Permanent magnet assisted motor coil designs

[captainpecan]

Thanks for the circuit drawings.  Considering 2 coils in series with the source of the MOSFET and the 10 k resistor towards the gate of the MOSFET,  the induced voltage appears across the gate-source I suppose. And the reed switches still operate from the small control magnets and they short the drain of the MOSFET to the gate, right?   And there is the body diode between the drain and the source, which may start rectifying the induced coil voltage and the horizontal line you indicate with a questionmark in the scopeshot above may be the indication of a DC component, what may be the supply voltage for a moment?  maybe this is what is happening.

You could use one scope probe across the gate-source when the other scope probe would remain across the same two coils and see the waveforms. (Both scope channel ground clips should be on the common negative of the circuit.) 
Temporarily you could remove the wire coming from the reed switch to the gate the moment you switch off the supply voltage and the rotor starts decelerating from the earlier full speed, to see if the spike disappears. I am watching the video now.

Edit: for the time being, no more comment, try to check the above things.

Thanks for all yout help so far. Okay, I still haven't wrapped my mind around it completely, but you are correct. I was doing some different things to try your suggestions, I noticed it disasappears when I disconnect all other coils from the ground except for the one I have probed. So it is an induced current from one of the other coils gating that transistor. I placed an extra wire to allow me to disconnect the probed coil from the transistors completely and go coil only as soon as power supply is switched off. I get a normal sign wave as it slows down then. Now that I know for sure how to replicate it as well as make it go away, maybe I can fully understand it. Thank you.

EDIT:.. Now, since I have 2 channels on this scope, I think I will figure out which is tripping the gate of this pair of coils and see if I can see it happen and understand why.

[captainpecan]

I am now working on the best way to recover the fly back ONLY. The way these coils work with an embedded magnet, helps boost during a drive pulse, but the opposite side is they are terrible for generator coils. So much Lenz drag with that inner genie field popping out. Just using the normal recovery circuits are usually really a half wave rectifier of sorts grabbing the whole negative half wave. I just want the extra spike so it won't drag down the rotor. I plan on generating from this on a different side of it and let these coils only drive.
I found something that seems to work, and does not slow the rotor at all. I dont know if this is the most effective or efficient way, but it's working. I simply hooked a capacitor in parallel with the charge battery, an 18650 at the moment with a resistance between them. The recovery circuit is dumping into the Capacitor which in turn dumps to the battery. I adjusted resistance between the cap and battery to find a balance where the cap stays at about the same voltage of the waveform, but still dumping any extra to the battery. This appears to be very effective. This is kind of like a smoothing capacitor I guess. It seems to do the trick and help me snip that extra out without using the drive coils as generator coils for a half wave, and doesn't change rotor speed at all. I'm trying to learn more as I go. Here is the difference in the circuit as I connect it.

[gyulasun]

Just for clarity: the coils you use now and recover their flyback energy have no embedded magnets? or they do?   

Whether the method your using a resistor between the puffer cap and the battery is efficient (or not) can be estimated by hooking up a scope probe across this resistor and read the Mean voltage value of the waveform (it will be kinda saw tooth like).   And knowing the resistor value and the Mean voltage across it the dissipated power comes out. And you would learn about the mean amplitude of charge current the charge battery receives. 
In this test you mention did you use the power supply or a 'run' battery to feed the setup? 
Avoid ground issues if there would be any when estimating the voltage drop across the resistor. 

[captainpecan]

Thank you for the response. Yes, they do have the embedded magnet. And I am posting a quick view of the circuit. The switches are n channel mosfets but you get the basic idea of the circuit here. I will do what you suggested when I get home tonight. I think I understand it. I am thinking to help with the collapsing field, maybe i should use 3 more diodes and pull it down from each coil instead of a coil pair? Also, when using a puffer coil like this, doesn't the energy higher than the caps voltage just flow right over it per say? So it's not actually even touching the cap is it, just the loss through the resistor? Or am I thinking of this wrong, or is it actually charging and discharging that extra voltage over the top?


I really appreciate the suggestions everyone makes. I'm trying to follow them and learn from you guys as best I can. Thanks again.

[gyulasun]

Well, you can attempt using 3 more diodes as you describe I never tested whether there is a small or significant difference between letting the flyback pulses add up on 2 or more series coils or collecting them individually. 
In theory, the two flyback spikes of the two series coils should add up as if you connect two batteries in series but this could be checked with scope probes, checking individually the amplitudes of spikes across each coil, one at a time while the setup is running.  (care for any scope ground issue)
IT is worth considering that in the present circuit drawing each flyback current of a coil (out of the two in series) goes through the other coil's copper resistance. With the extra 3 diodes this would not happen. Which causes higher loss: the 3 extra diodes or the copper losses?  checking can tell this.

You showed around 80 v peak for the unloaded flyback pulses across two series coils and if the flyback pulses indeed had around 40 V peaks from each coil, this still sounds a little low amplitude. 
This may be due to a slower switch-off speed of the MOSFET concerned or maybe due to some asymmetry between the coils (I suspect the first, not the latter.)   
The slower switch-off speed can come from your using 10 kOhm gate-source resistors for the MOSFETs, replace them with say 2.2 k resistors (or less) to discharge the gate-source capacitance more rapidly, the reed switches should work well with the 7 mA current instead of the previous 1.4 mA or so.  But then you would have about an extra 20 mA draw from the input (for the 3 reeds) 
If you use MOSFET driver circuits, the switching speeds will increase significantly versus the reed switches + the gate-source discharge resistor,  though these circuits also need some (small) input power to function.   

I assume you do not wish to connect the 2 coils in parallel that are in series now, to avoid the higher input current draw involved with this.

Please clarify what you mean on "using a puffer coil like this" ? you wrote it in your last but one sentence.

If you meant puffer capacitor, then the answer is no, all the energy coming from the flyback pulse goes into the capacitor as long as the cap voltage is lower than the spike's peak amplitude. Of course, some part of the flyback pulse will go towards the resistor + the charge battery and how much part is involved depends on the series impedance of the resistor+battery and how this is related to the (normally lower) capacitor impedance.