transistor voltage rating Vs supply rails
with rails of around +/- 50v, are bc556 and bc546 ok for Q7 and Q8 ?
with rails of around +/- 50v, are bc556 and bc546 ok for Q7 and Q8 ?
Attachments
Last edited:
Q7 & 8 can potentially have the total supply voltage across them. They are rated at V CEO 65volts. Probably be OK but I would choose a better component voltage wise.
As Jon says...
Simulate the amp at full output and then look at the voltage across each. It will be close to 100v peak. I would say you need a Vce rating of at least 120v here. Q8 is simply a current source and so could be something altogether larger like an MJE350. It doesn't need be a complement to the other one.
Simulate the amp at full output and then look at the voltage across each. It will be close to 100v peak. I would say you need a Vce rating of at least 120v here. Q8 is simply a current source and so could be something altogether larger like an MJE350. It doesn't need be a complement to the other one.
Beware that while your rails might be 50V any glitches on the mains that get through could be a bit higher.
I always add 100nf and 100uf across the rails close to the amp for insurance.
There should be a cap between R5 and ground or your dc offset will be unstable.
I always add 100nf and 100uf across the rails close to the amp for insurance.
There should be a cap between R5 and ground or your dc offset will be unstable.
Last edited:
thank guys, using ltspice and taking Q7 for example, the emitter is at -47v, but the collector swings close to 46v or so at full output.
well spotted, they are omitted for this ltspice cct, as was the input filtering.
well spotted, they are omitted for this ltspice cct, as was the input filtering.
Running 65V transistors at 100V is asking for trouble, expensive trouble. 150V plus parts are a much better idea. 2SC1845 / 2SA992's are a possible pair (only 120V though).
Or cascode them with high voltage devices.
Or cascode them with high voltage devices.
2n5401 2n5551 are usually in stock over here. 150 v parts.
VAS works hard in my amps, I usually use MJE15028/29 with a heat sink. Heat loss in a TO92 package is a bit iffy.
VAS works hard in my amps, I usually use MJE15028/29 with a heat sink. Heat loss in a TO92 package is a bit iffy.
> with rails of around +/- 50v
At full power, these parts will have essentially 100V across them, because you need to slam the output stage's input ALL the way to get full power.
At full power, these parts will have essentially 100V across them, because you need to slam the output stage's input ALL the way to get full power.
I think this thing will die when Q7 and Q8 are replaced with suitable bjt's.
As soon as one of the mosfet's switches off, there's is this 470nF (C4) fellow.
To be charged-discharged-charged by the MJE's. No fun.
C3-R13 should be addressed as a zero, but with overall fb might provoke a nasty pole popping up in the response as a high pitch doorbell alarming the dogs instead. C2 will not suffice here.
Why R7 not 560E like R5? Better DC nulling at output to avoid suggested ground-cap.
An extra small cap (100p) to ground there is not unneccassery.
As soon as one of the mosfet's switches off, there's is this 470nF (C4) fellow.
To be charged-discharged-charged by the MJE's. No fun.
C3-R13 should be addressed as a zero, but with overall fb might provoke a nasty pole popping up in the response as a high pitch doorbell alarming the dogs instead. C2 will not suffice here.
Why R7 not 560E like R5? Better DC nulling at output to avoid suggested ground-cap.
An extra small cap (100p) to ground there is not unneccassery.
Alas high output capacitance power transistors add a lot of HF distortion at the VAS stage, since Cbc is very non-linear. Cascoding or emitter-follower in the VAS can allow low power devices to be used without excessive dissipation and will perform better. Ideally the Cbc of the VAS device should be totally swamped by the compensation capacitor, or the VAS device should not see voltage swing between base and collector, or both.
Using a single power transistor for the VAS is a bit like paralleling the compensation cap with a high value varactor! A small signal VAS transistor is like adding a very low value varactor across the comp cap.
Cascoding means the Cbc current is dumped to the rail, not the VAS input.
Emitter follower again means the Cbc current is isolated from the input section by the first transistor (it reduces it by a factor equal to the gain).
So in either case the second transistor can be less performant, but higher voltage and higher power, the linearity is basically defined by the first transistor.
The VAS places big demands on transistor performance, that's a given!
Use 2N5401/5551 , period.
NO WAY 65V transistors can stand 100V even for a millisecond, because they will breakdown way before that.
Imagine Zener but in a destructive way.
NO WAY 65V transistors can stand 100V even for a millisecond, because they will breakdown way before that.
Imagine Zener but in a destructive way.
Actually 65V transistors have a manufacturing spread, some may appear to survive, but reliability is terrible. Transistor voltage rating depends sensitively on process parameters so the actual rating is the worse case value for the batch, not any individual device. This is why people get away with select-on-test for 2N3055's in high voltage circuits.
Zener's are binned for breakdown voltage, completely different story.
Zener's are binned for breakdown voltage, completely different story.
Generic transistors are often binned for breakdown voltage, too (ie, TIP41A/B/C). If the entire batch passes 100V, the it gets the C, if 80, then B. But you still get a spread within the batch - you could get full rating, you might not. But other parameters are pretty well correlated with breakdown - like gain. The full 100 volt parts would tend to be on the low side of the gain spread. The same is true for parts that aren’t typically binned (or may be binned in some less obvious way) - higher vceo, less gain. Selecting a nominal 65V part for 100 volts puts you on the low side of the gain distribution - which may defeat the purpose of choosing that part in the first place. You wouldn’t end up putting the “best” BC546’s in there - you would end up with the “worst” ones.
C3 and C4 are rarely seen 'solutions' to bypass ac over bias resistors. When things run close to to limits, perfect poles creates sound meltdowns.
However, one can find small caps over regular bias networks ('rubber' diodes), say 100p, maybe 1n.
However, one can find small caps over regular bias networks ('rubber' diodes), say 100p, maybe 1n.
You missed the
qualifiers 😛
The "Zener" mention was an analogy to make the OP understand better the "transistor breakdown" meaning.
"Explaining the unknown by comparing it to the known is a common didactic procedure" 🙂
In any case, the meat of the sandwich was "use 2N5401/5551"
Transistors can and often do stand somewhat more than rated specs, some BC546/556 could be trusted with , say, 5V or (quite more difficult) 10V above rating, but that is BAD ENGINEERING and 100V is definitely out of their league.
That maybe 1 in 200 (and I should see that to believe it) stands 100V does not make them "usable" in any real product; perhaps in some Lab curiosity.
I'm definitely on JMFahey's side. Never ever even consider using components beyond their standard specs. It's a recipe for disaster.
The key phrase in a datasheet that reflects this is "absolute maximum rating, never exceed". Tequila bottles could do with something similar I feel!
Funnily enough, we did have a case of a commercial amp here once that inadvertantly ran a transistor noticeably past its rated Vce because whoever drew the schematic had done goofed and the board layout unfortunately followed that to the letter. They did survive for surprisingly long, we're talking decades.... I guess a lot of people just don't listen that loud after all.
And I know of one instance (a very good amplifier, BTW) where a 2SD313 was used as a driver on +/-80V rails. Service manual said it had to be specially selected. If I was rebuilding that particular amp MJE15032’s would go on there.
Back in the 70’s, BGW, Peavey, Bose, and others used to select 2N3773’s and 2N6259’s for 200 volts, to use them on up to +/-85V rails. They had better reliability than switching types that were actually rated 250V or more. Some tried 2N3055’s but that didn’t go over so well. Using 3773’s at 200 V isn’t that much of a stretch - considering their rated SOA at 100 volts. I ran 3773’s up there for years, until you could get a 15024 for the same money. Back when it was common practice there was no MJ15024. There is no need to do it today.
Back in the 70’s, BGW, Peavey, Bose, and others used to select 2N3773’s and 2N6259’s for 200 volts, to use them on up to +/-85V rails. They had better reliability than switching types that were actually rated 250V or more. Some tried 2N3055’s but that didn’t go over so well. Using 3773’s at 200 V isn’t that much of a stretch - considering their rated SOA at 100 volts. I ran 3773’s up there for years, until you could get a 15024 for the same money. Back when it was common practice there was no MJ15024. There is no need to do it today.
- Status
- Not open for further replies.