Electronics > Power & Renewable Energy

MOSFET battery switch consistently frying

(1/2) > >>

felixer:
I need to connect a 12S LiPo (50.4V fully charged) to a power bus. This bus has over 6,000uF of capacitance on it and can be at higher voltage than the battery itself. Ideally this would be done as quickly as possible, in the 1-2 mS range. The switch needs to support about 300w steady state.

I currently have an Hbridge driver abused into making an isolated 10v for an isolated gate driver.

The attached circuit has been my most robust thus far but only stays working when the turn on time is up in the 200+ ms range. When I reduce the gate resistance and capacitor to the shown values it turns on in ~32ms on the bench. Then when connect to the battery and about half the capacitance is still works. But when finally connected to the full level of capacitance it fails instantly (FETs short and stay closed). I'm really confused because the measured inrush current is FAR below the rated amount for the FETs, I have tried P ch, Nch, IGBTs, GaN everything. It does get more robust as the FETs current capacity goes up but nothing has just worked. I have looked for dedicated current controlled drivers but haven't found one that works with a bidirectional switch AND at these voltage levels.

I have killed A LOT of silicon trying to get this working and i'm at my wits end, I would really apricate any advice. Maybe I need to be turning the FETs on faster not slower?

FET PN: AUIRF7749L2TR
Driver PN: ADUM4121-1ARIZ-RL
Hbridge driver for the isolated 10v: MAX13256ATB+T
10v LDO: S-1142BA0I-E6T1U

uer166:
1-2 ms is infinity in terms of blowing up MOSFETs. Most likely you're violating the SOA of the FET since it spends so much time in linear range, you don't have to exceed the peak current rating to blow up a switch with that. E.g. the AUIRF7749L2TR SOA curve says at 10ms it can handle only 1A at 10V Vds. You're running it at 50V, which is so useless it's not even on the curves.

You'd do well turning those modern FETs as quickly as possible at the correct speed. What I would do is: switch the FETs in under 10 microseconds with a real gate driver. Of course that may create > 1kA inrush current, depending on impedances, so you can also add a little bit of dedicated resistance in series to limit inrush to some guaranteed level, probably in the couple hundred amps range. Generally the actual silicon doesn't matter here, your current approach wouldn't work with any FETs since it doesn't take into account the safe operating area ratings.

felixer:
That is what I am thinking. does the ADUM4121-1ARIZ-RL part count as a real gate driver? I think it is going to be easily capable of switching these FETs on within a few microseconds.
Are you recommending series resistance with the gate drive or with the battery?

Also thank you so much for taking the time to look at this for me, I really appreciate it.

uer166:

--- Quote from: felixer on September 13, 2021, 11:10:26 pm ---That is what I am thinking. does the ADUM4121-1ARIZ-RL part count as a real gate driver? I think it is going to be easily capable of switching these FETs on within a few microseconds.
Are you recommending series resistance with the gate drive or with the battery?

Also thank you so much for taking the time to look at this for me, I really appreciate it.

--- End quote ---

Yes and yes, at 2A it's plenty to switch even large FETs in us range. A gate resistor is a good idea to control the speed, you don't want it too fast, especially for turn-off (in case the harness is inductive). You'll need to generate a floating rail as you did before, like with a charge pump or a transofmer based DCDC, but obviously connect to gate driver VCC instead of directly to the gates.

f4eru:
Use a pre-charging resistor, with a separate switch, switching on first.
This way, the energy is not dissipated in mosfets.
Why 1-2ms, and not a bit more ?

Navigation

[0] Message Index

[#] Next page

There was an error while thanking
Thanking...
Go to full version