I'm getting read to wire up my mosfets and I wanted some tips for making sure I don't have too much inductance. The signal path will be up to 6 inches in opposite directions (may be closer to 4 inches each way). This is driving multiple mosfets so I'm putting my driver in the middle and plan to solder a signal wire directly across all the parallel mosfets.

I understand I should probably add small value gate resistors to them all to help prevent HF oscillations, but as far as the wire goes. Would it be best to use a small gauge twisted pair wire and ground one of the conductors at each end and the middle?

Switching frequency is ~20khz. And I'm not going to make some eccentric shape to mount my mosfets they are going on a long copper bar/heatsink, doubling as the current path for all of them as well.

Hope this is enough info.

 

Well, if you put them in a circle with the driver(s) in the center, then the gate traces could all be of equal (and minimal) length.

 

Well, if you put them in a circle with the driver(s) in the center, then the gate traces could all be of equal (and minimal) length.

Then I will let you design and build me the heat sink for that .

 

Actually, just use a wide trace on the PCB. If you're doing a double-sided board, try to avoid having ground planes or other signals running parallel to it on the other side, as you'll get capacitive coupling, which will increase the load on your driver.

 

I decided not to use a PCB for the output

The left circuit their will have a PCB attached. Along with the other circuity of the PWM and power supply on a custom PCB I'm going to etch. The plate they are mounted too is going to carry the Main positive rail so its going to be insulated on the back side and mounted to a larger heatsink or water block. Going to build a mount to have a negative rail right below the mosfets and to support the assembly. Gonna run a gate signal wire across the mosfets with the 4.7 ohm resistors soldered right to the mosfet pins.

 

At 20 KHz, there will be very few stray inductance problems arise. Your greatest concern is simply moving charge onto and off the gates of the FET's to minimize dwell in the ohmic region..

 

Hows this for adding lots of inductance hehe.
This is just a test anyway.

 

What's your peak current going to be on the gates again?

Remember, solder has about 5.5x the resistance that copper does. Wherever possible, you want to have a good physical connection before you apply the solder.

 

The driver delivers 9amps peak.

I left the resistor wires long to try to get as much physical contact between all the wries before soldering. I also pinched them together with needle nose piles as I soldered. Then cleaned up the appearance some.

 

Rough looking, but just wired up the driver and tested it. Extremely sharp square wave. The rise and fall is less then 200uS at 20khz. My scope isn't very accurate at this low of timescale but I would estimate rise/fall of around 100-150uS by looking at the readings. This was only tested at 13V 25Amps. Will scale the test up to higher voltages and currents tomorrow.