Sharing ideas on how to make a more efficent motor using Flyback (MODERATED)

[Magluvin]

Hi Mags, thanks for your post. You're right, the frequency drive controlled motors would not work for what I'm suggesting.
I think the motors I was thinking of is the BLDC which have hall sensors to trigger the 3 mosfets.


Luc

Hey Luc

Well the bike  hub motor is a bldc and uses halls to detect timing. Just really getting a hang on how they really work. The outrunners use self detecting driver modules that dont need halls or other timing circuits from what I understand, like my bike motor could probably run on one of those driver circuits and eliminate the halls.

What I want to try is to separate the windings so they are not together at all. And possibly be able to capture what isnt there in their normal config by having separate drivers for each independent coil.  This may involve only having the coils operate at 1 polarity and shut off while the others do their thing and so on. Like my Tidalforce bike motor, it has 7 stator coils, but none of them are connected like a typical 3 phase as shown earlier. They have their own separate drive circuits, kind of. Will explain later as I get into that. 

Like Brad was saying, and Im thinking he is correct, when the transistors of my hub motor fire a coil, when the power is cut, the circuit immediately switches polarity to the coil, not giving bemf a chance to be captured.  Thinking about it most of today, the whole pwm while the throttle is less than wide open, there are transistors that connect one coil to neg or gnd zero as it is 48 v dc in, and the plus side transistor does the pulsing. Then when time for a phase shift, another transistor grounds the other end of the coil and yet again another transistor applies plus to the other end of the coil. So there is a lot going on there in the 3 phase setup. 

So what I want to try to do is change the circuitry to allow bemf capture. It may require a lessening of performance due to off periods described above while the others work.  With all that pulsing going on during one phase, im seeing a basic switching supply that should have bemf or field collapse output capabilities, while driving the magnet end of the motor. 

Its nice to work with well made consumer/commercial motors, if viable to changes I am suggesting, because we are enabled to already have a very good base to begin with, tolerances and all.  Like Brads rt, it would take a bit of work to make the motor exactly like that, when it can be had readily for little investment. Plus, having nice tight tolerances for field paths should enhance the bemf output like a decent switching supply.   

I have a 3 phase automotive cooling fan motor that has a bell rotor with 3 mags inside the bell and a stator that is inside the bell with 3 coils. As basic as it gets. Im going to disconnect the windings from each other ad see what I can come up with there as a test bed. Has heavy windings and has a small magnet wheel with alt poles to trigger 3 halls. 

What you have shown with the pulsing and loading the motor without increase in has inspired me.  It should be a goal to expand on that and test the limits and such. Like would there be an increase in if the motor were not allowed to turn at all while pulsing. If so then I am inspired more.

Mags

[verpies]

As i stated before Verpies,it would only be you of the EE guys that would take note of what others have to show when it go's against these known laws--oh and the books lol.

I would like some other EE to verify:
- my calculations,
- measurement methodology,
- conceptual errors.

Seeking a second opinion is good science.  It not wise to rely on the expertise of only one person.  I am only human.

[itsu]

It sure does.

Going by the zoomed scopeshot:
Input energy during the ON-pulse:  E₁ = 4.070W * 4.36ms = 4.07 * 0.00436 = 0.0177452J = 17.7452mJ
Recovered energy in C2:  E₂ = ½ * 10μF * (99V)² = ½ * 0.000010 * 9801 = 0.049005J = 49.005mJ

E₂ / E₁ = 49.005mJ / 17.7452mJ = 2.76159 = ~276%

These welding rods is good stuff !


P.S.
The sawtooth current energy estimate yields:  E₁ = 660mA * 12.4V *4.36ms * ½ = 0.660 * 12.4 * 0.00436 * ½ = 0.017841J = 17.841mJ  ( pretty close ! )

Hmmm,   so where is the flaw, as there must be something wrong, Right?

Guess i need to empty the C2 cap before each new spike capturering, so will try first with the reed, then with your
nice circuits using a MOSFET / driver if the reed does not work.


Itsu

[itsu]

The one you use "green tab" give me the below. The red tab give me the next one with no data.

So by using the software I don't get anything better. It's the scopes limitation compared to yours which must be better.

Luc

Luc, seems strange to me as your scope is newer (i still have a 1.44MB floppy and 2007 Firmware).
Perhaps a scope Guru can step in and advice a different setup.

Itsu

[verpies]

Hmmm,   so where is the flaw, as there must be something wrong, Right?

So far I cannot detect any mistake of mine.  That's why I'd like somebody smarter to check me.  Cyril Smith (a.k.a. Smudge) has not commented yet.

I have received one constructive criticism via PM, namely that the voltage drop across the CSR affects the C2 voltage as measured between the points B and D.
This is a valid observation, because you are not measuring this voltage directly across C2 (between +V_1A and point D), however at ~1A currents, this error is very small, approximately ~100mV for C2 peak voltages around ~100V ...so the error is only 1:1000 or 0.1%
I was aware of this quirk when I designed the scope probe placements.