"Jim Lesurf" wrote in message

In article ,
Serge Auckland wrote:

[snip]

One oddity I've noticed. On the SILK attenuator pages,
they say of conventional resistor attenuators "That's
why at high attenuation level (low signal out, low
listening level) the bandwidth of such volume control is
very poor and the bandwidth is not constant over any
switch position."

Can you give the URL for the page?

http://www.sacthailand.com/transSilkVolume.html

"The benefit of replacing plastic type volume control to stepped attenuator
is obviously the accuracy of tracking and the quality of selector switch
plus low noise resistors can reduce the signal loss to some degrees. However
the regular stepped attenuator featuring resistor type fails to offer better
impedance matching over plastic volume control. That's why at high
attenuation level (low signal out, low listening level) the bandwidth of
such volume control is very poor and the bandwidth is not constant over any
switch position. "


I'm puzzled as I'd
expect the maximum output resistance of a conventional
attenuator to occur at about the -6dB setting, and fall
as the output level is reduced below that. i.e. not have
a "poor bandwidth" at low settings, but at relatively
high ones.

SAC Thailand seems to be challenged both technically and linguistically.

Is there some reason why the transformer referred to
*doesn't* also have a bandwidth that "is not constant" as
the attenuation is altered?

Magic!

If nothing else, I'd assume
that the inductances, capacitances, and resistances,
would all tend to change, and affect the bandwidth...

Of course. We both know the drill...

This is only relevant if using very long
high-capacitance cables. If using cables of normal
capacitance, and of normal domestic lengths, it has no
relevance as the bandwidth is already well in excess of
the audio bandwidth. Although I think a multi-tapped
transformer as an attenuator is a bit of overkill, it is
a theoretically sound solution as the output impedance
will reduce as the attenuation increases.

It's ironic that tapped transformer attenuators have achieved some
popularity in whole-house loudspeaker systems, but resistive attenuators are
making a come back based on wider bandwidth.

Will the input impedance not also tend to change as the
o/p tap is changed?

With transformers, lots of funny stuff happens. One reason why a pro audio
mic preamp with transformers is usually considered to be a kind of EFX box.