Properties of DACs in order of importance to SQ

Audio Gear Talk
1

Hi Guys,

People hear differences between DACs. I want to figure out what’s more likely to impact SQ, and suspect it might not be the usual poster boys (Sample Rate, THD, RND).

I wanted to start a discussion about this, hear your respected opinion and hone my own.

Start with a friendly disclaimer about this post:

  1. SQ Focus: As the title conveys - I wish to focus here on Sound Quality.
    Features/Properties are relevant here only if they are important somehow to SQ.
  2. Pls no “Holy wars”: It’s there in any “geek field” I know. I wish to avoid them here and if possible stick only to measurable theoretical properties and knowledge (however limited or incomplete) in the digital audio domain that is available to us, whether first hand or quoting experts. Surely it is only a discussion and individuals will always have their own set of preferences that give him/her joy!

So, the list of SQ relevant DAC properties I’ve gathered so far (unordered, let’s fix that :slight_smile: ) -

  • DAC Chipset
  • Input Type (HDMI, USB, Optical, RJ45)
  • Jitter
  • Sampling Rate
  • Upsampling
  • Format Support
  • THD
  • Dynamic Range
  • Crosstalk
  • Volume Control
  • Balanced outputs
  • DSP options (Inc. Room Correction)
  • Integration (Internal wiring/soldering/galvanic isolation/capacitors/transistors/DC coupling/etc)
  • Power supply (quality/isolation/separation)

Missed something? Let me know.

Let’s order them for general importance to Sound Quality so we can all be more educated choosing our compromises!

What’s your take?

2 Likes
2

I get where you’re coming from and it would be interesting to see the responses – but for me it is impossible to order this set in a cogent way. If I’ve leaned anything about DACs, it’s all about integration.

This seems to hold true today as much as ever. I recently tried the budget sweetheart du jour (Project Pre Box S2 Digital) for a couple of weeks, which seems to tick most of the fancy boxes spec-wise, but turned out to be a bland and anaemic performer in my use case. I’ve replaced it with a Meridian Prime, which is nothing special spec-wise, but boy – does it sound good.

I guess my answer would be in the right implementation of most of the above. :slight_smile:

2 Likes
3

If I was putting a list togethor, these are the 4 most important.

Analog output stage
Power supply(s)
Clocking
Filters

4 Likes
4

Meanwhile, my take on each:

DAC Chipset
Probably a large impact on sound quality between different classes of quality.
For chips within the same quality class - my current opinion is that SQ won’t differ and if there is an audible difference (if there was an identical DAC with different chipset choice) it would be a matter of personal taste and not measurable SQ per se.
There are known DAC Chipsets (like the ESS Sabre line) that are used in many DACs models from different vendors. (**Talking here only on the actual D to A conversion, upsampling, sample rate, formats - all might be derived from the chipset but are discussed separately)

Jitter + Input Type
Joined these for convenience.
Some audio inputs have different bandwidth limitation which might affect SQ but since Sample Rate is covered separately I’m more interested in Jitter implications and have learned that it might have the potential to be really impactful on SQ.
Here is an article I enjoyed reading about this topic (aimed at HDMI but explains the entire subject along the way):

Some DACs perform re-clocking, jitter reduction, etc.
Some use RJ45 to get the digital stream and feed it internally to the DAC so the clock could be controlled by the manufacturer.

My opinion: I don’t know.
I know HDMI is inferior in this regard because of the audio clock issue and I guess the RJ45 with internal clock feeding is ideal?
Unfortunately, jitter data is not commonly provided and the source is also important as it is generally the clock (where products like SOtM claim to fame is).

But what is audible?
Would an HDMI device with acceptable Jitter sound audibly worst then a fancy re-clocking USB/internal clock RJ45 device? I have no idea. Do you? What does the theory say?

Sampling Rate
Well, that’s a huge debate that transcends DACs. Regardless, I guess anyone would like his DAC to support the highest sampling rate available in his collection.

My current opinion is that Sample Rate conversion, or the elimination of which, is more impactful/audible than the actual Sample Rates above a certain kHz/bit depth.

The DAC’s implementation of said Sample Rate* should be taken into consideration and ideally we should feed the DAC it’s best (not necessarily highest) implemented native rate to reduce processing on the signal.

Again not a commonly documented property of a DAC.

Upsampling
So coming from the previous item my opinion is that quality difference in upsampling implementation might be audible and vary in Sound Quality.

My opinion: Personally I’m an HTPC geek and prefer doing Upsampling in software with the likes of HQPlayer/Roon’s DSP just for the sake of removing the DAC’s upsampler from the equation for more freedom if not also higher SQ overall.

Format Support
PCM, DSD, MQA, etc.
Doesn’t have enough knowledge here to form an opinion.
For me in an ideal world I wouldn’t need to - Let the software player (which can be updated, upgraded, replaced, etc) handle the formats and not DAC.

THD
Well…
My opinion is that the differences here between decent and quality DACs are negligible compared to the analog parts of most systems (e.g. the Amp and definitely the Speakers). So my guess is: Not important .

Dynamic Range
Same as THD.

Volume Control
Well… I’ve read several reports that it could be impactful on SQ. How could this be measured?

Balanced outputs
Might be of importance for long cable runs / noisy electrical environment.

DSP Options
This is a tricky one.
MiniDSP is a great example of a company with devices of modest Sample Rate support but raving customer reviews. I firmly believe proper Bass Management, PEQ and Room Correction is of higher importance to SQ than all other items on the list, considered they are all above the acceptable audible range.
But, being an HTPC geek as I mentioned before means that I also prefer to handle these items in the PC and not the DAC. Not for everyone, I agree.

Integration (wiring/soldering/electrical components in the signal path)
I guess these would affect THD and Dynamic Range?
Anything else?
Assuming not, my opinion: As long as THD and DNR are acceptable than it shouldn’t matter much.
Question remains for me - “What’s acceptable?” :stuck_out_tongue:
Hoping you help.

Power supply
Similar to integration.

 .


 .

My Order:

  1. Chipset - Should be of a known good quality class
  2. Jitter - I’m guessing here out of elimination since I can’t understand how THD/DNR be impactful enough.
  3. Power - Reading it might also cause Jitter (Less about THD/DNR)
  4. The rest.

Caveats:
If can’t have DSP/RC/PEQ/BM in software:

  1. DSP

If can’t have upsampling in software:
2. Upsampling

Your turn :slight_smile:

1 Like
5

Although I tend to agree to @RBM and @DrTone as well I would like to add to the list:

Type of converting - R2R, delta sigma, DSD

6

I get what you are saying.

I just hold the opinion that in the digital and electrical domain the majority of SQ parameters could be represented in numbers and measured.

It might just be that the unfortunate reality of our field in which the properties that actually “count” for SQ are not well known by the consumers and thus not measured/provided by manufacturers.

I hope we can change that eventually and think that maybe even now we can do enough detective work to be able to “guess” the theoretical potential SQ range of different models.

7

I’m on the same page as Rene. But I guess it depends on your definition of sound quality. To me, it’s about my perceptual reaction to the music. And that sound is derived from the entire system: my mood, expectation, room, amplifier, DAC, transport, Roon, speakers etc. I don’t think you can measure this.

Some enjoy (prefer) vinyl over CD or valve over solid state. Arguably from a purely numbers perspective, one is “better” than the other. Yet, some will say that SQ is far superior for vinyl | CD | DSD | MQA … and they are correct … because it is a perceptual reaction.

8

And I’ll argue that we simply measured the “wrong” things or not all the “right” things. In that specific case you mentioned (vinyl over CD) we looked at noise, then sample rate, now Jitter… What next? All can explain better why, in numbers, some prefer vinyl over CD. Numbers are numbers but what numbers are we measuring and how those numbers are affecting the way we perceive sound is the key. We are keep learning, as humanity and society.
I think it is perfectly reasonable to assume we can measure what is likable a better SQ and what is not.

Of course, “In many conditions, placebo effect is a big part of the effect of the drug”. That’s OK too. It’s part of what being Human means.

9

I don’t think this is possible. What I like, you may or may not. Unless you are saying that SQ is accuracy of reproducing the original sound.

That’s why I mention mood and expectation. When we make that purchase we often have high expectations and this can cloud our judgement. But not in Rene’s case with the S2. :slight_smile:

10

From a well respected DAC (and amp) designer, the late Charles Hansen of Ayre:

3 Likes
11

I know, but can’t we categorize measurements into “likely to degrade SQ for most” and “subject for individual preference”?
I’m sure most would agree that high jitter sounds worse, everything else being equal.

With subjects like the “House Curve” we learn that flat measured response is not equal to flat perceived response, in light of frequency preferences of the Human ear, which also varies between Humans and the same Human in different conditions and stages.

I’m not going there. I am of the opinion that there are measurements and properties that would likely convey the SQ of said digital device better for most people than what is shouted out at us today.

12

:slight_smile:

I think there are too many variables and of the properties mentioned some only offer marginal gains. I’d rather enjoy the music!

13

Exactly what I wanted to discuss, Which properties you believe are most likely to be important for SQ so that me and others would be able to make educated decisions considering budget and other compromises, just by reading.

14

A post was merged into an existing topic: Pro-Ject Pre Box S2 Digital

15

Hi Ron
I can’t talk to your relevant properties list as, for me at least, I can’t A/B things on that list.
I CAN say from my experience the difference in DACS, to my ear, is as comparable to the difference in any source, as in tape deck, turn table, CD player. In other words huge as you would expect your source to be.
I upgraded several dacs until I settled on a Schiit Yaggi. With that I found a subtle difference using the AES input ( better) than using the USB input. Although after the most recent USB upgrade I found little or no difference.
In regards to the recording format input I have found no advantage between and sampling rate / upsampling / format. Its seem to be completely random based on the quality of the original recording. In my experience across all formats, digital, analog, vinyl, tape it always comes down to the recording. I have some of the latests DSD recording that are simply no better than some of my best 16 bit CD’s. I have an ancient mono vinyl Deutsche Grammophone that makes my neck hairs stand up if you can stand the sound of an oomph pa pa tuba being played by a man in lederhosen.
My larger point being all of the items on your list are obviously important but they work together and while I suppose a poorly designed product could have a weak link, thay are all equally important as long as they are of equal quality. I would think rather than trying to focus on the small details of the sum of the product it might be better to focus on the quality of the recordings in your collection, and simply buy the best DAC you can afford. I hope that two cents worth is worth some value to someone reading.
Cheers
Steve

16

This discussion reminds me…

The OSI network stack has 7 layers:
7. Application
6. Presentation
5. Session
4. Transport
3. Network
2. Data link

  1. Physical

But somebody added:
10. Religion
9. Politics
8. Finance

2 Likes
17

Well for multi-channel (I need it for Stereo content for software DRC and BM) there is an issue since the only non-HDMI consumer ones available are:

  1. miniDSP U-DAC8 @ ~$250
  2. exaSound e38 @ ~$3900
  3. Merging+ NADAC @ ~$11000

So… you can now see why I want to better understand the implications of something like HDMI Jitter :stuck_out_tongue:

Not sure a $3900 DAC is sensible for a ~$10k system budget-allocation-wise.

18

BTW I want to give a huge shoutout to Keith from Emotiva for writing a very insightful DAC internals post answering my question here -

http://emotivalounge.proboards.com/post/931517/thread

19

Hey Ron, reading Keith’s great post there, it seems like he is addressing DAC chips for the most part? Since he ends with:

"I would put up Sabre DACs as an example of a few of my comments.
They have a distinctive sound.
Some products that use them have more or less of that distinctive sound.
However, when it’s there, their distinctive sound is… well… distinctive.
Some people like it, and some do not, and some don’t notice it at all (but most do).
While it seems to be related to their filter characteristics, I have never been able to point to the specific cause of their characteristic sound.
(Overall, the characteristics of their filters don’t seem to be unique from everyone else’s.)"

But I believe the bigger picture is how the chip/s is implemented inside a box which provides analogue outputs (RCA or XLR). For that, my post above linking to Charles Hansen’s post probably gives a good summary. I’ve had a couple DAC designers tell me that post of Charle’s pretty much nails it…

Of course if your question was really only about the DAC chips themselves, then Keith nails it, for sure.

20

Thanks!! I think I’ve missed that link somehow, on it now! :slight_smile:

Regarding Keith’s post - OTOMH he was refering to DACs in general, including chips.