Internal Gate-source connection of a half-bridge SiC module (Wolfspeed)

Thread Starter

SiCEngineer

Joined May 22, 2019
444
Hello all. I am using the XM3425 power module from Wolfspeed, data sheet here: https://www.wolfspeed.com/products/power/sic-power-modules/xm3-power-module-family/cab425m12xm3/

It is composed of multiple parallel connected FETs. In the connection diagram, there is a 4-pin connector - two for the gate, and two for the source. Unfortunately, on my PCB, only one of these has been connected to source, with the other one intended to be used for the source left unconnected.

Of course, there is no information regarding how the devices are connected internally, but as a sanity check until I get back to the lab tomorrow, I wanted to know if anyone had any opinions on whether this would cause any issues. It is not clear whether all sources of the devices are connected together, such that it does not matter, or whether the sources are all common only when both of the pins are connected to the source plane on the PCB. Connection diagram attached.

I often see a single gate pin and a single source pin for high power SiC modules, so have yet to run into this issue before.
 

Attachments

ronsimpson

Joined Oct 7, 2019
3,223
The gate driver should be connected to pint 4-5 and 6-7. If you used 1 and 2 for the Source you will have more inductance in the Gate-Source loop. For fast switching the Kelvin Source pins are brought out for the Gate drive. You can use any Source pin you want but to get the cleanest switching use the right pins.
 

Thread Starter

SiCEngineer

Joined May 22, 2019
444
The gate driver should be connected to pint 4-5 and 6-7. If you used 1 and 2 for the Source you will have more inductance in the Gate-Source loop. For fast switching the Kelvin Source pins are brought out for the Gate drive. You can use any Source pin you want but to get the cleanest switching use the right pins.
understood. In the current arraignment I believe I have pin 6 connected as it should be (Kelvin connection) whereas pin 7 has not been connected to pin 6, such that it is probably connected directly to the source with no Kelvin connection.

This then begs the question about what goes on internally to the device - are the bond wires connecting all Kelvin sources together such that this doesn’t matter, or am I going to run into some devices switching faster than the others, or some not switching at all if this is not the case?
 

ronsimpson

Joined Oct 7, 2019
3,223
what goes on internally to the device - are the bond wires connecting all Kelvin sources together such that this doesn’t matter
At DC there is little difference. At AC it matters because every wire, every trace has inductance. On the switching edges the inductance from the Silicon to the PCB has volts across it. Wire up the Gate drivers wrong and this voltage gets into the Gate waveform.
 

Relaxe

Joined Jan 14, 2021
1
Internally, The Kelvin Source and the Source are connected right at the same point.
However, if you pass massive current, the Source bonding will exhibit a voltage drop, and the inductance will be high.
Controlling a fast and precise gate drive will be very hard. Just imagine what happens when the device suddenly start conducting!
To prevent all of this, there is a separate bonding wire for the kelvin source, minimizing the parasitic on the gate drive loop.
Since there is much less current going trough the Kelvin Source inductance, it is much easier to control.
1714391277558.png
1714391354576.png

As to "How much am I screwed?"
It should operate. Just maybe not at as high frequency as you wished it would.
 
Top