The Great DAC Mystery

@knownothing 

I have no experience with Chord or JRiver. I don’t doubt your observations are correct - they seem to match what is observed by many others. Chord M-scaler is clearly a design intended to improve upsampling filters over the conventional approach. I read that Chord designer, Rob Watts, believes that highly accurate extremely long tap filters make an audible improvement over the conventional approach. 

@shadorne this is a great thread, thanks for initiating.  Have you tried FPGA DACs like those built by Chord?  I have had several of their DACs and have come to appreciate what they do - low listening fatigue, excellent timing, good detail and insane placement of instruments and sounds in the soundstage.  Wondering how you think your discussion applies to that design solution.

FWIW, I have tried software upsampling with JRiver V32 upstream of Chord DACs and find that while that can improve performance of cheaper delta sigma DACs, it actually reduces the performance of my Chord DACs which are wonderful at handling red book files all on their own.  JRiver may be significantly less capable than HQP which I haven’t tried.  Any thoughts on this appreciated.

kn

@sudnh 

Yes for sure the analog side of a DAC is important - the output with or without volume control is where the rubber hits the road! Absolutely!

No..  I mean the analogue amp that is in every DAC. 
 

the implementation of the analogue amp also affects SQ. 

@sudnh 

Do you mean the type of output (rca or balanced), output level, and if there is volume control? Or the final analog output filter design and buffer? Or the degree of separation of digital from analog side of things? 
 

For sure there’s a lot there to unpack

I don’t see any mentions on the analogue side of a dac.

To me, NOS sounds more real than the upsampled, filtered options on my Pontus II. Listen to the echo decay of an organ in a cathedral. The decay with NOS sounds more realistic, 3D. Instruments have presence. 

If measurements were enough this hobby would be simple and we’d have healthier wallets. People who assume that everything we hear can be measured is a baseless assumption. Our hearing acuity is much more sensitive than current science can measure.

Not wanting to put down anyone’s system, but seems those who claim not to hear differences their DAC is not high enough quality, their audio chain is not transparent enough, or they simply haven’t tried demoing themselves.

Top DACs can be over 100k (DCS, Aries Caret, Wadax), with many top brands in the ~50k level (Total DAC, Lampizator, Linn).  Comparing DACs <5k level may show small/negligible sonic differences, but as you climb up a brand’s product line aka spend more for better engineering and parts, improvements can be heard

Like all components, it’s considered unwise to purchase a component beyond the ability of the audio chain (source, electronics ,speakers, cabling), money usually better spent elsewhere to lift sonics.  In other words, buying a quality DAC will not fix weaknesses in the audio chain, where the weakest link is the sonic bottleneck - the entire audio chain matters.

Thank you, @shadorne!

@desktopguy RCA long Black plates are goldilocks to my tastes - just right!

+1 macg19

Just bought a new Black Ice FX DAC and am currently burning it in. It already sounds quite good, but the real test will come when I drop it into my big desktop system with headphones, speakers, sub.

Current DAC in that system is the MHDT Labs Orchid with the best NOS buffer tube I tried with it (a Mullard CV5331 + adapter). It’s a great-sounding DAC that no doubt measures badly.

It will be very interesting to hear whether the Black Ice DAC, similar to the Orchid in use of antique chips, but ups the ante with a full tube (matched pair of 12AX7s) output section, sounds as good as the Orchid.

I’ve read enough positive things about the Black Ice to have invested in 2 matched pairs of pricey NOS 12AX7.(RCA blackplates; Brimar greyplates).

One response to this conjecture is there is no universal consensus on what is best dac topology, R2R, chip dacs and FPGA dacs all have their proponents and critics, much more than filters in question here.

I've also experimented a bit with HQPlayer, Euphony OS has embedded HQPlayer which allows one to audition for limited time periods, have not yet discovered settings that bettered bridged Roon setup, and Roon dsp a noticeable downgrade, artificial or hifi sound quality.

R-2R DACS?  I thought open reel tape decks were analog!!??

I don't get it. Guess I'll just jump back on that quilting forum. It's safer there.

This entire subject is way beyond my pay grade, but I definitely am enjoying the whole conversation.

Surprisingly civilized and much more enjoyable than the other end of our Bell Curve.

The 0.7 ms time delay correlating to inter-aural distance is an interesting point. Listening to the effect at -10 dB reminds me of what I hear in the phantom center on a typical stereo listening triangle. It's darker sounding compared to sounds panned to the sides. I've found that using a wider listening triangle works better for me. The first null is pushed down to about 700Hz, which is similar to the results of my experiment. But the benefit I think is better head shadowing. So the phantom center overall sounds brighter and clearer to me with a very wide speaker spacing.

This is a mathematical certainty - no hand waving at all.

Agree. I'm convinced it happens. I'm not convinced it's typically an audible problem. I just did an experiment with 0.7 ms pre and post echo loud enough to be easily heard. At -10 dB the effect is loud and clear. Besides being obviously louder, it also sounds fuller with more midbass energy and dramatically reduced treble brightness. Definitely not harsh and gritty. Using REW to combine artificially generated measurements to get one with a -10 dB pre and post echo at .7 ms shows comb filtering starting at about 200 Hz with minus 13 dB nulls occurring all the way up, leading to an average sound level above 200 Hz of roughly -6.5 dB compared to below 200 Hz. That explains the reduced brightness.

At -67 dB the comb filter nulls are about .015 dB. To call that subtle would be a vast understatement. 

@rbstehno 

So you concur that upsampling on a DAC chip is not as good as a manufacturer custom upsampling on a FPGA. I agree fully. 

FPGA days are the best (I won't buy a dac that isn't fpga based) for current and future use. DSD is the best. i2s or network are the best interfaces to a dac/streamer, network into streamer, i2s to the dac, or network straight into the dac.

USB is the worst interface. PCM is ok if you can't use DSD. 

@sns

R-2R in NOS mode will indeed not suffer from upsampling artifacts, as there is de facto no upsampling. Proponents of R-2R emphasize the purist approach to the conversion. That said, in NOS mode the R-2R DAC may still benefit from high quality HQP upsampling on a PC because this allows the output filter to better remove aliased sound above the nyquist (the upsampling having pushed the nyquist way higher and making the gentle analog output filter more effective)

While I'm not an EE or fully immersed in technical aspects of this hobby, what @shadorne ponders is how innovation comes about in audio equipment. Digital innovations are coming relatively fast and furious, this sort of speculation is what drives it. 

I also now have a better understanding of why I prefer the minimum phase filters to the linear on my Delta Sigma dacs. Also, now just getting into R2R dacs, some of the above may help to explain the unique sound signature of these dacs.

I have found the power cord can effect the quality of sound , the usb cable for sure,

when I am using a DDC reclocker running I2S cable the quality hereto matters quite a bit ,then youhave everything upstream  starting at your router modem combo and quality to start , #1 getridofthe Junk wall wart , Digital isnot grounded 

from house-to house .I use a Linear Tube Audio LPS power supply which made a Big difference a 5 amp slow Hifi tuning gold copper fuse plenty good ,and a budget Pangea sig,mk2 power cord , then not too much for the moment a decent EthernetHub which has a LPS a good temp control clock ,low noice regulators ,

the LHY  sw-8  nice hub same brand fuse and power cord , makes a very nice improvement ,leave no weak links in the audio chain.

@asctim

The late Julian Dunn - known for the invention of the “J-test” - postulated in an AES paper that pre-echos could be the reason why higher resolution digital audio sounded better. Julian wrote some of the manuals for Audio Precision on DAC testing. You can kind find some of his papers by googling Julian Dunn AES.

That a sinusoidal ripple in the passband of a filter will produce a pre-echo and post-echo in the time domain is a well known fact of Laplace/Fourier transform mathematics. This is a mathematical certainty - no hand waving at all.

https://src.infinitewave.ca/

Here's an interesting web app. to compare sample rate converters. I've definitely heard the weirdness that happens sometimes when digital goes off the tracks. If you play a sweep you hear very distinct echoes when this sort of problem is happening. It's like an effects generator.

When it's really bad you can definitely hear it. At -67 dB maybe some people can, but I doubt it's one of the main causes of audiophile dissatisfaction with digital

I can't find any mention anywhere of oversampling causing pre and post reflections. Can you site some references?

I suspect that DACs' share the same issues as software in general as it applies to computers in general....nothing is going to 'sync' as 'flawlessly' as the vendor applied in their tests.....if 'flawless' was achieved, which...? ^shrug*...

"Only to a point..." is what I read here, which does infer it's the miniscule programing details that might induce 'weirdness' into the listening experience...

For now, I've got other anomalies to pursue ....

@asctim

Yes I agree. Typical DAC 2x upsampling filter will have 0.7 msec pre and post echoes of amplitude -67db of main signal. It’s incredibly small but certainly in theory, this is in the actual humanly audible range, and considering the pre-echo will not be masked. I will explain below why I believe we can hear subjectively this tiny echo as harshness and as telling our brain the sound came from the speaker (will reduce quality of stereo image).

We are very good at locating sounds above 6kHz using the interaural difference in sound levels at each ear (our head being responsible for heavily attenuating higher frequency sounds coming from our right from reaching the left ear and vice-versa). The echo, although very small, therefore causes confusion for proper location of the stereo image. It gets worse; this echo at 0.7 msec is very close to the actual interaural time difference of hearing a sound at each ear from extreme right or left - so even sounds between 300 and around 2KHz may not image as well (this frequency range is where human hearing tends to rely on time arrivals at each ear to sense sound location). Now worst of all, this echo is a true echo - it is an entire reflection of the audio signal - it perfectly correlates to the music or vocals! So although very small, our hearing is well developed to detect it - especially as that is precisely what our ears/brain are listening for in order to detect sound location from left to right.

The above is also why speakers with baffles less than 9 inches will image very well and those speakers with baffles of 12 or more inches no longer “disappear” and no longer image so effectively (due to our ears detecting the echo from the baffle edge diffraction, being only perceived once it exceeds about .6 msec in delay). This phenomenon is well understood by speaker designers - the manifestation being the large number of speakers that have a narrow baffle just for the mid range or tweeter.

I did some reading on the subject. I read the echo is under 1ms offset and -60 dB or more below the main signal. That’s really quiet, and not a lot of timing offset. The thing to show is if under any conditions anyone can pick up on that, maybe not even hearing it directly, but sensing it somehow changing their perception of the sound. Perhaps some people can. I suspect I can’t.

I’ve tried mixing echoes of various types in to recordings to see what effects it might create. I quit hearing any notable change by -35 dB. Maybe even -30 or -25, depending on the type of echo and the timing.

If you want to hear something that really stands out, use a pitch shifter to slightly de-tune the echo. That’s a really bad sounding effect. Even that one disappears for me long before -60 dB.

@asctim What you refer to seems to be the conventional wisdom and accounts for 99% of the literature. However this Gibbs ringing is not audible to the majority of people as the frequency is at the transition frequency of the filter (the point where you have the filter acting) - and this frequency is almost always above 20KHz. This is a fact and therefore makes this pre-ringing or post-ringing irrelevant as far as our hearing is concerned. I don’t believe we hear it at all. We do hear of course the affect of minimum phase filters because they alter phase relationships and therefore the timbre. (Some of us might notice a very smooth filter that rolls off at 15KHz - those younger listeners with full frequency range hearing)

I propose that what actually matters is changes in phase (timbre) and the passband ripple. The passband ripple creates two echos ( a complete reflection of the entire audible signal ) - a pre-echo and a post echo. By changing the filter type and it’s steepness: the most audible impact will be a combination of two things:

1) the way minimum phase will change relative phase and therefore the timbre compared to the theoretical optimal of linear phase (which preserves phase/timbre)

2) the change in the ripple within the passband - which affects both the echo timing and amplitude.

High quality (heavy processor overhead) HQP filters allow for a passband ripple to be incredibly small (many orders of magnitude smaller than a DAC FPGA) so upsampling to DSD with HQP will eliminate the pre and post-echoes found in all regular Sigma Delta DAC’s (due to their limited processing power) An alternative is an R2R NOS DAC used in NOS mode (as this will not suffer from sinusoidal passband ripple with practically any correctly designed analog output filter)

An interesting article about DAC filters and ripple:

https://addictedtoaudio.com.au/blogs/how-to/how-to-pick-the-best-filter-setting-for-your-dac

I remember reading a Stereophile article way back in the old days. One of the reviewers had got into his possession a DAC of some sort that would let him write and apply his own filters. He was pretty sure that ringing filters were the culprit for what he was hearing that he didn’t like about digital. His conclusion after writing and listening to some horribly ripply filters was... filters weren’t the problem. He couldn’t make the DACs sound any better or worse to his ears by eliminating ripple.

I bought a NOS, filterless dac with an old chip because I wanted to hear it for myself. I couldn’t hear any improvement like I was expecting from that either. I think most of what people are hearing and referring to as things like microdynamics and timbre are happening on much longer time scales than any problems with timing that digital is introducing, or pretty much any electronic gear such as amps or pre-amps. However, anything that alters frequency response even just a little bit can result in all sorts of unintuitive perceptions about timing, I’ve tested my hearing on timing issues, creating signals that I thought would audibly reveal timing issues with slower bitrates compared to higher. Measurements showed the higher bitrate effect was actually coming through the speaker, but I definitely couldn’t hear the difference. That’s when I realized I was making timing effects that occur faster than 1/20,000 of a second. I can’t hear that high, so I can’t hear the timing that high either.

@shadorne 

I've always felt that software upsampling yields better results (to my ears at least) than hardware upsampling. I've had a number of DSD DACs (PS Audio--I was a beta tester for the original PerfectWave, EMM Labs, Playback Designs) in my own system and moved on to R2R designs with software upsampling (HQP). Just more musical IMHO.

Great insights shared here. All of them, including those who advise there is no holy grail to be had and just keep trying until you find what you like. I get all that but the nerd in me is always curious - especially as the differences I hear are much greater than I hear between amplifiers.

I experienced a variety of sound with PS Audio DSD DAC (each major software update changed the sound) - this was my first realization that the upsampling calculations had a huge impact(as hardware was the same and only software changed). I moved on from the PS Audio DSD and began experimenting with other DACs until here I am with Roon + HQplayer.

Anyone else go down the DAC rabbit hole and end up using Roon or another upsampler to high rate DSD (running on a pc) to feed their DAC and with much better results compared to letting the DAC FPGA do the upsampling?

I think most of us might agree with the OP that maybe there is a legitimate electrical engineering reason that some DACs sound better than others, but ultimately it is almost a fool’s errand to try figuring out which is the "best" DAC.

I mean, you’d think you could send a flat noise to a DAC from 20Hz to 20,000Hz and measure the output at all volume levels in say 0.5dB increments from zero to full volume.

Then the "best" DAC would be the one with the flattest frequency response. Also send pulses of every frequency in increments of maybe 1Khz of varying times say from a 1/1000th of a second to 1 full second and see how it responds.

But at the end of the day, would you think the "best" DAC actually sounds the best?

And come on, we are fooling ourselves if there is any expectation of reproducing exactly what the recording or mastering engineer is hearing in the studio or through their mixing headphones.

Buy a DAC you like the sound of and enjoy it. And keep in mind that R2R DACs shouldn’t add any artificial "soundstage and imaging" that isn’t in the original recording. The "best" that any DAC can do is to get out of the way and let whatever is in the recording come through.

Heck, I have a cheap Chi-Fi Fosi Class-D amp used for desktop computer sound and it is obvious they are doing some weird phase thing to make the soundstage artificially wide. But hey, for that use scenario, I’m not complaining, just noting it.

We all seek "truth" or "beauty". Sometimes they are not the same. Choose which makes you happy. 

@shadorne 

HUH?

Passband ripple is the elephant in the room that nobody talks about.

I’m pretty sure my BlackIce Tube DAC sounds great to me is because the measurements of the 1950-ish NOS tubes suck.

 

Measure twice, cut once. That's my deep thought about it

@baylinor nice one (Jack Handey would approve)

@shadorne, your item 3 has always struck me as a highly significant issue. When the Redbook standard was announced back in the late '70s, I heard about it from a physics teacher friend, and my reaction was that a 44.1 kHz sampling frequency was too low, probably be a factor of 2 or more. We agreed that virtually any filter would introduce phase shifts at frequencies well below 20 kHz, which would likely be audible as degraded transients. And I heard that from the dawn of the CD age.

There's another, related issue. The engineers who invented the A to D and D to A processes relied on the Shannon-Nyquist theorem to take the position that the DAC output would be indistinguishable from the ADC input. The problem is that the theorem, as it is usually stated, states that the sample rate must be at least twice the bandwidth of the signal. However, this is an incomplete statement of the theorem--in reality, properly stated it says the sample rate applied to a continuous function must be at least twice the bandwidth of the signal. And music is not continuous, so you can't rely on Shannon-Nyquist; there will always be some variance, sometimes audible, sometimes not. But in practice, both the application of a higher filter cutoff and a the concomitant higher sampling frequency both reduce the variance between input and output.

@corente plus1

You said before I did.😁

Remember the discussions about amplifiers and THD: the ones that sound better usually were the ones having not so good THD

@shadorne Thank you for your post, very interesting! Theoretically speaking, would oversampling all digital signals to DSD256 and feeding that to a non-oversampling, DSD-only DAC (or at least a DAC with a discrete DSD stage) provide a workaround to the issues you are describing?

@OP et alia: Passband ripple is an often-cited property of power amplifiers. How do the (calculated) amounts of passband ripple in considered DACs compare to those of the amplifier(s) in your system?

Measure twice, cut once. That's my deep thought about it.

If I remember Robert Harley was warned.by Moffstt that yagdrassil has bad measurements. Turn out to be a very good one, got stellar review from Harley.I don’t understand measurements I like the sounds of my 3 systems.

The subject matter is waaay beyond my comprehension but to all those saying “just go with what sounds good to you” I have these thoughts. As a longtime admirer of David Hafler, Edgar Villchur, and Peter Pritchard, engineers who made great sounding products that were affordable by the many music lovers, not only the few with high incomes, their goal was to discover what was musically relevant in circuit or material science, and apply clever methods to make products that excelled without needless cost.  Hence the PAS-3, Mark lll, AR XA, 3a, and the XLM, Sonus Blue Label, etc.  If “passband ripple” is reliably identified as a key source of audible signal degradation, and an economical way to rid it from the D/A process can be found, then we all benefit, at all price levels…at the Topping ChiFi and the dCS budgets alike.  That would be progress.

Interesting, way beyond my comprehension except for the understanding DAC’s aren’t able to convert, output signals that measure perfectly.  
 

I’ve had traditional DAC’s, R2R, FPAG, DAC’s with a Tube stage.  Currently have a PS Audio that is a FPAG, doesn’t have any of the off the shelf chips and converts all signals to DSD and outputs them in DSD form.  Believe the measurements aren’t that great on the DAC.  Designer talks about designing the DAC around being able to tune the DAC to shape the frequencies in a way that sounds best to our ears.  Interesting that following that format produced a DAC that for some measures poorly but to many sounds great.  
 

OP, how does your theory, principles you reference apply to FPAG DAC’s? 

I agree.  DACs (and amps) sound different.  Ears and brains perceive sound differently than microphones and machines.  Measurements do not mimic human hearing well.  Long listening periods help us become familiar with subtle changes and overcome human (and possibly other physical) variables. 

Remember the discussions about amplifiers and THD: the ones that sound better usually were the ones having not so good THD

Intense, but might explain why I like my R2R in NOS the best, and the DS in my second rig, that ASR measured @ -123 dB SINAD with filter #4, slow roll, linear phase. The least pre and post ringing mirrored impulse curve. 

Long post.  You appear find satisfaction and frustration in contemplating, philosophizing, and pontificating on audio engineering problems and solutions way beyond my level of comprehension.  My recommendation is to relax using whatever meditative or chemical means you wish (good bourbon or cognac for me) and leave these issues to the audio design engineers. Get back to the basics of this hobby, the enjoyment of music and the pursuit of the most natural sounding system within your budget.  Listen to live music, both acoustic and amplified.  I always calibrate at Carnegie, the MET, the NYC Ballet, or jazz and rock bars in the Village.  Establish listening principles to develop your perception of live music.  Develop good auditioning methods (much is published on this). Forget about measurements (except for determining system compatibility such as impedance matching) and design. Find a good dealer that is knowledgeable to assist in system compatibility and set up.  Use your ear-brain connection to determine what equipment meets your perception of good SQ.  Use reviews only for preliminary research to find equipment that may meet your goals and perceptions.  Build the best system within you budget that sounds right to you.  Chill and enjoy the music.  
 

As Nonoise stated , some of the best DACs reside in an amplifier.   My Cyrus i7XR has a great sounding DAC built in.   I love that amp, very easy to listen to for a long time. 

Late last night was one of those special times that all audiophiles know what I'm talking about. Whether it was low usage on the power grid, atmospheric pressure or special alignment of the stars, the music was just flowing and every audio aspect was nirvanic perfection. Record after record sounded better than the one before and knowing that I would soon have to call it a night and that it would be quite awhile before these conditions would come again, I went deep into my archives and pulled out some of my favorite measurements and immersed myself in numeric bliss.