[bruceme]

Originally Posted by steven4601

Total igbt loss 320 watts. or seen from 10kW output, 3.2% loss.

Not bad, but i think you might find better switchers with little effort.
A similar powerloss result but for 17 USD worth of switchers can be realized with the : STGW30NC60VD

Double-E for dummies question.. The power rating for the IGBT you spec is 250 watts, that's only in regards to the switching losses or the total throughput energy?

Thanks

-Bruce

[steven4601]

Sorry, forgot to mention , in my previous response id had 5 of these paralleled. One mosfet is about 3 to 3.5 usd totalling at 17 USD.

5x 250 is 1250 watt maximum heat-loss.

Your question about losses however is slightly more complex when looking in detail. Purely looking the maximum power dissipation number only gives an idea of what is possible. Its not a guarantee.
This is because junction temperature depends strongly on the length of the pulse.

[valerun]

Originally Posted by steven4601

This is because junction temperature depends strongly on the length of the pulse.

Yes but wouldn't you spec the design in this case to have peak current to be under continuous rating for components? Esp given paralleling of many smaller parts and resulting need for additional derating?

V

[steven4601]

Continious rating as in DC?

That way you'll be safe ofcourse, but depending on the application, the silicon can be pushed further as long the junction temperatures are satisfied. The graphs describing the thermal resistance vs time are a great way to estimate the junction temperature for high current pulses.

[bruceme]

Originally Posted by steven4601

Your question about losses however is slightly more complex when looking in detail. Purely looking the maximum power dissipation number only gives an idea of what is possible. Its not a guarantee.

So.. keep going bigger till it's not smoking hot

[valerun]

Originally Posted by bruceme

So.. keep going bigger till it's not smoking hot

LOL. Well, I have learned, thermal management is nothing to laugh about. Just most recently, I had to deal with my freewheeling diodes blowing up at the rated current just because I used thicker thermal tape instead of my usual one (had to use thicker tape for better electrical isolation of 1000V spikes on IGBT turn-off in the new 700V design). So had to mount diodes metal-to-metal to copper heat spreader first and then spreader to sink through tape. A bit of a hassle... Again, who decided that cathodes should be electrically connected to baseplate on all the power diodes???

V

[valerun]

10 200A dual IGBT modules showed up in the mail today. Will be testing out the PFC and new, 17kW / 100A output version of the charger later this week. I think this will be the ultimate charger, capable of fully utilizing max rating of the level 2 J1772 output. Stay tuned.

[bruceme]

Originally Posted by valerun

10 200A dual IGBT modules showed up in the mail today. Will be testing out the PFC and new, 17kW / 100A output version of the charger later this week. I think this will be the ultimate charger, capable of fully utilizing max rating of the level 2 J1772 output. Stay tuned.

I want one of those

[valerun]

ok a bit slower than promised but I finally got to finishing the PFC stage for the charger! (if you are wondering why took longer than expected, check http://www.diyelectriccar.com/forums...ock-60262.html ;-)

Total parts cost ~$130 if you get them in bulk - including a used 200A half-bridge IGBT module from ebay. Of course, if you use this PFC stage with the charger, you are saving input bridge and at least one elcap that you now don't need for the voltage doubling. So technically net parts cost is probably around $100.

Tested to ~5kW so far. Waveforms are pretty clean - I am only limited by the heatsink at the moment as this thing does generate a bit of heat and my minuscule fanless heatsink just doesn't cut it ;-)

Recap of some parameters:
1. IR1153 PFC chip with overcurrent, overvoltage, undervoltage, etc protections. Switching at 22kHz - ideal for IGBT. Custom PCB
2. Same IGBT driver design as in the main charger unit - on a new custom PCB mounted directly on IGBT leads
3. 400uH inductor - same type as in the main charger unit.
4. In a half-bridge module I am using (200A 600V), one IGBT is used as switch, another one has G-E shorted to serve as freewheel diode. That way, stray inductances are minimized and thermal management is simplified. None of that stupid cathode-to-heatsink electrical connection to deal with in all the 'normal' power diodes...

Will post schematics and PCB files to the site soon - I'd imagine we'll add this to our kit options.

V

[valerun]

anyone has schematics for GFCI circuits? I want to build one into this charger. As a few rightly mentioned, since it's non-isolated, we need some additional protection. Instead of using off-the-shelf $100 50A 240V GFCI, seems more sensible to just add a bit of circuitry to the existing design.