Whatever it is, the same concerns about sequence apply. The data sheet says nothing about this. This has to be one of the most unreliable chips ever made.
My experience in far too many years is that if not actually frying, only ICs exposed to external transients from outside the equipment, fail often
I think the problem is that the TDA1541, any variant, actually dissipates quite a lot of heat overall - c.0.7W constant, in a 28pin DIP - and an awful lot of implementations didn't help this; skinny traces of thin copper on single-sided PCBS etc - neither did Philips suggest any heat-sinking.
Naim's early implementations have it sat on a lot of thick, wide copper in the PCB (compared with competitors), and so perhaps like Mark's, my CD2 is at c.25yrs in with no issues. Perhaps rare.
Also,
perhaps, Phillips didn't much care for longevity; this dac family was after all their first couple of steps, the first to market for 16 bit for domestic digital audio - and it bought time during which Philips invented& developed 'bitstream' as a route to a
much cheaper process (no more precision diffusion required for current division by emitter-scaling) and offering a route to far higher accuracy, from single supplies, vastly lower power consumption via cmos, massive logic integration etc.
And funnily enough - bitstream, sigma-delta, or whatever you want to call the approach and its closely-allied versions like PWM - remains the only real route forward to this day.
ETA:
Supply sequencing isn't a thing in the 1541; the internal current bit sources are between the -5 and -15 v supplies; the +5 supplies the ECL that generates the output above the -5v supply (in fact AC current on the +5v supply is the complement of the output signal) Dropping any one - or more - of the supplies, does not damage the chip; and since everything is a current source, or a complementary form of, sequencing (even voltage tolerance on supplies..!) does not appear to matter at all here. Agreed,it can be fatal for many cmos/bitstream dacs.