I quite believe that the shunted arrangement you have there would be optimal, the signal effectively passes only through the Shinkoh which is only 1, very high quality, resistor.
I recall a discussion with AndyR from SNA about this some years back, it might have been in relation to an AKSA preamp and involved a TKD pot shunted with Vishay Z-foils.
I know what you mean about the Dales, my previous stepper was a ladder type with Dales and it did lean to the pale and washed out side of things.
My present arrangement is similar to your arrangement in that the signal passes through only 1 resistor on the way in.
You will need a clearance from DCA before you fire up those OTLs
Audiophool,
A point of clarification in regard the shunt style arrangement being discussed.
The leg to ground (the pot) as actually a very significant, and probably the most significant 'flavouring' component in this setup. The current path is from source thru the series resistor acting as a voltage divider thru the pot to ground. The voltage seen at the grid (which is a high input impedance) is the resultant I*R across the pot. Thus the signal that the grid is seeing is actually the vlotage appearing across the the pot from the current flow thru it.
I noted that GAMVE commented that the pot seemed to be paralleled across the tantalum resistor which suggests that the shunt arrangement drawn earlier is probably not how it is aranged.
Assuming that the pot is indeed parallelled across the tantalum, it is possible that the tantalum itself is the 'shunt' leg as you discussed above and there is another fixed series resistor, in which case the pot will act to vary the shunt leg to ground in a continuous manner. In this arrangement the sonic influence of the pot increases as volume increases.
Option B is that the series resistor is the tant and the pot is paralled across it using a fixed shunt to ground ie also a variable voltage divider. This is a less likely setup because it creates a greater variation in the input resistance. BUT the sonic flavour is more consistent with volume because the leg to ground is unchanging.
Stepped attenuators will provide lower noise injection than a carbon film pot. That is an additional attraction to the greater channel balance that you mention.
One also then has a defined sonic flavour from a particular style of resistor and have the ability to play.
A series shunt (switching both series and shunt) is the best way as there are only ever 2 resistors in the pathway at this position.
Food for contemplation,
Cheers
Rawl