Transistor choice for switching inductive loads

Den fredag den 5. februar 2016 kl. 15.55.20 UTC+1 skrev Cursitor Doom:

why bjt?

-Lasse

Yeah we use a FET to switch a big inductor. I guess you want to think about what happens when you turn the switch off.

George H.

Mainly because it's a vintage piece of test equipment and I'd like to keep it original if poss. That's what they used a third of a century ago, you see. ;-)

It depends on what type of precautions are taken in configuring base drive and load line tailoring in the intended circuit.

These precautions will recognize realistic load lines for the SOA required in the parts that are candidates and will provide recommended base current profiles and eb junction bias, through suitable impedances.

Bipolar devices have published SOA and Isb ratings, when relevent to the intended application.

As switching is a lossy event, repetion rate and Tj may be important in applying the mfrs' data.

RL

Then quote the original part number so a replacement can be suggested.

RL

Well, maybe beta, and certainly Vcbo limits. Safe-operating-area is usually not a switching concern, but if the inductor turns on slow, it will be.

SOA becomes important if the freewheeling diode isn't there (or in the wrong place). Been there.

Specifically RBSOA, which isn't usually provided unless it's a switching type transistor.

What's interesting about RBSOA is, as charge clears out of the junction, the breakdown voltage rises. This is a real effect, observable on BJTs and diodes (and relevant to SCRs/TRIACs, but with the compounding effect that, when those turn on a little bit, they turn back on and latch hard, and they can also be triggered by dV/dt, with increased sensitivity right after turn-off).

In any case, all the usual parameters are involved in such a decision, so if you know the switching application, a selection of parts soon follows.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com

Yes, there is such a diode in circuit, but I must remember to check it as its failure could have caused the switching transistor in question to fail. Thank you for your timely reminder!

A BSW68 NPN transistor was the originally-specified part. The originality aim is NOT critical; nice, but NOT critical.

You're perfectly correct as ever, of course. But patience was never my strongpoint so I've inserted a 2N2219A in place of where the burnt out BSW68 was (thank god for sockets!) and the scope is working fine again now at 2/3 supply volts at least. I have it on a variac at the moment and will crank it up to 230V in due time, if it doesn't go 'phut' beforehand. The voltage headroom of this replacement transistor is only 5V and nowhere near the generous 100V or so the designers originally allowed for with the BSW68, but at least it proves that was the part that failed and I can try to find a better match now secure in the knowledge I'm not wasting my time on a knackered scope.

Up to full supply volts now and the case temp of the subbed part is 54'C which I'm not happy about. I could wrap one of those little heat sinks around it of course, but still feel I'm pushing my luck a bit too far if I left it like that. I'm guessing the life of that transistor wouldn't be very long.

Sigh. I was right. It's gone phut already. :(

BSW68 looks very hard to find, there is UK stockist of BSW67A (120V rather than the 150V BSW68) but price is GBP13 !!

I checked UK websites RS and Farnell and got these possibles for you to consider:

TIP47 ST13005 KSE13003 TS13003 FJP3305

2SC3902 BU406 BUX84 MJE15030 MJE15032 KSC2690A FJE3303 (hfe a bit low?) STX83003 2SC4614 STBV42 (hfe a bit low?) KSC2383 ZTX455 ZTX857 (*)

None are in TO-5 or TO-39 cases. The TO-126 and TO-220 ones could be easily heatsunk if needed. If the BSW68 was running bare (no heat clips) then the ZTX parts should also do and possibly the TO-92 ones with low-enough Vce-sat.

The MJ13003 / 5 family can be got free from inside a dead CFL.

piglet

Why thank you, Piglet. Your extraordinarily helpful response makes you a credit to Usenet! I have some of the parts you suggested already in my vast junk box. It's vexing to have so many obsolete/discontinued transistors, yet none of the type that actually failed in this instance. It would have been nice if the 2N2219 had just a bit more Vce capability, because it's a TO-39 with the same pinout as the failed part and just plugged straight in. But maybe a more robust tranny is called for than the original, even. I shall check out the ones you've suggested and have a play. Thanks again.

Well, that's absolutely nothing to worry about, being almost 100C below the limit! Actually more than that, because it's a metal-can type.

So what *did* cause it to fail? Overvoltage? Current spikes? Did you scope the pins?

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: http://seventransistorlabs.com

With this first attempt at transistor replacement, the scope would work fine if wound up to full working voltage on a variac, but the replacement transistor would blow if switched on directly to mains voltage. As I said there was only 5V leeway with this device and I'm guessing the surge accompanying a sudden switch to full supply volts exceeded that leeway. Any way you look at it, that conveniently-cased replacement isn't robust enough for the job; I only tried it because it fitted physically, wasn't too far out electrically, and I had 50 of them to play with so could afford to lose a few. :)

Hmm, well without any better ideas (waveforms, load characteristics, etc.), it may be a good idea to chuck a TVS on there, when you get the properly-volted replacement. *shrug*

Tim

-- Seven Transistor Labs, LLC Electrical Eng> So what *did* cause it to fail? Overvoltage? Current spikes? Did you

With this first attempt at transistor replacement, the scope would work fine if wound up to full working voltage on a variac, but the replacement transistor would blow if switched on directly to mains voltage. As I said there was only 5V leeway with this device and I'm guessing the surge accompanying a sudden switch to full supply volts exceeded that leeway. Any way you look at it, that conveniently-cased replacement isn't robust enough for the job; I only tried it because it fitted physically, wasn't too far out electrically, and I had 50 of them to play with so could afford to lose a few. :)

There's already a snubber diode across the inductor as part of the original design. I know you think my approach was unscientific, but it was Sunday and I was only goofing around out of boredom. I wouldn't have been happy to leave that sub-standard part in there as a permanent solution.