I'm sweating this one.

[Duo]

Well, this is a problem. I have almost entirely built an amplifier of my own design. I'm happy with the design and have implemented it in some units I've built for friends for Christmas presents and such. They are pleased with the outcomes and they have so far been reliable.

The problem here is that I'm making one for myself and I missed one important transistor specification until I was done mounting them in the case. They are extremely difficult to change so I've basically decided to leave them in. In the worst circumstance I can change them, it's that I really don't want to.

What I'm asking (for those more experienced in the amplifier field) is if I can get away with 140V complementary transistors in the output stage on +-70V rails. This +-70 is what the power supply reaches under no load and I have observed it to drop a few folts under a test load. I would imagine that during high power swings on the amplifier, the power supply could droop a fair bit, thus limiting the voltage the transistors would have to withstand.

Does anyone think that I should simply just get them out of there/change the power supply voltage? Or should I be able to swing it at such a close measure?

[Charles Hansen]

You should probably be OK, although it is not the best design practice. Depending on where the part is in the schematic, it probably only sees the full rail-to-rail voltage (140 volts) when the unit is driven to clipping. This probably isn't too common in actual operation. Secondly, anytime I have put a good name brand transistor on the curve tracer, the typical breakdown voltage is usually *at least* 10% - 20% higher than the rating, and sometimes as much as 50%.

The thing to be mildly concerned about is high mains voltage, although this is far less common in most places than is low mains voltage.

Good luck.

[ChocoHolic]

You are talking about complementary bipolars?

I would not be so confident that they are working reliable in your design. Besides the simple voltage rating you might run into trouble
with save operating area.

My two cents:
Try to get the proper transistors or reduce the rail voltage by
about 15%.
Max output swing will be reduced, of course.
Let's guess from +/- 60V down to +/-50V.
For your ears this means 1.6dB less, just a small step.
But for the reliability it will bring a remarkable step.

Cheers
Markus

[Charles Hansen]

Sorry, I missed the word "output" in the original post. I was thinking of drivers. If they are the output devices, then I agree with Markus. The Safe Operating Area (SOA) will not have enough safety margin.

[Duo]

Yeah, you see, I normally would not do such a thing, and it was a simple mistake to make since I had brought a set of outputs with me to make the amplifier not remembering that they were only 140V.

They are Onsemi MJ15003 and 15004 transistors. I know that they are highly reputable transistors.

I'm thinking that it would be easier for me to reduce the power supply that it would be to change the output transistors. They are mounted in such a complex way and really are buried in. I've employed lots of protection and the construction is really solid. This of course doesn't stop over-voltage from getting to the transistors in the first place.

Like I said before, I know that at clipping into a normal load, the power supply would not be able to sustain +-70V at all. But back emf is my consideration here. The reactivity of a typical speaker is what really bugs me.

[traderbam]

These transistors are guaranteed for 140V so they will survive a little above this. Therefore, I wouldn't be too worried about 70V unloaded.

For the back emf issue, make sure you have diodes between collector and emmitter of each transistor - like 1N4007 - to ensure the speaker output does not exceed the power rail voltage.

Like Markus, the SOA would be my biggest concern here. You can only draw 2A max at 70V Vce and 0.5A at 140V Vce or the transistors will be damaged. Are you using several of them in parallel?

[K-amps]

Be safe if you want... however me thinks you want to hear what you want to hear so here goes.

I have ran Mj15003/4 (140vce) from MOT (not other sources) at rails of +/- 90vdc and ran the amp into clipping where it clipped at over 65volts VAC into 4 ohms. I had no issues at all, the amp worked fine without any issues. Granted I had 16 devices per channel..... so yes it can be done. Heck I have seem many published designs with MJ15003/4 with designed rails of +/- 68vdc and I know of one manufacturer, Aragon that used the Mj15003/4 with 140 v rails (+/- 70).

So if it's a Mj15003/4 from MOT, go for it, if not.... take the safe route.

[Stuart Easson]

Perhaps you could add a low insertion loss regulator, chop the rail volts down to the fully loaded drooped value minus a couple, and then you should achieve close to the full output without stressing the output transistors at idle and low loads...presumably you have already measured the droop under load.

Another idea would be a cascoded output stage, effectively more output transistors, but perhaps overall it would be simpler...

Stuart

[jackinnj]

Originally posted by K-amps

So if it's a Mj15003/4 from MOT, go for it, if not.... take the safe route.

just a note -- there are no more MJ15003/4 from MOT -- they spun out the discrete semi biz (well, a big chunk of it anyway) as ON-SEMI. It's the same fab and everything, just a different corporate entity.

[Duo]

Well, some ideas here have highly gripped my interest. I'm not into cascoding really, since I don't have any more mj15003/15004 left and I don't feel like cascoding with mj21193/21194 or something weird in the circuit.

I'll reiterate that my transistors are onsemi devices and something does make me believe that they'll handle a lot of beating.

I do, however, like the idea of adding regulation to the power supply, since I have a lot of big power transistors hanging around that would be good for that. I'd imagine that dropping 10V or so would make things more than acceptable. I'd also be able to deal a bit with drooping problems and perhaps have a better sounding amplifier.

Also, yes to the question about running transistors in parallel. I have four pairs in each channel. And a very large heatsink with the ease of additional active cooling and thermal shutdown for I hate the thought that my amplifier should die of overheating.

What makes a regulator so appealing to me is the fact that I have a ton of room left in the case for such additions. I could easily add all kinds of goodies in there.

I'll think about this some more but I do like the idea of regulating the works. Thanks for the suggestion.