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

[listener192]

Attached shows progress on my new build.
Driver boards for MOSFET switches, and 30 x isolated 15V DC supplied for all HSS.
Boost H bridges qty 5 per plank.

First plank constructed, and tested remaining two planks will stack vertically on first.
Software completed.
Wiring to stator coils then follows coils in opposed parallel.
For current rotor pole width, I will be using 7 x 2 coils on at a time, 7 coils per pole.



L192

[gotoluc]

WOW L192... you sure give it all you've got

Thanks for sharing
Luc

[listener192]

WOW L192... you sure give it all you've got

Thanks for sharing
Luc

Hi Gotoluc

The idea is to make it fairly bomb proof.
These switches will handle all the current you want, so in the case of my 30V supply, that would be up to +/-30A through each switch. The MOSFETS are rated at 210A so the RDS is very low. The input current is limited by the high side isolation diodes @ 30A.

Recovery current is rated up to 30A (average) per bridge.

The input rail is up to 60V and recovery up to 100V (limited to 48V by the super cap bank).
If you want to go up in voltage then just change the LSS, for example to 100V. The HSS are isolated by diodes rated at 100V.

In practice, the boost voltage manifests as current drive to charge the cap bank. The cap bank then supplies current to the  much smaller value input cap via the 4 ohm resistor.
I may still be missing something in the architecture, but the building blocks are good.

The HSS's use the on board opto drives and are slow switchers and just control bridge polarity.

The LSS's are driven by Toshiba drivers that have a fast on time and a controlled off time to limit the voltage transient when you turn the LSS's off.

Some may wonder why I didn't just use charge pump MOSFET drivers? and the answer to that is they work OK when driven with a periodic waveform but when your duty cycle starts varying the LSS have to be switched sufficiently frequently to charge the capacitor to maintain a reasonable gate voltage. I wanted to remove his a potential source of problems ,so went for isolated drivers.
Still searching for a 10KW 3 phase motor (not Chinese), that will provide a stator thickness two thirds of the teeth depth.

L192

[bolt]

looks like a 1970s Moog:)

[FixedSys]

Here is the answer to the rotor flux problem.
My M16 steel stator was saturating at 1.5T(purple color on previous posted plots).A linear B/H plot from FEMM demonstrates this below.


If I apply a stator model equivalent to Pierres stator i.e. 33% pole length 66% solid steel, then 1.4T is achieved with only 5A through the coils, without any other input on the rotor.
The attached flux plot shows 1.4T or 2.8T swing with 1A of counter flux representing a 236W load.There is also more flux capacity on the stator and rotor.


Pierre's  stator is unusually thick and I think we will not be able to replicate a working device without the thicker stator.

No matter how the flux is generated, if the stator steel saturates, the flux will not be fully coupled to the rotor.

So the stator cross sectional area must be greater than or equal to that of the rotor? Why aren't the pole ribs a flux limiting bottleneck if they have the smallest area?

If I have digested your posts correctly, then the coils at either end of the rotor are used for EMF collection. So can the same results be achieved with 120 degrees of stator (9 coils) and a half length rotor, or even 90 degrees (3 stator coils) with extra coils also at both ends of the rotor? I ask because this could cut the cost of the electronics by 60 or 70 percent. It could also allow for simple fabrication of the array of stator segments. I'm thinking bars machined over the length with a bull nosed cutter then welded together in a radial arrangement so the coil field focal point is the end of the half length rotor.