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

[verpies]

I don't know about the C2 resistance and why I gave you 2 different samples in case one is better then the other.

I had in mind the quality of the capacitor - not only its capacitance.  Itsu is using a polypropylene cap, which is an excellent pulse cap.

As for the diode, did you not see the quality of diode I'm using?  Ultra fast RHRG5060 paired which gives a 0.34v forward voltage.

I did not check its specs. However 0.34V is not very much, indeed.

This is the second time you mention wiring starpoint!... I never heard of this? what is it and can you show me a picture of what it would look like.

Similar to this one

The MOSFETs are IRFP4310Z  120A and RDS 8.4 milliohm resistance, so I don't think this is the problem.

A good 100V MOSFET with low RDS_ON and large 6.8nF gate capacitance.
Let's see its gate-to-source voltage waveform (not simultaneously with the other waveforms because of scope ground hazard)

I can order some CD4047
Can you provide the complete circuit so I can make sure I order everything needed at the same time.

Give me some time to build it so I don't make a mistake in a purely theoretical schematic.
But the other parts I expect will be non-electrolytic caps, multi-turn trim pots, which you probably already have.

Could a on the fly pulse width adjustment capability be added to this?... and if so, what components are need to be ordered for that option?

Yes - that is half of the purpose of such circuit.

I have 2 of the below outrunner motors. Do you think a circuit could be built to power it and collect the flyback to power the second one which would be attached to the same shaft?

I don't know offhand, I'd have to think about it.

The magnetic circuit can be source of large loses, too.
I don't remember what coil's core and magnets Itsu is using but if it is a ceramic ferrite and ceramic magnets then this would make a difference because NdFeB magnets are internally conductive and allow Eddy current to develop inside them just like in steel.

[itsu]

I don't know? the scope I have access to is a TDS 2024B which by the way has very low resolution screen shots captures compared to your and why I started taking pictures of the screen to show more details.

Luc

Luc,

I use the Tektronix "OpenChoice Desktop" free application to capture / display my screenshots which can be found here:
http://www.tek.com/oscilloscope/tds210-software

I hooked up my scope via its ethernet connection, but you can do it with your USB connection too and get the same screenshots i do.

Itsu

[itsu]

Hi itsu (and Hi to all people who do decent measurements like itsu):

To find out if the energy in the spikes can be "fed back" to save power, one should do the following measurement:

1) You do the input power measurement with the motor running like you do in your video https://www.youtube.com/watch?v=dANzzD6X75E&feature=youtu.be

2) But then you do not stop the motor, but you disconnect the path to the capacitor (by disconnecting the diode). This will prohibit that power is fed back from the spikes. And now you measure the input power. Is it less or more?

I tried that several times with different pulse motors and different drive circuits. I never could prove that input power to the motor was less if one tries to feed back power from the spikes (from the back EMF of the drive coil).

I think that the power in the spikes is forever lost.

Greetings, Conrad

Hi Conrad, 

i did that test, but there was only a very small increase in input power after removing the diode (177mW to 180mW), but i don't think that the present circuit that i am using
has any capability to feed back any spikes.

Itsu

[partzman]

Partzman,

my understanding is that your "red Math Mean -125.8mVV" value is the mean value from the total screen (so the 10 horizontal divisions).
So not the the Math mean value of the area between the 2 vertical cursors which is what the question was.

You presently have displayed the values of the area between the 2 cursors of the yellow trace as amplitude and timing (upper right corner).
If you now activate the red math trace, in that upper right corner you will see the data of that red math trace (wattage and timing) inbetween the 2 cursors, but
i doubt it will be the calculated power delta as it does not match my full screen calculations when trying, but i will double check.

This might be interesting for Luc also as he has a similar scope.

Itsu

Itsu,

OK, I may be misunderstanding you but the measurement shown on the previous scope shot is the mean or average of the Math waveform between the vertical cursors and not the full screen width. I've attached another scope pix taken from the same test but with the cursors re-positioned to show the mean of CH2 which is 234.8mv but also notice the Math mean of -82mvv.

You can use the vertical cursors to define any measurement window in time that you desire with any of the available measurement resources on your scope. It is really powerful IMO!

partzman

[verpies]

2) But then you do not stop the motor, but you disconnect the path to the capacitor (by disconnecting the diode). This will prohibit that power is fed back from the spikes. And now you measure the input power. Is it less or more?

Please remember that we are measuring energy expended by the power supply only during the ON-pulse period and the "flyback" spike occurs outside of the ON-pulse period, so anything we do outside of this period will have minimal influence on this energy measurement.
During the 100μs that the current flows from L1 to C2, the L1 remains an electromagnet and is capable of attracting the rotor.
Once all of the energy is transferred from L1 to C2, the electromagnetic influence on the rotor stops, after which the energy stored in C2 is simply wasted in R2.

I foresee, that once we make sure that Itsu's energy recovery efficiency numbers are not a result of some measuring error, then we will try to loop the energy stored in C2 back into the power supply (...or L1).