Which DACs bypass digital filtering?

[barrows]

On 6/24/2018 at 11:56 PM, buonassi said:

probably not a helpful post, but I thought I'd share nonetheless so take it FWIW:

 

Regarding all 44/16 source material, I've found that some magic is lost when I upsample (iZotope via Audirvana) on my Metrum Amethyst (NOS DAC).  No matter what the settings, it just doesn't sound as good to me.  Usually, I prefer the interpolation done by upsampling to that of the OS DACs I have, and can settle in on a configuration that maximizes the sound to my preference.  This is the first time that I couldn't improve upon the sound (to my ears) by upsampling.  So what gives?

 

I think there's something unique to DACs that are designed to not oversample.  I've seen it described by a prominent DAC reviewer on the SBAF site (@hands) as "spooky good stuff".  I have to agree that at least this NOS DAC can present an eery realism to the sound - very organic/relaxed/natural, yet retains good imaging queues and holographic staging.  So yeah, somewhat "spooky" in the regard that it sounds unexpectedly real sometimes.  Especially in the reverb/impulse response characteristics of the room the music was recorded in.

 

I think there's something inherent in the interpolation process that kind of "robs" the music of musicality.  I posit that it is because the original discrete samples with their "more precise" amplitudes are ultimately discarded in favor of an interpolated waveform that is superior (in terms of distortion).  So through an OS DAC, you're hearing a wave form that's been transcribed two times (once via ADC decimation and another through DAC reconstruction).  My theory is that by omitting the latter, you get one step closer to the original wave form.  Yes, at the expense of IMD nasties, etc. 

 

This is my opinion only and I can't back it with any science.  Just the ramblings of one dude in the midst of new toy syndrome.  

2 things to consider:

 

1.  Do you connect to your DAC directly to a computer which is doing the oversampling?  If so, then the extra computer resources needed to oversample will cause more noise to get from the computer to the DAC, and hence, you may prefer the less noise approach of NOS, despite the technical benefits of oversampling.  The solution for this is Ethernet connection via an Ethernet Renderer (like Sonore, etc), as this allows the computer doing the oversampling to be far away from the audio system, where it can do no harm, regardless of how hard it is working (processing power required for oversampling). 

 

2.  If you oversample using integer rates (2x, 4x, 8x, 16x, etc) the "original discrete samples" are not discarded at all, and their ""more precise"" amplitudes are not lost.  Although, with computer based oversampling with much more precise algorithms than what is typically done in DAC hardware based oversampling I do not perceive even non-integer oversampling as a problem.  In any case, with integer oversampling the idea that the original samples are "discarded" is entirely false, they are intact at exactly the same value as before.

[barrows]

I can also confirm perfect DSD 256 performance with Amanero USB interfaces running on 1099c via Sonore Rendu renderers (linux based).  This is with Native DSD (no DoP).

[barrows]

12 minutes ago, bilboda said:

apparently not

This from Sonnet's description sounds like complete BS to me:

 

" The result of this process is an extremely high linearity, right down to -140 dB, which gives our products a realistic 24 bit dynamic range.  It is musical, honors the Non-oversampling principle and simultaneously incorporating the technological progress that has been made over the years."

 

Where are the actual measrement results for this DAC?

[barrows]

10 minutes ago, bilboda said:

lol, I dunno, not sure I value your doubt over the designers pedigree...and I am not sure that measurements at this extremely inaudible level will tell you how it sounds either...asr aficionado?

 

Haha, not at all.  But I am skeptical about any company which prints what is clearly inaccurate claims about the linearity of their product.  No DAC in the world actually achieves 24 bit resolution at its output.

As to how it sounds subjectively, yes, for sure, it could sound wonderful, it is just that false claims like that make me very uncomfortable. 

[barrows]

10 minutes ago, bilboda said:

We could go way off topic, dark matter?, fancy cables, optical network? No measurements there either, I'm sure, and who are u to say what's impossible? Suffice it to say that his dac meets  op's requirements and I think it's terrific.

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.

[barrows]

4 minutes ago, bobflood said:

The useful dynamic range is probably 20 bits or less. I think the best ever measured was 21 bits but I can't remember which DAC that was. It is very hard to get much better than 20 bits just because of the inherent noise of moving electrons around.

Yes, exactly.  Which is why it makes me cringe when this company claims 24 bit resolution.  

[barrows]

2 hours ago, bilboda said:

Well, I view your statement as unsupported, likely submitted for controversy only and thus false,until proven otherwise.🤨

The burden of proof would fall to the company making the claim, especially in a case where the claim is it for far better resolution than is understood by to be possible with currently available parts.

[barrows]

36 minutes ago, bobflood said:

. I don't think most appreciate that any sound that our brains can process (including music) must first emerge from that background din (noise)

Actually, I do not think this is correct at all?  Humans can hear sounds below the noise floor (the in room noise floor), as the brain can sort these sounds from noise, right?  Think about LP playback, one can certainly hear sounds at levels well below that of the surface noise: this process is usually helped if the sound in question is well known by the brain already (not necessarily the same sound, but a "like" sound, such as the sound of a bell).

Also consider dither, I am sure @mansr has a good explanation of this.

[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).

[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.

[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.

[barrows]

1 hour ago, fas42 said:

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

 

[barrows]

Topping D-90 offers NOS for DSD only in "DAC" mode.  This has been confirmed directly by Topping.

[barrows]

1 hour ago, asdf1000 said:

What Topping needs to confirm is “DSD Direct” mode, as per AKM DAC chip datasheet..

 

Yes it does, I confirmed directly with Topping.

1 hour ago, asdf1000 said:

DAC mode does not mean by-pass DSP.

 

Yes it does with DSD input, this has been confirmed directly by Topping.

[barrows]

21 minutes ago, asdf1000 said:

 

Can you copy and paste exactly what you asked them?

 

And copy and paste exactly their reply?

 

I have long since trashed the e-mail thread, and do not have the patience to go searching through my trash bin to suit your needs.  As you seem to doubt my verity, i suggest that you contact them yourself for verification.  Topping was very responsive to my inquiry and answered my questions in less than 24 hours.

[barrows]

I feel the Bricasti M3 and M21 DACs should be on the list, for their discrete DSD conversion side only (PCM is always OS on these DACs).

[barrows]

24 minutes ago, semente said:

 

I looked in the manuals and it seems like they can be run digital filterless when fed DSD (NDSD DSD mode).

Can you confirm that?

Unfortunately its input is limited to DSD128 but that doesn't mean it won't perform brilliantly.

Yes, DSD is processed directly by the discrete DSD conversion stage on the M21 and M3 (if one chooses that stage for conversion).  These DACs also currently accept DSD 256 input with the most recent digital input board (Mdx) and firmware.  I run DSD 256 into my M3 all the time via USB from my Sonore Signature Rendu SEoptical acting as NAA for HQPlayer.  Love the sound from the M3 with ASDM7EC modulator going to DSD 256!

[barrows]

@semente, I would add: "only for the discrete DSD converter" as a caveat to the Bricasti M21/M3 info above.  These DACs do not allow for NOS with PCM content

[barrows]

6 hours ago, opus101 said:

 

I suspect 'NOS' no longer has its original meaning of No OverSampling. Any DAC using an S-D modulator cannot be NOS. As far as I'm aware all AKM, Wolfson, Cirrus and ESS DACs are S-D architecture and therefore cannot be NOS in its true sense. (I'm speaking only in the context of PCM here)

This is incorrect.  Just because a DAC chip has a SD modulator on board, does not it mean the signal passes through that SD modulator in every configuration: AKM, with their "direct DSD" mode are an example of this (this can be easily understood by looking at AKM data sheets).  My Bricasti M3 has two SD DAC chips on board in a dual mono configuration, but the DAC also has a discrete DSD converter which completely bypasses the DAC chips, so when one sets the DAC to use the discrete DSD converter there is no SD modulator in the signal path.

[barrows]

T+A DAC 200 with DSD at least.  this is a separate product from HA 200.

[barrows]

@hopkins:  To my mind, the purpose of this list is to have a verified source of DACs which will natively convert an incoming sample rate.  The reason an audiophile might want to have this information is for those audiophiles who might want to take control of the aspect of digital filtering/OS through software (like HQPlayer) rather than being "stuck" with what the DAC manufacturer, or even just the DAC chip manufacturer provides.  Many have had the experience that HQPlayers much more sophisticated oversampling algorithms are superior to that offered onboard in most (perhaps all) DACs.

Certainly my experience, listening, suggests that HQPlayer's approach to oversampling in demonstrably superior to most DACs of my acquaintance.  This is not really that surprising, considering the processing horsepower required by HQP oversampling.  The I9-9900K processor I use in my server for this is many magnitudes more power fun than any onboard processor in a DAC, and HQPlayer takes advantage of this additional computational power to allow for markedly better sound quality in playback.

So far the only DAC with its own onboard oversampling which has come close is the Mola Mola Tambaqui, which uses multiple AD SHARC chips to apply its oversampling.