through the 1 ohm resistor, what a poor setup to measure Pin/Pout. !

 At 300 MHz sampling, I rather doubt your notion that the current signal is being "swamped out", but a demonstration by you would be instructive...  Please use a 300 MHz DSO (or faster).
  I would truly like to know -- what setup you would propose?

The current set-up developed/evolved over some weeks as discussed primarily in these two threads which you may wish to read to catch up:

http://www.overunityresearch.com/index.php?topic=717.125  (the LTJT circuit)

http://www.overunityresearch.com/index.php?topic=710.0   (average or RMS?)

In particular this post was helpful (its accompanying schematic was provided here, see my first post (left schematic) )
--

However, it is possible to calculate that output without inserting a third measuring transistor between the ground and the emitter. In fact, we should remove CSR2 and have the LED directly connect to the ground, then the new Pototal can be found by integrate [V3*V2] over time, and this time the Pototal includes the output energy spent on both the LED and the transistor. Moreover, this time the current is estimated by V2 on one single CSR1, so value errors in CSR1 won't be a factor anymore, the calculated n would be EXACT!!!

Thus, I would say yes for the sake of simplicity! Here is the proposed method:

1. take CSR2 out and connect the LED directly to the ground (or just leave as is and don't measure V4, measure V2 instead if you were using two scopes).

2. measure V1, V2, V3 as before.

3. calculate Pitotal by MEAN [V1*V2], which gives the time average input power.

4. calculate Pototal by MEAN [V3*V2] over time, similar to finding the Pitotal.

5. then n = Pototal/Pitotal, which is EXACT as the error in the value of CSR1 is balanced out in the way n is calculated.  

This seems to be as simple as it can be. Also, it only leaves tiny energy leaks out of consideration, namely the leaks in the 1k resistor and the coils.

cheers,

lanenal

Read the rest of the thread for further applications.
So you will see the method for measuring Pin/Pout was presented earlier after considerable thought.  

Nevertheless, the goal of this thread is to determine a valid and reliable way to evaluate Pin, Pout and n -- and I have raised specific questions -- and to determine whether or not a simulation can adequately describe this simple circuit -- and so, I am open for USEFUL input and suggestions.

Rigor without rancor.     Thank you.

[poynt99]

Professor,

One factor that is not well known in the ranks of DSO users, is something called "record length". All digital storage scopes have a limited record length, and many offer this as selectable.

Your model, the TDS3032 has a record length of 10k. That means the scope can only display 10,000 samples (memory locations), regardless of how high your sample rate is (2.5Gs for that model). The scope uses those displayed samples in its calculations. So what EM is saying, is that as you widen your time base to see many cycles, there are less samples "applied" to each cycle.

In my experience however, as long as you display many samples, the "error" (i.e. missed samples) gets averaged out, and the measurement is accurate. If you however can produce one, or just over one cycle on the display, you can try using the "cycle mean" measurement rather than the mean measurement we have been using.

.99

[PhysicsProf]

Professor,

One factor that is not well known in the ranks of DSO users, is something called "record length". All digital storage scopes have a limited record length, and many offer this as selectable.

Your model, the TDS3032 has a record length of 10k. That means the scope can only display 10,000 samples (memory locations), regardless of how high your sample rate is (2.5Gs for that model). The scope uses those displayed samples in its calculations. So what EM is saying, is that as you widen your time base to see many cycles, there are less samples "applied" to each cycle.

In my experience however, as long as you display many samples, the "error" (i.e. missed samples) gets averaged out, and the measurement is accurate. If you however can produce one, or just over one cycle on the display, you can try using the "cycle mean" measurement rather than the mean measurement we have been using.

.99

I agree.  I will be glad to record "cycle mean" on my next trip but I have already zoomed in on the signals, and the cycle mean was also observed to be close to zero, for Power Out as defined above.  
I hope someone will address the true concerns I have raised here, in particular, that the Mean Power Out is giving approximately ZERO (whether over one cycle or a number of cycles) and this is clearly in disagreement with the reality that the LED is brightly lit.
And the request for a simulation of the circuit.

[poynt99]

Professor,

Here is how to make the Pi measurement using the RC/DMM method. Pout is made with the scope as before.

.99

[EMdevices]

I scrutinized the circuit and picture some more and I can't believe you are using CSR1 for both input and output power measurement.

Think about it, that waveform indicates current for both input (when transistor is on) and output (when transistor is off)

To answer the RMS question, a random waveform has mean of zero, but a finite RMS value.

Tomorrow I'll say more from the office computer.

EM

[PhysicsProf]

@.99 -- thanks, I'll study it.

@Emd -- You may wish to finally read the thread I referred to (gave you the URL even) where this circuit evolved... also, pls tell us where you would put the current-measuring CSR's (if you insist on plural).  You may have a better idea (would be glad to hear it!), but I suspect it will look a lot like the one on the LEFT in my first post on this thread....

[poynt99]

I scrutinized the circuit and picture some more and I can't believe you are using CSR1 for both input and output power measurement.

Think about it, that waveform indicates current for both input (when transistor is on) and output (when transistor is off)

EM

EM,

You are probably going in the direction of having a CSR for each component you want to know the power in. Yes, if you look back through those threads, I have already proposed that. However, I did do a simulation using the method where CSR1 is the only CSR, and although the power output calc was not exactly the same, it was low by about 3% or so, compared to the separate CSR method. See attached.

It does seem to work, but it won't hurt to revisit this issue.

.99

[Itsu]

PhysicsProf,

I don't think that simulating this "groundloop special" circuit is that easy.
The problem lies in the toroid, as my sim db does not contain a suitable component for it.

I made an attempt to simulate, by using an "iron core xformer", as your toroid, but it gives me no possibility to enter your measured inductance etc.
Separate inductors (for which the inductance can be entered) does not perform the coupling needed.

The sim won't run in this configuration, but that could be due to my inexperience with sim's

Learning a lot lately,  thanks to all, 

regards Itsu

[PhysicsProf]

PhysicsProf,

I don't think that simulating this "groundloop special" circuit is that easy.
The problem lies in the toroid, as my sim db does not contain a suitable component for it.

I made an attempt to simulate, by using an "iron core xformer", as your toroid, but it gives me no possibility to enter your measured inductance etc.
Separate inductors (for which the inductance can be entered) does not perform the coupling needed.

The sim won't run in this configuration, but that could be due to my inexperience with sim's

Learning a lot lately,  thanks to all, 

regards Itsu

I'm learning a lot also... it's great.
Interesting that the sim won't run in that configuration, due to coupled inductors.  Thanks for taking a shot at it.  I'm hoping .99 or Humbugger or someone will take a stab at it also.
I note that your probe connections differ from mine (see my first post), but clearly that won't solve the problem if the sim won't allow you to enter the parameters for the coupled inductors -- which are what make this circuit interesting!  IMO.
Thanks, ITSU.  I'm following your work on youtube.

[feynman]

Okay, wow things are getting interesting here.

@PhysicsProf

First of all, thanks for posting your data.  This will help me as I try to confirm or disconfirm the Wikipedia JT oscillation formula.

PhysicsProf I know EMDevices from a few years back on overunity.com, and he is excellent in both his theoretical and practical abilities.  I hope he has the time to help us out with this.

@all

Also, this will NOT solve the "problem" noted above, since the frequently-out-of-phase pattern of Pout will still "cancel out" to about zero for the MEAN value of output power...  with the LED glowing very brightly all the while!
-PhysicsProf

Hahahaha... so either
1) The measurement methodology is wrong
2) There's something going on, energetically speaking, that we don't realize or ascertain

My money is on #1, but you never know, there's always that outside chance we could be in a #2 situation.  Perhaps we should talk to lanenal since he helped develop the power measurement protocol.  There are almost always multiple ways to do a correct scientific measurement... so perhaps there is a second power measurement protocol which can confirm/disconfirm these results... I know poynt has suggested an additional method, so this would be worthwhile.

The one part I worry though is that using a low-pass filter might destroy some of the high frequency harmonics (spikes/transients) which contribute to the AUC, but perhaps I am wrong about the significance of these waveform spikes.

[humbugger]

@feynman

Destroy some of the spikes, huh?  Using a low-pass filter made of physical parts will average the waveform, spikes and all in a very accurate way.  Why no objection to using scope math to get an average value?  Does that not "destroy" the spikes?

Destroy and Average are different things, Feynman.  The professor is looking for the net DC equivalent average power values; that involves averaging when an irregular waveform is the subject.  You can average using an 8-bit scope with sampling errors or you can average using a resistor and capacitor with zero sampling errors.

Either method when done right will take full accounting of all "spikes".  The scope method takes very extreme care to make sure the sampling is done right while the RC method does not.  I would think you would know that much.

Humbugger

P.S.  Did you take that half hour of yours to whip up the Ainslie hardware and put an end to the ridiculous speculations yet?   

And did you ever get any scope-shots from Rosie that were useful in studying the actual waveforms of the 1.5MHz oscillations?  Nope.  She never took any by her own accounting.  Just massively undersampled slow sweep traces showing the burst envelopes...completely useless for doing scope math based on sampling.

[EMdevices]

P99,

I like that circuit with a resistor in each leg,  however this can be done with only one resistor as well,  just not how the professor is attempting to.

That's why I keep repeating, we need to zoom in and analyze a single cycle and do the V*I integration manualy, and then average results from many cycle runs.    We just can't use a scope's averaging function here because:

1)   The voltage signal at the battery playes a role in both the INPUT and OUTPUT power.
2)   The single resistor also conducts both the INPUT and the OUTPUT current.

so when the MATH function is used to multiply the resistor voltage times the LED voltage to calculate the Output power, in effect we are mixing "apples and oranges", in other words a spike in the resistor voltage due to the current flowing through the emitter of the transistor is not current flowing through the LED and will produce erroneous results.   I hope this is clear to everyone who understands how the currents flow in the oscillator circuit.


So this is what I would do,   I would replace the LED with a resistor, that way we know the V and I relationship, and we don't have to monitor the battery voltage for the output power.    Also I would use P99's   RC filter schematic and the battery for Input power.     If this approach is adopted, then we can use a scope and try to average over many cycles, etc..,     but as educated folks,  I still advise to zoom in on the waveform and perform manual integration of the V*I waveform.

EM


P.S.  I know it's a pain, but this is what I mean by integration of the very sharp spike waveform, which I have expanded in the x-axis here as an example.  This is just a typical waveform and the almost linear left side can be integrated with a larger time interval, perhaps 7 * delta t.   

I've also been thinking of how to do this best with the least distrubance to the circuit as is, meaning keep the LED, since that is what Lawrence first proposed,  so I would suggest just another CSR connected to the LED leg.  I know this was suggested before, I'm just making suggestions here.

[muDped]

P99,

I like that circuit with a resistor in each leg,  however this can be done with only one resistor as well,  just not how the professor is attempting to.

That's why I keep repeating, we need to zoom in and analyze a single cycle and do the V*I integration manualy, and then average results from many cycle runs.    We just can't use a scope's averaging function here because:

1)   The voltage signal at the battery playes a role in both the INPUT and OUTPUT power.
2)   The single resistor also conducts both the INPUT and the OUTPUT current.

so when the MATH function is used to multiply the resistor voltage times the LED voltage to calculate the Output power, in effect we are mixing "apples and oranges", in other words a spike in the resistor voltage due to the current flowing through the emitter of the transistor is not current flowing through the LED and will produce erroneous results.   I hope this is clear to everyone who understands how the currents flow in the oscillator circuit.


So this is what I would do,   I would replace the LED with a resistor, that way we know the V and I relationship, and we don't have to monitor the battery voltage for the output power.    Also I would use P99's   RC filter schematic and the battery for Input power.     If this approach is adopted, then we can use a scope and try to average over many cycles, etc..,     but as educated folks,  I still advise to zoom in on the waveform and perform manual integration of the V*I waveform.

EM


P.S.  I know it's a pain, but this is what I mean by integration of the very sharp spike waveform, which I have expanded in the x-axis here as an example.  This is just a typical waveform and the almost linear left side can be integrated with a larger time interval, perhaps 7 * delta t.   

I've also been thinking of how to do this best with the least disturbance to the circuit as is, meaning keep the LED, since that is what Lawrence first proposed,  so I would suggest just another CSR connected to the LED leg.  I know this was suggested before, I'm just making suggestions here.

Agreed.

There seems to be some difficulty in confining
the mathematical analysis to the specific timings
of the Input and the Output portions of one complete
cycle.  Keeping the waveforms separated for
calculation.

When input and output are time division "multiplexed"
perhaps manual calculations are a necessity.

Input from the source is available for both half cycles.

Output from the coil is added to input from the source
during the "Output" half cycle to power the "load."

Unless carefully separated and accounted for the
output power can be elusive.

[EMdevices]

Try this schematic Professor.


PS  If you chose to use the scope averaging functions instead of integrating by hand, use the MEAN, not the RMS, and make sure the scope is sampling at it's highest rate for best resolution and accuracy.

[PhysicsProf]

@EMDevices:

Try this schematic Professor.

Umm...  The circuit you propose is not the circuit I am testing and for which I have given data (and schematic) in post #1 of this thread.   So I'm attaching again the schematic of the circuit under consideration in this thread, and I respectfully ask you to show where to connect the scope leads to get output power (all of it) -- and then how to calculate n based on how you've connected the scope leads.

I do not say this antagonistically, just asking that you apply your method to the circuit under consideration in this thread.  Thanks.


PS -- .99's scheme of using a Cap in series with a 10K resistor makes sense -- basically using the cap to do an integration, in order to evaluate the current.  His schematic also reflects another (simple JT) circuit, but the jist  of his schematic is of course the current measurement.

There can only be ONE current flow finally in the circuit -- One electron goes in, one comes out -- correct me if I'm wrong. 
However, what we're trying to determine is whether there is a boost in voltage along the way, not just at the battery (a KNOWN) but perhaps also in the inductors & transistor circuit, somehow.  That is the question.

[EMdevices]

ok, then try this circuit.     (I still like the previous one though because it has less components.)

Note:   I flipped the Diode.  This is how it should be to conduct the kickback from the coil.

[PhysicsProf]

EMD and Dumped:  I have done manual integrations of power waveforms numerous times, over a single cycle-- my logbooks are replete with these.  And I have presented results of hand-integrations from other circuits -- particularly from the L. Tseung circuit, which is quite different (!) from the one under consideration here.
Integration is easier when the waveform-section is either triangular or rectangular, of course.

Attached is the output power waveform for this circuit, which I have called "boost resonator" or "transistor resonator" to distinguish the circuit from the JT or JT-LT circuits -- call this one BR.


The question is, how do we evaluate the output power when the waveform has a significant AC component?  as is the case I presented in post #1 of this thread.

Here, let's zoom in on one cycle (out of a few) and I'll ask you guys (especially EMD and Dumped) to do the manual integration, I'll do the same and present my result later -- see attached.  (Got to get back and finish my taxes!      )
The zero line is clearly delineated (by the "M arrow") and the scale is 200 mVV per division.   The horizontal division is 1.0 microsecond.  
NO NEED TO LECTURE ME ON DOING THE ZOOM WITH THE SCOPE RUNNING or expanding the scales -- I'LL DO THAT next time I'm up at the University (a LONG drive). I know this is crude, given the pixelation, that's not the point -- --

The question is, again:

How do we evaluate the output power when the power waveform has a significant AC component and the current and voltage are not in phase?  
(BTW, the input power does NOT have a large AC component, relative to the DC component...)

Pls let me know what you get, and I'll give my result also, later.

PS -- I see that you have another schematic, EMD -- thanks -- I'll get back to you later on this, after we finish the simple exercise delineated above which should help us in communicating (I hope!).
I tried your idea of reversing the LED -- interesting -- it glows brightly EITHER direction!  That was a good idea...  I can hardly wait to get back up to the Tek 3032 at the University to look at the waveforms, zoomed in for detail, for the two cases...  

[PhysicsProf]

  me:

PS -- I see that you have another schematic, EMD -- thanks -- I'll get back to you later on this, after we finish the simple exercise delineated above which should help us in communicating (I hope!).
I tried your idea of reversing the LED -- interesting -- it glows brightly EITHER direction!  That was a good idea...  I can hardly wait to get back up to the Tek 3032 at the University to look at the waveforms, zoomed in for detail, for the two cases...  

So I added another LED diode parallel to the first, but in the opposite polarity -- and they BOTH light up!  the LED in the direction you show appears to be somewhat brighter (both red LED's).
  Good idea, EMD...