Hi,
I am sure moods make some difference but does not explain what actually happens to a component during warm up. Since not much university scientific information has been presented, I thought I would expand on such for better understanding of heat related/warm-up concerns. Also check RCA Radiotron Designers Handbook, written by 26 engineers.
An audio component is actually quite dynamic during warm up, as we shall see. For ease in understanding, I am simplifying the discussion and not covering every issue. I am starting with parts, then negative feedback and comparing it to no feedback (open loop), as these have a major bearing on the discussion involved.
I think most understand that gainstage and passive parts conditions vary with temperature change, especially using bi-polar transistors, electrolytic capacitors, even transformers etc.
For example, under the most basic conditions, with no protection at all, bi-polar transistors are extremely heat sensitive and the hotter they become, the more collector current flows. "Thermal run-away" is the condition. This means that if enough power is available, "run-away" bi-polars will actually burn up (not including emitter followers). Of course such unstable conditions cannot be allowed to exist in any design but protection must be implemented. This information gives one at least an idea of what one has to deal with warm up issues with SS designs using bi-polar transistors. The problems are certainly no superficial. Tubes also require time to heat with the elements expanding and characteristics changing as well as passive parts changing characteristics.
Attempts to correct open loop variations (no feedback), such as harmonic distortion, frequency response, phase integrity etc, by applying some musical feedback voltage to correct the variations can help in certain respects, but also presents its own problems concerning warm up.
First, linear global feedback does not change the internal dynamics of a component per se. In theory it only samples the output of a signal, inverts it (180 degrees out of phase with the input signal), and feeds the inverted sample into the input, thus attempting to correct problems inherent in the open loop design. Global negative feedback can vary from little to very large amounts.
Linear global negative feedback reduces the higher gain portions of the audio band with respect to the lower gain portions of the audio band. Thus the bandwidth for a given -db is increased over the open loop design (no feedback applied). Imagine viewing a saucer bowl upside down on a table, viewing from the side. Now lower the top of the bowl (negative feedback applied). The bowl will appear flatter over a greater area, flatter frequency response over a greater range. Negative feedback "creates" a wider/higher frequency response as well as influencing other variables. (I hope the example helps one to understand.)
The actual effect of negative global feedback signal changes from cold to warm up conditions. Depending upon the open loop gain, bandwidth, phase response etc of the amplifier (which changes as it warms up) normal mid-band negative feedback (180 degrees out of phase with the input) actually becomes positive feedback at some high and low frequencies. Let's stick to the higher frequency example for ease.
Luckily protection measures are implimented to prevent oscillation from strong positive feedback at very high frequencies. However, just above mid-band frequency, "A" the negative feedback signal becomes only 170 degrees out of phase with the input signal, a little higher frequency "B" 120 degrees, at higher frequency "C" zero degrees out of phase and into positive feedback territory (although not oscillation).
As the "warm-up" temperature rises, points "A", "B", and "C" shift/change (Using single points simplifies the explanation, but the phase shifts occur over many octaves in the audio frequency range.), with many variables changing as a result. So feedback is not just "static" by any stretch of the imagination. Neither are active and passive parts.
If steps are taken to limit each individual stage's warm up shifts/changes, such as using local current feedback, then gain is reduced which means more stages, more parts, and "more power supply", which are not sonically perfect by any stretch of the imagination. With more stages, more problems such as higher order harmonics, phase shift problems, feedback from stage to stage through the power supply, etc results. (And remember, linear negative feedback reduces all harmonics by the same proportion, not selectively. And the higher the harmonic, the greater the weight it has with relation to the second harmonic.).
If steps are presented to increase gain but limit DC changes, such as
bypassing cathode/emitter/source resistors with electrolytic capacitors (or even other parts), then we have added more DA and DF capacitor problems or other part problems which themselves are, or can be temperature sensitive problems. (Almost forgot, DA and DF problems include the huge power supply electrolytic capacitors as well as film capacitors, although to a lesser degree.)
So warm-up not only changes individual active and passive part characteristics, but varies the relationship between feedback, especially global feedback, and the variables in the open loop portion of the design itself.
As mentioned earlier, another problem that some may be encountering is system masking of inner detail, distortion, and spacial Qs. In such a case, musical changes may not be noticable with temperature rise but then the system is not resolving all the musical information either.
As one can appreciate, an in depth engineering analysis reveals that what is actually occurring as a component heats is quite complex involving many parameters, and secondly, a component is a vibrant, dynamic device, not "static" in any sense.
Cheers.
