Rosemary Ainslie circuit demonstration on Saturday March 12th 2011

[WilbyInebriated]

like sands through the hourglass, so are the days of OUR lives... 

[Rosemary Ainslie]

Guys - more good news.  A friend of mine has done a full on replication.  I've downloaded waveforms.  I'll post it later.  All negative values.  Two fets attached to a common heatsink running a standard 9 Ohm 12 volt automotive soldering iron.  It was cooking - but we didn't have the means to measure the heat.  Actually we did.  Just forgot to do this.  Too hot to handle - in any event.  Probably in the region of 80 degrees or more. 

Anyway he did a short video on this.  I'll upload it when I get my computer back.  And I'll upload the screen shots tomorrow morning.  Very simple config. I think it's going to be very easy for you guys to replicate this.  Crocodile clips - crude connections - but everything working.  And everything permanently in NEGATIVE - as with our own tests.  You should also be able to see the required waveform on your standard oscilloscopes. 

Anyway.  That's the first FULL ON replication. 

Kindest regards,
Rosie

Sorry I should have added.  He used our functions generator to drive the switch.  But he's getting his own.  But it's JUST SO EASY everyone.  Simple and easy.  He wired the required 'legs' together and then applied the signal from the generator to both those gates.  IT WORKS FINE.  And the oscillations are PERFECT.  Again, interestingly, better results at higher frequencies.  He was running it at 800 Hz.

[Rosemary Ainslie]

Abject apologies guys.  It seems I didn't get that download onto my flashdrive.  Am mortified.  I'll see if we can redo his test this evening or at the weekend.  I've asked him to become a member here - which will be WONDERFUL.  Then he can do his posts directly.  He has many ideas to take this to application phase.

But I'll get those waveforms to you as soon as I have them. 

Kindest regards,
Rosemary

added : BTW - I'll take the trouble then to measure the temperature.  I've got a little 'testo' probe that I use for ambient temperature which I forgot about.  Not keen on dismantling the temperature probe on my artefact - because it's been fixed there to keep a reference for our controls.

[Rosemary Ainslie]

Just as a general discussion related to the energy in that circuit vs battery 'mass' that I've been referring to - here's something of interest.

The first 'load' on that replication test was a simple light bulb.  Can't quite remember the required voltage but I think it was 24v or thereby.  In any event - this did not result in those 'negative' cycle mean or mean average voltages.  Interestingly the math trace did stay negative which is a discussion all on its own.  But the light filament itself is very small.  And it was barely alight.  Flashed at low frequencies - steady at high frequencies as expected.  But very little evident light output at either switching speeds. 

Then when we applied the soldering iron as the load - there was an immediate reversal of the numbers that the cycle mean and the mean average AND the math trace just moved into and stayed at NEGATIVE.  Not only that but the level of heat dissipated on the iron was extreme.  And it could be adjusted in either direction.  Resistance of that iron was 9 Ohms.  In effect the amount of wattage dissipated appeared to be greater over the load that also had the greater mass.

I wonder if there is a ratio to applied voltage and the actual material mass of the load.  I suspect so.  It seems that the greater the mass the greater the energy available.  Not so much related inductance - but actual mass.  Then maybe there's more energy available to 'return' during the 'off' period of the duty cycle.  Just a thought.

Again, kindest regards
Rosie

[Rosemary Ainslie]

Guys.  I've just thought of an easier way to describe this eccentric MOSFET positioning.  It's not back to front - not upside down - it's HAND TO HAND.  lol.  In effect what was done in the replication was this.  Two mosfets put side by side and attached to the surface of a shared heat sink.  Q1 on the left -  Q2 on the right.  Legs facing the same same direction.  Then using approximately 6" of copper wiring (about 0.25 gauge) x 3.  The first wire soldered to the gate of Q1 (far left leg) then connected and soldered to the source of Q2 (far right leg).  The second wire solder to the drain of Q1 (middle leg) to the drain of Q2 (middle leg).  The third wire soldered to the source of Q1 (its right leg) and then to the gate of Q2 (its left leg).  Sort of holding hands.  And if I had my PC I'd have given you either a sketch or a photograph.  But it's been taken away for forensic analysis!  lol.

If you can visualise that - or even sketch it - then that's the required positioning and linkage between those legs.  THEN.  Put the lead from the signal generator Positive to the GATE of Q1 and Ground to the GATE of Q2.  THEN , nota bene guys or Poynty et al will SCREAM DENIALS - put your shunt resistor in series BEHIND that FG's ground and directly in series with the negative rail of the battery supply.

Then.  Adjust your duty cycle that it's OFF for a longer period than it's ON - and then SIT BACK and enjoy the show.  You'll see everything just oscillating and oscillating.  Finer adjustments still required - but that's a good 'kick off'.  Thereafter you adjust your 'offset' to get the optimised power values.  That may take some time to assess.  But if you get to this point then perhaps I can help you out here.

I confidently predict that this is going to be MUCH, MUCH easier to put together than the 'single' spike number that we showed earlier and it'll be easier to prove those gains.  And I absolutely do not think you need to get married to that IRFPG50.  Use anything that has a zener body diode.  On that replication we had a maximum peak to peak voltage from the supply at 3 times the average supply voltage.  And on the 24 volts applied that gave us a peak to peak of about 72 volts.  That's well within the tolerance of most of those NFETs.

The only downside, perhaps, is that I can't give you an alternative to the functions generator.  Not yet anyway.  The one that we tried gave us a rather messy waveform and dubious returns.  But I know that with the talents available here - then one if not more of you - will find some solutions.

Hope that helps.
Kindest regards,
Rosie

added.  And if I may quote what my friend kept saying.  'It's so easy.  I can't believe it.  There it is.  It's so EASY'.  lol