New Coilgun Kicker project + IGBT Problems / Electromagnetic Projectile Accelerators / Forums

Forums

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

« Previous topic | Next topic »

New Coilgun Kicker project + IGBT Problems

1 2 next

Move Thread

LAN_403

Gianluca

Thu Jan 24 2008, 10:23PM

Registered Member #1261 Joined: Thu Jan 24 2008, 09:11PM
Location: Milan, ITALY
Posts: 5

Hello Everybody! I'm new to the forum... I hope this will become an interesting and active thread

I am building as a grad project an electromagnetic kicker for a robot. It's basically a coilgun but the projectile is a 10cm long pure iron rod, 1'' in diameter and 0.4kg plus the required velocity is 10m/s.
The coil is almost done and it works (not looking for efficiency, just results at the moment!) but I have a few problems with the electronic board i have done. Especially the IGBT so ... PLEASE HELP!
I have attached a very simplified schematics of my board:
Everything is controlled by a PIC that generates the PWM for the boost converter and gives the high signal to the to IGBTS. One is for discharge to an high wattage resistor (680ohm) and the other is for the proper coilgun activation.
The IGBT's are driven by tc4432 drivers by Microchip capable of 1.5 A and Vdd-1V of voltage (in this case i give 23 volts to the gate of the igbt).
The cap I have used is an Epcos 3300uF/450V (charged at 400V).

Now THE PROBLEM IS: is it ok for the IGBT to driven by a Vge greater than 20V? From the datasheet it doesn't look so but many say it's better with high currents. (the coilgun discharges at 100A peak approx).
I have mounted on every IGBT and diode a simple C-shaped heatsink. It seems ok for the boost converter but what about the coilgun IGBT????Is that enough??? The pulse width is around 20ms.
I am asking the latest two questions because I have already burnt two of these quite expensive transistors (They are around 20$ each here in Italy)!!! The first lasted some testing (maybe 10 shots) where the lasted never worked and I assumed is broken since I checked with the oscilloscope the rest of the circuit! And I am wondering why! This project must be reliable!
The datasheet of the IGBT I am using is:

By the way, if anyone is interested in the project I have lots of pictures and data on it!

P.S. It's not relevant to my question but the values of the voltage divider are wrong! The 390k resistor is there but the smaller one it's wrong! It should read close to 5V when the cap is at 400V. Can't remember the value

Thank you very much for the help

Bjørn

Fri Jan 25 2008, 01:43AM

Registered Member #27 Joined: Fri Feb 03 2006, 02:20AM
Location: Hyperborea
Posts: 2058

It says +-20V continious and +-30V pulsed. You need to check the waveform while it is in operation, maybe there are spikes you don't kmow about. The same goes for Vce.

The heatsink should be fine unless it is rapid fire.

Shaun

Fri Jan 25 2008, 02:04AM

Registered Member #690 Joined: Tue May 08 2007, 03:47AM
Location: New Jersey, USA
Posts: 616

For the actual firing of the device, you might consider (for cost purposes) using SCRs. 100Apk is nothing for them; any TO-220 device could handle it (I have some that are good to 420Apk).

Plus, you don't need gate drivers or anything; just a resistor and 3VDC which could be provided by a regulator.

Gianluca

Fri Jan 25 2008, 10:03AM

Registered Member #1261 Joined: Thu Jan 24 2008, 09:11PM
Location: Milan, ITALY
Posts: 5

I have surely considered SCRs but the problem is that they don't turn off. I must be able to control in some way the speed of the kicker and I managed to do so by closing the IGBT at different timings. By doing so the cap doesn't discharge completely into the coil and the kicker achieves different speeds.

I have checked the output waveform of the igbt driver and it looks clean. Obviously the IGBT mounted is dead so this check is a bit useless cause with a functioning one everything could change.

Any suggestions on how to protect my igbt? The one from the boost converter has an integrated CE diode and should be fine. But for the firing one I can't efford to blast a couple of them just to check if there are spikes some where!

Thanks!

Electroholic

Fri Jan 25 2008, 08:09PM

Registered Member #191 Joined: Fri Feb 17 2006, 02:01AM
Location: Esbjerg Denmark
Posts: 720

you are abusing silicons...

two better ways to do it
1, diagonal halfbridge(will need highlow side drivers)
2, use SCRs and control the voltage in the cap.

Gianluca

Fri Jan 25 2008, 10:43PM

Registered Member #1261 Joined: Thu Jan 24 2008, 09:11PM
Location: Milan, ITALY
Posts: 5

Thanks for your advice! Maybe I haven't explained my requirements well enough and you got me wrong:
This kicker is needed on a soccer playing robot (at least a prototype) and it has to be able to kick a ball from 0 to 10 m/s as the robot requires. To do so i fully charge a capacitor and with a look up table I convert the ball speed requirement in timing for my igbt. Everything is done in a couple of ms at most. That is why your second option can't work!

I must use IGBT because timing and turning off is vital.
The half bridge design is a much more refined way to go with. It shortens the fall time in the solenoid and recharges the cap... but the power handling requirement for the single IGBT are still the same that in my one-igbt configuration, right? The igbt still has to handle 100A at 400V for 20ms or so....

My problem is: I have got 3 IGBT already burnt and I am trying to understand why and if I am doing something wrong! so this are the points:

1)Driving the IGBT at 22v with a TC4432 dirver. Is it too much??? The absolute ratings are 20V continuous and 30V pulsed. How long is "their" pulsed?? Many people project coilguns with igbt working at 25v-30v of Vge because they say it works better... why mine just stop working?

2)Going for a lower Vge (like 12v-15v) are there still elements I could use to protect my IGBT? I just put a freewheeling diode parallel to the coil and that's it. Could I use a diode on CE??? On my simulations it doesn't seem to work by adding it. Everything perfect without.

3)I just use a small C-shaped heatsink on the IGBT. Is it enough? Could be high temperature that damages my devices? The IGBT is rated at 330A pulsed and pulse lenght is limited by temp. I can't think of a 20ms on-off-on time destroying it. Only part of the pulse as such an high current!

Have a look at my graphs (MultiSim)... and give suggestions!

THX

Bjørn

Sat Jan 26 2008, 03:29AM

Registered Member #27 Joined: Fri Feb 03 2006, 02:20AM
Location: Hyperborea
Posts: 2058

It is possible in theory that you have damaged the devices by static electricity. If you have dry air such things happen from time to time.

Figure 16 in the datasheet lets you calculate the junction temperature based on pulse length. I don't know what walue of PD to use but using the full 290W I got about 75 deg C for a 20 ms pulse. If I did it correctly then your heatsink should not ba a factor at all if the die has time to cool between shots.

Shaun

Sat Jan 26 2008, 04:47AM

Registered Member #690 Joined: Tue May 08 2007, 03:47AM
Location: New Jersey, USA
Posts: 616

I don't know if this will work in this application, but the way I'm used to protecting IGBTs is Zener diodes G-E and TVS diodes C-E.

Or, try a different IGBT. Sometimes one certain type just won't do what you want it to, and another will work miracles.

Dr. Dark Current

Sat Jan 26 2008, 06:22PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Gianluca wrote ...

I have surely considered SCRs but the problem is that they don't turn off. I must be able to control in some way the speed of the kicker and I managed to do so by closing the IGBT at different timings. ...

Here is your problem! As the IGBT turns off, the coil voltage can skyrocket to a few kilovolts by inductive action, this can lead to complete destruction of the device (IGBTs have usually none or poor avalanche rating).
The solution is either to use a MOV (metal oxide varistor) across the IGBT with clamp voltage below the Vce max. of your IGBT; or use SCR and control the velocity by adjusting the firing voltage of the cap.

TheMerovingian

Tue Jan 29 2008, 01:24PM

Registered Member #14 Joined: Thu Feb 02 2006, 01:04PM
Location: Prato/italy
Posts: 383

Glad to see that here in italy i'm not alone in designing, coilguns and their cousins. Jmartin is right, the inductive kick will kill your IGBT, just use a freewheeling diode or another type of snubber (RC network can work well if you size the resistor correctly, using the worst case estimation formula from the max IGBT C-E voltage knowing the peak current, and the cap voltage V = R*Ipeak+Vcap < VmaxCE -> R < (VmaxCE-Vcap)/Ipeak

Hope this helps, and welcome in the world of coilgunning.

(ITA: mi fa molto piacere sapere che non sono l'unico in italia a costruire/progettare i coilguns e i loro cugini. Jmartin ha ragione, la sovratensione generata dall'induttanza brucia i tuoi IGBT, utilizza un diodo di protezione in antiparallelo oppure una rete RC , la quale funzionerÃ egregiamente se calcoli correttamente il valore della resistenza di potenza necessario, utilizzando una formula "pessimistica" a partire dalla tensione massima tra collettore e emettitore dell' IGBT , la corrente di picco, la tensione di picco, in base a questa formula V = R*Ipeak+Vcap < VmaxCE -> R < (VmaxCE-Vcap)/Ipeak

Spero di essere stato di aiuto, e benvenuto nel mondo dei coilgunners

1 2 next

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