I'm a bit late on this thread, but getting back to the microprocessor burn-in issue. The burn-in process during manufacturing has nothing to do with performance change or break-in, at least not as far as I've ever heard. There are two failure modes that are brought out during this process. The first is infant mortality - basically fab defects in the die. The second is ionic contamination. This situation causes mobile ions in the transistors to collect at the junctions when the transistors are powered up and run at high temperature. These extra ions cause the junction characteristics to shift which can cause out of spec behavior. The burn-in temperatures depend on the type of voltages present on the chips. I've worked with 105C and 137C burn-in. The time is also dependent upon what the IC specs are. These are derived from some MIL standard. I spent 4 years working with burn-in processes and never once heard about performance-variation being a factor in burn-in - and this was with with analog and digital circuits for pacemakers and defibs, which are a bit more critical than audio equipment. If it was a real phenomenon, I would think that it would be a factor in that industry. Just my .02
Pete
