Flux transfer coil calculations

[hallkbrdz]

Although I have a background in electronics, I never had experience with designing motors and how to calculate the amount of turns of N guage wire at X voltage and Y current to produce Z Gauss for a fast pulse to produce flux transfer.

My initial target is a set of control coils to produce 14,200 Gauss with a MicroMetals 40 Mix Iron Powder Block Core. Wire size, voltage and current (within rational mosfet ranges) are open. Switch time ideally would be 620 ns or less (single pulse). Core ID 3"x"1/4" at present.

Initial rough prototype idea of what I'm thinking about building...

 

[fechter]

You might try this calculator:

https://www.omnicalculator.com/physics/solenoid-magnetic-field
or
https://www.accelinstruments.com/Magnetic/Magnetic-field-calculator.html

It's the amp-turns you need to calculate. You have flexibility in the actual number of turns as long as the amps x turns is what you want.

620ns seems like an incredibly short pulse. The inertia of the rotor will make it hard to notice.

 

[hallkbrdz]

Thanks.

I hope I can get that much shorter. I can enlarge the motor and make the rotor iron wider, which will lower that if necessary.

I've been playing with FEMM, such as it is, to get some idea of what geometries make the best magnetic sense. Wider and short seems to be the key for this type of an arrangement. I plan to start with a linear path since it's easier to construct and then compare the outputs.

 

[fechter]

Is that some kind of switched reluctance motor?

If you want shorter pulses, you want fewer turns to reduce the inductance.

 

[hallkbrdz]

If everything works in testing, eventually it will be a switched flux PM motor. Magnets and coils in the stator with an iron rotor. While you cannot "turn off" a PM, you can use coils to concentrate the flux either on the keeper end or the rotor end, nearly doubling the Tesla value when doing so. The coils only need to pulse N/S or S/N in relation to the PM to make this happen, they don't need to stay on since the magnetic forces want to keep "flowing" the same direction once changed. I haven't figured out how to get the software to do this, so the forces shown here are about 1/2 of what they should really be.

This is based on the work by Raymond Radus. Charles Flynn took that a step further with the parallel path magnetic circuit, actually leaving the coil energized the entire time for even more power, although with ceramic magnets. However, I think the simpler single magnet Radus pulse version is best suited to minimize the battery requirements for a given power output using N52 magnets.

 

[hallkbrdz]

To flip the flux direction I need to quickly energize the coils. That needs lots of voltage (cheep battery and wire size compared to current).

What does that today? Coilpacks on a combustion engine. They produce strong, quick pulses to fire spark-plugs. Timing of the example circuit below allows for up to 400Hz operation between 2ms and 500ns. Here I could use a similar arrangement (I believe) to instead fire the coils next to the magnet, staggered such that one makes it N/S, and another quickly following as S/N. At high speed, the second may be redundant since when the coil fields collapse they automatically reverse - I'll just need to remove the reverse bias diode circuit at that point.

Still... a neat concept but definitely an early work in progress. I have a lot to learn.

https://www.instructables.com/Ignition-Coil-High-Voltage-Display/

 

[fechter]

I found this description of the Flynn motor:
https://www.flynnresearch.net/technology/PPMT Technology.htm

Not sure I totally buy the concept, but if it works, great.

Seems like you'd want to discharge a HV capacitor into the coil to get the pulse. Use IGBTs that can handle the voltage. The size of the capacitor can be chosen to reach the desired peak flux.

 

[hallkbrdz]

Note, although it is using a similar idea to Flynns control coil use, I'm using a quick spike to change flux direction, not drive the coils "continuously". This is much more like how Radus setup the astronaut boots.

I've done some more research on how to drive the control coils next to the magnet. What I need to build is a MOSFET H-Bridge with microcontroller control, an automotive ignition coil, and the set of control coils. The control coils connected inversely to each other and in parallel with the ignition secondary. Primary voltage can be low, 12-14v with low amperage. Secondary... ?? 30 or so KVs with almost no amperage.



Instead of this type of circuit driving a DC motor, it will be used to control the charge and release of the primary of the ignition coil. M1 and M4 on, M2 and M3 off, will "slowly" charge the coil one direction. M3 and M4 on, M1 and M2 off, will very quickly drain the coil causing an induced voltage spike on the secondary (just like a modern engine ignition circuit). That spike is also repeated on the control coils, changing the direction of flux. M2 and M3 on, M1 and M4 off, will then charge the primary in the other direction. M3 and M4 on, M1 and M2 off, will again very quickly drain the coil causing an induced voltage spike on the secondary, and another change of direction of flux. Then the cycle is repeated.

I think the concept works on paper, but as they say, proof is in the pudding.

 

[hallkbrdz]

Scratch that... concept is good but that circuit won't work because the primary charge time is too slow. Just activating the coils directly works better, at least in simulation.

If nothing else, this is fun and is helping me to get back into electronics after working too long in IT.

 

[fechter]

The flux in the core will be a function of amp-turns and also time in the case of a pulse. Getting the flux up to the saturation point of the core, or close to it, is the most you need to shoot for. You can pick the number of coil turns. To get the flux to go from zero to saturation is going to take some time due to the inductance of the coil. There's no way to make it go to near saturation instantly. If you use a lot of coil turns, you need a higher voltage, but you will also have a lot of inductance. I'm sure there's a way to do the math on the rise time for various coil geometries. My guess is it would be better to use fewer turns and higher current vs. many turns and a higher voltage. For one, it makes the driver circuit easier to build since you can use a lower voltage.

It may turn out that regardless of the turns count and voltage, that the time it takes to reach near saturation will be the same and will be much longer than 1ms.

I have a battery spot welder that uses FETs to switch a 12v lipo battery. Peak current is around 3k amps. I once made a coil from 12ga wire and about 6 turns (no core) and hit it with the welder output. The coil expanded itself and jumped off the bench. This is with pulse of about 8ms.

The thing I don't follow about the description of the Flynn motor is they say the magnetic flux "keeps going in the same direction" after the pulse ends, which makes no sense to me unless the core has residual magnetism, which would be lossy.

 

[hallkbrdz]

Good points, thanks.

"The thing I don't follow about the description of the Flynn motor is they say the magnetic flux "keeps going in the same direction" after the pulse ends, which makes no sense to me unless the core has residual magnetism, which would be lossy."

This is true. You can change the path of the flux with the coil and it will keep going the same direction after the power is removed. No, the core doesn't have residual magnetism - it's more like a stream of water that just wants to keep going the same way once changed, sort of having momentum in that direction. That is what I'm basing this all on, Radus called the effect flux transfer. See link below at the 20:15 mark:

http://www.pbs.org/opb/historydetectives/video/1513607701//

I'm going to try it with relatively low voltages and lowish amps with different numbers of turns of (what I have on hand) 19 AWG wire to see what works. I do need to pick up a digital o-scope so I can see the actual rise times. The 4-channel Siglent SDS1104X-E one looks like it may be good enough for my needs.

 

[fechter]

I think the link for the video is wrong. It was about space boots.

It seems like it should be easy to test whether this effect works or not by using a hall effect sensor on one of the pole ends. As far as I know, magnetism doesn't have any "inertia" and can change very rapidly under the right conditions. I might have to try this.

 

[hallkbrdz]

Yes, space boots, using PMs with coils. Same configuration for the motor design.

I know it is not inertia - but I don't know how else to explain the flux effect.

 

[hallkbrdz]

I now have a rough take for a starting place with experimentation. This uses MOSFETs to pulse the coil in one direction and then the other, using code to pulse the coils timed with the rotor and speed of course. The resistors to ground may or may not be required, but here they help to drain the coil quicker than the internal diodes in the mock MOSFETs in the simulation can. I also plan to use driver chips instead of the other transistors, but those aren't available in the simulator.

[youtube]sUAgFzgAKVo[/youtube]

 

[hallkbrdz]

The 1" N52 cube magnets showed up. This is the flux field with a magnet and iron mu metal.



However 7 turns of 19 gauge wire were not enough even at ~12v / 700 amps (bypassed resistor). However I did learn that even at that current level the wire does not even get warm when pulsed quickly. I need more turns of wire...

So I bought some thinner gauge wire and plan to try 50-200 turns, adding layers as I go. I'm still trying to find a calculator for inductance of rectangular coils on a non-air core so that once I find out how many turns I need, I can come up with a plan to keep the inductance as low as possible (parallel coils) to meet a < 33uS LR time constant. At 3TC that will electrically allow for a 10K RPM motor.

BTW, yes that is pink nail polish. Several posts suggested using that to bind coils, and that is what my wife didn't care for... Next time U channel and masking tape, as that take too long to set.

 

[hallkbrdz]

I took this still of the "space boot". It looks like there are about 17 turns and 5 layers, for ~85 turns in total. That's for what I would assume (60s) are larger ceramic magnets of about 1/2 the strength of my N52 cubes. So somewhere around 200 turns might be about right. My plan is to start high and then remove turns until it stops working.

 

[hallkbrdz]

Magnetism is an interesting study where only some normal electronics knowledge applies. This is a set of coils of 26 windings each, with 8 layers - all electrically separate and then connected in parallel. What I think should happen:

1. Resistance placed in parallel divides by their values. This proves true for the coil wires. ~0.45 ohms / 8 = 0.056 ohms measured.
2. Coils placed in parallel divide by their inductance value. Coils of about 88uH / 8 = 11 uH. At least when stacked on each other (physically) this proves completely false. 88uH / 8 = 82.9 uH... not what I expected at all! Now this still should be much lower than one single coil of 208 turns (26x8), but unfortunately not the 11uH or so I was hoping for. And low inductance is important to keep resistance down and the time constant low for optimum battery efficiency. Voltage wastes power and makes heat.

This will require much higher voltage (up from 6.2v) and require me to add back in more resistance (ironically) to reduce the LR time constant down to 270uS. Now with 25.6+vdc and an added resistance of 0.25 ohms I should produce about 207 lbs of pulling force in twice that time (2TC). Time to add the magnets and keepers and break out that second truck battery to test this theory.