Overrated transistors

This whole thing begs an interesting question for me as a hobbyist. Using BJTs in parallel to distribute the current load is fairly easily handled with emitter degeneration resistors and otherwise simply tying bases together and collectors together. I see this topology frequently, when more current capacity is required. And it works reasonably well.

But what would work if, instead of wanting higher current, one wanted much higher voltage support than each BJT could handle? Stacking them and distributing the voltage across them doesn't seem nearly as easily managed. I'm curious if there is a simple configuration or if this would require rather a fair degree of supporting circuitry to achieve well.

In this application, if he OP actually does want 50V or maybe even more and does want to use these particular parts as well, the SOA chart shows that 25V @ 20A could be sustained at DC (with appropriate heat sinking.) So the 50V @ 2A should be easily possible with, say, two such devices if the Vce could be held to about 25V. But the two would need to do a credible job in sharing the load voltage.

Is there a 'standard' and not overly complex toplogy for this?

Jon

Can't disagree with that. However, I think Nico just wants to check his mod to his HS at this point.

Not more than two days ago, I saw that topology cited, probably here, too. Though I vaguely recall that it was a digital 'on off' application rather than linear.

It was IIRC, a common emitter transistor, with it's collector connected to the emitter of the next transistor 'up'. Seemed real plausible, though I didn't explore their biasing scheme at the time.

--Winston

I probably am too ignorant to know what I read, then. I don't recall it. But if it is on-off and not linear, then I'm not nearly as curious about it.

It's the biasing part I'd care about. The starting connection is obvious. But the negative feedback to stabilize is the question. For example, if I split the design into two BJTs, I could easily arrange a resistor divider to get the center voltage and then compare it with the collector-to-emitter point of the stacked BJT pair, but the scheme, while simple at this vague point, becomes just a little more complex in considering exact details. It gets more interesting stacking more.

Anyway, I'd be interested in seeing a discrete part topology for it that I could actually use to figure part values and then test out here for fun.

Thanks, Jon

My original dummy load worked that way. 4 TO3 transistors in parallel with 0.15 Ohm resistors in the emitter to balance the current. This setup handled >300W without problems. A SUT (Supply Under Test) with some nasty spikes on the output killed it.

Well, 50V is actually the highest I can get from the power supply I used. The actual design target is 400V / 20A (but not more than 400W in total). That has suited my testing needs so far.

Anway I ditched the transistors and ordered 12 MOSFETs to do the job. Easier to control and no need for current balancing resistors.

--
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
nico@nctdevpuntnl (punt=.)
--------------------------------------------------------------

"Nico Coesel"

** That is SOOOO wrong it is funny.

Operating power ( switching) mosfets in parallel groups requires careful device matching AND the use of ballast resistors.

The tempco is about 5 times more than with BJT too.

Nothing easy to control about that.

.... Phil

Well, emitter degeneration is only approximate balancing, anyway. And I suppose there are modifications possible to help deal with known spike events. Everything has its limits.

Understood. You don't need to present inductive and capacitive loads, either, I gather. (For example, testing the Supply Under Test with a large transformer winding or motor.)

Darn. Now everyone will see you don't need any further help to stay nicely within the SOA of your BJTs and that will reduce temptations for an answer to my new question!! ;)

Jon

TO3? Well, those were from the good old days and everything was better in the good old days, or so people say ...

But you did check the SOA graphs for those, did ya? :-)

If it's any comfort, my 1kW RF dummy load that I built as a kid rusted out (no joke), so I threw it away last year.

--
Regards, Joerg

http://www.analogconsultants.com/

"John Simpleton"

** LOL !

The term " non- repetitive " normally alludes to pulse conditions being repeated only such that the chip does not heat much.

.... Phil

(...)

(Googles PMPO)

Yech. That's a new one on me.

--Winston

Back to the seventies and "6 x 9"s and other than name brand (middle class, early hi-fi years) manufacturer specs.

There were a handful of companies that made good gear and released real specs (Pioneer, Marantz, etc.).

Most of the car stereo makers are where the big lies were on amp specs.

"Jon Kirwan"

( snip)

** All you need is a resistive voltage divider - assuming NPN and collector output, equal value resistors connected from base to base and then the top collector. Drive the bottom device only.

The output BJTs need to be darlingtons or have drive transistors so current gain is high enough to prevent loading the voltage divider.

.... Phil

Methinks that the power rating applies only when operated in an LN2 bath with a big heat sink.

?-)

We wish. At about 1/10th of "rated power". &^Y*#)_@%&*@#^ specsmanship.

?-)

Not that simple.

It's apparently a 'Secondary Breakdown' limit, Vc > 26 V. Please see the middle of page 4 "FORWARD BIAS SAFE OPERATING AREA":

Also section labeled 'Second Breakdown' in:

--Winston

I have seen working totem poles of three transistors tall to the load. Getting good bandwidth is rather challenging.

?-)

"joseph krackpot kook"

Methinks that the power rating applies only when operated in an LN2 bath with a big heat sink.

** The full rating ( ie 625 watts) is available up to 25 volts Vce.

Then drops off rapidly due to " second breakdown ".

Rather poor * continuous* Pdiss at high Vce is characteristic of switching BJTs and most Darlingtons ever made.

However, the device in question CAN take 4500 watts for 10mS and 9000 watts for 1mS - single shot.

Not too shabby.

.... Phil

Hello Winston,

You are correct, I interchanged the voltage and current scales. Thank you for correcting my values!

to Nico: It means that a single device can't stand safely 100V 2A and that there is at least a DC SOA problem. When I extend the graph, 100V would result in just below 1A (for 25 degr case temperature).

Sorry for the confusion,

Wim PA3DJS

Please remove abc first in case of PM