Rosemary Ainslie Circuit Demonstration, June 1 2013

[picowatt]

She has apparently resorted to either outright lies, or her ability to read and comprehend is sorely lacking: 

Guys - I'm not sure that this is clear.  But let me see if I can explain this.  What picowat is ACKNOWLEDGING is that the current from the battery supply CAN'T move through the source leg of Q2.  But he's proposing it doesn't need to.  He CLAIMS that it simply bypasses that drain/source thing - and moves through the function generators terminal - and then comes OUT at the probe - which, in turn is on the GATE of Q1.  Then it IGNORES the negative signal AT THAT GATE OF Q1 and continues - merrily - down to the negative terminal of the battery. 

Picowatt actually states that when Q2 is turned on, current from the battery supply DOES pass thru the source of Q2 and thru the function generator.

What picowatt (two "t"'s), as well as many others, has actually "acknowledged" and stated, over and over, is the following:

Referring to the schematic in the first paper, when the function generator output is a positive voltage, Q2 is turned off, and if the function generator output is above the threshold voltage of Q1, Q1 turns on and conducts current.  Current flows from the battery positive, thru Rload, thru Q1, thru the CSR, and returns to the battery negative.

When the function generator output is a negative voltage, Q1 is turned off and Q2 is turned partially on.  When Q2 is turned on, DC current flow is from the battery, thru Rload, thru Q2, thru the function generator, and returns to the battery either directly or thru the CSR, depending upon whether the function generator signal common is connected to the CSR or the battery minus termnal.

Q2, configured as a common gate amplifier, is only partially turned on (biased on) by the negative voltage applied to the Q2 source terminal due to the FG's 50 ohm output impedance.  Q2 is therefore biased into a region of linear operation.

When Q2 oscillates, the bulk of the AC currents flow primarily thru the intrinsic capacitances of all 5 MOSFET's (a smaller amount of AC current also passes thru the function generator).

Now, from this simplified explanation, how could anyone arrive at the nonsense she writes?  Mark E., in a recent post at PESN, even drew her a nice shematic showing the DC current flow thru Q2, as has .99 and others as well.  There is no mystery regarding her circuit's operation to anyone but her.

Her posts read like an act of desparation.  She should simply admit the error regarding Q1 not turning on when it should in FIG3 and at least retain some degree of personal integrity and honor.

It is all becoming very sad...

PW

[TinselKoala]

PW said,

She has apparently resorted to either outright lies, or her ability to read and comprehend is sorely lacking. 

These are not exclusive. She has "resorted to" outright lying many times before, AND her ability to read and comprehend "EE speak" is sorely lacking.

She does not understand that a mosfet can function as a linear amplifier! She thinks they must be ON with zero resistance (she neglects Rdss, doesn't understand what it means nor does she understand power dissipation in this resistance) or they must be OFF with infinite resistance. So how can she ever even begin to understand the oscillations in Q2s or the current paths?

She is willfully ignorant and overweeningly arrogant. Her statements prove this, because NOBODY HAS EVER SAID ANYTHING ABOUT THE Q2 MOSFETS BEING DEGRADED IN ANY WAY. (I'm shouting for the benefit of the willfully ignorant and mendacious person who wants to pretend that we have.)
It is only the Q1 mosfet that is put in stress by her arrangement including 6 batteries at 72 volts plus (ironically indeed, because the "rejected" schematic diagram that originally appeared in the second paper, first version, would not have done so) and it is ONLY THE Q1 MOSFET's FAILURE to perform that is indicated in the Fig 3 scopeshot.

It seems that you are proposing that our IRFPG50's have degraded to the point that they cannot pass any current coming from the battery supply source.  Your evidence is based on the fact that the applied signal from the function generator is at 12 volts.  Which under all other circumstances should then enable the flow of current from the battery supply.  Yet we clearly do not have the flow of current during this 'on' period of the duty cycle - coming from that supply.  Therefore do you propose that our MOSFETS's are no longer functional. 

The question that we suggest you should be asking yourself is HOW it is possible for those transistors to pass any current at ALL - IF, as you propose - those MOSFETs have been so degraded?  Because, self-evidently - there is a great deal of current passing both to and from that battery during the 'off' period of that switching cycle.  We would be MOST intrigued to see your answer to this.  If you take the trouble to explain it then that would be considered appropriate as we are now going to some CONSIDERABLE lengths to show you that indeed we DO what you claim is IMPOSSIBLE.

Liar. Ainslie is a sarcastic insulting ignorant liar.
Only the Q1 mosfet is in question. Picowatt has never said or implied otherwise and neither have I.
"Considerable lengths." It took me less than fifteen minutes to make my demonstration of the problem, including showing what correct and blown traces look like, and even including testing the mosfet. But all Ainslie can do is insult, whine, lie and make excuses.... and further postpone what she says is "easy" to demonstrate.

[TinselKoala]

@Tinman:
I see that you have asked about mosfet types. I have tested the previous single-mosfet circuit with many mosfets; I found that the 2sk1548 worked the very best of any I tried in that circuit, much better than the IRFPG50, and it was about 1/3 the cost and was carried in stock by my local supplier. Great fast HV collapse spikes due to much cleaner switching than the PG50.
For the present circuit with 5 mosfets, I don't have any k1548s any more but I've got a handful of PG50s, and I did a long comparison using IRF830 and a shorter one using IRF530ns. The smaller mosfets have greatly reduced gate charges and so will make the circuit oscillate at a higher frequency-- and that's the only difference, as long as current and heat dissipation limits are respected. The 830s oscillated at around 4MHz, iirc, whereas the PG50s in exactly the same config (wirelengths, etc) oscillated at 1.2-1.4 MHz. I even got a version to work with 2n7000s.
The key is the long lead wires to the mosfets to make and sustain the oscs, and increased inductance in the battery wiring to increase the amplitude of the oscillations. If you use "conventional" tight layout, buswires and short jumpers, you may not be able to get spontaneous oscs at all. I didn't with my first version, with all mosfets on buswires and in a circuitboard space about the size of a business card.

[TinselKoala]

@PW:
I hope you didn't miss my post about the heating element and the possibility that it might contain an overtemp thermal fuse. It's just barely possible that an RV water heater element might be designed with a bimetal thermo-fuse that will trip open when the max temp is exceeded and close again after things cool off. (Many kinds of equipment have such fuses, like microwave ovens and even my old Tek RM503 scope has one mounted to its chassis.A RV water heater element might be a very logical place to have one.)

Obviously, we need some assurance that the heater element does not have any kind of temperature-sensitive switch built into it -- or formed inadvertently somehow -- that could open its circuit when it gets really hot.

I'm not sure what an open load will do to the rest of the scope traces, though.... I'll have to check quickly to see if the idea of an open circuit at the load is even viable, given the presence of the oscillations.

Later...:
OK.... I've just tested TarBaby, and with the load disconnected it won't oscillate. So for my build, at least, the presence of the oscillations indicates that the circuit through the load is not open. So I suppose that eliminates a thermal cutoff switch as an explanation for the Ainslie Figure 3 trace.

[picowatt]

TK,

The reason I have not responded with regard to a method for testing Q1 is that if she duplicates FIG3, it will immediately reveal that she has resorted to deception.  She no longer argues that the FIG3 scope capture is being read incorrectly and agrees that +12volts is indeed being applied to the gate of Q1.

She knows that she cannot apply +12volts to the gate of a functioning Q1 connected as per her schematic without Q1 turning on and passing current.  To show otherwise would require an act of deception.  So why would some test of Q1 after the fact not be part of that deception as well?

I agree with your prediction that the FIG3 demo will never happen.

If she has her circuit all set up and ready to run, why doesn't she just admit there was a problem with Q1 in FIG3 and repeat her tests? Doing so would at least allow her to retain some degree of dignity and honor.   

PW