Will this optical trigger circuit for a coil gun work?

[user1453]

Hi I've got a circuit for triggering a mosfet based coil gun, the power source is 300v capacitors. I am trying to design an optically triggered mosfet based coil gun

the coils will be 7ohm, therefore with a 300v source the max current is below the max pulse current of the mosfets which is 40A

My question is a simple one will my circuit work?

as the projectile accelerated along the beam it will bolack the heat from an ir diode this will cause to phototransistor to stop conducting, raising the voltage that is enetering the comparator, when it is high enough this triggers the mosfet drive circuit.

My main concern is that the gate drive circuit will not switch the gate fast enough, anyone have a comment about that?

 

[jrz126]

I dont quite follow your triggering circuit. What actually makes it fire? The way you described it, the projectile will already have to be moving for it to trigger. Unless you are having multiple stages?

I think you might have too much resistance on the gate of the fet. Do you have a datasheet for the fet? Also, you'll need a free-wheeling diode connected across the fet as well. The back emf will kill that fet.

Check out www.4hv.org, there's an electromag. coil gun section of the forum.

 

[user1453]

yes it will be multiple stages, the first stage will save a switch between gnd and the phototransistor

here's the datasheat
https://www.ortodoxism.ro/datasheets/irf/irf740.pdf

 

[Roff]

Your FET is dissipating about 675 watts while it is on, so it had better not be on very long. If the projectile stops in front of the sensor, your FET is instant toast.
What is the inductance of your coil? This will determine the minimum time required for the FET to "charge" the inductor. If I am interpreting fig. 11 in the datasheet correctly, your FET can only tolerate a pulse width of about 1 millisecond max. This corresponds to an inductance of about 2.5mH max.
I think you should have a monostable between your comparator and the FET driver to limit the ON time.

 

[user1453]

the coild inductance will be around 1.9mH I would have assumed that
1) 35A could be sustainded for longer
2) becuase the pulse will only be once a minute or so the pulse length could be longer
3) the pulse is only going to be at max current for part of the time

 

[Roff]

the coild inductance will be around 1.9mH I would have assumed that
1) 35A could be sustainded for longer
2) becuase the pulse will only be once a minute or so the pulse length could be longer
3) the pulse is only going to be at max current for part of the time

Your time constant is L/R=1.9mH/8.5ohms=224us. Therefore the pulse will be near max current within about 700usec. 675 watts will heat up your junction REALLY fast. Look at fig. 11 in the datasheet.

 

[user1453]

ok, I've modified my circuit to include a variable length single shot pulse, the pulse cannot be retriggered until 1second has passed, how long is the maximum the pulse should be?

 

[Roff]

ok, I've modified my circuit to include a variable length single shot pulse, the pulse cannot be retriggered until 1second has passed, how long is the maximum the pulse should be?

Looking at fig. 11, and working backwards to get maximum pulse width:

Device current=300V/(8+Rds)~35A
Device instantaneous power = PDM = I^2*Rds, PDM ~ 623W
From fig. 11: Tj=PDM*Zthjc+Tc
Tjmax=150C (from specs)
Solving for Zthjc,
Zthjc=(Tjmax-Tc)/PDM
Assume Tc = ambient temperature = 30C (Tc=case temperature)
Zthjc=(150-30)/623, Zthjc=0.192
Using the bottom curve of fig. 11 (duty cycle~0),
maximum pulse width~2.5ms.
This will, according to the spec, push the part to its limits. I would probably start with 1ms, and adjust it slowly upward until it fails, and then back off a little. However, if you go too far, it might still shorten the life of the part, so, for example, it might only last for 10 shots.
I don't think a heat sink will help, because of the short time involved.
Is 2.5ms long enough to appreciably accelerate your projectile? Seems pretty short to me. You might need something like an SCR with a capacitor-discharge scheme, or maybe an IGBT. Some of the other guys here have more experience with high-voltage, high-current applications than I do.

 

[user1453]

what about paralelling 2 mosfets, so each is only handeling 20A would that allow me todouble the time of the pulse?

or alternativly, could I use 4 mosfets witout heatsinks

they would be within 10A per mosffet for a few microseconds

 

[Roff]

what about paralelling 2 mosfets, so each is only handeling 20A would that allow me todouble the time of the pulse?

or alternativly, could I use 4 mosfets witout heatsinks

they would be within 10A per mosffet for a few microseconds

Actually, two MOSFETs will cut the dissipation of each to about 27% of the dissipation of a single one (assuming they are identical, which they won't be), and the maximum pulse width (from fig. 11) goes to about 50ms. However, the case temperature may rise appreciably in 50ms, so that is probably optimistic. Three or four devices would be safer.

Keep in mind this is all theoretical, and I have absolutely no experience with actual hardware in this sort of application.

 

[user1453]

would it be worth using heatsinks?

 

[Roff]

would it be worth using heatsinks?

Couldn't hurt.

 

[Dr.EM]

Your trying to dissapate a vast amount of heat energy very quickly, make sure you get an excellent thermal transfer to the heatsink if you are relying on them. Use something like silver CPU transfer compound and if possible do not electrically isolate (just be careful not to touch them if they are at HV. Seal in insulated box?)

 

[user1453]

ok so I will use 4 mosfets with the gates, drains and sources connected, I will mount all with a heatsink(20degs/W) using thermal grease snd I will keep the pulse as short as possible.

I'll initially have the pulse small, and increase the size until the mosfets die or I find an optimum pulse length.

anyone see any problems with this?