passive preamp

This is a highly opinionated comment, from me:

There is concern in the high performance audio community that potentiometers introduce a lot of their own special kind of non-linearity into the signal chain. Rightly or wrongly, the construction of a pot - with a band of resistive material, impressed on it with a multi-prong 'wiper', which taps the resistance in a relatively continuous smooth range of values - is also subject in surprisingly short time to noise, both in turning, and just "sitting still".

There is also (I think less than wisely) concern that the ganged-potentiometers in a typical stereo-capable single-shaft volume control are not necessarily very well matched in their tracking of resistance, and therefore volume-per-channel. This can be pretty easily measured with a simple digital volt-ohm meter: hook up to measure resistance, and turn the knob to various positions, measuring the shunt resistance per channel. Often differences exceeding 2% are measured.

So the thinking goes ... if the whole point of high-end audio is to present the most balanced and natural sound-stage through your amplifier - which incidentally is almost always "trimmed" for very-high linearity on each channel - then 2% resistance-ratio difference ... is quite a kick in the pants to the underlying ideals and theories of exactness.

Enter, the 24/36/48 step attenuators - which are just switches, with a whole lot of precision resistors on board. Now, with precision resistors, it is possible to get channel-to-channel attenuation matched closer than 0.1% ... it all depends on what you want to spend for the precision resistors and big old switch.

Personally, from listening over a lifetime to zillions of uber-high-end systems, some with really nice pots, some with crappy pots, some with sort-of-OK-pots, some with click-attenuators, some with entirely solid-state encoder-attenuators ... all I can say is that a 2% difference is inaudible. Not just "barely audible", but utterly inaudible. Truly so.

Consider the math of it.

2% "more" or "less" than a reference signal is either 0.98x or 1.02x. Do the decibel calculation:

20 x log10( 0.98 ) = -0.176 dB
20 x log10( 1.02 ) = +0.172 dB

Which of course makes sense: people can not hear 0.2 dB of difference in a signal. The dB scale was chosen, in fact, so that "1 dB" was just about the minimum difference in sound level that people can discern. I admit there are "golden-ears" out there than can definitely hear 0.5 dB and 0.4 dB in A/B switching tests ... but insofar as I know, there is no evidence that anyone can hear the difference below about 0.2 dB or so. So... +/- 0.17 dB is quite acceptable.

HOWEVER - before you finish with any conclusion, remember that those wiper-on-resistance-strip pots also can become noisy, both sitting static (for old pots) and when being changed. There have been plenty of ideas to mitigate this - but mostly it comes down to sealed pots (get rid of most of the dust), use of resistive polymers (for the resistance strip, eliminating dust), and having a multitude of "in parallel" wiper contact-fingers, which together eliminate any scratchiness, as well as average out the resistance-strip resistance, somewhat increasing precision.

Now you know entirely too much.

Go forth and either buy NICE sealed pots, or NICE reasonable-precision stepped attenuators. Either will do. You'll be able to buy 5 to 10 great sealed pots for the price of 1 decent precision attenuator. Is 0.17 dB worth that?

I'd rather have the nice pots.

GoatGuy
 
The big snag with switched attenuators is that you have to ensure no DC gets to the switch; otherwise you get clicks every time you change volume. That means coupling caps on both sides, and possibly a ground resistor on the output. Pots are better without DC too, but they can cope with a little DC - switches can't.
 
thanks GoatGuy and DF96....i certainly don't have 'golden ear' (wonder even if i have silver ear!) i would go with sealed POTs for sure. i see ALPS are popular but they cost a lot. any particular brand u suggest here? should it be linear or log POT for volume control?