Is DSD noise audible?

[mcdonalk]

Two of my favorite recordings are DSD256 downloads. Lately, I have noticed some audible anomalies. In one recording (chamber orchestra), violins in the left channel sometimes have a metallic character. It is not sublime; it sounds like metallic scraping. In another recording (again, a chamber orchestra), I hear occasional faint "crumbling" sound coming from the center of the stereo image.
At first, I thought that this was my speakers distorting. However, when I plug headphones into my preamp, I hear it via the headphones. When I plug the same headphones into the DAC, I hear it also, but it seems less pronounced. It seems to become more noticeable the farther it propagates down the chain, but that's hard to say for sure. I have tried two different DAC's, although they are both from the same manufacturer and use the same ESS chip.
The last major change I made to my system was installing a Benchmark HPA4 preamp about six months ago. The only recent change I have made was a few weeks ago when I replaced the high end 30-year-old Audioquest 18" balanced silver interconnects between the DAC to the preamp, and from the preamp to the power amp. The new cables are 3' starquad cables from Benchmark.
I have not heard this on any other recordings (whether DSD or PCM), which does not mean that it is not there. But, I will be paying closer attention for now on. It seems unlikely that the recording is the source, because they are relatively recent recordings, and it is hard to imagine that a consciously-engineered production would result in this defect being deployed.
Is it possible that I am hearing DSD noise? If so, I thought that DSD noise was out of the range of audibility.

 

[kemmler3D]

Is it possible that I am hearing DSD noise? If so, I thought that DSD noise was out of the range of audibility.

DSD noise is well outside the audible range, so that's not it, especially since you tried different DACs and amps.

I would see if you can find a different version of the same recording. And/or share a clip of the distorted section so we can take a crack at diagnosis.

 

[mhardy6647]

I have not heard this on any other recordings (whether DSD or PCM), which does not mean that it is not there. But, I will be paying closer attention for now on. It seems unlikely that the recording is the source, because they are relatively recent recordings, and it is hard to imagine that a consciously-engineered production would result in this defect being deployed.

On the strength of these comments in toto, it does indeed sound like an issue of the specific recording/master (i.e., version).
Tried any variants thereof?

 

[MaxwellsEq]

If you are playing a DSD file using a path that never uses PCM, the ultrasonic DSD noise should be suppressed by the last stage. If there is a conversion to PCM anywhere in the path, the ultrasonic noise may not be suppressed and can be quite high level.

Although not audible itself, this high frequency noise can upset some amplifiers (e.g. some Class D amplifiers such as Ncore) and some tweeters.

I have a couple of "hi-rez" commercial PCM releases which are clearly taken from a DSD master and the ultrasonic noise is at similar levels to the music's midrange and much louder than the music's high frequency content.

 

[Sokel]

Can't see how:

 

[dartinbout]

Two of my favorite recordings are DSD256 downloads. Lately, I have noticed some audible anomalies. In one recording (chamber orchestra), violins in the left channel sometimes have a metallic character. It is not sublime; it sounds like metallic scraping. In another recording (again, a chamber orchestra), I hear occasional faint "crumbling" sound coming from the center of the stereo image.
At first, I thought that this was my speakers distorting. However, when I plug headphones into my preamp, I hear it via the headphones. When I plug the same headphones into the DAC, I hear it also, but it seems less pronounced. It seems to become more noticeable the farther it propagates down the chain, but that's hard to say for sure. I have tried two different DAC's, although they are both from the same manufacturer and use the same ESS chip.
The last major change I made to my system was installing a Benchmark HPA4 preamp about six months ago. The only recent change I have made was a few weeks ago when I replaced the high end 30-year-old Audioquest 18" balanced silver interconnects between the DAC to the preamp, and from the preamp to the power amp. The new cables are 3' starquad cables from Benchmark.
I have not heard this on any other recordings (whether DSD or PCM), which does not mean that it is not there. But, I will be paying closer attention for now on. It seems unlikely that the recording is the source, because they are relatively recent recordings, and it is hard to imagine that a consciously-engineered production would result in this defect being deployed.
Is it possible that I am hearing DSD noise? If so, I thought that DSD noise was out of the range of audibility.

Purely out of curiosity, what are the recordings in question? As a DSD fan, for reasons outside of the veil of objective analysis (S's and G's really). I've never heard any degeneration of replay in any of my signal chains with DSD64, 128, 256 or 512.

 

[Pluto]

Two of my favorite recordings are DSD256 downloads. Lately, I have noticed some audible anomalies. In one recording (chamber orchestra), violins in the left channel sometimes have a metallic character. In another recording (again, a chamber orchestra), I hear occasional faint "crumbling" sound coming from the center of the stereo image

My approach would be to beg, borrow, steal or buy a PCM version of the same recording and see how you get on from there. I would guess that this recording used, in essence, an MS recording technique (as opposed to an AB technique) but I'm not going to explain my thinking on this point just yet ;-)

In the absence of other information I would happily stick $10 on the idea that you are hearing something on the recording, regardless of DSD or PCM.

 

[mcdonalk]

I don't think that it's the recordings. I suspect that I have a pretty good idea of the problem, and it's pretty weird, and I don't want to report on it until I'm more certain. I don't want to seem coy, but I'll have a better idea in about a week, when I'll report back my results. However, if you're still curious, the two recordings in question are DSD256 downloads (from nativedsd.com) of the following (I hope I can upload these two images):

(My taste in classical music includes primarily 20th century works, and Haydn.)

 

[tengiz]

If you are playing a DSD file using a path that never uses PCM, the ultrasonic DSD noise should be suppressed by the last stage. If there is a conversion to PCM anywhere in the path, the ultrasonic noise may not be suppressed and can be quite high level.

Although not audible itself, this high frequency noise can upset some amplifiers (e.g. some Class D amplifiers such as Ncore) and some tweeters.

I recall a similar known issue with early direct DSD DACs - the simple ones that used a simple passive low-pass filter fed almost directly by the DSD stream. While the noise was ultrasonic, residual energy could still push downstream analog stages into non-linear behavior, creating intermodulation that folded back into the audible range. Indeed, I remember that that’s one reason Putzeys designed early UcD amps with power bandwidth well beyond 20 kHz - to handle scenarios like this cleanly.

If the OP is using an old-school DSD DAC like this, that might be the cause - but it’s a long shot. I think one would likely notice similar issues on other recordings too.

 

[voodooless]

Can you be more specific? Track number, timestamp?

 

[Scytales]

If there is a conversion to PCM anywhere in the path, the ultrasonic noise may not be suppressed and can be quite high level.

It seems to me that, to the contrary, the decimation of DSD in PCM would suppress the ultrasonic content of DSD better than anything else, because any signal above the Nyquist frequency (which is relatively low in any PCM format compared to DSD) have to be sufficiently suppressed in order to avoid aliasing of said frequencies in the pass-band. Surely any designer of digital decimation filter know that, because it has been pretty much the textbook method the obtain PCM from delta-sigma modulated signals for decades !

An illustration of that kind of suppression of DSD ultrasonic content when DSD is decimated in low-sample rate PCM can be seen in NTTY's review of the Denon DCD-SA1 when the optional decimation filter is enable:

I recall a similar known issue with early direct DSD DACs - the simple ones that used a simple passive low-pass filter fed almost directly by the DSD stream. While the noise was ultrasonic, residual energy could still push downstream analog stages into non-linear behavior, creating intermodulation that folded back into the audible range. Indeed, I remember that that’s one reason Putzeys designed early UcD amps with power bandwidth well beyond 20 kHz - to handle scenarios like this cleanly.

If the OP is using an old-school DSD DAC like this, that might be the cause - but it’s a long shot. I think one would likely notice similar issues on other recordings too.

Which early 'direct DSD DACs' that used a simple passive low-pass filter do you referred to ?

The first SA-CD player produced by the Philips group, a player which was sold under the Marantz brandname, the SA-1, used a very short analogue low-pass FIR-filter with coefficients set at unity (several D to A converters, each delayed by a sample clock and whose outputs are summed in the continuous time domain) followed by an elaborate active low pass filter stage. The first SA-CD players from Sony, the SCD-1 and SCD-777ES, used a potent active GIC-type low-pass filter design. The first SA-CD player from Sharp, the DX-SX1, used the very first Burr Brown DSD DAC chip, the DSD1700, which incorporates an analogue low-pass FIR-filter longer than in the Marantz SA-1 and with coefficients set by on-chip resistors, not to mention a post-DAC active low-pass filter.

After this very first SA-CD players, almost every DAC chips capable of handling DSD that have been put on the market by Cirrus Logic, Burr Brown, Niigata Seimitsu, NPC, Wolfson, AKM, ESS or Analog Devices have applied digital low pass filtering on DSD streams.

 

[tengiz]

Which early 'direct DSD DACs' that used a simple passive low-pass filter do you referred to ?

I honestly don’t remember - it was almost 20 years ago when I interacted with Bruno Putzeys regarding his UcD-related research and his related paper. If I recall correctly, I learned from him about early DSD DACs and how their ultrasonic residuals were one of the motivations for designing power amps capable of gracefully handling that kind of output, especially from literal 1-bit DACs. I do vaguely remember the Sony DAC schematics you’re referring to, but unfortunately, not much beyond that.

 

[Scytales]

Thanks.

The point is, none of the early SA-CD players uses a simple passive low-pass filter following the DSD D/A converter. They already are much more elaborate devices than that. The only device I know of which has a passive, first order, RC filter following the D/A converter is the TASCAM DS-D98 pro DSD recorder monitor output, but the DAC used in this device is a Burr Brown DSD1700 that incorporates an analogue FIR low-pass filter, as already said. The TASCAM monitor output also has an optional GIC active low-pass filter that can substitute the RC filter.

 

[bogi]

All delta sigma DACs are using delta sigma modulator - a digital circuit which changes oversampled high rate PCM signal to low bit two or more level PDM or PWM signal, generally called as delta sigma modulated signal. This format conversion allows to use equally weighted conversion elements in the D/A stage itself. That's much cheaper than the R2R approach with binary weighted conversion elements (because it doesn't depend so much on precision of values of conversion element parts).

PCM to DSD software conversion is essentially doing the same thing like delta sigma DAC chip. DSD signal is one bit two level pulse density modulated or pulse width modulated signal. So asking if DSD noise is audible relates to the question if delta sigma DAC noise is audible.

There is no general answer for the opening post question since different DAC devices deal differently with incoming DSD signal. Some of them are able to work in direct DSD mode, where DSD signal enters directly D/A stage without rate or format conversion. Those can advance from running oversampling and delta sigma modulation in software like HQPlayer.

Other delta sigma DAC devices convert (demodulate and decimate) incoming DSD signal to low rate PCM first and then it goes through oversampling and delta sigma modulation like with PCM signal input. This added processing can add its own noise and distortion.

The newest delta sigma chips form ESS and AKM don't decimate the incoming DSD signal to low rate PCM, they do format conversion while keeping the input high rate and after optional volume control they remodulate the signal again. So complete DAC oversampling stage is skipped with DSD input, but DSD content is remodulated again.

So once again in short: There is no general answer for the opening post question since different DAC devices deal differently with incoming DSD signal. DSD noise is essentially the same thing like delta sigma modulator noise in delta sigma DAC devices (which may or may not support DSD input). DSD signal is one bit two level delta sigma modulated signal, so it is nothing exotic. DSD signal is a kind of delta sigma modulated signal all delta sigma DACs are creating and sending to their D/A stage.

 

[jayapple]

Both of my ESS DACs produce a crackling sound when switching from PCM music to DSD, but the AKM DCA doesn't have this issue. What could be the reason for this?
Even when switching DSD tracks, there's a crackling sound at first, but it stops after that.
By the way, the audio source is provided to the DAC through Roon.

 

[Scytales]

Other delta sigma DAC devices convert (demodulate and decimate) incoming DSD signal to low rate PCM first and then it goes through oversampling and delta sigma modulation like with PCM signal input. This added processing can add its own noise and distortion.

Let me add that, to my knowledge, there were only two DAC chips ever which were able to decimate DSD in low sample rate PCM, and this step was optional, not mandatory: the Wolfson WM8741 and WM8742.

Unless I have missed some or misunderstood their ways of operation, every other DAC chips having a DSD input interface ever built belongs to the category you described next, where DSD is actually digitally low pass filtered, which produced a multibit code at the DSD sample rate (or even higher in some cases), or the category you described first, where DSD drives directly the on-chip D/A stage.

That is not to say that a decimation of DSD in PCM is not done outside the DAC chip, in another processor chip, at the choice of the designer of a particular device. This situation is much more widespread, especially in the audio-video market.

 

[Scytales]

Both of my ESS DACs produce a crackling sound when switching from PCM music to DSD, but the AKM DCA doesn't have this issue. What could be the reason for this?
Even when switching DSD tracks, there's a crackling sound at first, but it stops after that.
By the way, the audio source is provided to the DAC through Roon.

I have read a testimony of the user of an RME ADI 2 and another one of an Yamaha CD-S2000 who encountered the same problem.

The noise can come from the fact that DSD signals contain a DC component which is signal dependant, because there is no "0" level in it, only two levels alternately switched. It must be said that, in the first SA-CD players built by Sony and Philips, the DSD decoders incorporated a digital fade out/fade in process of a few milliseconds duration which avoids clicks or other noises at the beginning of a track or when switching from track to track.

But I suspect a more probable cause might be a sub-optimal timing of resetting the D/A chip or muting function to avoid such noise artifacts.

 

[jayapple]

I have read a testimony of the user of an RME ADI 2 and another one of an Yamaha CD-S2000 who encountered the same problem.

The noise can come from the fact that DSD signals contain a DC component which is signal dependant, because there is no "0" level in it, only two levels alternately switched. It must be said that, in the first SA-CD players built by Sony and Philips, the DSD decoders incorporated a digital fade out/fade in process of a few milliseconds duration which avoid clicks or other noises at the beginning of a track or when switching from track to track.

But I suspect a more probable cause might be a sub-optimal timing of resetting the D/A chip or muting function to avoid such noise artifacts.

It seems that Roon can also offer fade-in and fade-out functions. I'll give it a try to see if I can eliminate the crackling sound.

 

[bogi]

Let me add that, to my knowledge, there were only two DAC chips ever which were able to decimate DSD in low sample rate PCM, and this step was optional, not mandatory: the Wolfson WM8741 and WM8742.

That is not to say that a decimation of DSD in PCM is not done outside the DAC chip, in another processor chip, at the choice of the designer of a particular device.

The situation with DAC chips may be like you mentioned. I agree that DSD to low rate PCM conversion is usually done outside of DAC chip.

The DSD to low rate PCM conversion is typical for R2R DACs which support also DSD signal at their input (both DAC chip based or discrete R2R implementations). Only few R2R DAC devices contain separate DAC implementation for DSD input (for example Holo DACs, Denafrip DACs probably too).

Chord DACs are known to convert DSD to 705.4k PCM. It's then followed with oversampling and delta sigma modulation, since they are delta sigma DACs..

 

[mcdonalk]

Abstract: The problem was a short term eustachion tube disorder which has been resolved, and I no longer hear the distortion.

Discussion: When I was hearing the anomalous distortion, and after listening through headphones at different stages in my system's audio chain, it seemed as though the recordings were the source of the distortion. But, I reversed the headphones, and the distortion remained in my left ear; it did not change to the other ear. I made an appointment with my ENT. Then, after fretting about this for a while, and reading on the internet the worst possible things that could explain it, I remembered an anomalous hearing experience about 6 months ago, when having stitches in my head tugged on my ears and changed my hearing. (You can read more about it here: https://audiosciencereview.com/forum/index.php?threads/a-tale-of-my-hearing-tmi-tltr.60711/)

So, I started tugging my ear in all directions, experimenting with my ear canal geometry. At one point, I heard a "crackle," and my hearing returned to normal! The distortion was gone! I nevertheless kept my appointment at the ENT. She examined my ears and asked me some questions and concluded that the cause of the temporary distortion was a temporary eustachion tube disorder caused by allergies. Yes, I do remember sinus congestion and an earache around that time. It's just that the violin's sounding like scraping metal seemed so real!