Self accelerating reed switch magnet spinner.

[TinselKoala]

Sense coil and drive coil (mosfet drain) signals.

The horizontal timebase is at 2 millisec/div. so the frequency shown is about 150 Hz, times 15 = 2250 RPM.
Channel 1 is the mosfet drain, DC coupled at 50 v/div, and the spike is off the screen at over 250 volts. Zero volt baseline is the center graticule marker, also indicated by the mark on the right.
Channel 2 is the sense coil signal, AC coupled at 1 v/div. Since it's offset by 6 volts (the 22k - 22k divider puts it at half the switch Vcc) the true baseline is off the screen bottom. The 6V level is the "baseline" indicated by the mark on the right of the screen.

ETA: The reduction in the ringing waveform is due to the fact that I have a core (a 1/4" steel bolt) inserted in the drive coil. This increases the RPM slightly but damps out the nice ringing oscillation completely if I insert it far enough.

Look at how accurately the comparator detects the "zero crossing" or rather the crossing of the 6V level by the sense signal! It is very precise.

[TinselKoala]

It would be nice to have the autostrobe work properly. But flashing on every magnet passage makes it flash 4 times per revolution, so you get a blur instead of a frozen image like you need. One flash per revolution will give the strobe effect properly.
So... digital logic, divide-by-four pulse counter? Or is there an analog way, perhaps using an RC circuit?

I can do it pretty easily using a 4017, I think.

[synchro1]

@TK,


       Unquestionably the highest state of the art in pulse motor circuitry to date. Congratulations on perfecting duty cycle control so elegantly. The speed up coupled with increased back power out defies customary logic. I wonder if a loop back to source would produce a self runner? The 250 volt back spikes are awesome! I bet JLN will bench test a replication.

[MileHigh]

TK:

It would be nice to have the autostrobe work properly. But flashing on every magnet passage makes it flash 4 times per revolution, so you get a blur instead of a frozen image like you need. One flash per revolution will give the strobe effect properly.
So... digital logic, divide-by-four pulse counter? Or is there an analog way, perhaps using an RC circuit?

"Well, how low can you go?" - asked the Limbo Man.  How about an ultra low tech solution for a high tech problem?

The top of your rotor drum looks like there is silver gaffer's tape on it or perhaps metal foil tape?  Perhaps you can remove it to make a clean surface.  Then I would suggest that you put just one dot lined up over one of the magnets.  Perhaps do it with White-Out correction fluid or a White-Out marker.  Perhaps even just a tiny piece of white paper glued in place.

That's it!  There is nothing else to do!  Just mark the TDC on the top of your red rotor for one of your four magnets and you are done.

A piece of cake!

MileHigh

PS:  Supposing that you put a white dot on the top surface of the drum at the outer edge.  Then if you wanted you could put another dot over the next magnet at TDC but this time it's not at the edge, you move radially inwards by a little bit.  Repeat the process for the remaining two magnets.

Now look at what you have created:  You have four separate and distinct "arcs" or "stripes" that you can see when the LED strobe illuminates the top of the spinning rotor drum.  So that means you can actually see what the timing looks like for the pulse firing for each individual rotor magnet.

[synchro1]

Quote from MH:

"That's it!  There is nothing else to do!  Just mark the TDC on the top of your red rotor for one of your four magnets and you are done."


Not so fast! All that will do is show the dot moving around all four different quadrants, not freezing at one!