EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: LightningPhil on February 20, 2024, 01:30:49 pm
-
Dear All,
Years and years ago I was into meddling with coilguns. There was a discussion regarding multiplexing switches to reduce the number needed to drive a lot of coils. I don't suppose anyone else remembers such? If so, answers on a post card :)
The goal is to launch 12 mm bearings, with many stages, for fun. It's catapult amo and is unlikely to go as fast as a catapult would launch it.
There are no plans to take this off the bench and will certainly keep the muzzle energy below the UK legal limit for a rifle. So safe and legal...
Main motivation is completing a project I dreamed of now I have the skill set to do it (and access to COMSOL). While using it as a good excuse to learn to use an FPGA to monitor projectile position, control the coil drive circuit and analyse the performance of each shot.
Since they have lots of pins... Keen on having above 50 coils, each 12 mm long. If each is driven by its own bridge, then that's 200 switches. But... the dusty memory suggested a multiplexed idea may allow a vast decrease in switches and associated drive circuitry.
While the max speed it could launch and still be legal is just shy of 68 m/s, keeping muzzle energy down to just a couple of J would also keep currents down and capacitor energy requirements also. This is for FPGA fun and ticking a teenage project off.. Something like 25 m/s would be great.
So if anyone has ideas of how to multiplex coils or info of the mysterious patent (which I can't find on Espacenet) that would be awesome.
Cheers,
Phil
-
Do you need to drive the coils with -ve voltage? If not, first candidate for switch reduction would be to go for an arrangement like a two transistor forward converter, which can only drive +ve voltage but is still able to recover the inductor energy back into the DC bus.
I thoroughly recommend isolated gate drivers for something like this, because the big switching transients will be all over the place, making ‘ground’ a moving target.
Something else I’ve always wondered about is a tuned LC approach, where each coil is fired by a single SCR / thyristor. Once fired there would be a resonant current rise, peak and fall back to zero at which point the thyristor turns itself off. Slight catch: the capacitor voltage would go from positive at start, to zero at peak current and then negative so you’d need to use film or ceramic caps.
-
Aaaand I've nerd-sniped myself.
Attached are some entirely non-optimised simulations to illustrate the point. Both deliver fairly similar current to the coil, but they use quite different methods.
Edit: for the resonant type, the next coil stage can just flip the polarity of the coil & SCR...
Edit: switches are totally wrong, I just used what I had lying around.
-
Thanks Jbb,
The two transistor forward converter is likely the way to go. And yes, only need to drive forward voltage and then turn off, ideally as fast as possible - which this is helpful with.
The multiplex circuit if I remember correctly operated like an array of two transistor forward converters, with coils aranged between them such that if there were N pairs of switches (forming a single effective converter with each pair), they could drive N^2 coils. Each coil has to have a diode in series to avoid being driven in reverse (I think). It's been about 20 years. Suprised I remember anything of it. Shall also try and draw it.
So with 8 pairs of switches, should be able to drive 64 coils.
Time to fire up LTSpice too :)
-
Attached is a circuit with 4 pairs of switches and 16 coils. Drawing more is faff... But should be fairly obvious how this can be expanded. Think I've remembered enough of the diagram from the patent now.. Still wish I could find it!
Anyhow, If it turns out it wasn't patented, then it can't be now ;D
So, with N (the number of switch pairs) being 3 or higher, more coils can be driven than there are switches. Rather keen on trying N=8 and going with 64 coils now.
We have to imagine gate drivers and a big fat capacitor involved in Pbus in the circuit, keeping voltage reasonably stiff during the chain of small discharges.
[attach=1]
-
Ah yes, that makes sense. Some care required about which order you put the coils in; might want to fire coil 2 even while coil 1 is still unloading energy back to DC link.
For top side switches I recommend you use isolated gate drivers with individual isolated power supplies. Costs a bit more but avoids a whole lot of headaches.
-
What voltages are you planning on using?
-
Ah yes, that makes sense. Some care required about which order you put the coils in; might want to fire coil 2 even while coil 1 is still unloading energy back to DC link.
For top side switches I recommend you use isolated gate drivers with individual isolated power supplies. Costs a bit more but avoids a whole lot of headaches.
Thanks jbb. I see you've overthought such things also ;) absolutely a good idea to use swing a bit of the magnetic field energy from one coil to the next as one turns off and the following one turns on.
Likely use the same insulated gate drives all around. Seems simpler. [Is there a thumbs up emoji?]
-
What voltages are you planning on using?
Good question. |I have a few 400 V, 1 A bench PSUs - wonderfully dangerous things! And some jucy electrolytics left over from an old project that are rated a bit north of that. So 400 V it is. I have a tube of IGBTs lurking about somewhere also. They're replacements for my onetesla coil that blew up a few times ;D
But might need more, so something like these should do the trick - FGH75T65SHD - can probably be pushed from the 150 A rating to double that or so for very brief pulses.
Will take a pause now as am on holiday.
-
No armamenets please.
-
No armamenets please.
Totally agree.
Coilguns are beyond useless as an armament :) Would need to be ridiculously long to compete with an actual gun - and then, well, it would just be a linear accelerator bolted to a bunch of benches in a science lab.
As mentioned, not aiming for speedy - just completing something dreamt up years ago. A few balloons might be put in front of it to demo it to my daughter before it's taken apart again.
For clarity, this is never getting off a bench an will perhaps be slightly more powerful than a pea shooter, but not much.
-
For a practical coil gun, it is the coils driving inductance that actually matters. Hence "remote" driving with lots of long wired that carry the high current drive pulse is very unlikely to work that well.
Given the low cost and very short pulse durations, i can't see individual drivers with local capacitive energy storage being that expensive, in fact, it's probably cheaper due to the reduction in high current (and hence large cross section) wiring required.....
Recovering coil magentic energy is also highly unlikely to be in any way worth while at these power levels as it will take millions of firing events to "save" the cost of the additional complexity components fitted