Hifi, why bother?

[Mark Hennessy]

But the trouble with a distorting valve amp is that you can't bypass the distortion.

 
Why not? I'm thinking (consider something like the Mullard 3-3 which has been mentioned in this thread), a 2-position switch. One position is as-Mullard; the other position removes the negative feedback, and also introduces an input attenuator to bring sensitivity down to the with-feedback level.


Distortion isn't bypassed, it's cancelled - but you get the same thing.

I mentioned in passing that some amps have a triode/pentode mode switch - they achieve a similar thing.

These are all interesting thoughts, but even when in "low distortion" mode, the distortion is still going to be orders of magnitude higher than a half-decent solid state amplifier. And the output impedance will also be considerably higher, so the frequency response will be affected by the impedance curve of the loudspeakers (I think I mentioned this earlier in the thread, but we've mostly concentrated on distortion so far). It might be better to sneak a solid-state amp in there, and switch between that and the valve amp. Just a couple of TDA2030s would do the job

Checking now, I see the Mullard 3-3 is around 0.5-1% THD (measurement conditions and measurement bandwidth unknown). On a pure 1kHz sine wave, 1% THD is readily audible if there's plenty of 3rd harmonic, such as you'd get with clipping, as we're especially acute at 3kHz. But it's surprising how much higher it needs to be if it's 2nd-order, as 2kHz is masked to a higher degree by the presence of 1kHz.

Until now, we haven't really touched on the audibility of distortion. It's a very complicated subject - one of those that gets worse the more you dig. As a starter, the nature of typical programme material needs to be investigated. I've done this sort of thing in the past, but a lot of it is hidden away on a forum that you need to join to read (that wasn't the case at the time) - suffice to say, it opened a lot of people's eyes, including the designer of a well-known UK brand of speakers. One day I'll put a shortened version of it on my website.

So while Andy reports a THD of 8%, I'd be interested to know how that varies with power output. A plot like that found on the [url=http://www.r-type.org/articles/art-003h.htm]Mullard 3-3 page[/quote] would be ideal, but perhaps a bit time-consuming to produce. If nothing else, results at 10dB and 20dB below maximum power would be telling.

If I ever manage to clear the workshop bench, I might experiment with a Hacker Mayflower. That's literally the only valve amplifier I have any experience of, having restored a load of them over the years. One time, I did a repair diary where I measured the THD with each set of component changes. The challenge would be finding the scrap of paper with all my measurements on!