View Single Post
Old 12-12-2009, 05:57 PM   #2739 (permalink)
Mark Weisheimer
Join Date: Aug 2009
Location: Columbus Ohio
Posts: 62
Thanks: 8
Thanked 13 Times in 10 Posts
Originally Posted by NiHaoMike View Post
Some IGBTs can be directly paralleled and some cannot. Check the datasheet. In general, Non-Punch Through (NPT) can be directly paralleled and Punch Through (PT) cannot.
In either case, it might be possible to use a bundle of wires instead of one large one so the wire resistances would tend to equalize the currents. Note that diodes generally cannot be directly paralleled, so it's a good idea to use a bundle even if you're using MOSFETs or IGBTs that can be directly paralleled.
You can use parallel IGBT devices if you have enough of them
and "match" them to each other.

The best way to use them is to test them for conductance and match
them as close as you can.
Say you have a box of 40 of the same type of IGBT device
and your design is going to use 12 of these devices.
To make these matcing measurements you will need a meter
that will read millivolts DC and
you would need 2 regulated bench supplies. (or, batteries and good regulators) One that will supply 15vdc at low current to drive the gate
and the second that will regulate current at 1 amp at whatever voltage develops across the drain and source. so a pair of 0-30VDC 0-1amp bench supplies would do nicely.

You would pick the 12 devices that fall within the smallest range using the following process that Nate at DIYElectriCar explained:

"What you want is the IGBT to act as close to its neighbor as possible.
Make sure the IGBT tests are all at the same temperature.
Test each IGBT individually but you should test them all on the same pad/heatsink.
Put 15V on the gate and 1 amp across the emitter and collector.
Measure the mV drop across each emitter to collector. Write it on the device.
Start each IGBT at the same temperature and monitor the temperature
rise over a set amount of time. What you want is consistency in voltage drop and temperature rise.
The final design should mount them on a common heat sink to help keep them thermally equal.
This is the basic way to test, you can vary the parameters but be consistent on all units."

Modules can be "matched" the same way, but should probably be tested using a higher current and a larger heatsink.

So, you can use what you have or can obtain at bargain prices.
There are plenty of EV's running around with parallel IGBT's in the controllers.


Last edited by Weisheimer; 12-13-2009 at 11:13 AM.. Reason: Correcting E/C for IGBT versus D/S for MOSFET
  Reply With Quote