Peak power rating of BC847

On a sunny day (Mon, 25 Aug 2014 06:01:11 -0700 (PDT)) it happened Klaus Kragelund wrote in :

There must be many transistors for >= 1 A, why must it be BC847?

I'm with Jan on this -- why use that transistor if it's not rated for the service?

In my forays into the realm of pulsed-power, I've found that there are devices out there that are rated for it (although maybe not transistors in SOT-23's). Generally if a device is rated for pulsed power its data sheet will have a chart of dynamic thermal characteristics showing a family of curves of max current (or max power) vs. pulse width for various duty cycles.

Have you looked for a transistor that's rated for pulsed-power service?

I think a bc817 is rated for 1A pulsed

-Lasse

Look at Fig 1. The beta is not specified at 1 amp, but is seriously headed towards zero. If you can get 1 amp through it, with tons of base current, Vce_sat will be huge. Fig 3.

Why not a bigger NPN, or a small mosfet?

This is not my design, one I am responsible for getting ready for production. I have pointed out the over-rating high current to the original designer, but the response is that it works, no discussion about if we are on the threshold robustness wise.

So my aim is to arm up for the "review battle" when that comes

Regards

Klaus

You might insist on checking it over temperature. And maybe deliberately reducing base current - simulate beta variation - and see if it still works.

Most people are reluctant to use a part out of its guaranteed limits, unless there is some huge payoff. In this case, I suspect that there isn't. "It works" doesn't mean the next hundred will work.

My point exactly. I always try to stay well within recommended ratings

Cheers

Klaus

There is (or was?) one that is characterised to 500mA:

But figure 2 does not look too encouraging. So when it gets hammered with base current it might work. There should be better candidates.

Also, you need to get it out of saturation within that microsecond and since that's slow you'll have to start earlier in the pulse. I've been (and am right now on the bench) pulsing fast RF transistors quite hard but there I had no choice, they don't build pulser transistors in the tens of GHz. Works great. But in the BC847 case I spend half a penny more and get a device that can take 1A.

I love it when they use the "it works" argument. Testing can't prove something works. Testing can only prove the presence of a fault, not the absence unless you can test all possible combinations of parameters, many of which you usually can't even control.

I was once asked to construct a timing analysis of the bus for a board being built. The software guys (meaning the FPGA coders) wouldn't cooperate with me so I had to find the source and reverse engineer that to get the clock cycle timing. I found the Flash memory was being used with a timing violation on one parameter. This was part of an existing circuit and had been through verification testing. I tried to explain that passing a test does not mean the part will never fail but ears were deaf... oh well it was only a military radio. Obviously it's not important or anything.

True, but temperature testing can find a lot of bugs and marginal cases. You can't think of everything at design time, especially when you use a lot of other peoples' ICs and things.

I'm not sure what your point is. My point is that you can't rely on testing to prove the design is bug free. It's not even a question of did you think of "everything". "Everything" is too big to be tested, especially for software... even if you could think of it all. The best shot is good design practices, but never assume that it is "correct" because it passed some test.

My point is that temperature testing can find a lot of bugs and marginal cases.

Thank you for that insight.

what sort of Hfe are you seeing from that part at 1A

A lot of the old ZTX series devices (Now Diodes Inc., originally from Ferranti / Plessey / Zetex in the UK) are very good at high current, with high current gain at many amps. I remember using these a long time ago:

I have more recently used this one from NXP, it can do 4.3A (or 8A for

Chris

That's a nice part, thanks

Cheers

Klaus

But be careful, only 20V. And with BJT they usually mean it :-)

production. I have pointed out the over-rating high current to the original designer, but the response is that it works, no discussion about if we are on the threshold robustness wise.

Make it clear (up front) that the designer is signing his life away by continuing to thoughtlessly defend the design and demand that management sign up on enforcing it.

?-)

Indeed! My personal opinion is that you shouldn't run a bipolar device at a current level above IKF... because you are current-crowding the emitter periphery and just begging for a failure.

In my chip designs I tend toward IKF/10 as a limit... otherwise the devices won't match and track. ...Jim Thompson