DAC Collections

why-me
again
money-pit

#1

To paraphrase Joni Mitchell –
    They took all the DACs and put 'em in a DAC museum.
    And charged all the people a dollar and a half just to see 'em.

Don’t it always seem to go? Like this. Five or so years ago, I didn’t even have or want a DAC. I mean, I knew what they were, sort of. They were a chip that was in most of the stuff I bought to play music since the Walkman. I recall looking at CDs for the codes DDD, AAD, ADD and so forth, and understood that music got coded from analog to digital and back again.

My understanding wasn’t even stupid level. I was in Scientific, Technical, and Medical publishing, and I’d read some journal articles about it.

Then at some point, I think I decided I needed a DAC. It was after I bought a Sonos Connect to stream music in my stereo. And had a soundbar on the TV with a TOSLINK output. I wasn’t super happy with not being able to plug in headphones, and getting a TEAC DAC/AMP solved that. It also had a cleaner DAC than the Sonos Connect, so there was a side benefit to routing digital out from the Sonos also.

Then I somehow got the idea that I deserved better sound from my phone. Where someone who grew up on POTS landlines got that crazy notion, I don’t know. I’d used little headphone amps for years because sound from walkman and mp3 players was not so hot. And a couple of years ago I got the Dragonfly Black when the 1.5 came out. Now I see I want a THIRD one for convenience in another location, and could forsee a future one for a different house or my office, where I’m usually too busy to listen to music.

Why Why Why do I seem to dither and maybe buy two so I can see what I prefer, and maybe start a collection of these little intermediate things that I never needed before??? It’s not like it was when I bought camera lenses for my SLR. Is it?


#2

That depends. If you ever got to the point where you owned multiple 50mm lenses, like one for portrait bokeh, one for low light capture, one for the built-in and silent autofocus motor, a pancake one for discreet street photography and a couple of vintage ones for some soft focus and vignetting, then yes, it’s like that :slight_smile:


#3

DACs are the last part of a chain to worry about I think.
So far I’ve found that synergy between DACs, amps and headphones/speakers plays a bigger role the just the DAC alone. Also the more features for controlling the DAC the more I like it for critical listening (RME ADI-2 DAC). But if I’m listening for enjoyment basic DACs are fine. I currently have 6 DACs (that includes combo units and mobile dongle DACs). Also currently have 3 on loan. They all have strengths and weakness, but I’ve found that trying to pair them has created just as much enjoyment as what @pwjazz does with EQ.


#4

My cameras were more manual than that. I had lots of filters for different effects. Out on the street, I’d just hang a twin lens reflex from my neck to take discreet shots, most people expect the camera to be at eye level. But I did have a few 50mm lenses, 135mm, 28-50 and 150-270 zooms. and and and. Silent Autofocus Motor? Try fingers and groundglass screen.


#5

I agree, DACs are the last part of the chain, just like wires…I mean interconnects.


#6

In the recent era (last 5 or 10 years), DACs have become much less important than before. However, their prior weaknesses were seemingly responsible for the bad reputation of “digital” music and computers as sources. LPs might not have maintained such a devout cult following if not for the very real digital issues of the 1980s and 1990s. [But, LPs are also about the analog experience and hands-on interaction…]

The first 1980s CD players had a variety of quality and stability issues (e.g., fluttering discs until they put a weight or clamp on top). Shortly after the first issues were noticed, they went to 2x and 4x oversampling to improve quality. And then I bought a Panasonic MASH 1 Bit CD player, as it was supposedly much better technology–well, it was indeed brighter and harsher with bad sources, yes… But, it was not more pleasant or noticeably better in actual use. Early “DDD” recordings such as Dire Straits “Brothers in Arms” and Sting’s various releases were supposed to work around digital to analog issues, but didn’t seem much different from AAD or ADD releases to my ears.

On the computer front, I had multiple generations of SoundBlaster audio cards in the 1990s. These went from 8 bit to 16 bit to 32 bit to 64 bit. I recall widespread disappointment with the original 16 bit card of 1992, as it increased audio resolution but did nothing for noise filtering. So, you could hear incredibly loud artifacts that had not been audible with prior products.

Similarly, the very first MP3 encoders ran the CD signal through a sound card (I think through the DAC before redigitizing) and dramatically reduced quality. They made a CD sound like bad FM or good AM radio, at best. I encoded a couple test tracks and then set MP3s aside until the ripping process changed.

Finally, even as recently as 5 years ago many computers had poor quality DACs and a replacement could provide a huge boost in high range detail and clarity (also, current iPhones). But, other computers were decent as provided. I still hear slight high range differences between the two DACs I use most often (older FiiO Alpen 2, newer FiiO Q5), but the differences are not important.


#7

I used to believe the same thing. And I have 3 top notch CD players (NAD, Marantz, Sony ES). Then I bit the bullet and bought a Schiit Audio Bifrost (The plan was to start collecting Hi-resolution music and I would need a good DAC). Boy was I wrong. Their catchphrase “For the music you own, not the music you have to buy” couldn’t be better. I am hearing a much clearer, more dynamic presentation from the CD’s I own than I would have imagined. I am still planning on collecting Hi-res music, but suddenly it isn’t as high a priority


#8

I should first emphasize that I have never gone down the EXPENSIVE DAC route. By which I mean fancy tube DACs (my guess is that the tubes are for audio output, not decoding) or DACs in the price range of $1k-$10k. Since electronics seem to still follow the principal that things get better and cheaper over time, I can get more by waiting a year or two. Still, the evolution seems (to me) to be starting to slow just a bit.

So the collectible DAC seems to be a matter of convenience, form factor first. Then the power I want to drive which headphones or other inputs. And what whiz-bang features I will actually use. I distrust gimmicks (see Gimmick thread) but can’t help being attracted by tinsel fluttering in a sunbeam.

Someday - if my ears don’t wear out - I may get a more expensive DAC. Probably to go with some kind of music server that caches my CD rips and any hi-def tracks I have purchased. I’ve been happy with Wyred4Sound equipment as a price-point compromise, but that could change if I actually pull the trigger on new speakers. Everything costs money, and I’m still working on that MIRC* designation.

*Member, Idle Rich Class. (Ref Bertrand Russell)


#9

Differences between DACs do exist. They can be very subtle … and often are - requiring audition-style listening to detect at all. There are cases where they are comparatively gross, too. And their relative performance with different sources can vary significantly as well.

One thing to keep in mind is that anything that the DAC (or any other source) doesn’t resolve or expose is lost - nothing downstream can bring it back. If the DAC can’t deliver on detail or micro-dynamics even the best amp and headphones in the world can only reproduce what they’re fed.

Headphones will always give the most noticeable change in sound. They’re the best place to focus your budget. But they can’t retrieve or render information they never receive. Nor can they fix issues of omission or commission from the source.

For example, some DACs do a really poor job with content that contains a lot of content at, or near, 0 dBFS. And such content is much more prevalent when using lossy-sources like Spotify, simply due to the way that the psychoacoustic encoding models and optimizations the codecs involved with these systems operate. At a technical level, the issue comes down to distortion resulting from filter-overshoot … otherwise known as “intersample-overs”.

A DAC that sounds fine with lossless content with most of its information not pushed right up against the headroom limit may be harsh and unpleasant when paired with high-energy lossy-content. As an example, I have four “budget” DACs sitting on my desk right now … one of these is unaffected by this phenomena, one is marginally affected, one is awful and the last is unlistenable (despite measuring well with lossless non-0 dBFS content).


#10

For us DSP users, is this solvable by simply applying some negative gain in the digital domain? I understand that one loses some dynamic range from this, but with 24 or 32 bit DACs it’s not something about which I worry.


#11

Yes … provided you don’t run into similar issues/limitations with the filter implementation in your DSP. It is entirely possible, however, for a software filter to cause similar issues.

Unfortunately most standard DAC measurements don’t illustrate the phenomenon. It is definitely measurable though, and just last week there was an AES presentation on the subject (“Digital Audio at the Limit”). The findings of that paper indicate that the issue is largely avoided by listening at -3 dBFS.

Some DACs account for this. Chord’s default “DAC mode” output on the DAVE is -3 dDBFS. Benchmark’s DAC3 is designed with appropriate digital and analog headroom. A recent update to the RME ADI-2 DAC (and Pro) added more headroom for this sort of thing too.

Most data-sheet/reference implementations of common (and popular) DAC chipsets do not.

With well-mastered losslessly encoded files/material the problem is fleeting and scarce and may not occur at all - unless the native levels are pushing above -3 dBFS a lot. With some lossy encoding schemes/settings it seems that content gets pushed up to 0 dBFS even if the source material wasn’t. There’s some logic behind doing so (utilization of the encoding space), but it causes other issues.

I know when listening via Spotify the issue seems to be much more prevalent than via Apple, Google or Prime Music. Whether that is due to the way Ogg/Vorbis encodes, how the encoder was setup, or what state the encoded masters were in, I don’t know. I do know that Spotify becomes very fatiguing, very quickly for me, even when I don’t know that it is the source in use. Far more so than other sources. And one of the things I’ll be doing is some actual measurement and analysis to see if this is down to intersample overs and encoding headroom or something else.


#12

Fascinating!


#13

I will see if I can get the author of the AES presentation to post about it here. He is usually stupidly busy, but it’s useful information and helps explain a lot (in my opinion) about the sometimes differential performance encountered with well-measuring data-sheet DACs and how they wind up sounding in practice.


#14

This is interesting, as it implies that when considering gain staging on the recording side, it’s probably safest to stay below -3 dBFS at each stage.


#15

I believe that’s been true since the old reel-to-reel days.


#16

It certainly has!

There is a big difference in the audible effects of analog overload/clipping vs. digital. Especially when it comes to gradual tape/head saturation vs. the gross mess that occurs when you clip in the digital domain. Also, this is a replay-time issue, not a recording-time one (though lower output levels in the delivered content fix it just as readily).

And in this particular case, the interpolation involved in the conversion from digital to audio can result in such clipping even when the source signal doesn’t actually reach 0 dBFS (which should be fine). The AES presentation I referred to shows that -3 dB is a reasonably safe starting point - but specific implementations can be worse.

And as this is exacerbated with lossily-encoded content - which is a) typically played via software that does’t have precise (or high-quality) digital attenuation capabilities and b) is more often paired with lower-prices DACs that seem to be particularly affected by intersample-overs.

For example, the Topping D50, on most of its filters, seems a lot more sensitive to this than the D30.

I expect the issue, as a whole, is going to get a lot more attention in the near future. Which will hopefully have the effect of forcing DAC designers to take steps to avoid it AND get the mastering-engineers to stop pushing things so close to 0 dBFS and back off with their dynamic range compression!