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

[gotoluc]

Dear Itsu,

I have been testing all the coils I have and cores like metglass, steel laminations, welding rods and so on.
I have carefully tested them and noticed thing that I have noticed and said in the past.
Not surprisingly the better results were had with the coils of low resistance but with a super low duty cycle like 1% (if coil is below 1 Ohm) but use as high of a voltage as possible but obviously stop when the cap reaches max voltage and drop it maybe a volt or two.

If you can redo your test at 10Hz instead fo 40Hz which should be easier to sync your rotor and drop your duty cycle and increase your voltage as much as you can all while keeping the rotor turning, I would be prepared to say the results could be even better.

Thank you for your great work

Luc

[tinman]

I have a lot of faith in Verpies and Itsu.  Yes I think an error of over 100% has been made and I am willing to bet you that Verpies and Itsu also believe that an error has been made.  My assumption right now is that they will continue this investigation and find out where the error was.  Things like this have happened many times in the past and they are bound to happen again.  I will just repeat that you are jumping the gun.

Let's see what happens.

MileHigh

I agree MH
There are two very bright people working on this, and yes-if there are errors, they will find them.

Itsu
Are these results still with the rotor in play within the system?
If so, what are the results without it in play?

Brad

[verpies]

How about trying another cap of the same value, or just trying another cap value completely, perhaps between 2X and 10X larger?  You get the new peak voltages, do the numbers still crunch to "over unity?"  I realize that you want to keep the cap voltages high to minimize the effect of the diode drop on the measurements.

Not only. Larger cap also lengthens the energy recovery period and that has a negative correlation with the L1 discharge efficiency, which decreases with longer discharge periods (times that are larger fractions of its Tau). 
To accommodate 10x larger cap while keeping the efficiency, the inductance of L1 would have to be increased too.  But I think we could get away with a 2x cap.

First of all, disconnect the MOSFET cap discharge circuit. 

I don't think the MOSFET cap discharge circuit has been implemented yet.
The latest scopeshots were made just with a 10K bleeder resistor in place of Q3, because Itsu found out empirically, that this resistance discharges C2 to 0V just before the next pulse begins.
I think Itsu's next version of the circuit will use a choked reed relay in place of Q3.

...my instincts would be telling me that the problem would be in the measuring the energizing of the drive pulse itself, even though at first glance that also looks pretty well done and quite straightforward and not prone to problems.

Do you mean something like scope probe attenuation set incorrectly in the scope's menus ...or the probe's 10x switch not making a good contact?

Also, if the CSR had resistance of 0.01Ω instead of the expected 0.1Ω, then the ohmmeter would not catch the difference and that peak ~600mA L1 current reading would really mean ~6A.
...but the problem with that theory is that Itsu has used his Hall current probe and the current readings were comparable with those given by the 0.1Ω CSR.

If you think of a good method to triple-check the L1 current waveform - I am sure that Itsu will be open to it.

[verpies]

Not surprisingly the better results were had with the coils of low resistance but with a super low duty cycle like 1% (if coil is below 1 Ohm)

Yes, and your coil appears to have superior L/R parameters compared to Itsu's, e.g. your coil's current is a straight up-ramp and it doesn't flatten like Itsu's current at the end.

[tinman]

Yes, and your coil appears to have superior L/R parameters compared to Itsu's, e.g. your coil's current is a straight up-ramp and it doesn't flatten like Itsu's current at the end.

Verpies
Once the current ramp flattens out,this is the point of core saturation-is it not?
If so,then it is possible for Itsu to reduce the current flow time,and switch off before or just at the point of saturation,and gain even a higher efficiency.
Do i have that correct?


Brad

P.S
Do you think that these results have anything to your theory you told me about in an Email some time back?. I believe you were in hospital at the time.