Hi Gyula
OK pretty much understand what you are saying now - the first pair of bidirecitonal diodes, when they turn off, send a back spike into the battery, and this can be gathered,  by having cap with its neg on the positive-rail....and then steering diode from drain to pos lead of this cap.
I dont know if this backspike will amount to much, since that mosfet-pair (AB)
is still ON when the other mosfet (CD) pair acutlly shuts off the circuit so I assume the "brunt" of the backemf/recoil exits through the mosfet C but I dont know that is just assumption...
I didnt try the other diode on cap over AB out like that,  out so thanks for the tip - I thought here must be some other way for backemf/recoil to "sneak" out of there - as you say the classical way is put AC legs of FWBR across the whole mess of mosfets and this should catch everything going which way...
having single diodes is better I think since then making it work in "diode plug" manner is possible plus less resistance/loss (perhaps) using single diodes also I bet the circuit is heavily biased backemf-wise on mosfet CD side of it, rather than mosfet AB side, so perhaps different caps and different diodes will work better for one side as compared to other side.
this might work out for bettew way to take out power from coil-shorting at peaks too with genrtor coils, since I use FWBR in  that, but maybe singel diodes situated like thiswould work better for that...

Hi Doug,

I have made another schematics from your original schematic to make it clearer for everyone.  Consider the first circuit on the left.  The two switches (each is a bidirectional MOSFET pair) are in series with each other so they both must be ON whenever you wish for any emf current flow in the motor coil i.e. you wish to use the coil as an electromagnet.  This is how you drew your circuit in your original backemfdirectional.jpg picture, which I simplified as you can see on the left, ok? (without the bemf recovery) 
IF you disagree with this simplification, please tell.
Now if you wish to switch the current OFF in the coil at a certain moment (duty cycle point of view) and capture the bemf  i.e. the collapsing energy of the motor coil,  you have to switch OFF the second series switch (consisting of MOSFET C&D) because you assigned Hall B to control duty cycle and Hall B drives MOSFET switch C&D.
BUT the moment you place the recovery diode to the drain electrode of MOSFET C (and it is exactly the connecting wire between the two switch symbols in my simplified drawing) and you connect recovery capacitor negative to the common source electrodes of MOSFETs C&D as you showed in your original drawing,  THEN you have to keep MOSFET switch A&B still ON while you switched C&D OFF to insure closed current path for the energy flow coming from the blue wire of the coil due to the collapsing field because  the voltage spike developes across the coil at switch OFF of course (I indicated its polarity in my yesterday drawing wrt the emf polarity) and the circuit from bemf point of view is closing via the 12V battery and drain-source body diode of the MOSFET D and meeting on the common source with recovery cap negative pole.  IS this ok?  Because this is how your original circuit works.

Now to reflect on your present post, I refer to my second drawing where I meant the recovery diode + capacitor yesterday and you got it correctly.  And the amount of the backspike can be high if you interrupt the motor coil current fast (and you do it fast by the 4421-22 chips)  no need to worry about the still ON A&B switch because the C&D switch in series with it is already an open circuit (this latter is which interrupted the motor coil current) so no any current can develop via C and/or D.

IF you were to use diode bridge across the motor coil then it would be worth thinking on whether the induced voltage by the approaching and leaving rotor magnets (which induction is interrupted by the ON time of the motor coil of course) could give any extra into the recovery capacitor,  I forgot to mention this yesterday.  This should be tested if induction really can give useful extra, justifying the bridge usage.
I do not see as you say "the circuit is heavily biased backemf-wise on mosfet CD side of it, rather than mosfet AB side, so perhaps different caps and different diodes will work better for one side as compared to other side"  because everything is in series from the motor coil point of view: the 12V battery, MOSFET D&C, MOSFET B&A (and the source of the voltage spike is the motor coil);  the 12V battery is a short circuit from AC (and DC) point of view, and D&C is OFF and B&A is ON.
Hope this helps those who may still have had some problem with my earlier post.

rgds,  Gyula

Hi Gyula
I'm a little bit confused now - the simplified circuit you show now has only one cap, and over on the "other side' of the circuit - it is what I didnt have in my original circuit - which is that diode and cap off the pos feed to coil (high side recovery circuit you could call it?)
So are you saying that this is "better" than the low-side recovery circuit, which I showed in my orignal circuit (with diode and cap coming off the C mosfet)
Or is it better to have BOTH what I had before, plus this cap and diode too??? Thats what I thought you were saying in your drawing before this one...
anyways I will have to try both ways - and see if they both will work at same time - -  is that what you are thinking? 
So for that pulse-width adjsut circuit with the two pairs of mosfets in series, there should be two caps and two diodes total - one diode and cap coming off the ground (low) side, and one diode and cap coming off the pos side as shown in this new simplofoed drawing?

Hi Doug,

Sorry for this, I did not mean using recovery circuit at two places,  only I meant across the motor coil.

When you use recovery diode + cap across MOSFET C as you showed originally then the current from the spike is directed through the 12V battery too and when the recovery circuit is across the coil only, this current goes directly into the recovery cap only.

I cannot tell you which method is better, it should be tested, I only analyzed your original circuit. 

I do not see much sense in using a diode + a cap across BOTH MOSFET pairs, if the the 'lower side' recovery turns out to be better vs the-across-the coil-recovery then the circuit you originally showed is ok,  because MOSFET switch A&B should be kept closed while you open MOSFET C@D  and by the time the collapse rings down and you open switch A&B there is no current in motor coil to interrupt.

rgds,  Gyula

So... I'm still waiting for parts and building this Muller dynamo thing to include bifilar coils and mechanical switching and I am just wondering-

Has anybody considered all the different possibilities that can be done with all these outputs?

For example; might it be possible to configure this dynamo in a "progressive" mode?
In other words, as each coil 'fires' the resulting generator output, or the BEMF energy, be fed to the next coil in line so it's already being 'charged' as the magnet approaches?

Or how about a 'cumulative' mode where BEMF energy is, somehow, collected from EVERY coil and stored in capacitor?

Or how about a combination series/parallel mode with some coils in series then sets of series coils wired in parallel?

Can one simply short one side of the bifilar to get a strong BEMF back out?
And would this require additional switching?

With 18 coils, each having 2 'input/output', and 72 wires to play with, which may be configured in MANY different ways in relation to each other, I am surprised some of these other configurations have not yet been attempted here. . .

Start with step one: First coil, or coil pair, to spin the rotor.
Step two: What should be done with the BEMF energy OUT from the first coil or coil pair?
Should it go to the next coil in line? A capacitor? Additional switching? Should other coils operate independently? Or another combination of things?
How many combinations might be accomplished with 72 wires and lots of BEMF potentialities?

And one thing I THINK I know about this device; it should be built as a complete unit to take full advantage of the off-set between 8 magnets and 9 poles.
If all your poles are not in place; will you obtain the same cogless effect?
And if all coils are not configured to operate in unison, or are complimentary to each other, what would this accomplish?

}:>