Pierre's 170W in 1600W out Looped Very impressive Build continued & moderated

[konehead]

 hi Jerdee
For the switched a.c. leg it needs to be bidirectional mosfest for example since it  is switvhing ac leg.
Perhaps you could do the delay switching of recovery.over on the dc side.of fwbr now.bidirectional not needed just use single mosfet.
Maybe Pierre does some sort of delayed "echo" recovery circuit through his computer code nobody has picked up on it yet I dont know
SSRs always slow especially heavy amp type but so it is slow just adjust timing to compensate.
The Chinese company Gold makes good SSRs.cheap...460v 40a rated for 16 bucks...20 amp ones about 12 bucks
You can dig out internal mosfet in them and put in your own bidirectional mosfets maybe put 4 in parallel cluster one side another 4 in cluster other side now very low resistance


Can't really beat brush commutators for low resistance switching .
Pierres relays similar as mechanical switching to them so ultra low resistance too...for peak coil shorting very low resistance in switching is required could be reason for success of DZ generator if it does have some sort of coil shorting effect when fields rotate and overlap for a bit of time....
What some people see as cancelling perhaps can be seen as short circuiting of coil and it creates not "nothing at all" but instead hyper ringing effect and this why his caps fill so fast


probably it is something important to mimic permanent magnet field not as N or S  blob of magnetism but instead N edge and S edges sweeping across coil one edge leading the other trailing...think of it like small hurricane spinning over coil and sweeping past coil too at same time to make power...


Fact you need resistive load on coil for it to make power might be key as if there is no Lenz law then the power normally sucked into Lenz law braking action can perhaps coheree with output into caps instead and reinforce not kill it.

[jerdee]

Here is a drawing of the diodes in action that Pierre drew on the back board in one of his videos.   You should see why we have to shut off the field in rotation now.  Keep the code the same using overlap mode.  This also explains why you do not want the coils in series.  You want to isolate the field(s) and you can run them in parallel.

[jerdee]

For the switched a.c. leg it needs to be bidirectional mosfest for example since it  is switching ac leg.

I understand the concept of echo, but not fully understanding why you would want to delay the recovery quite yet.  It's not sinking in for me quite yet.

Again, I have nice quality bi-directional FET boards that will do the exact switching we need for one ac leg, however, this will get costly.  I would need many of these built.  SSR's was a thought, but can see how they will be too slow. 

Perhaps you could do the delay switching of recovery.over on the dc side.of fwbr now.bidirectional not needed just use single mosfet.

Thanks. I think it might be best to do it with two instead on the d.c. side.  I like the idea of replacing two of the diodes from FWBR DC side to mosfets instead. A bit cheaper and easier to control.  I like your idea and direction with this.  You don't have to deal with dual switching of the DC side, only one polarity uses the normal diodes, but the other polarity requires both FETS to be triggered simultaneously.  Active recovery in one direction, not both, and only requires one pin from MCU.  Just trying to think about utilizing recovery for both polarities as much as possible when switching ac.

Here is another thought. You could even setup logic gates to trigger FETs in FWBR to be ON in the correct polarity.  This would require no more MCU pins. Each field group would have it's own dedicated logic gate to prevent extra pins from MCU to trigger the FWBR FETs. They only turn on when the field is off and in the correct polarity. You don't need MCU control for this to happen.

Jerdee

[pmgr]

I suggest we stick to discussing Pierre's original setup and find out how that setup was possible to work.

1. It had a 4ohm resistor between cap bank and FWB.
2. Relays were driven from FWB, not cap bank (dangerous if the cap bank is low on voltage as it will try to fill the caps through the relays and recovery diodes and will burn the relays)
3. He had 6 poles, NSNSNS, all biased in parallel, no poles in series
4. Original code had everything off at end of every loop

All replications trying to reproduce this setup have failed so far.... question is why.

PmgR

[Dog-One]

Here is another thought. You could even setup logic gates to trigger FETs in FWBR to be ON in the correct polarity.

It's called a synchronous or active rectifier and is a method to get past the normal diode drop allowing access to much more current.  Here's a part you may want to experiment with:
http://www.analog.com/media/en/technical-documentation/data-sheets/4320fb.pdf

What I was getting at, it seems you found with the "Diodes In Action".  That resonance between the base sine wave and the peaks I believe is your parametric resonance; that's a real gain you can take advantage of.  If you have this happening in every coil (resonator), those gains add up pretty quick.