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

[listener192]

Yes exactly
That's how I built it a few weeks ago.
For the coils to fully drive through please use PNP and NPN FET types.
I send you the circuit diagram again.

Hi Andy71,
I am using the BTS7960 bridge boards, so that it how they were connect from the start.
The main point of my post was not to show the waveform switches but to show how we could achieve the pulses that Pierre had riding on top of the sine waveform (see post 1186).

The only problem I see with your circuit is that you have no isolation for your MOSFET drives. One device failure could destroy an arduino digital pin, or even take out the whole board.
At some point in experimentation it is likely this will happen. I have had switch failures twice now as I have pushed current limits.

Regards
L192

[listener192]

No the field is not shut off in rotation. This waveform shows a pulsed signal riding on top of the output sine, these are not recovery pulses, although as there is pulsing, there could be recovery current via the switch body diodes.

Attached is a simplified drawing of how this could be achieved.
Only a few MOSFET switches shown for clarity. Coil connections and switches remain unchanged only the supply rail is modified.

This could be accomplished with one relay or two MOSFET's as shown.

L192

Further points on this circuit...
There is never a period where all switches are off, so coil current is maintained.This is achieved with +25V.
The caps in series provide +50V to drive an ON/OFF pulse above the sine waveform (when looking at the output waveform) i.e. a further pulsed increase in current through the coils.

When the pulse turns OFF, the input caps have switched back to the parallel configuration +25V, this allows substantial current to flow through the body diodes back to C2, as the flyback voltage is 25V higher than C2 voltage.

The problem with this is I can't test it at +25V input, as the BTS7960B over voltage protection will turn on the HHS at +27.5V, as advised in a previous post.
So I could only test at say +12V.



L192

[listener192]

Attached is the result of the experiment, as shown in the circuit posted previously.
The cap charge scheme did not work very well, with the top cap only developing about 5V, so the bottom cap has about 25V, you can see this in the dark blue waveform.When the caps are switched in series, about 30V is developed. You can only see a very small increase in current due to the top cap not adding much voltage to the stack.
When the pulse is turned off, the rail drops back down to about 25V, however a large recovery current is developed into cap due to the lower potential on the parallel caps.
In this timing I am switching the overlap coil off immediately after the one preceding it is turned on, then apply the pulse (width determined by delay X), then followed by another delay X, followed by the next sequence.
The next experiment will be to use the circuit shown below. This should allow a decent current pulse and be able to show a large pulse voltage on top of the stepped output sine.Possibly the caps on the bridge boards will absorb some of the voltage pulse, as they charge from +12.5 volt to +25V.
Note: the external recovery diodes shown as D1 & 2, allow the coil flyback to recover directly onto the lower cap bank.If you were using relays a higher DC voltage could be used (compared to the BTS7960B device limits).

L192

[listener192]

The experiment using the cap bank in two series sections was problematic with the top bank getting warm.
I changed over to the circuit attached, so just a single pulsed 25V DC supply.
The recovery voltage was enough to maintain the waveform between pulses.
In the second scope shot you can see the supposed recovery current however, I think this is actually just bleed through during the pulse period.
The first scope shot shows the output off load and the second on load. Not much in the way of voltage pulses.
I had the timing a little different in this run, so I will try also in the previous configuration.
Rotor induction is still poor, infact generally worse since I cut down the end of the rotor to 6 slots from 9.All this tells me is that I just reduced the flux linkage.


L192

[listener192]

Long Poles:
This test reverts back to long poles which  briefly tried before. 
11 slot span rotates and reverses polarity on 2nd pass (visible in current waveform), still with (short) overlap, so there are no interleaving poles to short out the flux.
Output voltage in blue input current in yellow.Scope shots for unloaded and loaded output.
Not a bad output for only 2A RM,S  26.5 DCV input. The 6 pole configuration was running 10 times this current for the same off load output.

Note the unloaded waveform and the  spikes due to the increased inductance of 12 coils in series. The coil resistance and double the inductance is the limiting factor for current.Note a 50/60Hz output is achieved and if you reduce the switching rate this output does not increase.

The bridge boards will over volt at 27.5V, so current cannot be increased by increasing applied DC voltage.
So the only way to increase coil current is to reduce coil resistance and inductance, this means parallel coils which means breaking up the 360 degree series chain.
Of course this is just a reiteration of what Pierre has already disclosed.



L192