Coilgun (kinda) for robots / Electromagnetic Projectile Accelerators / Forums

Forums

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

« Previous topic | Next topic »

Coilgun (kinda) for robots

Move Thread

LAN_403

cinexero

Fri Apr 06 2007, 01:18AM

Registered Member #362 Joined: Mon Apr 03 2006, 03:37AM
Location:
Posts: 4

Hey everyone, just to start out. Hi, my name is Andy, this is my first post. I am a computer/electrical engineering student at Georgia Tech. I have a design question that I have been beating myself up over for about 2 weeks now, and I was hoping somebody could help.

We (GT RoboJackets) are participating in RoboCup small size league this year. Basically, we play soccer with small (180 mm in dia) robots with a golf ball. The robots travel at ~5-7 m/s and the ball generally travels at ~15 m/s (35 MPH!!!). It is very cool to watch. It's like autonomous pinball.

We need to vary the amount of force we apply to the ball, so using a solenoid seemed like a natural choice. It takes about 5J to make a golf ball go 20mph (assuming no losses) so we are gonna run ~20J. I lifted a reference design for a photo flash capacitor charger from Linear (a LT3750 - very sweet part - I image it would be perfect for handheld or battery powered rail/coil/herf guns etc).

So the circuit that charges the caps (a bank of 480uF at 300V) is done, but not yet tested. So here's the question: How do I deliver this power to the solenoid? And, how much power do I need to handle with my switch (IGBT/SCR)?

I was gonna use a Fairchild IGBT, the SGR20N40L. Assuming 150A is enough to switch ~20J. I can still use that part, but I need to know how much current I need to supply with my optoisolator/gate driver. Oh, the gate voltage is 5V which is sadly too low for most (everything I've looked at so far) opto/gate driver combos.

I imagine I can use a discrete optoisolator and a transistor/fet/bjt/something to drive the gate of the IGBT, but I have no idea how to do that. It appears folks who do coilguns/railguns use SCRs. That would require more components, but if it'll work, I'm game.

Anybody with portable coilgun/railgun experience who can help?

Thanks

-Andy

PS. I am really more of a computer engineer/ computer scientist. I do computer vision and AI, it's just that our analog guy jumped ship (before it sailed) and I got thrust into this component of the project. Not really a big deal, I have experience with switching power supplies and some other analog stuff, but frankly, high (to me) voltage, current and power are out of my area of expertise.

...

Fri Apr 06 2007, 02:29AM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

Few things:
1. The current you need depends on the inductance/resistance of the coil you are using. Basically, since you are using a very heavy projectile, a large coil would probably be fine. But really, you need to experiment and finalise a design... Then you can calculate how big of a fet/igbt you need.
2. You may consider a higher power charging circuit, I do not know how often you need to fire, or how much power that charger can provide, but I would hate to be limited during competition because you cannot recharge fast enough...
3. It is probably a good idea to feed whatever switch you plan to use through an optoisolator (a good ole 4n33 should do the trick). You just pick a correctly sized resistor to feed the LED side from your controller, and then have an isolated 12v supply (a 12v lighter battery would work fine, you will draw very little current) that feeds the gate of the fet/igbt through the transistor side of the optoisolator. Note, you will also need a resistor across the gate of the switch, 1k would be fine.
4. You could use a SCR, but igbt's are soo cheap...

Have fun!

Steve Conner

Fri Apr 06 2007, 10:50AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

I don't see the problem with an IGBT having "Too low a gate voltage". You can usually run optoisolated gate drivers off lower voltages than they are designed for, and also the gate of a 5V "logic level" IGBT shouldn't mind being driven with more than 5V.

Also, why use such high voltage in the first place? Couldn't you just use a solenoid with fewer, thicker turns and actuate it with the same battery that drives your motors?

...

Fri Apr 06 2007, 02:32PM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

I would say the switch should be optoisolated even if you could drive it directly, from the sound of it they are using a $$$ controller, so even if the chances of the fet shorting g/d are slim I wouldn't want to risk it... Also, can you get big logic level fets? I thought they only came for low current stuff...

cinexero

Fri Apr 06 2007, 08:09PM

Registered Member #362 Joined: Mon Apr 03 2006, 03:37AM
Location:
Posts: 4

Just wanted to clarify; the solenoid's plunger strikes the ball. Its not a coilgun. We are building an entire team of these robots (6 or more) so when I make references to cost, its because we have to build a few of them, not because we are strapped for cash (well we are, but ... yea).

I started writing replys from the bottom, so they may be slightly confused.

> 1. The current you need depends on the inductance/resistance of the coil you are using. Basically, since > you are using a very heavy projectile, a large coil would probably be fine. But really, you need to
> experiment and finalise a design... Then you can calculate how big of a fet/igbt you need.

True. It's just that I have to get boards made to play with this charger, and while this is Rev 0, I would like to design as much as possible before we spend a lot of money.

> 2. You may consider a higher power charging circuit, I do not know how often you need to fire, or how
> much power that charger can provide, but I would hate to be limited during competition because you
> cannot recharge fast enough...

So I am charging with ~1.5A @ 9.6V which is stepped up to 300V at 85% efficiency which should give ~38mA at 300V. Ok, Since I = Q/t and C = Q/V, t = CV/I. I mean, there is some resistance and inductance in there too, which will affect things, but this is just an estimate anyway.

Ok, 480uF at 300V at .038 A gives a refresh time of 3.8 sec. That is a bit slow, but the cap bank may be a bit big, who knows. We may have to go to another battery pack for the kicker.

The charger is very cool since is is not monolithic, it can support much higher charging currents then what we are using. (when I say charging currents - I mean on the primary side of the xfmr). The spec sheet has a reference design for 10A on the primary side.

Also, you can sample the controller and the xfmr. True it is scary SMD, but cool none the less.

> ... 12v lighter battery ...

Good idea, I was gonna use a DC/DC converter (~$10).

> I would say the switch should be optoisolated even if you could drive it directly, from the sound of it they > are using a $$$ controller.

ARM 7, custom 4 layer board (that's the stuff I actually know about). The other guys are so paranoid they are insisting this be completely isolated. The HV stuff is galvanic isolated through the xfmr and the opto's power supply will be isolated DC/DC.

> I don't see the problem with an IGBT having "Too low a gate voltage".

Well, I thought if you drove the IGBT with too high a gate voltage, the dielectric on the gate would break down.

> You can usually run optoisolated
> gate drivers off lower voltages than they are designed for

I was gonna try that.

> Also, why use such high voltage in the first place? Couldn't you just use a solenoid with fewer, thicker
> turns and actuate it with the same battery that drives your motors?

Well, I assume you mean run the caps at the battery voltage. The motivation for higher voltages involve the amount of stored power. To get 5J at 9.6V, you would need 9 FARADS, and we want ~20 J. (E=0.5*Capacitance*Voltage^2).

> Also, can you get big logic level fets?

Yes, I designed a buck regulator recently with a 200V 100A logic level FET. It was about $8.

-Andy

ConKbot of Doom

Fri Apr 06 2007, 08:39PM

Registered Member #509 Joined: Sat Feb 10 2007, 07:02AM
Location:
Posts: 329

cinexero wrote ...

> Also, why use such high voltage in the first place? Couldn't you just use a solenoid with fewer, thicker
> turns and actuate it with the same battery that drives your motors?

Well, I assume you mean run the caps at the battery voltage. The motivation for higher voltages involve the amount of stored power. To get 5J at 9.6V, you would need 9 FARADS, and we want ~20 J. (E=0.5*Capacitance*Voltage^2).

Double check your math there, you would only need ~.43 Farads @ 9.6v for 20J, still not reasonable, but still not 9F.

Photoflash caps are definitely in the right size for what you're looking at doing, a few J each, easily available, and easy on the budget if you get a cooperative store.

cinexero

Sat Apr 07 2007, 01:42AM

Registered Member #362 Joined: Mon Apr 03 2006, 03:37AM
Location:
Posts: 4

Yea, you're right. I have no idea what I was thinking.

I was planning on getting a bank of 4 of these

edit: cant get urls to work right

Steve Conner

Wed Apr 11 2007, 08:32PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

cinexero wrote ...
Well, I assume you mean run the caps at the battery voltage.

No, I mean use no caps at all. Your battery presumably stores several kilojoules. It won't be able to deliver such high peak power as a 300V capacitor, but I argue that you may not need the peak power of a capacitor anyway, since you are launching a heavy projectile at a low speed. You can maybe make do with a shove instead of an explosion of power.

(For kicks, work out the time it takes your proposed launcher to accelerate the ball from a standstill to the desired speed. Divide the energy gained by the ball by this time, to get an estimate of the launcher's peak power requirement in watts. Compare this figure to the power draw of, say, one of your drive motors, to get an idea if the battery could support it without help from capacitors.)

P.S. The dielectric breakdown voltage of an IGBT or MOSFET gate is usually quite a lot higher than the nominal voltage to turn it on. I don't see a logic level IGBT being destroyed by 7V or even 12 or 15V. The datasheet should specify the absolute maximum gate voltage anyway.

cinexero

Thu Apr 12 2007, 02:34AM

Registered Member #362 Joined: Mon Apr 03 2006, 03:37AM
Location:
Posts: 4

We laid out a demo circuit yesterday. We used a disposable camera and a relay. We stepped up the capacitance between tests untill we had about 800uF. The power seems adequate (it was all the caps we had).

We then tried a NGD8201NT4G ON Semi IGBT in place of the IGBT. Worked fine. Nothing got even warm after about a dozen tests. I can rig up a test to try the battery, but I doubt it will work as well as the caps.

Time permitting, I'll put up a video on youtube or something.

-Andy

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