Scope shots of clean kicks

[bob.rennips]

Using Jason's driver circuit - driver TC4427/MIC4427 into mosfet IRF840.
No diode across the coil.

Scope probe set to x10, is attached to 1 side of the coil.
Coil is 1cm diameter.
Number of turns of wire is irrelevant, as is gauge of wire.
Core - I've tried both with and without a magnet - both times I can tune to a specific freq. to give a nice clean kick.

12V+ goes to coil and then to MOSFET. The mosfet is switching to Earth (negative).

Dual trace scope was used.
Top trace shows the input square wave - 5V/DIV
Bottom trace is the coil probe - 50V/DIV because of x10 probe.
Timebase is 2microseconds.
Frequency to mosfet driver is 146khz.

1st image - single kick each time the coil goes off. No real noise on input or output. Peak goes way off the top and is approximately 650V.
2nd image - same as 1st image except x10 magnification button is pressed for TIMEBASE on scope. This allows the detail to be shown. As you can see there really is no noise nor ringing.
3rd image - this is a typical scope shot before getting to the correct frequency - plenty of noise and oscillations.
4th image - hardware shot showing the coil and probe placement.

I think the kick SM talks about is the BEMF but at a frequency where the kick is nice and clean, as per my scope shots.

NOTE: The internal diode in the IRF840 stops the kick getting anywhere. This means coil connections must be carefully connected to ensure kicks are routed to other coils before going back to the mosfet - if this is what you want.

In the coil shown I removed the magnet and wrapped another coil through the existing coil. i.e. A coil wrapped at 90degree. This coil picks up the kick (100V peak) BUT NOT the turning ON of the coil.

However, if you feed this kick to 1 coil of a bifilar coil and have another pulse from your pulsing circuitry occuring at this time, then the kick gets added to the pulsing pulse to generate a larger pulse.

So how do you time the second pulse, to coincide with the kick. Easy! You just need a second pulse at twice the frequency!! This time you get a large kick. And if you feed this into another bifilar coil and pulse this coil at the right time... and again you would need to pulse this coil at 4 times the initial frequency. Which is very near to what SM said to do. Feed this pulse back into the 1st bifilar coil and you now have a runaway feedback in place. WATCH OUT!! It will also have a rotational magnetic field. This is what I'll be trying next weekend - some of us have to work!

Cheers, Bob.

[bob.rennips]

I've attached a schematic diagram of how I think the coils should be connected.

Each coil set is a bifilar coil (control coil) and wrapped around this at 90 degrees is the kick pickup coil (collector coil). In the diagram I've shown them as three seperate coil units, which should be placed 120 degrees apart around a circle.

Now consider how these would be combined together to give an SM style TPU.

1. The three kick pickup collector coils are stacked vertically on top of each other.
2. Around each collector coil is one bifilar coil placed 120 degree apart from each other.

This is EXACTLY what SM has described and is identical to what we are doing in the schematic. It will create a rotating magnetic field.

3. Wrap toroid style another coil to intersect the rotating magnetic field for the DC output.


I've shown the example using 40, 80 and 160Hz. In fact any combination of frequencies will work where the off coincides with an ON in the right order. You could use only two coil units rather than three and just invert the original turning on signal. Or you could have 4 coil sets and have 2 pair turning on at the same time on opposite sides of the circle. Again you'd only need the one signal and use the invert to turn the other pair on.

The 3 coil set I've described requires that the frequency and hence the width of the final pulse be smaller than the width of the kick pulse in order to get a runaway situation. When this condition is met the kick is still rising when the original 40Hz signal turns on - which obviously completes the feedback loop. This means if the 80 and 160 Hz signals are generate by a divide by 2 circuit, you only need one master frequency. And as long as this frequency is kept slightly off tune (where have I heard that before!) and you control how much of the kick from the 3rd coil set is fed back into the first coil set, then everything is kept under control.

[Earl]

Bob,

you are thinking along my lines.  Could you please graph three 40 Hz traces, each offset 120 degrees, followed by first and second harmonics of each of these three.  Nine traces in all.  Would like to see how these look like when 9 traces are lined up vertically.

If you could also do this for the 4-coil arrangement, with 4x times 40Hz, offset 90 degrees, with 1st & 2nd harmonics of these four, total 12 traces, I would greatly appreciate it.

Good week-end research, keep it up.

Regards, Earl

[Earl]

Bob,

I think I can draw up a schematic of a simple circuit to do the pulsing at the correct time, with no need to do coil relaying..  Will try to do it this evening.

Even so, keeping 90 degree coils, even tri-filar excitation coils allows more and diverse experiments and measurements.

Regards, Earl

[Grumpy]

MC14521B frequency divider - cost about 75 cents.

(could use a quad flip flop also, if you only need three freqs)