Transistor and Temperature

[lumanauw]

I just looked at a manual for one audio amp. It says that this amp equipped with internal fan (for cooling heatsink), that is adjusted to reach the "optimum" temperature for best audio reproduction. The manual don't say what temperature, but the fan will be off until this "optimum" temperature reached, the fan will be on if the temperature is more than the "optimum" range, and the whole amp will be off if the whole amp getting too hot.

Is there really an "optimum" temperature range for transistors (final transistor) to perform best audio reproduction?

What is happening in this optimum temperature? The transfer curve becomes more linear or what?

[anatech]

Hi lumanauw,
That beats me! I think more likely an optimal temperature range and optimal bias current. This reminds me of a Michael Elliot goof where the heatsink temperature was tighly controlled in a Counterpoint Natural Progression amplifier. He went to great lengths to ensure this behaviour.
I'm inclined to believe the amp has a small (cheaper for the power) heatsink and the fan prevents meltdown. In other words, a cost saving feature. The fan will cause large temperature variations across the heatsink as it cycles on and off.
What do you think?
-Chris

[HFGuy]

Like most products with forced air cooling the heatsink is too small for the largest possible heat dissapation for the amp. The optimum temperature that the fan kicks into gear is alittle before the transistors go into secondary breakdown or melt the junction.

[Mr Evil]

What HFGuy said is probably true in this case, but transistor characteristics do change significantly with temperature. Overclockers know this well, but I've noticed strangely little thought given to it in audio circles. Temperature can affect noise, bandwidth and gain, among other things, so there may well be an optimum temperature for a given circuit to meet its specifications. Exactly what changes and in which direction depends on the device, but it is generally true that transistors slow down with increasing temperature, and noise for all components increases with temperature.

[anatech]

Mr Evil,
I see your silhouette ahead of me.
I agree totally that most audio people do not consider thermal issues. It is amazing how long an overheated device will run. Still, the sound will change more with different bias levels than temperature, everything else held constant. The speed of the transistor is typically higher than the circuit it's in, so that shouldn't matter very much. The noise level is more critical in the first stage(s).
-Chris

[lumanauw]

So, its more like saving cost for heatsink than to reach optimum sound?

But I have another interesting example.
There is an amp, no fan at all, just huge heatsink. In the panel there is a led "optimum temperature". If the heatsink is cold, this led blinks all the time, if the temperature is reached, it spots on.

Is there really such thing as "optimum" temperature for best audio reproduction?

[anatech]

Hi lumanauw,
My theory is that if it's not running too cold (no condensation allowed please) or too hot (can not leave your hand on the sinks) it is fine. The cooler the amp runs, the longer it will last without failures.
I have noticed that FET outputs tend to want to run warm. That's apart from the higher idle current.
If the designer does not place the temperature sensor on the heatsink, or the preceding stages are temperature sensitive, the amp may have a "happy spot" temperature-wise.
I really can't believe that outputs and drivers have an optimum temperature that has not been documented. Motorola and others would definetly put out an applications note if this were true. The characteristic curves for these devices would show some abberation around an optimal range.
-Chris

[HFGuy]

Never underestimate the power of a marketing dept.

[Jonathan Bright]

Thanks for starting this thread. I have been wondering for a while whether there was an optimum temperature. Mr. Evil I am intrigued about transistors "slowing down" when hotter, can you point us in the direction of some data on themal characteristics? I'm mainly thinking power transistors/output stages etc.

[HFGuy]

Correct me if iam wrong, its been awhile since i took solid state physics. Increaseing the temperature effects the fermi level in transistors as well as decrease electron migration in the BJT. So yes temperature does have an effect on noise, mu , current leakage ..... but how large the effects are on the sound of the amp i do not know. I still stick by it being a marketing thing.