Hi,

We are planning for design 100KW 400KHz H-Bridge Inverter, i search many things in internet and confused, that, i have to go with PCB Based Inverter or Busbar. Inverter output current will be 315Arms@560VDC@400KHz switching Frequency for 100KW, this rating might now suitable for PCB Based Design, but again my inverter switches are discrete 4 Leg SiC MOSFET, So how to with busbar design.

Is anyone have experience on this type of Inverter design, then please guide us for better solution.

Regards,
Nandish Jasani.

 

You can't do 315A on a PCB. For that level of power DIY isn't really an option, you need to be talking to a company that does this all the time. Those SiC MOSFETs are not really suited to a busbar, you need power modules with screw terminals. What MOSFETs are they?

 

For 315A RMS you need copper conductors > 200mm2 at 50Hz for a 70degC rated cable at 30degC ambient, about 16mm dia. For 400kHz you'll need to allow for skin effect as well.

 

You can't do 315A on a PCB. For that level of power DIY isn't really an option, you need to be talking to a company that does this all the time. Those SiC MOSFETs are not really suited to a busbar, you need power modules with screw terminals. What MOSFETs are they?

Thank for your reply.

I am planning to use WolfSpeed make C3M0016120K 4 Leg SiC MOSFET.

 

For 315A RMS you need copper conductors > 200mm2 at 50Hz for a 70degC rated cable at 30degC ambient, about 16mm dia. For 400kHz you'll need to allow for skin effect as well.

Yes, skin depth @400Khz will be 0.11mm, i think i must used multiple 0.2mm copper foils. Is that good idea.

 

Yes, skin depth @400Khz will be 0.11mm, i think i must used multiple 0.2mm copper foils. Is that good idea.

So 315A needs 200mm2 of copper, skin depth = 0.11mm therefore track must be how wide?

How many of these MOSFETs were you planning to use?

 

C3M0016120K

I do not know what you are making but..............
This is a 1200V part. Have you looked for a lower voltage part?
This part really can't handle 100A. If you used three in parallel you will be at 100A each. Here is a graph of a hot part running 100A, 4Vd-s. That is 400 watts of loss each. (assuming the G-S voltage is over 9V) (100W x 3 = 300W of heat)

Here is the safe operating area graph. You should not run this part at 100A.

You can run 300A in a PCB. Not in traces but in plains.
You might think about this as 6 x 50A bridges.

 

The package you are looking at is old. It will be hard to make it run at 400khz.
Here is a higher current part. SICFET N-CH 750V 209A TO-247-4L
This package has lower inductance and I can make them run faster. Note the lower voltage and lower Rdson.

 

@ronsimpson yes, you went where I was going - though I was looking at 5 x 63A. The Rds and switching losses are quite significant. I was a little confused by your link to the SiCFET and then the image of the Infineon devices. Took me a sec to realise they were two different options.

Some new Infineon devices popped up in my inbox a few days ago (link to whitepaper) with top-side thermal paths which is going to make life a lot simpler for these types of projects.

 

Irving Every think about this project is hard. The PCB might have to be "Aluminum" to spread out heat. The PCB might have to have many, many thick layers of copper and very thin insulation.
There are some papers on "automobile motor drivers". They do not run at 400khz but the voltage and current are about the same.
One thing we do is to build a 50 or 60A version and work out the problems. Then stack up 6 or 5 of them in parallel.
Again: " Every think about this project is hard. " There is a year of learning and broken MOSFETs in the future.

I have broken parts that look like this. At $1000 each, accounting gets upset when I need more.

 

@ronsimpson Hence my comment in post #2 " For that level of power DIY isn't really an option, you need to be talking to a company that does this all the time. " Even a 60A version is non-trivial. Of course 400kHz is the MOSFET switching frequency, I'm guessing not the induction coil frequency? I did a 15kW grid-tie inverter recently, that was 3 x 5kW in parallel @ 100kHz mainly to use existing parts, but even that wasn't without issues.

NB link in post #9 now fixed

 

@Irving and @ronsimpson Thanks for giving me this much of information. Definitely i will think of 60A inverter and paralleling it.