11-05-2020, 09:23 PM
Several things come out of this thread.
It started with Diabolical Artificer saying he'd trialled several amplifier topologies, and found that after optimising, the better the spec, the more boring the sound. He preferred 8% distortion. And then went on to ask, Who needs hi-fi?
Mark H pointed out (post 3) that whatever the lousy amplifier did to make things sound nicer, it could in principle be done upstream in the studio process. And then, just use a low-distortion amplifier to drive the loudspeakers.
Then, we got onto what made things sound 'nice,' whether people would pay more for it, is DA enjoying something other than what the original artist intended, ethics of upping price for something of no technical merit (or possibly demerit), gullibility etc etc. However, looking at it from the design point of view, the trouble with setting out to produce an amplifier which would sound 'nice' to DA, is we just don't know how to write the spec!
Measuring frequency response, distortions, damping factor, is straightforward in principle. If our breadboard amplifier doesn't meet the target in one of these, it's down to finding out why, and then tackling it. We're likely to experience dead-ends and about-turns in the design process. But as engineers, we know what we're aiming for.
However, if we can't even write the requirement spec, the design cones to something not much better than having trained monkeys solder things together at random, we switch it on, and of those circuits which don't immediately start smoking, we feed a signal and see how it sounds. Given enough time, eventually, a circuit would be produced which sounds good to Andy. I'm not saying he was working anything like at random, but his surprise comes through, that the bad spec amplifier was the one he liked best. And that, to me, is most unsatisfactory, because who's to say whether we have reached the peak of 'niceness' or whether a tweak exists which can make it even better?
It started with Diabolical Artificer saying he'd trialled several amplifier topologies, and found that after optimising, the better the spec, the more boring the sound. He preferred 8% distortion. And then went on to ask, Who needs hi-fi?
Mark H pointed out (post 3) that whatever the lousy amplifier did to make things sound nicer, it could in principle be done upstream in the studio process. And then, just use a low-distortion amplifier to drive the loudspeakers.
Then, we got onto what made things sound 'nice,' whether people would pay more for it, is DA enjoying something other than what the original artist intended, ethics of upping price for something of no technical merit (or possibly demerit), gullibility etc etc. However, looking at it from the design point of view, the trouble with setting out to produce an amplifier which would sound 'nice' to DA, is we just don't know how to write the spec!
Measuring frequency response, distortions, damping factor, is straightforward in principle. If our breadboard amplifier doesn't meet the target in one of these, it's down to finding out why, and then tackling it. We're likely to experience dead-ends and about-turns in the design process. But as engineers, we know what we're aiming for.
However, if we can't even write the requirement spec, the design cones to something not much better than having trained monkeys solder things together at random, we switch it on, and of those circuits which don't immediately start smoking, we feed a signal and see how it sounds. Given enough time, eventually, a circuit would be produced which sounds good to Andy. I'm not saying he was working anything like at random, but his surprise comes through, that the bad spec amplifier was the one he liked best. And that, to me, is most unsatisfactory, because who's to say whether we have reached the peak of 'niceness' or whether a tweak exists which can make it even better?
