Rosemary Ainslie circuit demonstration on Saturday March 12th 2011

[FatBird]

1.  The V out depends on the supply voltage.  Example, if you used a 12 V supply, the Output V would be about + 11.5 Volts.

2.  The Output High is a Positive V.  The Low is about + .2 V, which is Ground for all practical purposes.

3.  Depends on the Mosfet.  An approximate average full turn on V for a 5 Amp, 200 V Mosfet is about 1 V.

4.  No, EXCEPT if her circuit had an inductance that would RING (oscillate).  In that case the ringing can indeed drop negative.
    If I remember right, her load resistor is wire wound.  If so, then it could ring negative.

Neptune asks = If this is the case , one could at that point disconnect the 555 from the gate and replace it with a small battery
                     giving a negative potential , and the oscillation might continue indefinitely .

5.  Probably so, but the circuit switching arrangement would be tricky & involved.
.

[cHeeseburger]

Hello All...Newbie here.

I've been following Rosemary's work on her blog page lately and I've noticed a couple of things that seem strange to me.

First, I can't find anywhere that she says what her measurements of input power or voltage or current actually are.  Never once is a number actually given for any of these values, as far as I can tell.  All the scope traces I've seen her present show an input (battery) voltage with over a hundred volts of AC at over 1MHz on it.  That can't be right.  There must be some large inductances or long wires being used in the battery circuit.  No decent battery would have such a high impedance on its own as to allow such a large AC voltage swing.  She says her batteries are brand new and very high quality types.

Second, she recently showed her latest circuit and parts values.  I noticed that the shunt being used to measure the battery current flow is made up of several quite long ceramic wirewound resistors.  She states the combined inductance as being 130nH or something like that and 0.25 Ohms combined parallel resistance.  With an oscillation frequency of over 1MHz, the inductance is the predominating part of the shunt impedance (by a huge margin) and the shunt impedance will be adding a large phase shift and showing much larger voltages across it than a pure 0.25 Ohm resistor would.

So, when the wildly oscillating AC "battery voltage" is multiplied (sample by sample) within her oscilloscope math by the phase-skewed voltage across her inductive shunt, the results will be totally unrelated to the actual DC-equivalent average power. 

Measurements made on this deeply-flawed basis could quite easily show a negative (reflected) power being returned to the battery when such was not actually the case at all.  Or they could easily show zero (or close to zero) power being drawn when, in reality, significant power was being drawn out of the battery.

I would suggest that a simple low-pass filter be applied on both the shunt voltage measurement and the battery voltage measurement in order to find the actual DC equivalent input power.  This will eliminate the false readings associated with the phase shifts and inductive parasitics in the circuitry and reveal quickly the actual DC net power flow either out of or into the battery.

This has been suggested to Rosemary many times by many folks on several forums but, so far, she refuses to do it and has ignored all such advice.  Adding fifty cents worth of R and C to form a simple first-order low-pass filter and then just measuring the results with a DMM is all that is needed. 

It's much easier than trying to change the batteries to smaller ones or run using a capacitor or DC power supply.  It could be done in five minutes at almost no cost and would give results that are far more ACCURATE AND TRUSTWORTHY than doing math on 8-bit scope traces which are wildly swinging around with huge imposed AC voltages far beyond what would appear across any decent battery or a pure resistive shunt.

This technique has been used for decades and is well-known to any engineer who has tried to make accurate DC-equivalent power measurements on circuits that have pulsed or high frequency AC current draw.  Multiplying phase-skewed values derived across inductive shunts and batteries hooked up with long wires and no bypass capacitors has no chance of ever yielding accurate DC-equivalent power numbers.

Doesn't anyone else here know this?  I have not seen it pointed out or heard similar suggestions on this forum.

cHeeseburger - to go!  (Hold the lettuce)

P.S.  Hooking two or more MOSFETS directly in parallel is well known to cause parasitic oscillations that are, in fact, difficult to get rid of when they are unwanted.  Rosemary is using a function generator and has liberally applied DC offset voltages to the pulse output and tweaked that offset to enhance the oscillations, so using a 555 timer circuit will probably not work the same way at all.  Anyone desiring to replicate should forget all about the earlier Rosemary Ainslie COP 17 schematics and use the latest circuit shown in her blog report.  Don't forget to use at least ten feet of wire to hook up the batteries!  And NEVER add any bypass caps ANYWHERE!  Oh...and use a long twisted pair of small-guage wires to run from the signal generator to the MOSFET gates.  That extra inductance and impedance mismatch can get a solid oscillation going even with a single MOSFET.

[markdansie]

@cheeseburger
Your modifications for testing as you suggest the logical way to go. I have spoken to a few other engineers who did not have ideas too distant from yours. I guess myself and others were dumbing it down to try and point out that nothing can be claimed to support her hypothesis if many of the variables or red flags are not eliminated especially the power supply.
@rosemary
In many ways you are just subjecting yourself to peer review here as you would have to in mainstream. We are in many ways much kinder. Please do not bring in emotional comments as there is no room for sentiment in scientific methodology and process.
However option B for you is adopting the "Ignorance is Bliss" stance.
No one is attacking you here...but many more qualified than me are speaking volumes here...please listen.
Kind Regards
Mark

[cHeeseburger]

Hi, Mark!

I believe the evidence points to the idea that Rosemary has chosen option B a long time ago.  To expect that any thinking person would accept her claims without even a statement from her regarding her measured input power seems just nuts to me.

After all, what else is there to measure?  She long ago (and with good reasoning supported by almost everyone commenting) decided that measuring the output power via voltage and current multiplication at the load would not work easily because of the load inductance and its inherent phase shift and non-unity power factor.  So the 'scope method was shunned in regard to measuring output power with not so much as a peep of objection from her or anyone else.

Instead, the thermal integration method was chosen...even in her original 2002 publication.  Comparing the thermal equalibrium temperature of the load driven by her circuit against the DC power from a bench supply that was required to reach the same thermal equalibrium temperature.  Perfectly acceptable and reasonable way to proceed.

What she and everyone else seems to have blindly overlooked is that the measurement of input power is plagued with the exact same complications of inductance-based phase shifts and power factor complexities that were deemed too difficult to overcome on the output side.  There is no difference!

Yet, thosands of forum posts and blogs and hundreds of thousands of words and arduous special arrangements to borrow fancy oscilloscopes and try to lern to use them and bitter heated arguments have since ensued all regarding the correct measurement of the inpuit power.  To date, no numbers have emerged!

All I am suggesting is that simple techniques that in no way alter the circuit operation, i.e. using a simple RC low-pass filter on the battery voltage and current shunt voltage will give the exact same advantage that using the thermal integratiion method of obtaining equivalent DC power provides in the output measurement.  Even better, there is no need for a "control" or comparison test at the input side as must be done on the output side.

The only reason to use an oscilloscope in this whole exercise is to "tune and tweak" the circuit for whatever characteristics Rosemary thinks are best.  Once that is done, the scope should be turned off and forgotten.

Measure the average DC-equivalent input voltage and current using a good DMM and the RC filter.  Multiply.  End of story!

cHeeseburger

[fritznien]

a dirty great pi filter would be even better but then it would show clear results.