Which DACs bypass digital filtering?

[Butterfly]

On 2/14/2020 at 2:40 PM, barrows said:

There is no DAC in the world which has 24 linearity at its outputs, period.  Find the measurements which say different and I will eat my socks!

Like I said, that does not mean this DAC does not sound wonderful, but it does mean that the company is misrepresenting its actual performance on their website, which makes me wonder about them.

Not wishing to get into a tech debate (not qualified) I could not resist the opportunity to (possibly) see someone eat their socks. That is if Martin Colloms (Ex HiFiNews Tech Ed for a while and contributor for 30 odd years – now HiFiCritic) qualifies as independent test source.

 

To declare commercial interest I have been dealing with (let’s say) Cees’ products for 7 years, so have a little background to all. Not much independent testing to confirm/question any claims. But Pavane is one reviewed by Chris Bryant 2015.

 

Buried in Critic’s Forums prior to review MC checked out Pavane on his test bench re what were called ‘outrageous claims’ re linearity/24 bit res & noise floor etc. He confirmed they were true. Ditto in Chris Bryant’s review :

 

https://www.chameleonracks.com/images/pavane/Metrum-Pavane-Hificritic.pdf

 

Further reading of 6 Moons Adagio covers next level Dac improvement but on specs only slight improvement re noise etc. – as I understand it getting to limitations of test equipment. There it covers limitations of resistors (about 14 bits) and their thermal (Johnson noise) self noise – making “32-bit resolution commercial hooey”.

Commentators do claim to hear improvements usually described as blacker background.

 

If Stereophile/HiFi News ever get their hands on a unit, maybe further test results might confirm claims.

[barrows]

@Butterfly, I see no measurements at your link?  Furthermore, even Metrum's own spec (no measurements shown) claims -145 dB but at 2 V output, for a device with 4 V output.  Unless those specs were based on the RCA outputs which I would highly doubt.  Where are the actual measurements?

 

The thing with measurements is one has to do them very carefully, it is easy for manufacturers, reviewers as well, to make the measurements look much better than they really are.

 

Not any criticism of the DAC itself, it could;ld be very, very nice sounding.  I will read the article from 6moons, looks interesting, thanks.

 

Oh yeah, one other thing to be careful about when looking at noise floor measurements, you have to have a signal present to accurately measure the floor on most DACs, not just powered up with no signal.  Some nefarious measurements use the no signal approach (a condition where a lot of DACs actually mute their outputs).

[Butterfly]

6 hours ago, barrows said:

@Butterfly,

Oh yeah, one other thing to be careful about when looking at noise floor measurements, you have to have a signal present to accurately measure the floor on most DACs, not just powered up with no signal. 

Hi – there’s no detail re testing regimes only statements from Colloms and Bryant that claims made were confirmed – “The first thing I did on receiving the Pavane was to check its linearity and found the most accurate R2R multi-bit converter that I’ve ever measured. The error barely moves off the reference line even at -120dB (a full 24-bits)”.

 

I'm no technophobe as mentioned so can only relay comments/statements made re test results.

 

Some reviews from Critic do contain Measurements or sometimes brief comments such as above. For Hex in a brief Lab report – “Beating the Octave the excellent resolution measured about 20 bits, a -100dB tone was resolved to a fine 0.7dB accuracy confirming the transparent and deep images heard”.

 

So much comes down to how much credence you give them (and their testing regimes). So I merely wanted to point out that some independent confirmation does exist for claims made. No matter who does what though, results can always be questioned. So in some respects a pointless exercise.

 

If the claim that 4 bits contain a lot of noise and therefore ditched, using 12 bits left (of most/least significant bits summed) is a logical way to arrive at 24 bit total. If it’s believed the 4 bits ditched contain valuable info/not mostly noise etc., then it won’t stack up.

 

In the end though hearing is believing and there’s no doubt that reduced noise floors reveal micro detailing, layering and separation etc. They are not the only products to achieve this by different means.

 

Available from Critic a top tier dCS (2009) system was put through its paces with a fairly comprehensive Lab Report showing … ”With substantial data averaging, the noise floor settles at a deep -140dB, despite the presence of that full level fundamental”. Again only if credence can be attributed to Critic’s Lab regime.

 

Over the last 20 years though, can anybody dispute the fact that Dac resolution has improved - is still improving - even if Lab results are questioned/incomplete/missing ?

[Miska]

8 hours ago, barrows said:

Some nefarious measurements use the no signal approach (a condition where a lot of DACs actually mute their outputs).

 

And it happens also if you play for example just zeros (digital silence). For example ESS Sabre engages output mute when there are enough subsequent samples of same value, regardless of the actual value. Which is reasonable because it also stops accidental DC outputs.

 

If you want to measure noise floor with PCM input, you need to play dithered silence. And check that it doesn't get accidentally truncated somewhere on the way so that the dither would get lost.

 

[barrows]

23 minutes ago, Miska said:

 

And it happens also if you play for example just zeros (digital silence). For example ESS Sabre engages output mute when there are enough subsequent samples of same value, regardless of the actual value. Which is reasonable because it also stops accidental DC outputs.

 

If you want to measure noise floor with PCM input, you need to play dithered silence. And check that it doesn't get accidentally truncated somewhere on the way so that the dither would get lost.

 

Ah Jussi, so true.  The "automute" function of the ESS 9038 is a bit wonky as well.  I was just troubleshooting a weird thing I discovered here with my ESS 9038 (DIY) DAC: I was playing a -90 dBFS tone, to listen to the noise floor around it, and the tone was cutting out.  Turns out I had automute turned on in the ESS 9038 settings, and the DAC was intermittently "deciding" to mute the output, stuttering.  I disabled automute and all was good.   

What do you think about J. Atkinson's approach of playing a 1 kHz -90 dBFS tone at 16 and then 24 bits to determine actual bit depth resolution, that seems legit to me.

[Bob Stern]

On 2/14/2020 at 8:25 AM, Jud said:

very well insulated house (we lost power and thus heat for 24 hours in a winter windstorm with 50-60mph winds and temperatures below freezing; indoor temperatures dropped from their setpoints of 69-70 degrees F to 67-68 degrees).

 

You're saying the indoor temperature dropped only 2°F after 24 hours with no heat?  That's unbelievable!  I'd be fascinated to read a description of its insulation techniques!

 

btw, I have a friend in Nevada who also custom-built a super-insulated home, but I suspect yours leaves his in the dust, so to speak.

[Jud]

1 minute ago, Bob Stern said:

 

You're saying the indoor temperature dropped only 2°F after 24 hours with no heat?  That's unbelievable!  I'd be fascinated to read a description of its insulation techniques!

 

btw, I have a friend in Nevada who also custom-built a super-insulated home, but I suspect yours leaves his in the dust, so to speak.

 

Hi Bob! (I owe you an email for your very kind note - life is busy, sigh.)

 

The insulation is nice and very well installed, spray foam with a lot of blown-in behind it. But likely what kept the house warm is the fact that we have radiant floor heat with a huge amount of thermal mass, and that thermal mass is backed with 2 inches of rigid foam and a vapor barrier. It gives up its heat very gradually.

[semente]

On 2/21/2020 at 3:55 AM, Butterfly said:

That is if Martin Colloms (Ex HiFiNews Tech Ed for a while and contributor for 30 odd years – now HiFiCritic) qualifies as independent test source.

 

His lab testing may be fine but his listening assessment is fatally biased by taste.

This review of the £96k Audio Note DAC5 is a good example:

 

 

Lab Report

 

As expected from a non-oversampled design, there’s no output filtering save for the natural high frequency roll-offs in the circuits and transformers, the latter nevertheless represents a certain benefit to subsequent components in the chain. Channel balance was excellent, holding within 0.08dB 20Hz to 20kHz, while the output, unaffected by loading, is sourced from a very low impedance of about 27ohms. Channel separation was fine: 39dB at 20 Hz, 76dB midband and 65.5dB by 20kHz. Distortion was unexceptional at 0.85% at full level, but rapidly improved at normal modulation levels, eg 0.08% at -10dB and 0.05% at -20dB.

 

With its single-ended triode output stage, the harmonic content was dominated by second harmonic, a little third harmonic and nothing else. With no digital filtering there’s no pre-echo, so accurate 1kHz square waves may be reproduced. The frequency response was flat in the primary central range +0/- 0.15 dB 30Hz to 5kHz; 20Hz was at - 0.25dB and 17kHz at -1.5dB – barely audible, since 10kHz was only -0.63 dB. At full level 20kHz the aliasing beat tones with the 44.1kHz sample rate were unattenuated (see graph), but as usual reduce quite quickly with reducing modulation level. At full level there is a 4kHz intermodulation product at -25dB, but at -20dB modulation that had fallen to -48dB and is likely to be inaudible. The valve output stage has a finite noise floor, so the ‘jitter gram’ for a pure 1kHz tone, while clean of itself, may only be resolved down to -115dB; no unwanted sidebands could be seen thus far.

 

Signal-to-noise ratios were more than satisfactory, albeit with some imbalance between channels, averaging 91dBA, 84dB CCIR (1kHz) and 83dB unweighted. High frequency intermodulation at full level was a poor -26dB, and very loud cymbal crashes might show slightly altered timbre, though I did not hear such, and this aspect improved rapidly at normal signal levels, eg to 0.35% at -10dB modulation. Linearity was good down to -75dB but some ‘expansion’ error was seen below that, with - 80dB reading -86.5dB. This ‘negative’ error continued to -100dB where the output finally dropped out and may be a function of the I/V method used. (this DAC chip has the potential for trimming the low level matching, but that is not done here.) One might speculate that this characteristic would attenuate and reduce low level detail, but as the listening showed this was far from the case.

 

As with the Metrum Octave results, it seems that much lab data for -70dB and below has little relevance to sound quality. Incidentally, I measured the RFI at the output up to 1.5GHz, and found that it fell to low levels above a few hundred kHz.

 

 

Conclusions
 

When this review was first proposed, my first reaction was hysterical laughter. UK houses cost a similar amount on average. How would a CD player like this sound in a complete system made up using comparably priced components? And what would such a system cost? A million pounds? I have tried to answer these questions by extrapolation and with an extensive set of listening sessions with a variety of audio components and listeners.

But my basic task is to observe and report, and my finding is that this CD ‘transport’/DAC is outright marvellous, newly defining the inherent quality of Red Book digital audio, which can be so much better than we had imagined was possible. It poses the question: “Why has it taken a specialist valve audio manufacturer to achieve this after 30 years of highly informed digital engineering by the audio industry as a whole?” While there is some hi-res audio material now, it is also clear that CD has been potentially ‘hi-res’ all along, but we never truly experienced it (and you never will with such low-fi CD playback equipment) Instead we’ve suffered the distortions and masking of digital filters, op-amps, and accompanying digital noise, for all these years. (WTH)

While very few examples will be made and sold at such extravagant prices, we owe a debt to the Audio Note UK team for showing just what is possible. Here is a CD player which really can give high end analogue a run for its money.

[Miska]

18 hours ago, barrows said:

What do you think about J. Atkinson's approach of playing a 1 kHz -90 dBFS tone at 16 and then 24 bits to determine actual bit depth resolution, that seems legit to me.

 

I think that should be good way. Unfortunately just about nobody seems to do similar checks in frequency domain. Many DACs hang around 10 - 18 bits resolution in that respect.

 

[fas42]

2 hours ago, semente said:

 

His lab testing may be fine but his listening assessment is fatally biased by taste.

This review of the £96k Audio Note DAC5 is a good example:

 

 

Biased by taste?? ... Meaning, if his "taste" is to get closer, subjectively, to what's on the recording, that's a fatal error?

 

2 hours ago, semente said:

But my basic task is to observe and report, and my finding is that this CD ‘transport’/DAC is outright marvellous, newly defining the inherent quality of Red Book digital audio, which can be so much better than we had imagined was possible. It poses the question: “Why has it taken a specialist valve audio manufacturer to achieve this after 30 years of highly informed digital engineering by the audio industry as a whole?” While there is some hi-res audio material now, it is also clear that CD has been potentially ‘hi-res’ all along, but we never truly experienced it (and you never will with such low-fi CD playback equipment) Instead we’ve suffered the distortions and masking of digital filters, op-amps, and accompanying digital noise, for all these years. (WTH)

 

Amen. ... He could have "taken the words right out of my mouth" ... 🙂.

[Butterfly]

2 hours ago, semente said:

 

His lab testing may be fine but his listening assessment is fatally biased by taste.

 

What makes you say that in particular and aren’t all subjective reviews biased by personal taste in some way ? Showing age I’ve been reading his reviews for just over 40 years and owned quite a few products confirming his assessments. Still own a pair of MF-MC2 Speakers as it happens.

 

At £96K I would expect something to be superb. His opinions coincide with pretty much everyone else’s re top components. The only bit of gear I can think of where he’s at odds with most is Devialet.

 

His Rating System is controversial though and you are not alone in questioning his opinions but he does drag in others during listening sessions to appraise. As many may know he’s also a long time colleague of JA from Stereophile and published many an article going back for them. So JA obviously has some respect.

 

Now could just be a patriotic trait (not alone in this) but there does seem to a heavy Naim bias from his team. With takeovers/connections etc. interesting to note that Brand has suddenly sprung up large at Audiostream recently.

 

Back to measurements : “As with the Metrum Octave results, it seems that much lab data for -70dB and below has little relevance to sound quality”.

2 hours ago, semente said:

 

 

[semente]

15 minutes ago, fas42 said:

Biased by taste?? ... Meaning, if his "taste" is to get closer, subjectively, to what's on the recording, that's a fatal error?

 

"to get closer, subjectively, to what's on the recording"...

 

Did you think before you typed this?

[semente]

15 minutes ago, Butterfly said:

 

What makes you say that in particular and aren’t all subjective reviews biased by personal taste in some way ?

 

 

They are.

Which is why they are worthless...

[semente]

16 minutes ago, Butterfly said:

At £96K I would expect something to be superb.

 

Yet it's rubbish. Your expectations have been defrauded.

[semente]

17 minutes ago, Butterfly said:

Back to measurements : “As with the Metrum Octave results, it seems that much lab data for -70dB and below has little relevance to sound quality”.

 

Cow. Dung.

[fas42]

7 minutes ago, semente said:

 

"to get closer, subjectively, to what's on the recording"...

 

Did you think before you typed this?

 

Indeed I did. What's your problem with what I said?

[semente]

Just now, fas42 said:

 

Indeed I did. What's your problem with what I said?

 

You have no idea of how the recording should sound like. You can't "get closer, subjectively, to what's on the recording".

[fas42]

1 minute ago, semente said:

 

You have no idea of how the recording should sound like. You can't "get closer, subjectively, to what's on the recording".

 

The recording should sound like a group of musicians plying their trade, as if positioned in front of a single or multiple microphones - assuming that's what the recording is of - rather than a watered down, pale imitation of that. Since that is how the recording was made, it's fairly reasonable to assume that hearing that should be possible.

[semente]

7 minutes ago, fas42 said:

 

The recording should sound like a group of musicians plying their trade, as if positioned in front of a single or multiple microphones - assuming that's what the recording is of - rather than a watered down, pale imitation of that. Since that is how the recording was made, it's fairly reasonable to assume that hearing that should be possible.

 

We live in different planets.

[barrows]

56 minutes ago, semente said:

You have no idea of how the recording should sound like. You can't "get closer, subjectively, to what's on the recording".

 

While I certainly agree with this for the most, it is not true for every aspect of playback performance, specifically detail retrieval.  If a component change enables one to actually hear details which were previously not heard, then one can be confident that the new component has increased resolution.  As to things like tonality, and such though, I agree.

[fas42]

1 hour ago, semente said:

 

We live in different planets.

 

Indeed we must ... if you have not heard what a competent playback is capable of presenting, then me describing the sensation will be of little value ...

 

38 minutes ago, barrows said:

 

While I certainly agree with this for the most, it is not true for every aspect of playback performance, specifically detail retrieval.  If a component change enables one to actually hear details which were previously not heard, then one can be confident that the new component has increased resolution.  As to things like tonality, and such though, I agree.

 

Typically, increased resolution, bringing out more detail, also contributes a good dollop of unpleasant distortion, courtesy of playback chain weaknesses. It takes attention to detail to eliminate the latter, while still retaining the detail ... this is what highly tuned rigs that people hear, which create 'magic' in the room are achieving ... it's rare that this happens, because so few individuals go to the lengths necessary to make this occur.

 

A simple principle ... the more the acoustic world changes, in front of, and around you, when you change the recording you listen to - the closer you are to the recording ... have a good think about it ....

[Butterfly]

Err lively discussion … Out of interest (as a newbie) Semente, are there any Audio-zines (let’s face it how most of us find out things) that you ascribe to ? All reviews are opinions – nothing more - even if several people involved.

 

As alternative here’s an oldie if JA is/can be accepted/taken seriously/believed :

 

https://www.stereophile.com/asweseeit/705awsi/

 

Colloms has listened to many original in-Studio Master Tape/recordings etc. As close as you can get to a recording (as a ref) – Pink Floyd’s Dark Side of the Moon for one if not mistaken.

 

To quash the Audio Note as rubbish a bit harsh surely ? Unless you’ve heard one of course. Personally I’ve only ever listened to Systems up to US$300K – were they worth it imo – no. Could I happily live with one – yes.

 

So expectations at £96K are zero - I’d have to sell the house to fund a commensurate system.

 

As for Cow Dung (to lighten the load) … ‘highly rated product’ in its field.

[Jud]

1 hour ago, barrows said:

 

While I certainly agree with this for the most, it is not true for every aspect of playback performance, specifically detail retrieval.  If a component change enables one to actually hear details which were previously not heard, then one can be confident that the new component has increased resolution.  As to things like tonality, and such though, I agree.

 

That's one reason I am more impressed by detail retrieval in the lower registers. While I believe I can tell the difference between detail and elevated high end frequency response in the mid and upper registers, it's a little easier to be confident the detail is real when it comes from a double bass, for example.

[semente]

57 minutes ago, fas42 said:

Indeed we must ... if you have not heard what a competent playback is capable of presenting, then me describing the sensation will be of little value ...

 

I'm jealous. 🙄

[barrows]

1 hour ago, fas42 said:

Typically, increased resolution, bringing out more detail, also contributes a good dollop of unpleasant distortion