Steven Mark`s TPU

[ronotte]

Hello Harty,

thanks for your kind words....

The 30...50V are referred to actual Power Supply settings. I'm using a separate programmable PS to keep under strict control the V & I delivered to Power Mosfets (in order to preserve them from thermal run-away). The associated current is then anyway always limited to 3.5 A.

As you have seen the power in that condition is not steady but rather is pulsing: so is the current synked. As many time sayd there is still no OU but I've had an improvement from the previous tests using ECD of almost 3:1: it does mean that I can light the lamp at the same brightness using 1/3 of the current previusly used with ECD. 

GREAT STEP TO CORRECT DIRECTION

Again the reason I posted the power clip is only to show the possibilities.....think that during that test the lamp FUSED...but no problem with Mosfets thanks to PS protection!!

Again think that about half of the dissipated power vanish in heathing the power Mosfet's Heat-sinks...even if I'm pulsing them at a very low duty-cycle: that's encouraging as I see groud for improvement.

So let us dream a little after a so long and stressed past time, just wonder what if I succeeded in ELIMINATING the Mosfets losses: there should be an OU of at least 1: 5.

Roberto

[Gustav22]

Hi Roberto,
you have hinted, that you already have an idea how to avoid the heating up of the mosfets.
I hope your idea will work.

Nevertheless I want to let you know my take on the heating problem:
Premise:
When the conversion in the TPU works, the three input signals somehow get coupled and transformed to create the 3sister-wave output.

Do you still use non-synchronized input signals?

If you still use non-synchronized input signals, the transformation of these signals into the 3sister-wave is probably only happening successfully, when all 3 input signals meet with  the correct phase relationship  - to each other and to the 3sister-wave output signal, which builds up in the collector and in the feedback- windings.

Also note, that the feedback windings are in a normal transformer relationship with the 3 CCs and constantly induce a signal into them.

The case of correct phase relationship between the signals fed into the CCs (coming from the mosfets) and the signal induced into the CCs by the feedback windings must be purely coincidental, no?
I think that if this phase relationship is not correct, the input energy is dissipated as heat in the mosfets, as it can not be transformed, because it "fights" against the signal from the feedback windings.

Maybe you will get much less heat and a much increased conversion factor, if you try with synchronized input signals with adjustable phase shift.

Or you can try to lock the phase(s) of the 3 oscillators to the output signal of the TPU.
Obviously I have no idea how to do such a thing.

In the end, with a closed loop, the feedback windings will force this phase-synchronization of output and input.
But at the moment the phase(s) of the input-signals are dictated by 3 INDEPENDENT oscillators. So all three input signals are out of phase
in respect to each other
and also out of phase
in respect to the signal from the feedback coils.

[ronotte]

@Gustav22,

Yes the 3 Freqs till now are not synched. Well the auto-synchronism process seems to happen when the 3 freq values are within +- 5 Hz: outside that window there's only current increase (in mosfets) without correspondent output increase -> hence there's only Heat production.

And if the phase relationship is not correct, the input energy is dissipated as heat in the mosfets, as it can not be transformed (because it "fights" against the signal from the feedback windings).

Maybe you will get much less heat and a much increased conversion factor, if you try with synchronized input signals with adjustable phase shift.

I'm glad that you raised a new perspective on how-to explain the energy dissipated as heat (In mosfet heat sink-- not coils). I've to think about that possibility. For sure the easiest way to prove it is just to put the 3 freqs in perfect sinchronism and observe the results!  This could be done by simply wiring a 555 followed by a chain of dividers :2, :5 and :7  (a far more sophisticated 'DDS swiss-knife' to play with synchronism, phase, etc  is in it's developing stage). The only reason why I haven't already did it is that the current setup allows for Chords testings.

If you still use non-synchronized input signals, the transformation of the input signals into the 3sister-wave is probably only happening successfully, when the input signals meet with  the correct phase relationship  - to each other

Why sayd process take-on is frankly unknown  as I did nothing to promote it externally......I never observed nor designed such a process!

The case of correct phase relationship between the signal coming from the mosfets and the signal in the CCs (induced by the feedback windings) must be purely coincidental, no?

Of course I don't think tha's purely coincidental!

Roberto

[Gustav22]

How about your unit? ...is it completed?

Magari
But I have to tell you that it is very bad style to expose the weaknesses of your friends before the whole internet.

Seriously: "I am getting closer".

[wattsup]

@Roberto

Good work as usual. I am curious, as may be others, to know the relationship between the power supplied to the mosfets, to the actual power applied to the TPU. Let's say you are providing a combined power of 150 watts to the mosfets, what is the actual wattage supplied to the TPU? Would measuring this be dangerous for a meter connected directly to the mosfet outputs. Maybe with a sniffer coil? (I'm guessing.) But if this was known and output of TPU was known, you could then have the exact performance of the TPU itself like a super efficient step up transformer or even better. This would be important to know if looking afterwards to make any potential modifications on the TPU and having a standard way of measuring the TPU performance regardless of the power supplied to the mosfets.

Also, is it possible to supply pulsed dc to actually power the mosfets. Maybe a pulsing DC into the mosfets would keep the heat down while still maintaining good mosfet performance. Also by pulsing the dc, there may be a way to recaptured power at every off pulse, increasing the efficiency of the main power supply.