common mode transients through transformers

Hi,

I have a small ~1Watt 100kHz ferrite transformer (EP13 core) that generates three floating ~15VDC supplies for powering the gate drive circuitry (opto plus fet drivers) for a high voltage 100kHz 3.6kW Hbridge, and I am having some problems with common mode transients from the Hbridge switching passing back through the gate drive supply transformer and into the isolated primary ground. The primary ground has an FPGA on it that is resetting due to the common mode transients.

The gate drive transformer is setup as a forward topology, with a 5V primary voltage. I am using 20ns mosfet switching on the Hbridge right now, I think reducing this switching speed ie. to 100ns will help, but also I am wondering about putting a common mode choke on the gate drive transformer primary and secondaries to absorb the common mode transients, ie. this one:

formatting link

?name=490-1069-1-ND

I measured the gate drive transformer interwinding capacitance and got these results:

29pF primary to secondary1

16pF primary to secondary2

11pF primary to secondary3 (must be wound furthest from primary as it has lowest capacitance)

I was thinking of switching to a split core bobbin to reduce the interwinding capacitance and reduce the common mode energy transfer back into the primary, from looking at the capacitances of the transformer I am using it looks like it is a single section bobbin with the coils wound on top of each other.

These transformers are from ICE components, I haven't asked them yet if they can make the transformer with a split core bobbin.

For the 100kHz Hbridge switching, will these lower frequency transients also pass through the gate drive transformer, as well as the 20ns/50MHz switching transients? I know the 20ns transients have a lot more energy than the 100kHz switching, but I am not sure if a lower frequency common mode filter would also be required for the 100kHz switching.

Also the "X2Y" caps may be able to filter out some common mode noise:

formatting link

?name=709-1009-2-ND

Also right now my FPGA is on a wirewrap board, so I am hoping that once I use a proper 4 layer PCB with GND/power planes it will have a lot more common mode immunity.

Any other techniques or suggestions for this?

cheers, Jamie