Hello! I have a quick question about PN diodes in a simple buck converter application for a single quadrant series DC motor drive (expecting load currents in excess of 600A, especially during acceleration). Rather than interleaving two Trench-Stop IGBT buck modules, I am wondering if simply placing two identical buck modules in parallel would be ok, since I want to avoid the isolation inductors (and more sophisticated controls) required with interleaving.

The free-wheeling diodes in the module (APTGT600SK60G) are soft-recovery PN junction diodes. Since they are minority carrier devices, a little voice is telling me that they might not share current equally, and this could potentially be a problem (not 100% sure). But I would think that the IGBTs might force the free-wheeling diodes to share current equally since the IGBTs should share current. Also, if the modules are “identical” then this might help with current sharing too. Again, not sure. The modules will be mounted on laminated bus bar. So the question is: is it ever ok to simply parallel buck modules if the FWDs are PN junction diodes? Thanks for all the comments and advice.

Jay, your little voice is probably correct. P-N diodes are hard voltage sinks, so any variation in the wafer processing can make a difference of several mV in its forward voltage. Schottky diodes, will not share equally either, but will stabilize in a current ratio set by their series resistances.
Short of a current sharing control circuit, it is risky.

Given that you are using buck modules, there must be a buck choke from each diode to the output, so the current in each diode should be forced by the respective choke.
At light load (when the chokes go discontinuous) then not sure but don't care - 1 low current so not a problem, 2 loop response would dictate minimum load to stay continuous or a compromise that can operate over the CCM/DCM boundary

I would recommend you to make a "soft switching" buck regulator. Your all problems will be gone and forgotten.

Hi Jay, some commercially available single quadrant motor drives also use P-N diodes for their FWDs. They might not share the current equally because of their inherent characteristics but you might get away with it by mounting them in a bus bar that would more or less let them track their temperatures. The difference in their current might be acceptable in your application. Although it might be hard to just work with two modules if you are working with peak currents of 600A or more. It might be good to monitor the sharing characteristics at lower current levels before you go full load.

I think several people have not understood Jay's application - it is a DC motor controller so there is no buck choke - it is in the motor!!! This fundamental issue is certainly something to be very careful about. You will need to de-rate wrt current rating to ensure you don't have an issue. You should also thermally couple the modules as closely as possible. FYI internally to the module it is quite likely that several diode die are used in parallel to get that current rating so there is already paralleling going on. You could of course go for a higher current rating module in the first place from Powerex, Fuji, Dynex etc.

I take the point about the application, its a Buck like converter but in essence a switching power amplifier & the diodes are coupled directly in parallel.
From the thermal perspective the dies in a pack are all bonded to the same case metal ( thermally coupled) & selected by the manufacturer to guarantee a degree of similarity in the doping - the sharing won't be perfect but good enough.
for 2 separate packages neither case applies the forward voltage drop has to be different between the 2 & will change for each new motor driver.
Thermal drop from junction to heatsink (bus-bar) could be ~30K (perfect sharing)
Ultimately one of the diodes will heat & take more & more of the output current.
I think that all silicon & schottky diodes that I have looked at will fail in this manner.
For a similar case where we are driving a 50Hz pulsed current into a solenoid load we are using silicon carbide & above ~ 6A they do show a positive thermal characteristic ( on the datasheet) for this application 600A & unspecified voltage ( I guess ~30-40V) not applicable but it would be the only way to give a degree of passive sharing.
The diodes still need to be rated for the maximum current as at start-up nothing is definite.

A current transformer could be used to force the sharing of current between many different devices. Just run each signal into the same side of the current transformer with the other sides tied together. Watch that the volt-second product of the core is high enough to prevent core saturation. The people who make variable transformers (variacs) have been using this technique for at least 80 years. They may have app notes to describe it. The concept is simple, the impedance of the transformer winding to the devices running less current changes because the volts per turn has to remain the same on each winding. Everything balances in the core. Why are semiconductor guys so adverse to magnetics?

I basically agree with Larry. I have used relatively small magnetic devices to force current sharing in switched semiconductors for years. To force two diodes to share, use a center tapped winding. Connect the diodes to the ends of the winding and take the current out of the center tap. The coil will support the voltage difference between the diodes in order to force the currents to be equal in the two halves of the winding. The coil must have sufficient volt second rating to support this voltage difference (usually less than 100 mv) for the conduction time of the diodes and the off time must be long enough for the core to reset.

Jay / Larry - I have often wondered about using a CT to force current sharing but have never tried it. However I am guessing the practical implementation gets quite bulky when you are into the hundreds of amps unless I am missing something. Single turns formed from the bus bar outputs is no issue but how would you form a balancing choke (as I believe this configuration is often referred to)? Any thoughts?

Charlie, usually the "coil" is a single turn or one pass for each half through a toroid for cables or a planar core for bus bars.

Hi Bruce - Thanks that is interesting. I have used single turn chokes formed using planar / toroid cores to act as chokes to help with sharing but never thought to magnetically couple them to improve sharing further. Isn't power electronics a wonderful multidisplinary area to be in with all sorts of cunning tricks to discover ? The day I stop learning will be the day I get bored and decide to do something else. Very unlikely to happen :)