parallel FETs

nope, won't do... if using them digitally, you might parallel them as they are, or put a very small resistor in each source and actually feedback them which is the more complex way to go

to

I'm sure you are already aware of "thermal runaway" in BJTs. But there have been some comments about using paralleled hexfets on sci.electronics.design, a few year ago, I believe. I gathered that safely paralleling power MOSFETs only works well when switched completely on/off, where V(GS) is substantially above the zero thermal coefficient point. Somewhere in the discussion about the problems of the temperature coefficient for Id, given a fixed Vg, the Hitachi

2SJ160 and 2SK1056 family was mentioned as attractive because of their zero/negative coefficient in this regard.

There may also be serious oscillations to worry about, particularly when operated linearly, as well, if memory serves. I'm no expert in this, at all. I haven't tried to analyze this problem; and never ever tried to do something like you are going to try -- hopefully someone who has can comment. But controlling large inductances at large currents with paralleled HEXFETs (or MOSFETs or BJTs) and where the actual motor load/acceleration can vary in spurts and jerks, I'm sure, will take some careful attention to get working well -- probably combining mechanical and electrical means.

You might use google to look up some threads in sci.electronics.design and/or otherwise post this question there.

Jon

Hi, Fets . The first requirement is to develop the gate drive pwm that has a rise time of at least 100ns and a range of zero to 12 volts. The fets should be of the same part number from the same mfg and connected close together as possible on a heat sink. The source leads as short as possible to the common connecting point. The most important requirement is the value of the series gate resister for each fet. If you are using To-220s use 33 ohm resisters. If you are using To-247s use 15 ohm resisters. The above is if the number paralled is below 10. If you are paralleling more increase the above resister value by 20% for each additional 10. The gate drive return should connect as close as possible to the source of the fets to the center point of the common connection. Also a large filter cap should be placed as close to the fets and the free wheeling ( fast recovery diode or schottky diode ) path. The motor can be located at length away but it is best to keep the switching transistors near the battery. Bipolars. No one has been able to parallel bipolars successfully without careful matching. They can be purchased in modules that work great but they use large die and are very close together.

I hope this is of some help.

Ray

to

to

Have you considered using the FETs as on/off switches and controlling their duty cycle to control the speed of the motor? Much more efficient. Fewer heat problems, among other things. If driven hard enough, load sharing should be a minor issue. Switching frequencies of a few hundred hertz should be adequate.