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

[r2fpl]

I opened one and found just by using some solvent spray I was able to restore contact resistance in the milliohm range. It may be possible to drill a small hole (without removing relay from the board) to inject some solvent.

My two relays I used for polarity reversal did get very hot, as they were carrying the whole load. To actually catch fire, I think more than 10A would have to be applied. So much for the theory that 3.5 to 5A is enough.

Regards
The192

good idea )

[listener192]

The last experiment I conducted before relay failure was to remove the "coil always on" in code, so each sequenced "on" delay "off". Due to the inductance then seen, the recovery is much higher and 10 switch cycles per rotation are seen. With the sequence running at 5Hz this provided better induction, developing about 75V off load.

This puts a greater strain on the relays and without suppressor caps would be asking for failure but the same could be achieved by switching the +rail with a MOSFET, in between relay switching and keeping the coil always on, so the large current change would always be though closed contacts.

L192

[listener192]

Is there field self excitation? Some feedback from the armature back into the stator windings? Has this been either confirmed or denied?

Power is recovered into the super caps from the switching process,also when looped the linear power supply is powered from the rotor AC 115V 60Hz output  (note we have not seen this waveform).So yes self excitation.
L192

[d3x0r]

Yes, but they are quite cheap relays, maybe a few of them had worse conduction and burned. I have one strip from another manufacturer and only 2/8 are good!
There was never more than 10A for ALL.

Ya I was noticing that too.. one board (6-8) has really strong currents that cause a magnet to turn VERY quickly... but the others are fairly weak... some are very weak; but I blame that on my poor soldering skills   Some coils might have cold-ish joints and I mean to walk though each to check...

[listener192]

I stripped the relays off 1 board, opened them, cleaned and adjusted the contacts now all are 10 m ohms or less. They all have to be this way to achieve the max current.
If the 60Hz were generated purely by what we can see...  10 sequences x 10ms =100ms per rotation.

That's 10 rotations per second, 6 polarity reversals in each rotation, so 10 x 6= 60Hz.
The max time to switch is 10ms, so in practice you would be lucky to get 5ms on time.
I would conclude that only 30Hz would be achievable.
L192