Preventing bjt breakdown

[Stretto]

I have a simple emitter follower with Vcc at around 330V. The bjt is rated for 350V. This is pushing it and the bjt's are dying. The circuit is designed so that for steady state the bjt should never have a large voltage across it(about 200V max).

The problem is startup. Right when the circuit is powered up V_ce = Vcc. As the base voltage increases(it has a capacitor on it so it takes a bit) it will reach 1/2Vcc and so the V_e = 1/2Vcc and so V_ce = 1/2Vcc(for DC, there is an AC swing of about 50V).

Is there any ways to save the bjt on start? I have a rather large capacitor on the base and so V_ce is rather large for a significant amount of time. Since Vcc is never above 350V I don't know why the bjt's are dying so easily but I guess the true max is about 300V and I'm over that for probably a few milliseconds as the capacitor charges up. But even without a cap the V_ce will be too large for a short period of time.


Obviously having higher rated bjt's would be preferable but unfortunately it's hard to come by cheap low power high voltage bjt's(and I tend to see more pnp's than npn's).

I might try to switch to a mosfet but it will have the same issue(I think I have some 400V but they cost about 10x as much as the bjt's).

Maybe a 200V zener across CE will help? Any ideas?


Heres an actual circuit of what I'm doing:



(I added the top bjt after the bottom bjt started blowing)

[infinia]

Insure Ic is 0 mA just after turn on or slow start the supply at Vc.

[Stretto]

infinia:

lol... that wasn't much help... I imagine a soft start is more complex than stacking transistors. Do you have any actual circuits that are pretty simple that will solve the problem(and not just shift it).

[digits]

Rod Elliot has a few you can look at on ESP.

[janneman]

Are you sure it is the start-up? Even if the full 330V appears across c-e, a 350V transistors wouldn't just immediately die.
Could it not be an overcurrent instead? If the current is too high momentarily, the transistor could die of SOA overload even with lower than max Vce.

jan didden

[CBS240]

The 10uf source cap for the J-fet requires some time to charge, drawing extra current at start up. 10uf is awlfully low AC impedance for such a small Iq. BJTs don't take this kind of stress at such a high voltage very much. Is there not enough AC gain so as you don't need it? Also you could try a lower Goss (Gm) J-fet.(less Vgs at the Id your operating.) There are plenty of cheap N-ch enhancement mosfets in TO-220 that can take this voltage, and driving it as a common gate from the J-fet is the way I would proceed. BJTs just don't like lots of volts because of secondary breakdown; probably the reason your BJTs keep failing. No use to continue melting sand IMO.

[infinia]

you have 3 time constants, make sure Vcc comes up last
Idss, Vz, Vc. Since Idss flows almost instantly , make the Vz much shorter and add an RC feeding the load. or you could supply a start up gate reverse bias. don't you have it in spice use DC transient?

[infinia]

Originally Posted by Stretto

infinia:

lol... that wasn't much help...

Well I thought I was helping you (ie learning to fish rather than catching them for you)
hopefully my second post will clarify my first.

[Stretto]

janneman: um, no. At startup the bjt is initially off so very little current flows. The peak current is 350/500k < 1mA and peak power is about 1/4W.

CBS240: Yes, but the peak current is still limited by the 500k collector resistor. I don't see how even a 1F cap would make much of a difference. It effectively remove the bypassed resistor for a significant time but the current will still be < 1mA. The large bypass cap is to reduce the low frequency cutoff point. I can't change it to much but even a factor of 100 shouldn't change things. I could increase the resistor values in proportion but then I start to run into other issues.

As far as I can see it is a voltage issue and not a current.

infinia: Well, I have no control over Vcc and even if I did create some soft start it would probably be much more complex and just as error prone as the original circuit. (I'd still have to worry about overvoltages)

It will be easiest to get properly rated transistors. Excluding that, I think stacking bjt's will be the 2nd best option. It's like how one creates higher rated capacitors and resistors. You put them in series. Not the best solution but in practice works well if done right.

[jcx]

junctions can breakdown/avalanch and not be damaged if there isn't enough current to cause excessive heating, C load can be charged up before the semi breaks down and provide the damaging high current/energy

totem pole/cascade resistor diver string may not charge up Q parasitic C fast enough if the circuit is switched to the precharged V+ rail, likely no problem if the supply ramps up slowly enough