I don't have high end attenuators
According to
its product page, the KGSS uses a pair of
DACT stepped attenuators. That's about as high-end as things get without getting into audiophoolery.
DACT claims 0.05 dB tracking accuracy. You're claiming that even that is not good enough? ("...even the KGSS pot unbalances fast...")
My current Amp is a KGSS, and it doesn't have gain control.
As I hinted above, you may be able to obtain the schematic and then modify its internals to lower its gain.
No, don't ask me, I've never built one. Take it up as a separate thread.
Good stepped pots (that have high granularity, especially at low levels)
I suspect you're confusing two separate concepts here.
There are "stepped" potentiometers, but they're just regular pots with a ball-bearing based click feature that makes it harder to hit certain points along the continuous resistance range inside the pot. A stepped pot is only valuable when you need to count clicks in order to hit some predetermined point. They're popular in
sound reinforcement products, for example.
Consumer equipment with clicky volume knobs are typically using some kind of
rotary encoder, which in turn drives a digital pot. You could drive the PGA2311 mentioned above that way.
Then there are stepped attenuators, which are entirely different things from potentiometers. They use discrete resistors instead of a continuous track of resistive material, so that they have jumps in attenuation like a
digipot, but they're purely passive, like a potentiometer. They're just a many-position rotary switch with resistors in between each switching position. There are subtleties in SA design, but that's the essence of it.
All attenuators have some amount of tracking error. It's purely a fact of the way they work. Every resistor has some error in it, so the longer the resistance string, the greater the potential error. It is a
random walk type of phenomenon.
The only reason stepped attenuators tend to be better than potentiometers is that you can hand-match the resistances at each step, reducing the channel mismatch at each stage; but, errors still build up over the string! With a pot, you have a stripe of resistive material with a continuous series of errors, so that short of using a process like laser-trimming, the random-walk errors will build up in individually unpredictable ways.
I'm not aware of any laser-trimmed commercial audio potentiometers.
There are some very expensive rheostats and linear pots, but those are expensive because they're used in industrial machinery that has to work reliably in harsh environments for decades.
Today, the best solution I know of for passive attenuation with low tracking error is the stepped attenuator, and you're telling us that even well-executed examples of that technology aren't good enough for you.
Good dual concentric pots
Just say "stereo audio pot." Every stereo pot is "concentric." That term just means that one control shaft moves both pots at the same time.
The best commercially-available stereo audio pot I'm aware of is the ALPS RK27, and as my audio attenuator article's data shows, it's still not as good as the DACT SAs at low levels.
Good pots that do not unbalance at low levels
As I've said, there's basically no such thing. Pots are not meant to be used that way. It's like asking for a bicycle that doesn't tip over when you ride at low speeds. Just as you must get a bike moving at a certain minimum speed for stability, you should be running pots up in their well-matched ranges.
And perhaps a basic guide for making an enclosure properly
Something like
this?
somewhere that sells good RCA and 1/4" jacks
Right at the top of this forum is a link to a list of articles answering standard questions like this. These two articles seem like what you're after:
http://www.head-fi.org/a/do-it-yourself-links
http://www.head-fi.org/a/diy-cable-info-and-resources
As long as I'm building it, may as well add a monitor port
Bad idea.
First off, passive attenuator boxes are already problematic from an impedance matching standpoint. You have to know the output impedance of the driving stage, its current handling capability, and the input impedance of the driven stage to properly design it. This is why virtually every preamp in existence is active, and "passive preamps" are something you generally only see from the audiophool section of the hobby.
Now you're talking about putting in a wye, which will shift the impedances again. No. Don't do that.
If you want an attenuation stage with a wye in it, make it active, as with the PGA2310/2311 mentioned above. But, don't expect it to be transparent. Like any amplification stage, it's going to color the sound somewhat.
I've tried that, and it unbalances at low levels, like every other pot I've tried
That's because that's all it is: a pot in a nice enclosure, with some nicely-done captive cabling.
except the one in the Schiit Asgard 2
Because tracking error is a random phenomenon, you can get lucky, and find a "perfect" pot where all the errors just happen to null out. If you want to buy a bunch of pots and hand-test them for tracking error, you can come up with another uncommonly-good exemplar of the art. Or, you could get unlucky again, and come up with a batch that all have reinforcing errors of some sort.
If you were willing to go to that kind of effort, I'd say it's better spent building your own SA from discrete parts,
hand-matching the resistors to high tolerances. As that article explains, that's expensive and time-consuming to do, too,