Coilgun circuitry (massive staging architecture) / Electromagnetic Projectile Accelerators / Forums

Forums

4hv.org :: Forums :: Electromagnetic Projectile Accelerators

« Previous topic | Next topic »

Coilgun circuitry (massive staging architecture)

1 2 next

Move Thread

LAN_403

BigBad

Mon Apr 13 2015, 04:35PM

Registered Member #2529 Joined: Thu Dec 10 2009, 02:43AM
Location:
Posts: 600

OK, so an ideal circuit for a coilgun would have:

a) the ability to drive *lots* of stages
b) put high(ish) voltage across the coils on and...
c) reverse the voltage across them to turn them off again ASAP
d) carry and be able to drive lots of current
e) computer controlled (arduino or whatever)
f) as cheap as possible and no cheaper ;)

OK, so my idea to do this is:

a) have an inverter (a 2 phase inverter), That's 3 half bridges with IGBTs or FETS. powered from an array of electrolytic caps.

b) a big crosspoint switch using well snubbed TRIACs for each row and column, one for each phase. So 12 TRIACS can handle 6*6=36 coils, 24 TRIACS, 12 for each phase = 72 stages.

TRIACS have decent current and volts. You can also run them in parallel to double their current capacity. They also switch themselves off when the current goes to zero (if they're well snubbed); and you can put negative volts through them from the inverter, which means the coils can be switched off much faster.

I've actually already got this circuitry running for another application, with less TRIACs, so I basically know it works.

Comments?

hen918

Mon Apr 13 2015, 05:56PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

what do you mean by switching the coils off by applying a reverse voltage? surely that would just reverse the field and the force would still act on the projectile.

BigBad

Mon Apr 13 2015, 06:05PM

Registered Member #2529 Joined: Thu Dec 10 2009, 02:43AM
Location:
Posts: 600

No, when you switch a coil on (put volts across it to ramp up the current) you have to switch it off again shortly afterwards to avoid suckback. If you use a half bridge, then you're applying zero volts across the coil and the current decays. It's much faster to apply negative volts across the coil until the current decays through zero amps.

If you're powering it through one or more TRIACs, when it transitions through zero current, the TRIAC will automatically disconnect it (provided your snubbing is working).

hen918

Mon Apr 13 2015, 06:52PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

wouldn't that cause a massively high current to flow through the TRIAC (about four times that of the current caused whilst the coil turns on)? wouldn't this waste twice as much power than letting the voltage die away?

(if I am understanding correctly)

klugesmith

Mon Apr 13 2015, 07:08PM

Registered Member #2099 Joined: Wed Apr 29 2009, 12:22AM
Location: Los Altos, California
Posts: 1714

You might want to reconsider assertion a). Complexity carries a cost.
I wouldn't design for more than, say, 5 stages
before laboratory evidence that I could do better with 5 than with 4.

How are you going to manage the gate drives for an XY crosspoint in which the voltage between X and Y rails reverses? It might be enlightening to simulate a 2x2 case, even though that uses just as many switches as 1x4.
I have used a Spice model of basic SCR (made out of BJT's), found on the Internet. You might be able to find something similar for a basic triac.

What would triacs do that you couldn't do with SCR's, if the current pulse in each coil is to be unipolar?

BigBad

Mon Apr 13 2015, 07:24PM

Registered Member #2529 Joined: Thu Dec 10 2009, 02:43AM
Location:
Posts: 600

hen918 wrote ...

wouldn't that cause a massively high current to flow through the TRIAC (about four times that of the current caused whilst the coil turns on)? wouldn't this waste twice as much power than letting the voltage die away?

(if I am understanding correctly)

No, the inductors have huge inductance, so the current doesn't change at all quickly; it just can't it's an LR low pass filter.

What does happen is you get voltage swings as the inverter half bridges switch on and off, the Ldi/dt creates voltages.

But inverters get that anyway, that's why there's beefy reverse/clamp diodes across the FET/IGBTs. As they conduct they recharge the diode on the DC bus, so far from wasting power it actually puts it back where you want it.

klugesmith wrote ...

You might want to reconsider assertion a). Complexity carries a cost.
I wouldn't design for more than, say, 5 stages
before laboratory evidence that I could do better with 5 than with 4.

It turns out it doesn't matter a lot, the TRIACs are the inexpensive bit. It turns out that I already have a board that could drive 9 coils.

klugesmith wrote ...

How are you going to manage the gate drives for an XY crosspoint in which the voltage between X and Y rails reverses? It might be enlightening to simulate a 2x2 case, even though that uses just as many switches as 1x4.
I have used a Spice model of basic SCR (made out of BJT's), found on the Internet. You might be able to find something similar for a basic triac.

Yeah, I've already modelled them, and even built them. It wasn't intended for this, but I just realised it would probably work.

klugesmith wrote ...
What would triacs do that you couldn't do with SCR's, if the current pulse in each coil is to be unipolar?

They're cheap and fairly easy to drive, and they're pretty robust.

hen918

Mon Apr 13 2015, 07:40PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

what you have, in effect, is an inductor and a resistor in series. When the power is cut to the inductor, the voltage will swing negative anyway, applying a negative voltage to that is only going to exacerbate it.

BigBad

Mon Apr 13 2015, 07:53PM

Registered Member #2529 Joined: Thu Dec 10 2009, 02:43AM
Location:
Posts: 600

You get twice the volts across the inductor, so it switches off twice as fast.

I also put some reasonably huge clamp diodes on the side of the triac that's connected to the coil. Otherwise I found I had some nasty resonances with the snubber and other potential issues during turn off. They don't need heatsinks, but they take large currents.

hen918

Tue Apr 14 2015, 04:52PM

Registered Member #11591 Joined: Wed Mar 20 2013, 08:20PM
Location: UK
Posts: 556

BigBad wrote ...

You get twice the volts across the inductor, so it switches off twice as fast.

I also put some reasonably huge clamp diodes on the side of the triac that's connected to the coil. Otherwise I found I had some nasty resonances with the snubber and other potential issues during turn off. They don't need heatsinks, but they take large currents.

sorry if I sound obnoxious, but if you have diodes in parallel with your TRIAC, you might as well use thyristors.

BigBad

Tue Apr 14 2015, 05:12PM

Registered Member #2529 Joined: Thu Dec 10 2009, 02:43AM
Location:
Posts: 600

Perhaps, there's certainly a million ways to do this kind of thing, although the diodes are just 13 pence each.

I had a board fabricated to hold the circuitry, and it held 6 TRIACs. It turned out that the board was about 80% of the cost.

In any case I actually needed the TRIACs for my other application anyway.

It may well be possible to use SCRs instead, and it might or might not be slightly cheaper, but probably not a lot different.

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob