Shallco offers both ready-made attenuators at custom Z values, as well as the naked switches. I'm using a Shallco-made 5k Ohm bridged H-pad (which is a differential path T pad). According to Nelson Pass, the Pass Labs amps with SuperSymmetry actually sound better with differential sources, rather than single-ended ones, and he confessed to me that he has never been in a mastering studio that wasn't entirely wired differntially.
The use of bridged-T or -H pad lets you minimize the number of resistors in series to 2 (per leg), at all steps of the switch. So, very quiet settings will not have more Johnson noise (; than higher loudness settings, as is the case with series-types. Also, the impedance seen by both the source and the destination does not change when switching steps with this type of circuit, so one can focus on the sound of the loudness differences without much interference from other voodoo.
The Lavry M•DA-824 can drive a 600 Ohm load, and Dan told me that 2k Ohms would be the theoretically ideal loading for that DAC. So, I had a 2k Ohm balanced passive attenuator made by GoldPoint Audio which worked nicely for a while. I preferred the sound of the CraneSong Avocet to that. But after DC insisted that passive, when loaded properly, is "essentially blameless," I splurged for the $600 Shallco bridged H. Of course DC was correct - at least according to my listening tests. The Shallco 5k H pad is sonically invisible! Loving the lack of color.
After I sold the Avocet, I had over $1500 left over to buy other things with. I did implement a source switch, as well as an independent mute switch for left and right channels. 5k is what DC recommended as a good compromise between performance and flexibility, and it is not much worse than 2k for human hearing requirements, as long as the cable is ultra~low capacitance and the leads are as short as necessary to reach the amps, and 5k does provide a little extra resistance for wimpy sources one might wish to audition on the same setup. (Fwiw, I'm all Mogami 3080 right now, which is 14 pF/foot).
Andrew
P. S., I'd like to know if anyone has modified their differential inputs and outputs on mastering gear to be single-ended and what was entailed as well as any gotchas.... I was once instructed to bypass the balancing stage (as well as anti-RFI components) of the inputs of a Dominator II by one of our resident electronics gurus, fwiw. This was only to improve the purity of the sound, which I think it did. But, it was on a multiband clipping limiter, so sound quality is a relative thing, right?
Since mastering studios are usually quiet and specialized environments, why is so much mastering equipment differential? There are a few notable exceptions for off-the-shelf gear, such as that made by Pendulum, which is deliberately single-ended.