Parallel Transistors?

[George]

I'm building a linear current regulator using 5 x TIP35C NPN Transistors, they are each rated at 25A continuous and 125W dissipation. I'm wanting a current regulation of 100A and requiring a power dissipation of around 300W.

I'm wondering about running these in parallel, I know it's quite common practice to balance paralleled transistors with emitter resistors, but due to the lack of headroom and high currents I'd rather not if I can avoid it.

Is it possible to use balancing base resistors instead of tieing all the bases together? Or will Hfe differences stuff me up.

If I do have to resort to emitter resistors - what voltage drop should I be looking at (bear in mind my total headroom is only circa 2V)

Thanks

[chuck]

The only way there would be emitter resistors , These make sure that the transistors carry equal current other wise one would carry more than the other and this would casue one big bang

I would use Mosfet's by using these you would not need the emitter resistors plus they carry current a lot better

[George]

Thanks, but aren't FETS more of an issue to use as a current regulator? As I understand it, an NPN transistor is basically a current amplifier (so to a point self regulating). If I was just switching then no doubt I'd use a FET but I'm never going to be saturating the base leg, just holding them in some sort of in between state, I thought transistors did this better than FETS? Also the reason I chose these transistors was because they had 125W of power dissipation available - I don't know of many FETS that cost $2 and can get rid of that much power.

It's a power supply for a laser and efficiency is not an issue. The current regulation needs to be bang on else I'll be making some very expensive bits of dust.

Thanks

[wjsmarine]

Hi George, members.

Dusting off the cobwebs I seem to recall the emitter resistors oppose hfe as emitter current increases due the higher volt drop taking place - this is their purpose and allow each parallel device to balance.

Given we are talking .6v turn on you can see that not much voltage is required to throttle back the transistor drive, so scale your emitter resistors accordingly e.g. if 10A flowing then only 0.06 ohm will give you 0.6v but still dissipate 6W of heat.

Also don't forget those transistors will be cooking too with Vce*Ice Watts each and are subject to thermal runaway if not controlled. Mosfets have a positive temperature coefficient and therefore don't runaway, instead shut down as they get hotter. I hope this helps.

Regards,
Bill

[wjsmarine]

Oops! Wrong information.
Like all semiconductors Mosfets also have a negative temperature coefficient NOT positive like I mentioned previously. But they still tend to shut down as temperature rises and therefore current share better than bipolar transistors.
Cheers,
Bill

[George]

But still - are FETs as easy to control in a current regulation situation? Also - are there many that can dissipate over 100W of power? I've never used a FET before to limit current and am not really sure how easy it would be to get a good steady current regulation from one.

I was thinking - what if I use a length of copper wire say 0.1 Ohm from each of the emmiters, the copper would naturally increase it's resistance as it gets hot and let less power through the transistor - so then it would have a semi dynamic leveling system rather than just a static resistor?

[wjsmarine]

Hi George, members.

Mosfets are capable of big power, no problem, obviously heat sinking is a factor. I would be careful using pcb tracks as a resistor where you are relying on heat as an aid in resistance changing - there is a fine line (no pun intended) between getting hot enough and a fuse, but if you decide to do this see Roman Black's site
http://www.romanblack.com/pcbtemp.htm
for a utility to allow calculating track sizes.

I have had good success with an arrangement that controls Source current by robbing Gate drive. Use a Source resistor (N channel device) to ground and place the b-e junction of an NPN bipolar across it. The collector goes to Gate with a 1k. This will rapidly switch so the average Source current is controlled by the Source resistor - value determined by 0.65/desired current e.g. 0.065 ohms will give 10amps average.

If this meets your needs I would start small and scale up accordingly.

I hope this helps.

Regards,
Bill