Transistor Power Dissipation

Hi,

what transistor are you talking about? Give a link to the manufacturer site datasheet.

When you talk about "very short pulses" you need to give a time and a duty cycle, so we are able to validate it.

A BJT specified for 60V and 1A does nothing say about allowed power dissipation and temperature rise.

ElecDesigner said:

Basically I have the scenario where the dissipation is about 0.5W for a few 100 us every 100ms and am struggling to believe i need a transistor rated for 0.5W dissipation.

Click to expand...

You definitely don´t need a 0.5W one. But a 0.5W * 100us / 100ms = 0.5mW one is too small. You need something inbetween.

Klaus

In the absence of SOA curves (small signal transistor datasheets down provide anything like that), what approach do we take.....

Click to expand...

For applications in pulsed opperation this is an essential information, there is no alternative. Depending on the material with which the package is manufactured, as well as die size chip, among others, is that the manufacturer is able to characterize the performance of thermal diffusion by conduction in the silicon for amounts of energy of short duration; there is a limit above which the gradient will overheat the p-n junction, even if the case is only slightly warm.

Are you going to violate Icmax or Vcemax during pulse period ? Or just power ?

My gut tells me find a transistor with pulse ratings.


Regards, Dana.

ElecDesigner said:

In the absence of SOA curves (small signal transistor datasheets down provide anything like that), what approach do we take.....

Click to expand...

I see pulsed power capability curves for many small signal transistor from quality manufacturers, e.g. Infineon.

ElecDesigner said:

Say I have a small signal rated for 60V, 1A but the power dissipation spec is something like 200mW (continuous).

Click to expand...

Do you have a specific transistor in mind?

Your pulse width fall in an uncomfortable range, for
analysis. It's not short enough to be "adiabatic" (though
this analysis produces the most uncomfortable
numbers) yet it's not wide enough to be fully settled
against the likely thermal time constant of the mass.

A 1/1000 duty cycle would likely make the DC equiv
dissipation look fine.

Questions include "where in the device, is the dissipation?"
(collector is a good answer because it's large, the dissipation
likely is diffuse and it's well bonded to the paddle for a high
heat capacity; a small region surrounding the base, might
put that 0.5W*100us=50uJ into a much smaller volume
that doesn't respond well to (say) time at 800C. Not that
this is the case, necessarily, but you'd need to know device
geometry and intra-device power dissipation distribution to
say what is.

A rule of thumb I have used, and gotten away with, is if
peak power is less than 1/10 the DC power / current rating,
and duty cycle averaged power / current is less than DC
rating, you're OK.

But maybe I just haven't found the challenge that breaks
that thumb, yet.

Infineon BC847 shows about factor 100 peak power for 100 us pulse with 0.001 duty cycle.

I'm thinking of parts such as the BC817-40 or any of the Diodes Inc offerings. Diodes Inc datasheets are very brief generally. The dissipation will be in the collector.

Maybe the answer if to use parts that do show this info in the datasheet.

Hi,

there are other brands of BC817 ... with more detailed datasheets.

Klaus