The problem of volume
Got one of those fancy new DACs? You know, one with all the nifty features — including a volume control? Well, if you don’t, chances are pretty good that you’ve seen or heard about them, and have wondered to yourself — “Self, why do we need that expensive preamplifier if we have this little doodad? Won’t this doodad do everything a preamplifier does? And isn’t it an obvious Audiophile Truism that the ‘simpler the path, the better the sound’? Self, I bet we can remove that preamplifier, convert completely over to digital or computer audio, and finally achieve system nirvana and couch potatohood all at one fell swoop!”
It’s a tempting line of thought (provided you aren’t prone to doing it out loud; that’s a bit mental) — yes, very tempting indeed. Honestly, the preamplifier has always been a little suspicious. I mean, what does it really do? Fine, it hooks in all my sources — but I only have source-s (em-PHA-sis on the plural there) because I’m a panty-waisted fop that hasn’t managed to commit to the Glory That Is Computer Audio. But now, I can! Ha ha! Now is the time! Ha ha HA!
I saw this trending over on Computer Audiophile a couple of years ago. Like then, I still think this is a bit akin to magical thinking: “I’ve finally seen behind the curtain! And lo! What wonders that have been hidden from me! Now, I know and understand something that the masses of my hidebound fellows have continuously and completely missed! If only they knew the Truth that I now Know! I must tell them! It’s time to engage the CAPS LOCK!”
You’ll pardon my skeptical expression, I hope.
I tend to think of this “paradigm shift” as the “DAC-Direct” approach. The idea is pretty much exactly as you’d expect; the DAC itself performs all of your signal attenuation and therefore can completely replace your line stage. And because doing this results in a simpler system — eliminating an extra component and at least one set of wires — it’s almost certainly better. At least, that’s the theory. I’ll put aside the whole “why a preamp” question for now. Let’s assume that this line of thinking makes sense and move on.
I’ve never been crazy about the idea of multi-function devices. A preamp with a phono, much less a DAC? It stands to reason that at least one of those features will suck. Too many compromises! But when I finally upgraded my beloved PS Audio DLIII DAC, I was stuck. I ended up with a DAC with a volume control. I didn’t want one, but I wasn’t given an option — the volume control was there whether I wanted it or not. Apparently, they’re cheap to implement and in a world dominated by competition, any differentiator is a good differentiator. Gotta check the box, dude! “They” have it — and so do we! They don’t have it? Well, we do! Woohoo! Whatever. Cheap isn’t always — or even usually — good.
And that the rub, innit?
Let’s start with your digital source; be it a computer or disc-spinner (you Luddite), it sends a signal to the rest of your system. This signal, passed directly to your speakers, will be … loud. Like “instantly blow your head apart” loud (where “head” = “your loudspeakers’ driver assembly”). Let’s call that output level “unity” — aka, “full volume”. At unity, your system is unusable — one note and you’re dead. Even if the debris fleeing the region of your now-atomized speakers doesn’t render some nonzero percentage of you into a moist red mist, or failing that, if your family didn’t then slay you outright, well, either the zombie hordes or the giant sandworms will most certainly find you pretty much pronto-burger. There’s just no way this ends well. Enter: attenuation.
This is where your preamp normally steps in — it takes that full-strength signal and turns it down, aka, attenuates it. How this works will depend — there are lots of approaches — but the most common is to put some kind of variably resistive load in the path that signal, allowing you to make it smaller. Smaller signals equals quieter sound. With me so far?
Now, digital volume controls don’t work on the analog signal at all, per se, they work on bits (that is, the signal that gets carried on an analog wave form). Bits — that’s interesting. Why? Well, one reason is noise: any signal, when amplified, has any noise that is riding along with that signal amplified right along with the signal itself. The signal is louder and easier to hear — but so is the noise. That’s bad! Worse, this is exactly what most preamps do — they amplify (even if it’s only a little bit) — before they
de-amplify attenuate. But with a digital volume control, that doesn’t happen. With a digital volume control, whatever noise that gets carried along with the signal is simply discarded when the signal is received, that is, when the digital signal is extracted from the incoming analog carrier. Ta freakin’ da. So, they say, digital volume controls rock and the rest can suck rocks.
The problem is that I don’t listen at volumes where this superiority is entirely obvious. You probably don’t either — or, rather, you shouldn’t, you tool. Here’s the issue, in a nutshell — digital volume controls destroy the signal. Said another way, they “toss bits”. That is, for attenuation to work in that clever little chipset your designer used to make you your toy, the DAC takes the number of bits found in the original stream of data it receives from your computer/player and “truncates” (aka, “tosses”) stuff off the end of that stream — that’s how it represents the drop in signal output. The more you attenuate, the more truncation is necessary and therefore more bits go bye-bye. Every bit is “worth” about 6dB. Here’s what all this looks like:
- One byte of audio data: 10001011
- One byte of audio data, attenuated by 6dB: 01000101
- One byte of audio data, attenuated by 30dB: 00000100
Note how the string of bits keeps shifting to the right; by doing so, it was that trailing “1” in the first line that took a dive, allowing the volume to be cut in half. Need more attenuation? Take more bits.
If that concerns you overmuch, it shouldn’t (they say), because most designs insert a “pad” into audio data which allows truncation to happen inaudibly. Happily, CD data is only 16 bits “in-depth”; given that most of the DAC chipsets with digital volume controls work at 24 bits, there’s “room” for an extra 8 bits that can be used to “pad” the incoming data stream, and again, it’s this pad that gets tossed out. Well. Tossed out first. So, if I have 8bits of pad and each bit allows me to attenuate 6dB of volume, then it follows that I have about 48dB of attenuation I can do — all before I need to touch any of my Sacred Bits of Actual Musical Content to bring the volume down further. This is good! 48dB is a lot. For example, my reference DAC offers 60dB of attenuation — which means I ought to be able to get the volume way down before Bad Things Happen.
- One byte of audio data, with 8 bits of pad: 10001011 00000000
- One byte of audio data, with 8 bits of pad, attenuated by 6dB: 00001011 10000000
- One byte of audio data, with 8 bits of pad, attenuated by 30dB: 00000100 01011000
So, you’re feeling pretty comfy at this point, no? Frisky, even. Ready to go DAC-Direct and live the Good Life of digital attenuation? Not so fast.
Any sentence that begins with “In theory …” really ought to make you stop a second. It’s usually a pointer that Something Has Been Overlooked. So, when I say “In theory, bit tossing is inaudible to the average listener”, your first thought could justifiably be, “umm, that doesn’t sound right”. And it shouldn’t. Tossing information — any information — out of a stream of encoded data representing an audio experience should be considered an act of hostile intent. Bad Things Are Happening. The question of whether you can hear it is very different from can the average person hear it. Remember that old Carlin bit about how dumb the average person is? Now, remember, that means that half of them are dumber! So, the fact that the average listener cannot discern something says nothing about you. You might. Why risk it? That’s one concern. Here’s another: Rob Robinson thinks that, depending on the DAC and it’s implementation of volume control, you might be able to hear any digital attenuation — and the more you do, that chance increases dramatically. In fact, Dr Rob says, anything more than 2 bits — even from a pad — and you’re playing with fire.
This is a fun game! Let’s play dog pile, shall we?
Lets chat about those horns of yours. The more sensitive your loudspeakers, the more “bit tossing” could be an issue. Not because those speakers are more transparent, per se, but rather because you’ll need more attenuation right from the start. Maybe a lot more.
Take your 84dB loudspeakers. They need quite a bit less attenuation than my 96dB loudspeakers. Going DAC-Direct, that 12dB of difference (when level setting) in the speakers will need to come out of the DAC, obviously. Which means two more bits fly right out the window when I switch over to my high-sensitivity system — before you even start worrying about turning it down. Going DAC-Direct with your 108dB horns will entail a loss of four bits in comparison with your more modern, insensitive speakers. Again, that’s before you turn anything down.
So, what about all that high-res music you just downloaded from HDTracks? Most of that wasn’t 16bits — that was all 24 bits. Look Ma, no pad! Which means that, for all that super-duper hi-res material running through the DAC, a digital volume control acts as a wood chipper from the first dip down from unity, cheerfully grinding through my files’ content and blotting out detail, dynamics, tone and timbre, all in the pursuit of a volume I can actually bear to listen to. Whoops.
Now, how bad can this be, you ask? I mean, you like your tunes loud, right? Well, if you’re thinking about playback at an average volume of eleven, then we’re talking about the speaker sensitivity plus the gain from your amp. If you wanted 80dB for that session, well, that would start with those average 87dB loudspeakers speakers, add the gain from the amp — about 26dB (average) and net you about 113dB. Think “power tools in your face”. You’d need to pull things down ~33dB, or about 6 bits. So, on my 24bit file, I’m only hearing 18bits of resolution. Not exactly bad, but ….
Rob, in a discussion found on his site called “Getting Started With Computer Audio” under the topic “Digital Volume Controls”, suggested the following rule of thumb: anything more than 2 bits worth (i.e., 6dB) of attenuation — even with a pad — and it’s possible to hear the changes being made. If you stay within -6dB of unity, and you should be roses — pretty much always. Which still means … animal stunning volumes, unfortunately.
I think this may well be why vinylophiles get their dander up about digital audio; digital audio really might not sound as good as an all-analog chain where sufficient care has been taken to address noise and it’s suppression — especially at “realistic” volumes. It really is quite possible that there is less there there on many DAC-direct systems.
Sure, there are ways around this. Some clever designers work with 32bit DACs, and this gives them room to pad even 24bit “high resolution” files. With better algorithms, I’m sure it’s possible to preserve the waveform with ever greater degrees of fidelity, even during heavy truncation. I’m going to insert a lot of hand-waving here, and simply grant that “things can be done”. This is where quantization (which is bad) and dither (which can be good) come in.
From Dr Rob:
If you have seen a Netflix video or any video on the Internet, in very dark scenes there will be visible “blockiness” or pixellation in shadows. This is quantization error: the actual black value does not line up with the digital value, but are forced to a certain brightness / color because there is no other available digital value that is closer to the true value. This is more apparent on dark scenes because you are getting to the resolution limit of the converter (video converters are typically 8 bit in compressed video, 8 bits for each color, 24 total).
Now imagine that there is some “noise” like on an old analog TV. The digital blockiness will be less visible, because the LSB [least significant bit — the one most likely to be dumped with more attenuation, or the one now exposed after attenuation] of the digital data is being “jiggled” randomly. The result is less annoying. You are a good photographer, and so you are surely familiar with noise reduction software for images. There comes a point where (to the trained eye) over-application of the noise reduction looks ugly and unnatural, color washes take on a pasty appearance (because the noise reduction requantizes the image); so it is better to have some noise in there to make the image look more pleasing and natural to the eye. Same thing ….
Dithering in audio does as much, masking the effect of quantization error, in effect slightly increasing the available resolution.
For a more technical explanation, the entry on Wikipedia for dither / quantization error is more precise and in – depth, and talks about the pros and cons of dither, the various kinds, and the expected results.
Whether dither will help, or “fix” the audibility of any negative impact on a digital volume control, will (again) depend on … well, everything. The point, however, is that digital volume control is not transparent. It’s destructive. Which means it’s like an analog volume control! It’s all the same … but different. Ahem.
Anyway, some might find all this to be a problem.
But it’s not the only problem.
There is another, and it’s more related to cost than technology. It’s a truism that everything gets built to a “certain level” and that not everything can be “cost-no-object”, so I’ll just take all that as read. Given the constraints of actually making a profitable, or simply viable, product to take to market, concessions are regularly made to keep costs from spiraling.
One of the things that regularly gets whacked, at least on a DAC, is the output stage. I mean, who needs it, right? Har har. The magic is all in the chips. Har har har! Got the input down, tackled jitter, added our filters, and … we’re done! Har har har HAR.
Well, let’s just say that in DACs below $10k, there’s a better than average chance that there were shortcuts taken to get that product to market, and that of them, an output topology required to drive amplifiers — and drive them well — may have gotten a haircut in the final specs. How do we know? Well, short of being a designer yourself, you won’t. But a clue might well turn up in the output impedance of your nifty new DAC.
This is a funny number. A high output impedance may not indicate an issue with your current rig … but it might. If you’re using amps with low input impedance (by which I mean anything lower than 45k ohms, for example), this could be an issue. The big Hegel H30 monos –which are amazing — have rather low input impedances: 20k for balanced and 10k for single-ended because in this config (monoblock) they’re actually a stereo amplifier running bridged, a topology that halves the normal input impedance while it increases the output power delivery. This low number isn’t catastrophic, but sending a signal from a device (DAC or pre) with 1 kilo-ohm output might not be exactly ideal.
And if you’re a fan of passive preamps … well, 1,000 ohms of output impedance would be a good number.
Referring to the Big Book of Audiophile Wisdom, it says that an impedance ratio between the amp and what’s driving it really ought to be 10x — that is, the input should be 10 times higher than output — but this is the minimum and more is better. Preferably, way more. Exactly why those numbers are the magical ones probably requires math, and since we abhor that, we’ll wave our hands here (again) and just pretend we all know the whys and wherefores.
But, even barring this egregious lack of specificity, we can ask what happens when the impedances are not matched well. And that’s an interesting question. Here’s what I’ve learned from my own experience and from talking to some old geezers, who saw this sort of thing regularly “back in the day”: it’s the frequency extremes that suffer. That is, bass gets flabby or bloated, mid-bass gets muddy, and you lose some measure of treble response, sparkle and/or detail. Now, that’s the “could happen” — does any of it really happen? Good question. Put a pin in that sucker and we’ll get back to it.
Say hello to The Buffer
One of the joys of writing on the Internet is that I get to be as long-winded as I want and you can just suck it. It’s all very
Stephen King Neal Stephenson of me. Anyway, consider all this a very long intro into why Rob made the Buffer in the first place. But first, what it is.
From the manual:
A low-noise, fully balanced design featuring an ultra-wide frequency bandwidth of 20MHz … the Seta Buffer is designed to be used as a back-end signal conditioning stage to facilitate setting the correct system gain of your digital audio playback system.
At its simplest, the Buffer is a fixed attenuator. Sounds simple, right? Well, it is. What it actually does is two things: one, it introduces a simple, static, high-quality resistive element to the signal path with the express purpose of knocking the volume down to some manageable, listenable, pre-set level. The attenuation introduced by the Buffer is adjustable by as set of internal dip switches that lets you set the attenuation level anywhere from -30dB to -6dB, all in 6dB increments. A large, and handy, configuration guide is printed on the brass plate mounted on the inside of the top cover of the Buffer. The resistor network is extremely high-quality and channel-matched to an exacting .1% tolerance. With the Buffer inline, you can/will then use the DAC’s digital volume control (and it’s remote, if it has one) for dialing in/fine tuning to taste — and thereby keep your DAC from ever digging too deeply into resolution-damaging bit tossing. Oh, and the whole thing runs on batteries — “off grid”.
For those that have a need for it, currently shipping models allow you to remove the resistive elements from the path entirely (that is, it can be configured for -0dB “attenuation”), effectively making the unit a truly transparent pass-through device — for use with an external preamp, say. Why you’d consider still using the Buffer without the attenuation has to do with the Buffer’s second main “function”: impedance matching. The Buffer has a 2 megaohm input impedance and < 20 ohm output impedance — electrically, it’s a “dream load” for any audio network it gets introduced into. So, if you’re happy with the volume control scheme in your purely passive preamp, but would like to “fix” it’s terrible (high) output impedances, stick the Buffer behind it and set it’s attenuation to zero and Bob’s your uncle.
Anyway, that’s it. That’s all it does.
Perhaps you’ll recall that I introduced the “beta” Buffer back in December; the design has been minorly tweaked a bit since and is now in full production. What’s obviously new is an additional set of outputs — in this case, a pair of RCA/single-ended outputs — so you can now drive your single-ended amps from your balanced DAC.
For more, Rob humored me with some Q&A, which I’ve included here:
Why did you make this product? Was there a particular aspect of the digital experience that you found lackluster? A piece of equipment you wanted to tweak? Or was the problem theoretical and/or philosophical?
This resulted from looking for a solution to a common and important problem in a digital playback system: setting the system gain structure to be able to use a digital volume control at its optimum settings.
An easy and relatively inexpensive (less than $100) way of setting the system gain structure employs an inline passive attenuator, to reduce the signal level presented to the DAC. We evaluated this approach with a range of DACs, costing as low as a few hundred dollars to well over $10,000, in our reference system as well as in various demo systems used in our audio expo exhibits. We noticed that with this device certain DACs sounded better (because of needing less attenuation from the digital volume control) whereas others took on an unpleasant hardness.
Trying to understand the differences between the behavior of the DACs, the logical conclusion was that the DACs which didn’t respond well to the attenuator were having trouble driving its relatively low (about 1000 ohms) input impedance. (Designing a passive analog attenuator that doesn’t introduce significant noise or frequency response issues requires using low impedance components.) The solution was an attenuator with high input impedance. However, this requires active circuitry, otherwise the output impedance of the attenuator would be too high to drive downstream components and interconnects without introducing noise, distortion, and frequency response problems. But the active circuitry used must be completely transparent to the signal. This is the design principle of the Seta Buffer (and we take that notion even further, using rechargeable batteries for the low noise power supply).
Some folks might be thinking that what we’ve done is merely design a sort of line stage preamplifier. Therefore, why not use a regular line stage preamplifier which has a more precise and wider range control of the volume? That is a valid point, but designing an attenuator / line stage with many more, finer volume “steps” with the same quality of components used in the Buffer would greatly increase the complexity and cost. On the other hand, there are less expensive and easier to use alternatives such as off the shelf resistive ladder attenuator volume control chips, but we feel that these would compromise the signal, and require circuit changes.
For example, the channel tracking in off-the-shelf volume controls is adequate for a single ended preamplifier but not close enough for a balanced circuit. A commonly used work-around to this problem is to convert the balanced input to single ended and use a single ended volume control. A single ended volume control also could be implemented with a custom attenuator and precision high quality components similar to the Seta Buffer, essentially cutting in half the component count needed for the attenuator. However, to do this, additional active circuitry must be introduced (for converting from balanced to single ended for the volume control, then back to balanced for the output) plus the added issue of common mode noise pickup (avoiding which is one of the primary reasons for using balanced circuitry). This would diminish the elegance and simplicity of the design.
Accordingly, we have distilled the balanced attenuator circuitry down to the bare minimum, keeping the important goal (transparent adjustment of system gain structure) in sight rather than creating a full-blown line stage necessitating compromises and increased complexity. Used properly, a digital volume control is at least equal in quality to an analog volume control, but we must arrive at its sweet spot without introducing other issues. The Seta Buffer accomplishes this while being a “set it and forget it” type of device.
Some audiophiles will state that simpler is always better. Isn’t there a concern about a drop in fidelity with any additional component being inserted, much less one we were trying to remove (to go DAC direct)?
It depends on the way that component is designed, and the trade offs (does it introduce more problems than it is intended to solve)? Here, it is designed to solve one specific problem: setting the proper system gain structure, which is a pervasive issue in digital playback systems that must be addressed if the goal is the best sound quality. It does this in a completely transparent (to the signal) fashion.
If we are talking simplicity, then certainly the Seta Buffer is a simpler alternative to a line stage preamplifier, the traditional means of setting gain structure / volume adjustment. The Buffer is not as simple as a passive attenuator, but a passive attenuator trades its simplicity for unacceptable side effects. And the simplest setup of all, connecting DAC to power amplifier with no analog attenuation, will work properly only if the system gain structure happens to be correct in the first place, which is unusual.
Any issues with throwing a bunch of resistors in the signal path? No one threw up on you when you suggested it?
Bunch of resistors? 🙂 I remember when I was a kid making my first forays into electronics, and wondering why someone in their right mind would want to use “resistors,” which only impede the flow of electricity. How wasteful! But they are the most basic electronic component and needed in the design of all active analog electronic devices. Resistors are extremely linear from frequencies of DC to at least tens of megahertz (if carefully chosen) and over a wide range of signal amplitudes. Resistors introduce negligible noise (in the values used in the Seta Buffer) and no distortion. The resistors are matched to better than 0.1 percent (essential in a balanced circuit) and are all metal film, large package size 1206 surface mount types chosen especially for their low noise and high linearity. The result is that the distortion introduced by the Seta is at or below the measurement threshold of most test equipment. In fact the measured frequency response is determined more by the characteristics of the test source used for the measurement than the Buffer itself. Because of the active circuitry combined with carefully chosen components, it is truly just a “straight wire with attenuation.”
What’s that big copper plate thingy?
It is a solid block of high conductivity copper with a cavity machined in the interior to house the active circuitry, for shielding (the circuitry has a -3 dB bandwidth of 20 megahertz) and also used as a heat sink / thermal reservoir to keep the circuitry in thermal equilibrium (temperature fluctuations are one source of noise). This part of the device actually has undergone substantial revision and improvement (in addition to adding a 0 dB setting and a single ended output) since we shipped your review unit to you last year. 🙂
Ahem. And now you know what’s new. Meep.
Why only balanced inputs? Is there a technical reason for this?
Most high quality DACs have balanced outputs. DACs with single ended outputs can be connected with a simple inline adapter. Balanced circuitry provides lower noise and distortion and improved noise rejection and signal headroom.
The single ended outputs are directly derived from the low impedance balanced outputs and are true single ended outputs preserving (as much as possible) the low distortion, noise and noise rejection of the prior balanced stage, while also exhibiting a low (less than 40 ohms) output impedance for capably driving downstream interconnects and power amplifiers.
My personal reference DAC is a somewhat-venerable Alpha Series 2 ($5,000) from Berkeley Audio Design. This DAC has a built-in volume control and comes with a wee little remote. With dual (live) outputs for both single-ended and balanced connections, the Alpha (when paired with the matching Alpha USB converter box) is still one of the best-sounding DACs on the market and clearly offers the best bass performance of any DAC up to and beyond twice it’s price. That’s my opinion, of course, but I’m happy to stand by it.
I’ve had a Da Vinci DAC ($20,000) on loan from Light Harmonic for the last few months. This DAC also has a built-in volume control and comes with a not terribly wee remote. It too uses dual (live) outputs for both single-ended and balanced connections, but needs no outboard converter to do its work. In fact, the volume control only works via the on-board, heavily filtered, USB port. The Da Vinci is the best sounding source I’ve ever heard, and clearly betters not only my extraordinarily expensive vinyl front-end, but every other front-end I’ve ever heard. Quite simply, it’s incredible — it’s also the first DAC that made me seriously reconsider my Alpha. Of course, it’s also a cost-no-object design, and as such, it’s a bit (!) outside the norm for what an audiophile is ever likely to see or hear. That said, it’s here, so I used it.
I also got a chance to use the Buffer on an Auralic Vega DAC via a courteous loan. This DAC is a remarkable piece of work, and at $3,500, includes all the latest features — including being able to convert DSD music files. Given its quality sound and it’s feature set, this unit is my current standard for mid-priced SOTA DACs, and is an easy recommendation at its price point. It’s also easy to use (I’m a “Mac Guy”, so, no drivers are necessary), and as I mentioned, it sounds great — and with DSD files, it sounds outstanding. This isn’t a review, but I encourage you to check out Brian Hunter’s review when you get a chance.
The most profound impact that the Buffer had was on the Auralic Vega, and on that wonderful day, I was able to also play with a Hegel H30 amplifier and a pair of Raidho C3.1 loudspeakers. In case you were wondering, I was shooting for a system that was “revealing”. Ahem. Well, the big Raidhos are startling, both in looks and sound, and with the Hegel driving them, the sound was fast and very precise. Like the new YG “house sound”, Raidho falls on the “detailed” side of the tracks, but veers (if only a bit) more toward “warm” than YG seems to believe is required — and sounds quite a bit like my big TIDAL Audio Contriva Diacera SE, if you must know, while the latter seems more rich and full over the C3.1. Just to locate them on the map, all three of those speakers are far cries from the warm, rich sound of my DeVore Orangutan O/96. Right. Feeling rooted? Properly oriented? Okay — now, the Hegel amp is an ocean of power, effectively bottomless, and capable of lightning responses, with plenty of texture and nuance. I haven’t had it on hand to compare, but I believe the sound to be fairly close in texture to what I’ve heard from Soulution. Linear. Speedy. Quiet. Great detail retrieval. And with an astounding bottom-end impact. Let’s just say that the Raidho and Hegel combo is about as good as it gets in audio’s high-end these days. Hi-Fi, but blessed with life.
Anyway, setup with that gear meant fiddling with the dip switches on the Buffer; I started at -30dB and worked my way to -18dB, where I settled in. This gave me enough output and I was able to tweak the Vega’s volume control + or – 6dB to get the sound levels I wanted. Inserted between that DAC and a Hegel H30 amplifier, the resulting sound was almost dark. But it wasn’t. Say rather that the DAC “naked” was, instead, perhaps, bright — but now was correct. Note that I didn’t feel this way before the Buffer was inline — I was quite enamored of the sound of that system with the Vega driving the Hegel.
To wit: Diana Krall’s cover of Tom Waits’ “Temptation”. This is a 24 bit / 96 KHz download from HD Tracks that I play mainly to annoy fellow Part-Timer Mal Kenney. Okay, fine, I’ll confess: I really enjoy this track, even though I recognize that it’s probably blasphemous to say that — on several grounds. Moving on! With the Buffer in place, piano on this track has more piano to the piano. Har har har. No, what I mean is that the timbre feels more natural — played straight through, there was tinkle but less bite, and Krall’s voice was smoother, with less huskiness. Laid out like slices of carpaccio on a serving board, this sounds rather more clean than I noticed during playback, so take this description as suggestive.
The Vega’s output impedance (balanced) is supposed to be quite low (4.7 ohms), so that wasn’t the issue. But it kinda seemed like it was, because “the everything” below the mid range seemed to take a healthy step down. Quite frankly, I’m getting a little tired of Chris Jones’ “No Sanctuary Here”, but hey, it works. Played back through the Buffer, that big ass bass line was punchier, more meaty, with surprisingly more << throb >> and general menace. Mid bass, an area I certainly hadn’t been complaining about, gained texture. The music simply felt more tactile, more dimensional and yes, the treble was more airy and sweet. Decays felt more articulate. All in all, it was an impressive improvement, with subtle to obvious improvements across the board. Full stop. Read that back? Yes. Let me say it again — improvements across the board. And … done! That’s a wrap! Let’s pack it in ….
With the Alpha, however, the improvements were more subtle. I’d started with a general A/B with the Auralic and the Alpha Pair (Alpha + Alpha USB) to get a handle on the main differences (FWIW, the Alpha Pair made the Vega feel a bit like a bantam weight boxer) before inserting the Buffer into the chain. Here, I did not get the same ah ha! experience that the Vega pairing brought. I had expected more dramatic differences, honestly, especially given the (unpublished and rumored) output impedance of the Alpha being a high 1,000 ohms, and if that was ever gonna bug me, I figured it’d be with the Hegel amp. Now, there was that textural richness fleshing out the presentation, and more mid-bass complexity was discernible, with more snap to the lowest bass, but these differences were somewhat less profound than what I had experienced with the Vega. Real, present, and observable, but not overwhelmingly so: overall, the Buffer added a bit of grace, elegance and charm to an otherwise adept player. With the Vega, we got a super-charger and racing slicks. With the Alpha, it was more like a tune up and some spanky new performance tires. Same stuff, just not quite as night-and-day, if you follow.
Where I heard the most difference was not in “critical listening” mode, however — it was with my too-rare “night-time listening”. When I needed the volume much lower-than-normal (in order to let the Wild Things sleep upstairs), the Buffer delayed the inevitable impact that low-volume tends to have with dynamics and texture. I’d been accustomed to the idea of “quiet listening” being pretty much all about pleasant distraction, but the system’s resolution with the Buffer was quite a bit higher than was used to getting, played low. Wildly ballparking the experience here, but I believe that I was able to get the same level of resolution at half the volume. And at night, that was just sweet.
With the Da Vinci, however, I spent way too much time scratching my head. In the end, I simply gave up. The output impedance on the Da Vinci is something like 12 ohms (aka, “insanely low”) and the volume control is 64 bit (this is a big, big pad). With high-res material, played into a First Watt J2 stereo amplifier and then into DeVore Fidelity Orangutan O/96 loudspeakers, the difference of Buffer/no-Buffer was really hard to pick out — even at night. Oh well. On a DAC that costs upwards of $20k, I’d call that fair. All the elements that the Buffer seeks to correct have been expressly addressed in the design of the Da Vinci, so the value prop there is unclear.
Interestingly, during the review period, I managed to recreate a situation where the batteries were mostly discharged. In other words, I left the unit to slowly discharge over the course of a couple of weeks of travel-induced neglect. When I inserted the Buffer back into the chain for some “critical listening”, I didn’t actually charge it up first. I just stuck it in there. The auto-charge feature didn’t automatically come on (this circuit isn’t defeatable), so there must have been “just enough” to make this work. In normal operation, the charging circuit only engages after a signal hasn’t been detected for 20 minutes. With the Charge Lock engaged, the batteries are always on the trickle charger; the Buffer has a little blue LED attached to let you know you’ve engaged the charger and are no longer “off the grid”.
Anyway, in this weird corner case, the sound of my system suddenly came back to life — when I hit the Charge Lock. All that texture was still there, but with the Charge Lock activated for those maligned and only partially charged batteries, the sound stage opened up, both in width and depth. Transients, which had been blunted (over the “naked” DAC’s performance), regained their bite and the overall dynamics improved dramatically. Sparkle and air, which I hadn’t noticed were dulled when running “on empty”, came back for a happy-happy dance party. Some advice? The Buffer likes full batteries. Given that there’s something like 24 hours to a full charge and that a full charge is ever only a few hours away, leave the unit plugged in and let it manage its own power levels. It just works better — and with the circuit set up to auto-charge after a 20 minute “quiet period”, it’s totally set-and-forget, with no switches or knobs. Interestingly, I didn’t really notice that much of a difference between Charge Lock on and Charge Lock off during normal (read: “fully charged”) play — so if you’re feeling fiddly, feel free to hit it for those extra long listening sessions. Since the charging circuit is galvanically isolated from the AC line anyway, what’s not to like?
I’m still not getting rid of my vinyl front-end, so don’t think I’m going all weak-in-the-knees here, but that said, I do feel compelled to say this: computer audio is way better than it was even five years ago. The ease, the flexibility, and most importantly, the sound, really ought to be taken seriously by even the most ardent vinyl-o-philes. Has computer audio achieved parity with a purely analog system? No — because in many important ways, it’s way past what even the most expensive vinyl rigs could even theoretically achieve. Again, don’t read too much into that — I still believe that truly great computer audio sound is a far more elusive destination than truly great sound from a vinyl front-end. But … things are balancing, though, and I suppose I could say that I’ve seen writing on the wall. Whether or not parity will actually matter is another question entirely. Collecting vinyl records will always be far more fun than ripping CDs or downloading files — and a huge digital library will never look as cool as a wall of LPs. But whatever.
For those of us embracing what the SOTA has to offer the would-be computer audiophile, I think there are some serious pitfalls that can be rather easily avoided. Cheap DACs are flooding the market in a way that even the densest of us cannot ignore — something is clearly “going on” — and more audiophiles are dipping their toes into the water oh-so-carefully with one of the newest, chock-full-o-features, units and … wondering what all the hubbub is about. And that’s a shame. Because there is a lot to get excited about, but getting there still isn’t cheap. Not yet.
The SOTA seems to be more rapidly changing in this segment, so if you have any serious budget sensitivities, don’t move on anything that you’re not going to be okay with dumping in a few years as it becomes completely obsolete. Know what I mean? If you aren’t burdened with “normal people” budgets, or maybe you happened to have found $10k lying around in a coffee can, then feel free to ignore all this, but my suggestion is to find something with all the features you think you’d want, something that lets you experience the joys of computer audiophile-dom — and there are quite a few. With DSD now coming (back?), I’d shoot for a DAC that can handle it, and happily there are quite a few on the market ready to satisfy a variety of budgets. Have fun.
And when you’re ready for more, call Channel D.
In many ways, I think the Seta Buffer can and will help you get the most out of your DAC. So, before you throw out that old DAC, or start an expensive trade up, try the Buffer. Better still, I think that the Buffer will also allow you to get a less expensive DAC — again, one with all the features you might have been curious about — and level that bitch right up to the heavy-weight class in one step.
I don’t think the Buffer is a panacea, which is unfortunate, because we could all use more than a few. So, to the caveats:
- If you have a single-ended DAC, you’re pretty much out of luck.
- It could be that whatever it is that “ails” your system might not be addressed by the Buffer at all. Hey, it could happen — some DACs simply won’t respond as well as others (which may say good things about their design).
- Lastly, and perhaps most obviously, the $3,499 price tag hanging on the Buffer (and don’t forget that second set of cables you’re going to need!), is not trivial. So, no, I can’t see anyone seriously looking to add this to a $500 DAC. I think you’d probably be blown away by how well that DAC would perform with a Buffer — a quality output stage on a DAC may well be the most expensive (and most commonly missing) part — but even I can acknowledge that such combos just don’t make real-world sense.
Now, if none of that applies, you’re in luck. If you are running DAC-Direct now (or wanna give it a try) — you have to hear what your system can sound like. The Buffer sorts out some of the most common and most hidden performance gremlins bedeviling computer audio’s aspirations in the high-end. You want a “more analog” sound? Try vinyl. You want a more natural, full-range, high-resolution audio experience that not only rivals but generally smacks analog around like a red-headed step child? Well, now you’re talking the Buffer’s language. Booyah.