A little idea I had for my dual full bridge design

[Overlordcheese]

Recently, I was redesigning my dual full bridge for a future QCW DRSSTC, when I realized the dimensions of the board were eerily similar to that of a 92mm fan (This is my first delve into the fan realm, why the fuck is there a ninety... two mm fan???) and I came up with an idea. As you all already know, fan covers are designed to help a fan with uniform distribution of air, but some energy is lost on the way. But what if the pcb was the fan cover? If that were the case, not only would the IGBTs be more directly cooled, but the bus bars would also be cooled!. Anyway, i was just wondering what you all think.

[Mike]

It's certainly novel. The two concerns I have are 1) the tabs of the IGBT's are facing away from the airflow with no way of coupling them into the airflow. This will result in a relatively high thermal impedance and with no heatsink there's very limited transient capability. 2) the layout of the power traces have a lot of compromises to allow the form factor. There are no power planes, no where to add bus capacitance on the board and traces are forced to be relatively thin. Given that the correct selection of (I assume) TO-247 IGBTs mean you otherwise have a 500A bridge, this layout might not let you get there.

I wouldn't go so far as to say it won't work, but I'm not sure it provides much benefit over a more traditional layout. Of course you could choose to build it anyway and prove me wrong, it certainly wouldn't be the first time!

[Overlordcheese]

It's certainly novel. The two concerns I have are 1) the tabs of the IGBT's are facing away from the airflow with no way of coupling them into the airflow. This will result in a relatively high thermal impedance and with no heatsink there's very limited transient capability. 2) the layout of the power traces have a lot of compromises to allow the form factor. There are no power planes, no where to add bus capacitance on the board and traces are forced to be relatively thin. Given that the correct selection of (I assume) TO-247 IGBTs mean you otherwise have a 500A bridge, this layout might not let you get there.

I wouldn't go so far as to say it won't work, but I'm not sure it provides much benefit over a more traditional layout. Of course you could choose to build it anyway and prove me wrong, it certainly wouldn't be the first time!

So a few things. First, there will be a heatsink on each IGBT, I just didnt get them into the 3d model kicad thing. Secondly, there will be bus bars, so the traces won't have to suffer alone. Yes, you're right about bus capacitance on the board, but... do people actually do that? why? just to reduce impedance? and finally, not exactly with the 500a. It's a QCW DRSSTC design, so not quite that high but I always like to plan for the worst.

[Mike]

Honestly if you have the time and money to experiment I recommend doing it, no matter the outcome your guaranteed to learn a lot.

Yes, you're right about bus capacitance on the board, but... do people actually do that? why?

Yes they do, my bridge is very similar to yours in topology (8 IGBT H-bridge, though for DRSSTC use in photos below before adding 40mm fans) and ended up with 8.1uF which is arguably on the light side. The capacitance on the bridge both provides the instantaneous current to the IGBTs when the gates turn on (while the current from the bulk cap is ramping up for to the parasitic inductance), saves the electrolytic from the worst of the high frequency components (which both cause heading and they struggle to provide for to their construction) and provides an energy store to help reduce the village rise caused by did/dt transients and stay inductance.

If you can turn your IGBTs around so the heatsinks are in the airflow I think that would be a big improvement unless you were planning on some sort of "counterflow" arrangement where the air exit vents force air to double back past the heatsinks.

All of these effects are worse the higher the current goes. If you're only running ~100A peak, you probably have more leeway to get away with less ideal construction, there are always tradeoffs involved in engineering.