Overdriving Gates on ISOPLUS IGBTs / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Overdriving Gates on ISOPLUS IGBTs

1 2 next

Move Thread

LAN_403

Daniel Kramnik

Fri May 20 2011, 07:54PM

Registered Member #3885 Joined: Sun May 15 2011, 12:47AM
Location: Newton, Massachusetts, United States
Posts: 94

Hello all,

I am currently building a small-sized DRSSTC for an AP physics project that is due this coming Wednesday. Thus far, I have all of the parts and I have soldered together the driver (based on Steve Ward's DRSSTC3 design with primary feedback and OCD) and wound the secondary coil. However, I could not afford the more expensive SOT-227 IGBTs he used, so I settled with the much cheaper TO-247 IXGH40N60C2-ND variant from Digikey.

I have read that overdriving IGBTs with 33V on the gates can reduce lifespan, but that this increases performance enough for it to be justified. One person commented that since we don't need them to run for 100 years, we might as well overload them a bit. However, because TO-247 IGBTs are not as robust as their SOT-227 counterparts, is it still wise to run them with so much voltage on their gates?

I do have some gate resistors that I can put in, but I don't want to do this if it will compromise performance. On the other hand, if my bridge explodes while I'm demonstrating, it could be a slightly problem. Does anyone have any advice or experience that could help?

I don't want to solder in gate resistors if I don't have to because this will inevitably increase leakage inductance and the coil's performance is already being limited by the weaker TO-247 IGBTs.

Thanks!

Daniel Kramnik

Steve Conner

Fri May 20 2011, 08:19PM

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

As far as the ability to withstand excessive gate voltage is concerned, there's no difference between TO247 and SOT227.

If you're determined to shove way too much current through the devices, then overvolting the gates might actually increase the life. (From 5 minutes to 10

)

If you attach TO247s straight to their own individual live heatsinks, they can give the SOT227s a run for their money. The heat dissipation of the SOT227 package is limited quite severely by the alumina washer that isolates the device electrically. (to be fair any isolated package shares the same drawback, you can't beat the direct silicon-to-metal contact of a live heatsink)

Daniel Kramnik

Fri May 20 2011, 08:21PM

Registered Member #3885 Joined: Sun May 15 2011, 12:47AM
Location: Newton, Massachusetts, United States
Posts: 94

Thanks a lot!

I'm going to use two separate CPU coolers for the H-Bridge, so isolation won't be necessary

Steve Conner

Fri May 20 2011, 08:25PM

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

You need three isolated heatsinks for a H-bridge.

If this is the device I think it is, the plain TO247 can actually handle more than the SOT227. It's the same chip in both versions, but the TO247 has better cooling.

Wolfram

Fri May 20 2011, 08:26PM

Registered Member #33 Joined: Sat Feb 04 2006, 01:31PM
Location: Norway
Posts: 971

The reason people overdrive IGBT gates is to prevent the IGBTs from going out of saturation at high currents. So it's more of a precaution than getting them to perform better. The VCEsat will also drop somewhat at higher gate voltages, but the difference between 24V and 33V is going to be negligible. The risk of gate breakdown is significantly higher at 33V though, so I would go with 24V. Extrapolating the admittance curve in the datasheet, these IGBTs should be able to handle around 800A without going out of saturation with 24V on the gate. There's a good chance that the IGBT itself won't survive 800A, so 24V should be enough.

ScotchTapeLord

Fri May 20 2011, 08:38PM

Registered Member #1875 Joined: Sun Dec 21 2008, 06:36PM
Location:
Posts: 635

If it's a small-sized DRSSTC then I don't think you'll need much overdrive. They are rated for 200A, and although I don't see any saturation curves in the datasheet for the particular device you've listed, I would assume that you could get away with a a lower voltage than 30V. If you aren't going to use gate resistors, then I'd definitely stay under 30V due to overshoot. If you don't have gate resistance, then the current is limited by the leakage inductance of your system, no matter how small. The energy stored in an inductor is .5*L*I^2, and that energy goes into overvolting the gate, so gate resistance lowers the I much more than increasing the L, and since the energy stored is much more dependent on I, it helps a lot.

I'd use 20-25V with 3-5 ohms of gate resistance, trying to stay in-line with the test parameters in the datasheet, while keeping the keep current within 200A, which I think would be enough for a small coil (though we all have different definitions of small).

I found this:

Figure 1 shows the saturation characteristics. When the plot flatlines, that means it has desaturated and will explode. As you can see, a small increase in gate voltage allows you to run a lot more amps. Excessive gate voltage will allow your device to run cooler but if it is at the expense of your gate oxide layer than it isn't worth it in my opinion, unless you plan to exceed 200A peaks.

Wolfram

Fri May 20 2011, 08:40PM

Registered Member #33 Joined: Sat Feb 04 2006, 01:31PM
Location: Norway
Posts: 971

Also, if a TO247 and a SOT-227 have similar part numbers (example: HGTG20N60A4 and HGT1N40N60A4D), they are likely to have the exact same IGBT die inside, so they will be the same electrically. The major differences are thermal, in that the TO247s perform much better (often half of the thermal resistance compared to their SOT227 equivalents, so twice as good at getting rid of heat). The beefier TO247 IGBTs also often lack the internal reverse diodes (as seen in my part number example above, the TO247 variant lacks the D at the end of the part number), simply due to the fact that there isn't room for a diode in the TO247 package together with the big IGBT die. And TO247s need to be mounted on separate heatsinks, unless you use an insulating washer, but an insulating washer will really increase the thermal resistance, making them worse than SOT-227s.

ScotchTapeLord wrote ...

If it's a small-sized DRSSTC then I don't think you'll need much overdrive. They are rated for 200A, and although I don't see any saturation curves in the datasheet for the particular device you've listed, I would assume that you could get away with a a lower voltage than 30V.

There is an input admittance curve, and it shows that the device should handle just under 200A at 10V on the gate without going out of saturation. Extrapolating the curve, 24V on the gate ends up around 800A, though without any guarantees that the IGBT will survive this current.

Inducktion

Fri May 20 2011, 09:02PM

Registered Member #3637 Joined: Fri Jan 21 2011, 11:07PM
Location: Buffalo, NY
Posts: 1068

On a completely different note, I found these Isotop fets

Which seem pretty close to the ones you all know and love..

Daniel Kramnik

Sat May 21 2011, 12:37AM

Registered Member #3885 Joined: Sun May 15 2011, 12:47AM
Location: Newton, Massachusetts, United States
Posts: 94

Ok, I'm going to try running the gates at 33V, but with 5 ohms of resistance. One more question: I hadn't realized that my IGBTs don't come with internal diodes, would a pair of 600V 20A RFUS20TF6S work, and if so, how much heatsinking would they need? In my ordinary SSTC, I use MUR860 diodes, which experience no heating during operation, but then again, my MOSFETs aren't switching hundreds of amps. Unfortunately, I have already assembled the power section, so it could take some doing to make room for extra diode heatsinks.

Daniel Kramnik

Sat May 21 2011, 01:28AM

Registered Member #3885 Joined: Sun May 15 2011, 12:47AM
Location: Newton, Massachusetts, United States
Posts: 94

Quick correction - the RFUS20TF6S are rated for 20A average rectified current, but 100A pulsed. I think that this should be good enough to get started (at first, I'll have my OCD set to around 100 or 150A), but the question of heatsinking still stands. I only have 2 of these diodes left, so I don't want them to explode!

1 2 next

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