Shunt resistor quality

[tpatton]

A resistor-based volume control (RVC) uses a fixed Input to Output resistor Rs and a number of switchable Output to Ground (shunt) resistors Rg.  I've seen a recommendation to use a very good resistor for Rs and a roughly $35 pot for the various Rg's.  This seems to imply that Rg is less crucial to sound quality than Rs.  Is there any reason to believe that?  Is there good reason to disbelieve it?  I use RVC's in two systems, and felt that I had to use very good resistors as Rg as well as a very good one as Rs.  In my case, as readers of a recent post of mine will know, both are Texas Components Corporation TX2352's.

Is the crucial fact here much like the crucial fact bearing on my question in an earlier post, about why the input load resistor (which plays a role like Rg) is important to sound quality?  I think so, but would like to get confirmation from someone who really knows.

[Paul Joppa]

... would like to get confirmation from someone who really knows.

Go over to the Asylum and you'll find many who really know. They won't of course know the same thing, but they will be certain. :^)

Try it yourself, and you can be one of them. You can even be better than most of them.

Seriously, try it with a TX2352 in series, again with a TX2352 in shunt, and again with TX2352's in both positions. Repeat this at several volume settings, since it may depend on which resistor is larger. Then you can make a conclusion about the TX2352 in combination with whatever cheaper resistor you used.

If you want to be a guru, you can then apply this conclusion to all other resistor combinations, for example by dividing resistor into "good" and "bad" types. But if you want real answers rather than unfounded certainty, you'll have to repeat the experiment with other combinations of resistors. And of course if you want it to be applicable to other people's ears and brains, you'll have to include an adequate sample of those other people.

Incidentally, there is plenty of scientific evidence that you must do the listening tests without knowing which combination is which, lest your unconscious mind lead you to conclusions based on your expectations rather than the actual perception.

Nobody does the complete experiment because it's way too expensive and time-consuming.

OK, so I'm being a little silly here. But the point I'm trying to make is, nobody really knows, certainly not those who profess most loudly to know. Also I want to make the point that you can't rely on theory, even plausible theories (such as those mentioned my myself and others in that other thread), to make conclusions about something you have not heard yourself. Not until the theories are substantiated. All components including resistors have imperfections, but there is precious little scientific evidence as to which of them (if any) are correlated with subjective impressions. I don't doubt others' experiences, or that some components sound better than others in particular situations - I've heard those things too many times, and too clearly, to have any doubt about it. But I am deeply skeptical about the various theories purporting to explain those effects.

[Grainger49]

I remember a poster's suggestion at the old site.  He suggested a top notch Rs and building with common MF Rgs throughout.  Then find the positions used most often and replace those with designer resistors.  

Cheap and makes sense to me.

My read on the requirements for Rg is that this resistor needs to be linear, it should not have an odd resistive curve relative to heat, for use in tubed devices.  It should not not be inductive as that would mess with both the passing of high frequencies and the phase angle between the current and voltage waves.  I wonder if that reflects back into the junction of Rs and Rg?

[tpatton]

Many thanks to both Grainger and Paul J.  I hadn't dreamed that a clearcut simple answer would be lacking here.  I could easily experiment, and probably will, since I have a lot of "common" resistors on hand, all the way from Holcos to the ones that come stock with Bottlehead kits.  I'll be unhappy to learn that an elite Rg is not required, since I've invested much of the family grocery money in a shockingly large number of them, between two all-TX2352 shunt RVC's in my two good systems.  But better to know the truth, I suppose.