It’s not that the DAC can’t do this role — it can. And quite capably, too. DAC-direct, bypassing any need for another, complex and audio-degrading piece of equipment, is probably sonically preferable. At least, for some amps.
There are two, potential, problems with this approach.
First, not every amp is all that happy with a 1kohm output impedance from it’s upstream neighbors.
I wish I had a good reference for the audible artifacts of an amp/preamp impedance mismatch, and what values will get you in the most trouble, but I don’t. In fact, I’m not sure anyone does. Just lots of rumor and innuendo on various forum sites … hmm! This is interesting, in itself, but more so when you re-read some past reviews of amps or preamps, especially those where the conclusion is that the component in question is accused of having a weak bass, thin mid-bass, or have a dull sound, is boring, or produces less-than-stellar dynamics. Was it a flaw in the character of the new gear? Or was this result due — at least in part — to an impedance mis-match between the reviewers gear and the gear under review?
In short, the Internet Trolls hold that the bigger the ratio, the better. The output impedance of the preamp needs to be low. Preferably in the double-digits, though I think the most common numbers I see are between 100-300 ohms. Solid-state preamps tend to be lower than tube preamps, where I’ve seen output impedences float as high as 1,000 ohms.
The input impedance needs to be at least 47k ohms (I think this is the industry standard), but some modern pieces push that up to 100k ohms. Some not-so-modern pieces … well, let’s just say they can be all over the place. Anyway, with a 100 ohm output impedance and a 100,000 ohm input impedance, the ratio is 1/100 — pretty much universally acclaimed as non-problematic for the audible range.
I don’t have a rule-of-thumb with which to beat this particular problem, but I think a 1/47 ratio is perfectly fine. Most forum folks seem content with at least 1/10, though, were it me personally, I’d be suspicious if the sound was anything less than perfect. Better safe than sorry, and all that. Again, I’d shoot for 1/47 — or higher. Higher is better.
Every DAC maker has his own pet theory on how to make a completely transparent volume control. Details vary — details always vary — but the idea goes like this: you have to change the structure of the file stream. When you change the bits in the stream, you run the obvious problem of degrading the structure — the quality — of the resulting audio. Why? Well, you throw out bits!
Okay, well, sort of. Most DACs pad the incoming stream with “all the bits necessary” to make this eventual bit-tossing irrelevant — or so the story goes. That said, many folks have reported that DACs with digital volume don’t sound as good — when the volume is turned down low, to “normal listening” levels.
The way this was explained to me, most digital attenuation has a sweet-spot for its application — which is about -10dB off of “unity” (zero attenuation, or, “full output”). More than that and bit-tossing starts getting pretty rapid. Again, not every DAC does this — just most. If you have a DAC that you want to run direct into your amps, and you’re greater than 10dB off of unity at your happy place for volume, then you could have a problem. Do note that -10dB is still deafening if you’re using most solid-state amps — with my Plinius amp, -10dB coming from the Berkeley DAC is bone-breakingly loud. In that setup, I’m usually -30dB!
Channel D’s Seta DAC Buffer
Anyway, here’s a possible solution that showed up a Chez Moi this week: a Channel D Seta DAC buffer ($2,799). This in-line all-analog unit sits post-DAC, and provides a configurable level of volume attenuation — your DAC’s high-quality digital volume control can then be used to make the finer adjustments, and stay within its sweet spot for operation. Attenuation in the Buffer is available in fixed 6dB steps from 0dB to -30dB. A solution to your DAC-direct volume control “issues”? Hmm!
The box is a fairly plain hunk of aluminum, but houses a 2,000,000 ohm (!!!) FET input buffer that feeds into a fully-passive attenuator network with .1% precision channel-component matching to maintain the benefit of the all-balanced topology and its native common-mode noise rejection. This network then feeds into an RF-bandwidth (aka, very wide) buffer with output of 20 ohms (again, my Berkeley Alpha has an output impedance of about 1,000 ohms, so this is quite a drop).
I should emphasize that the design is, at present, 100% balanced — and only available that way — though Channel D’s Chief Wizard Rob Robinson tells me that a single-ended version is still in the works. That would be sweet — it’s the tubes that tend to muck up the impedance matching thing, anyway.
I just got this unit in, so I don’t have anything to report on it as yet — but did want to show it off. I’ll be using it inline with my Berkeley DAC (high-ish output impedance) with a pair of Pass Labs XA-60.5 monoblocks (low-ish input impedance) — this should change my ratio from 1/47 to about 1/100,000. Heh heh.
More soon — but ain’t it cool?