🚀 go-pugleaf

In article <se0biINNaaa@network.ucsd.edu>, mbk@lyapunov.ucsd.edu (Matt Kennel) writes:
> kludge@grissom.larc.nasa.gov ( Scott Dorsey) writes:
> : 
> : How are you measuring distortion?  I gather that you are talking about a
> : THD rating, but a system that has a poor slew rate may have excellent THD
> : and still sound poor because it doesn't follow transients well.
> 
> I don't understand.  How is having a 'poor slew rate' any difference from
> high-frequency dropoff in response?  Intuitively, they seem to be
> two ways of describing the same thing, though "poor slew rate" provides
> a bit more physical intuition.
> 
> Or perhaps "poor slew rate" is a code word for "distortion at high
> frequencies"?
>                                       
As I understand it, poor slew rate is the cause of the audiophiles hero of
a few years back, Transient Intermodulation Distortion. The idea is that
if an amp is fed 2 or more signals simultaneously, 1 of them ultra-sonic,
the amp will not be able swing its output (slew) fast enough, so the 
differential input stage will become non-linear due to the late arriving
feedback. Hence it acts as a mixer, and produces sum and difference frequencies
in the audio band (intermodulation products), which sound yucky. There was
a recent article by Cherry in J. AES on different forms of IMD. There
is also the IEC high frequency difference test for amplifiers, which is a
pretty good objective method of looking for it. 

IMD is a good candidate for CD distortions for two reasons: Some (particularly
expensive audiophile types) produce lots of RF on the output, which could
upset amps further downstream. Second, the analogue stages in the CD player
have to cope with step-like output from the dac. This is why the op-amps
on the TDA 1547 dac are rated at (from memory) 200V/us, as opposed
to 13 for an NE5534 or 0.4 for an LM741 (hopefully such devices are no longer
found in audio amps).

It can't be measured simply as high frequency distortion, because the
distortion products resulting from a *monotone* will be so far out of band
that they will be killed by the output filters. However, the 19kHz/20kHz
IM test result is printed by most decent magazines: the differences are fairly
evident. It is quite possible such things are audible, since the test tones
are inaudible (to most everyone) and the difference tone (distortion)
is generated at 1kHz, totally unmasked by the original signal.

Interestingly, CD players with excellent IM results (Yamaha CDX1050,
Technics SL-PS 900) are usually bagged for sounding 'lifeless' or 'boring',
while those which measure to be inferior (eg Philips) get praised for
'dynamism', 'pace', 'rhythm' and 'timing'.

Graham B