help BFG135A

Didn't they show a maximum current on the plots? Since it's beta curve doesn't show much drop with current, it might work OK above that current (and the part also has emitter ballasting, which means it shouldn't have current-hogging second-breakdown problems), but to be safe you should stick to that current. It's a small part, you could use some in parallel. Again, the added internal emitter resistance should help in simplifying your parallel connections. And this will help improve the high-frequency pulse response.

As previously mentioned, you may find that you're dissipation limited or Ic max limited.

Ic max is on the data sheet. Although it's been suggested you might exceed this with a low duty factor usch as your 2%, I personally wouldn't suggest it.

You don't mention the supply volts on the collector - so you'll have to calculate device dissipation ( I assume you can do that ? ).

Then you'll have your answer.

Graham

Just to add a comment, normally paralleling power BJTs requires individual external emitter resistors to insure one part doesn't hog the current (as it gets hotter its Vbe drops, insuring that it takes a larger current share and gets hotter still). I plotted the BFG135A's transconductance curves, and concluded the effective integrated emitter ballast resistors amounts to about 1-ohm for a transistor (i.e. hundreds of higher-value resistors in parallel, equaling 1 ohm total). The datasheet maximum current is 150mA and we note that 150mA across 1 ohm is 150mV. That's a lot, enough to moderate the local-heating current-hogging effects from one single transistor. Whether we have one ohm of integral emitter resistance or not, it looks like it's enough to allow for direct paralleling of these transistors.

----- make that "guesstimated" rather than "concluded" ------

----- make that "one hopes" rather than "it looks like" -----

This is clearly a case for some bench measurements to determine the missing datasheet parameter.

My first experience with "integrated emitter ballast resistors" in a power transistor was with special SExxxx parts made by Fairchild in circa 1969, IIRC. They offered both NPN and PNP versions, and their performance was spectacular, basically comprising a large number of small high-frequency transistors each with current-equalizing emitter degeneration, in parallel on one chip.

With these fabulous transistors I designed a high-performance audio power amplifier that pushed upwards to the 1MHz region, and with Paul laid out PCBs, and a dozen or so were built (by others - I never made one for myself). The amplifier bested top-name professional amps in a series of tests at the Tweeter lab in Cambridge. Sadly Fairchild discontinued the transistors (too expensive to make I suppose - they had better uses for their IC fab lines than make large die-area power transistors) before I was able to purchase a lifetime supply.

Although I've carefully watched in the intervening 35 years, I haven't seen anyone else offer similar parts.

I have the old Sanken datasheet (now discontinued by them?). And a Mospec datasheet. What do you have?

"Winfield Hill" a écrit dans le message de news: snipped-for-privacy@drn.newsguy.com...

??? What did they had so fabulous that some todays parts can't do (do you still have a datasheet?).

I have made really amazing amplifiers with Sanken-Allegro LAPTs (I think linear audio power transistors) 2SC2922-2SA1216. They're very rudge and, from the curves, are probably ballasted transistors (I can post datasheet if needed). I guess Hitachi and Toshiba made some too.

I wouldn't touch Mospec parts with the proverbial bargepole. Our manufacturing subcontractor substituted theirs for genuine Motorola MJ15003s and 4s in one batch and they just expired if you looked at them wrong.

I sawed the lid off one and the die was easily half the size of the Mot part.

Graham

p.s. I could never work out how Sanken made such fast and rugged devices ( I'm thinking back around 1980 now ). Motorola and RCA had nothing to compete. I guess it's the fab method you mention.