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

[listener192]

Attached is a snapshot of Falstad running a boost circuit for a single coil. Simulation is attached.The shot shows the boost charge in progress and it has got to the point where the cap bank starts to charge the 220uF cap to a voltage higher than rail.
Across the bottom  left to right we have..

Voltage across 27F cap bank:
Voltage across 220uF cap:
Current through 5ohm resistor acting as a load: - normally the load would be coupled by the coil to the rotor winding.
Current through the 4ohm resistor: - the 4 ohm resistor is needed for the circuit to work. It isn't just a cap bank charge limiter.

If you leave out the diode connected between the cap bank and the 4 ohm resistor, then  the cap bank charges first through the 4 ohm resistor.
DC supply is 25V and has a diode isolated pre-charge of 25V for the cap bank. Note this is just to speed up the simulation, in practice a resistor needs to be in series to limit the current.

If you imagine that you have 20 single coils active, you can see there would be continuous boost and that 4 ohm resistor would get very hot.Also attached are circuit examples for uni polar and bi polar boost circuits.
L192

[e2matrix]

I think this statement from pmgr in the Figuera message thread may be relevant here also:
"EMF = d(flux)/dt = d(L*I)/dt which only simplifies to L*dI/dt in case L is constant (and this last simplification is the only thing we are taught in school).Adding or subtracting windings to an inductor changes L itself and thus L*dI/dt no longer applies. Instead d(L*I)/dt should be used. And with that it is very simple to obtain an overunity system as long as the amount of energy that it costs to change L is less than the amount of excess energy you obtain with the system.The more difficult part of this is to design a system that will do exactly this and which can be built in practice. The Figuera device is such a device.PmgR"

[listener192]

Attached are few scope shots of a discrete switch running in boost mode.
The switch is driving 2 stator coils connected in inverse parallel, rotor in registration no load on rotor.
10.5V DC input

Not shown, the coil current is actually over 8A due to the boost configuration.
The recovery current/power is shown. Note: use average power reading.
In a working system, 10 pairs of coils would be powered at any one time, which in this example would be over 600W.
You can see this configuration has no problem achieving high power levels.

I did try to use the BTS7960B H bridge boards, with modifications however the voltage rating of the MOSFET's in the devices was not high enough and this type of converter is prone to switching node transients.

Tests conducted at 20V DC input, the  last photo shows voltage at the node, after a snubber was added immediately after the recovery diode.
Originally, the transient was 100V in this example, now reduced to 60V. This really needs to be reduced further perhaps by slowing the switching speed down, trading a little heat in the MOSFET.


The due to track inductance and working space difficulties, I was unable to reduce the switching transient on the BTS-7960B lower than 40V  @10V DC input, which is too close to the maximum rating of the device. The over voltage protection also proved to be problematic.

A discrete H bridge design is the only reliable way forward.




L192

[listener192]

The first photo shows the improved rising edge transient suppression after snubber addition.

Now only a 10v overshoot at 25V input, so a 60V device with a low RDS (4 mohm) can now be used.
This will show a current drive improvement over the 28 mohm 200V MOSFET used in this example.
An H bridge built with 4 of these MOSFET's will have a combined RDS of 8 mohm plus, the isolation diode volt drop.

The second photo shows another problem that you can get with boost converters, falling edge ringing. Some MOSFET driver devices do not like the negative excursion.

Pmgr suggested  another diode is added in the source to ground path to stop this, although this will add to effective path resistance  and limit current.

The driver currently used does not exhibit any problem with this -4 volt excursion, however I do have 18V zener protection between gate and source for extra protection.

Attached is an experimental circuit to determine if a discrete H bridge boost converter can be constructed from these cheap Chinese boards.
The intention is not to use P channel MOSFET's for the HHS, as these have higher RDS than N channel devices. By maintaining very low device RDS, it will not be necessary to employ large heat sinks.
The boards need some modification, with removal of some parts and addition of others.. with some external diodes etc.
The two HSS would have complementary switching, 15 arduino outputs go to one HSS and same lines are inverted via TTL and these go to the other HHS. These only switch once every half cycle. The LSS's are controlled by individual arduino outputs (30), 45 outputs in total.
The floating supplies I have used are rectified & regulated 15V rails obtained from individual winding's on a central pot core inverter stack. This allows for a compact low cost design.

L192 

[listener192]

Attached are current waveforms for two sets of anti parallel coils (180 degs spaced on stator)25V DC input voltage.
The coil sets are momentarily overlapped, no load on rotor.

Shown..Input currentcombined coil currentcombined recovery current.

Note the recovery duration for the last coil  is greater however, this coil turns completely off, something that wont happen in the complete circuit.The first coil recovery represents what would be achieved by each coil pair turning off.
Note current is a respectable 15A peak.

L192