What tells me the max SUPPLY voltage for a transistor?

[midwayfair]

I'm confused about something in a datasheet: The Collector-Emitter voltage (or Drain-Source). Is this the maximum supply voltage for a circuit, or simply the difference between the collector and emitter voltages. Meaning, if I set up a buffer using the 2N5088, which says 30V for the collector-emitter, and put 48V on the collector, a 10K on the emitter, and put the base at 25V, I would expect the base to be around 24V (probably lower in real life), so the voltage between the two pins is only 24V.

I have a circuit that could be greatly simplified if I just ran the whole thing on 48V, so I'm hoping it's the latter. But I also know that's probably unlikely and that I'm going to need to make a regulator to drop the voltage.

Woops, forgot something a little more specific: The 2SK170 says "-40V" -- http://www.mouser.com/ds/2/408/6909-57550.pdf -- I'm guessing that there's not much functional difference there?

[PRR]

I've run 30V transistors on 300V supplies.

You just be SURE your bias scheme will pull enough current to cause enough voltage-drop in the series resistor to ensure low voltage on the transistor.

BTW, over-volting is not always fatal. Assume 300V and a 100K collector resistor. Say your "30V" transistor really breaks-over at 50V. So 300V-50V= 250V across the 100K collector resistor, which implies 2.5mA. Then 2.5mA through times the 50V across the transistor means 0.125 Watts of heat in the transistor. As all modern Si parts can throw 0.3W fine, the transistor will not melt. (They make fair "Zeners", except the Vce breakdown is very very poorly known in advance, and tends to be hook-shape.)

[PRR]

Or more visually-- how can the transistor "know" what the supply voltage is?

All it knows is what it feels on its legs.

In this case: 30V transistor emitter-follower bias at 24V on the Base: it idles OK but I assume there will be Signal. It can go up near 48V fine (transistor Vce falls toward zero). But as the signal swings down, at 18V on the base (6V negative swing) the transistor Vce is reached and it may break-over. Since in this case it probably won't melt (0.054W peak dissipation), it simply clips the signal when negative swings reach 6V peak.

It won't be this bad. "30V" on the data sheet is the reject point. They probably cook the dough for >40V breakdown, and 90% of parts will come out well over 30V. But you can not be sure. They may have had a big order of 40V parts the day before, and were scraping the barrels picking out 31V and 32V parts to sell under a 30V spec.

Why bias in the middle? The resistor can take 150V. For a 30V transistor the maximum you can swing is 30V or +15V and -15V. Bias the base 15V down from the +48V, at +33V. Now you can swing near 15V either way, double what you got with 24V bias. And surely more than you ever need to handle.

[midwayfair]

Or more visually-- how can the transistor "know" what the supply voltage is?

All it knows is what it feels on its legs.

In this case: 30V transistor emitter-follower bias at 24V on the Base: it idles OK but I assume there will be Signal. It can go up near 48V fine (transistor Vce falls toward zero). But as the signal swings down, at 18V on the base (6V negative swing) the transistor Vce is reached and it may break-over. Since in this case it probably won't melt (0.054W peak dissipation), it simply clips the signal when negative swings reach 6V peak.

It won't be this bad. "30V" on the data sheet is the reject point. They probably cook the dough for >40V breakdown, and 90% of parts will come out well over 30V. But you can not be sure. They may have had a big order of 40V parts the day before, and were scraping the barrels picking out 31V and 32V parts to sell under a 30V spec.

Why bias in the middle? The resistor can take 150V. For a 30V transistor the maximum you can swing is 30V or +15V and -15V. Bias the base 15V down from the +48V, at +33V. Now you can swing near 15V either way, double what you got with 24V bias. And surely more than you ever need to handle.

Thanks as always for the help! Stepping down the voltage it is, then.

[greaser_au]

...And don't forget to take care of cold-start conditions, e.g. a coupling capacitor with 0V across it pulling your bias point down close to zero unitil it charges, perhaps...

david

[R.G.]

... and turn off conditions where coupling and bypass caps are charged up to whatever their operating voltage is and keep that if the power supply goes down fast, or is shorted. That's what kills three-terminal regulators if there is no reverse discharge diode connected around them. And that happens with a voltage that is well within the max spec of the device.

[midwayfair]

... and turn off conditions where coupling and bypass caps are charged up to whatever their operating voltage is and keep that if the power supply goes down fast, or is shorted. That's what kills three-terminal regulators if there is no reverse discharge diode connected around them. And that happens with a voltage that is well within the max spec of the device.

I'm probably using a 100R -> 10K||BD681G (100V power Darlington) with a zener off the emitter. A 24V regulator can't take 48V input (even with a metal tab). I can either drop it to 33V and then use the 24V regulator to drop it to 24V, or, I suppose, just go with the 24V zener.

I did find some transistors that can run on 48V on Mouser and by googling the datasheet of everything in my parts bin ... 2SK30, several power Darlingtons, and ZTX653. The ZTX653 and BD681 can take 120V! That's low plate voltage. I added a few to my cart ... I might just skip stepping down the voltage if they're quiet enough transistors. (I have no idea what a power darlington is going to sound like in an audio circuit, I've never used one before.)

[R.G.]

Have a look at the LM317HV. IIRC it will accept up to 60V input without damage.

Otherwise, pre-regulate with a MOSFET/zener down to something a three-terminal will accept.

[PRR]

> with a zener off the emitter

Emitter is potentially "infinite current".

You would usually use a few mA into a Zener to set the *Base* of the transistor.

>48V transistors are very readily available (if you look outside our 9-Volt world). Not-too-long-ago, Radio Shack even stocked TIP120 (60V). 300V and 500V parts are standard in some fields. MOSFETs have taken much of this, and MOSFETs can be good audio.

As R.G. says, some look-around will find many other techniques. HV regulators. Pre-regulation.

OTOH, just setting a 30V emitter-follower's NPN Base to a 33V bias (47K+100K base divider, with 10K on Emitter) will keep your transistor in its happy-zone without complication, and handle quite huge signals.

Putting ~~2K series in base or collector will limit fault currents, including start-up transients and mild stage accidents.

[midwayfair]

> with a zener off the emitter

Emitter is potentially "infinite current".

You would usually use a few mA into a Zener to set the *Base* of the transistor.

Woops, yes.