Will a true class B have thermal runaway?

[dmfraser]

Not if it has gone into thermal runaway when running at high power and already blown up because the heat sinks were too small. If it was that easy to get around the rules, we would have done it in the 60s.

[Richard C]

The output transistors overheating and failing due to insufficient heatsinking under load isn't the same as thermal runaway.

[millwood]

the answer seems to critically depend on how you define "thermal runaway". some of us defined it as increasing Iq with no signal present, while others consider a signal.

[sam9]

"The output transistors overheating and failing due to insufficient heatsinking under load isn't the same as thermal runaway."

It may be hard to tell which is which though, if you happen to be in a situation where you are right at the edge with either one or both. A rare situation, of course. One which is now filed under "Interesting Ideas That Didn't Work Out".

[x-pro]

Originally posted by sam9

"The output transistors overheating and failing due to insufficient heatsinking under load isn't the same as thermal runaway."

It may be hard to tell which is which though, if you happen to be in a situation where you are right at the edge with either one or both. A rare situation, of course. One which is now filed under "Interesting Ideas That Didn't Work Out".

I would agree with Richard C - thermal runaway in a normal understanding, i.e. positive thermal feedback resulting in a continously increasing bias current (eventually leading to overheating and failure of the device(s) - is impossible if there is no bias. Overheating as a result of insufficient heatsinking is a completely different matter - it only may happen with a signal present - if you remove the signal then transistors will cool down (if they not destroyed yet ) .

Small note - in BJT's there is a possibility of an internal "runaway" called secondary breakdown. It occurs if the voltage AND current simultaneously go over a certain limit (usually shown on a graph in the datasheet), thought the power dissipation is not yet over the device limit.

x-pro

[azira]

Originally posted by dmfraser
Don't waste your time.
...
Bias circuits with thermal feedback have been in use for over 30 years and thermal runaway is one of the easiest to fix problems in a power amplifier. On the other hand, crappy sound from trying to have NFB fix a pile of distortion is something that won't go away.

I can't understand why my question is being shot down and ridiculed. It's just a theoritical question about thermal runaway.
If this hobby was called "CIY: Copy it yourself" I suppose I could just cut and paste from my textbook. I'm just asking a question that the answer to is not easily found because the solution has been in practice for so long that some people even forgot why it's there.

[sreten]

Originally posted by azira
Lets define for this post that a class B is a a NPN back-to-back with a PNP in push-pull with no diodes or VBE multiplers so that it will have the 1.2V of deadzone and xover distortion.
Since the transistors aren't biased on with a fixed voltage, and say I use feedback instead to correct the xover, will it have thermal runaway?
--
Danny

Your question is not being "shot down and ridiculed".

As defined above :

a) Thermal runaway due to DC biasing is not possible.

[b) Local hotspotting of the output transistors is possible
leading to destruction, caused by too much current and/or
insufficient heatsinking, this applies to all transistors and
topologies, it is not the classic thermal runaway of aB stages.]

c) Myself and others are just pointing out feedback cannot
"correct" the xover distortion, only reduce it, and not very
well at all in practise at high frequencies.

The topology could be used in a subwoofer amplifier, and
your question is relevant, TR is not an issue, heatsinking is.

For general hifi amplifiers the topology is not a good idea
due to the poor distortion performance, an optimally biased
aB output stage has far lower distortion, and avoiding TR is
relatively straightforward.

sreten.