Why SACD Disc playback sounds better (less jitter?) than its equivalent ripped DSF file / streaming playback

[MasterApex]

I believe you but the problem is not the disc vs. playback but presumably some glitch with the Oppo.

Take a look at

Marantz SA-11s2 Review (SACD Player; Seiko NPC SM5866AS Monaural DAC)

Marantz SA-11s2 Review (2008-era SACD player with Seiko NPC SM5866AS Monaural DAC) Disclaimer: This review is limited by the limited availability of physical SACD test/diagnostics discs. DACs are all about precision Before there was hi-fi, the battles of engineering and craftsmanship lay in...

audiosciencereview.com

Post in thread 'Marantz SA-10 Review (SACD Player & DAC)'
https://audiosciencereview.com/foru...-10-review-sacd-player-dac.31686/post-1532634

Measuring the "sound signature" of two different digital sources (DSD 2.8MHz)

Measuring the "sound signature" of two different digital sources (DSD 2.8MHz) https://www.audiosciencereview.com/forum/index.php?threads/marantz-sa-11s2-review-sacd-player-seiko-npc-sm5866as-monaural-dac.37986/ Vs...

www.audiosciencereview.com

Two reasonable DACs with a large PK Metric when tested with real music...

I've previously shown that my SA-11s2 and Panasonic UB9000 sound similarly and measure similarly with DeltaWave. In this post, I'm sharing some preliminary data showing a) a measurable PK Metric difference between two reasonable DACs with real music b) the DAC with a ~10 dB worse 1 kHz SINAD...

www.audiosciencereview.com

Today, I use the UBP-800M2 running my ripped SACD. It’s great.

If I had infinite money, I really like the SA-10 from a drive noise, remote control and sound quality. I sold the SA-10 to fund my speakers.

I am 100% behind you in saying that the SACD playback is superior to file playback on Oppo, since you feel it is true and there is a claim of ABX comparisons.

I am 100% sure that SACD files and proper DSD decode/PCM decode is perfect. You can get proper DSD to PCM decode with any Sony Blu-ray player. DSD decode is done in many AVRs. If you want luxury SACD player, get the the SACD 30n or SA-10, understanding that they do great with files also.

This is a very helpful feedback...Thank You
The above experiment was run in my friend sound room/Speakers/Amps/Pre-amp equipment.
My contribution was SACD Disc and its ripped DSF file using Oppo 105 (known procedure in the industry).

I have Marantz AV7706 and Sony UBP-X800M2 4k blu-ray for HT movie surround sounds.
When I play USB DSF file on Sony UBP-X800M2 (there is no audio analog output), I connect X800M2 HDMI to AV7706 so the USB DSF file playback will rely of AV7706 DAC to decode.

I use the Oppo 105 as SACD Disc player XLR output to Benchmark LA4 stereo pre-amp driving a pair of Levinson 536 and B&W 800D3
I can play USB DSF file on Oppo105 with two output options :
- XLR , so using Oppo internal DAC (which is the experiment we did but in my friend room/speakers/amps/pre-amp)
- HDMI to feed to AV7706 internal DAC (did not do this)
I recently setup a PC with JRiver that can also feed various digital files to DLNA players connected in network.

Do you recommend I get / experiment with SA-10 or used Oppo 205?

 

[GXAlan]

I use the Oppo 105 as SACD Disc player XLR output to Benchmark LA4 stereo pre-amp driving a pair of Levinson 536 and B&W 800D3

Measurements of Oppo BDP-105 Blu-ray Player

At our last gathering at a friend's house where we were measuring the Schiit Yggdrasil DAC, I was also asked to measure an Oppo BDP-105 player as a DAC. This player despite its age has a very positive reputation as far as fidelity. I made a complete set of measurements. I won't bore you with...

www.audiosciencereview.com

The Oppo 105 doesn’t measure well.

My bias is to go with the Marantz AV10 and use the Sony as your SACD player.

Your Levinson amp is 26 dB of gain, and your 800D3 is 90 dB for 2.83V at 1m. This means that you only need to 0.14V from the preamp to hit 90 dB at 1m.

This will also be the way to go as Blu-Ray music discs become the alternative to SACDs.

I do not know of a better setup for your Levinson/800D3 than a premium AV processor for handling encrypted high res PCM over HDMI.

Since you already have the Sony, it’s even easier to try it.

 

[stvnharr]

Interesting thread. Perhaps the difference you hear is in the Oppo 105 player capabilities.
However, too many posters have said that "You didn't do it right", which I find hilarious, though that tends to be the nature of ASR.
Given my two cents, along with some time with someone else's Oppo 105, and being the owner of a Marantz SA Ki Ruby, I would be inclined to think that perhaps you need a Ruby or SA10 player. Both do everything in DSD with low pass filtering to analogue. Ruby has the advantage of DSD256 file playback. SA10 has an advantage of balanced output, if that is important or needed, though only plays DSD128 files.

 

[Hayabusa]

Over the weekend, we had a chance to rip a few of our own SACD discs into DSF file.
We then compare the playback of SACD Disc vs the Ripped DSF File on the same audio system.

Player: Oppo 105 SACD player (has USB input to playback the DSF file from original SACD Disc) - using XLR output
Pre-Amp: Denafrips Athena
Amplifier: Two Mark Levinson 333 Bi-Amp
Speakers: A Pair of B&W N801
Room: Acoustically treated
Playback is level match with pink noise track so both SACD Disc and DSF file playback are the same volume SPL.

The ripped DSF file sounds good, but when compare to SACD Disc playback of the same album/song, the Disc playback has slightly better dynamic and transparency.
We could tell with our eyes closed when one of us switches between Disc Playback vs DSF file playback from the same Oppo 105.

Why is that?
Is it because the Disc playback are clocked signal inside the SACD player circuit so it has less digital jitter than the USB file interface playback?

USB in is asynchronous so the same clock of the OPPO will drive the D/A

 

[MasterApex]

Interesting thread. Perhaps the difference you hear is in the Oppo 105 player capabilities.
However, too many posters have said that "You didn't do it right", which I find hilarious, though that tends to be the nature of ASR.
Given my two cents, along with some time with someone else's Oppo 105, and being the owner of a Marantz SA Ki Ruby, I would be inclined to think that perhaps you need a Ruby or SA10 player. Both do everything in DSD with low pass filtering to analogue. Ruby has the advantage of DSD256 file playback. SA10 has an advantage of balanced output, if that is important or needed, though only plays DSD128 files.

Thanks....I sort of came to conclusion to purchase another SACD player that is better than oppo 105 , was leaning towards a used oppo 205 .
After the revelation or observation that SACD Disc playback is superior to DSF file playback in Oppo 105.
and after experimenting with a few friend's external DAC (using HDMI-I2S eBay adapter) vs friend's Technics SACD player

Why a new dedicated SACD player has more noticeable sound quality difference than HDMI-I2S with new external DACs?

Over the weekend, we brought together our gears and run experiment with HDMI-I2S on a few external DACs to hear the comparison among: Oppo 105 SACD Player XLR output Oppo 105 HDMI-I2S to S.M.S.L. D400EX - XLR Output Oppo 105 HDMI-I2S to Denafrips Venus II 12th - XLR Output Oppo 105 HDMI-I2S to...

www.audiosciencereview.com

When I ripped SACD disc with oppo 105 into DSF file, is the file DSD128 or which DSD is it?

 

[-Matt-]

When I ripped SACD disc with oppo 105 into DSF file, is the file DSD128 or which DSD is it?

Well, according to:
https://en.m.wikipedia.org/wiki/Direct_Stream_Digital

and:
https://en.m.wikipedia.org/wiki/Super_Audio_CD

If your file is a bit perfect copy of the SACD it should probably be DSD64 (with a sampling rate 2.8224 MHz).

But really, if you are going to claim an audible difference between SACD playback and the DSD file version, the absolute minimum requirement for rigor is that you ensure the DSD file is a bit perfect copy of the disk. If you make such a claim you should be fully on top of such things and not asking us!

 

[Ktacos]

It was not an assumption, it is a repeatable distinguishable audible sound signature that we (me and friends) observed.

IME, a group of people who are trying to listen for something will come up with differences they claim to hear even when there is none. Humans are kinda silly.

bit of an edit - For some time I've wanted to create a setup with various filters that my audio friends can play with and give feedback on, but in reality the filters are a lie and do nothing. Only way for this to work is to not tell them before hand but I have some ethical concerns about this so I will never do it, and we've got plenty of info on placebo at this point. I've definitely watched some of my friends effectively guess as to what their hearing. I'm definitely not above it, you haven't lived until you've been in charge of checking monitors on stage, give the ok to FoH that the changes they made are good, only for them them say over the whole damn pa that they haven't done anything yet.

 

[tmtomh]

Interesting thread. Perhaps the difference you hear is in the Oppo 105 player capabilities.
However, too many posters have said that "You didn't do it right", which I find hilarious, though that tends to be the nature of ASR.
Given my two cents, along with some time with someone else's Oppo 105, and being the owner of a Marantz SA Ki Ruby, I would be inclined to think that perhaps you need a Ruby or SA10 player. Both do everything in DSD with low pass filtering to analogue. Ruby has the advantage of DSD256 file playback. SA10 has an advantage of balanced output, if that is important or needed, though only plays DSD128 files.

"You didn't do it right" might be an inartful and not-super-helpful way of stating it, but it's not hilarious, and it's in "the nature" of ASR because it's shorthand for the most sensible answer: Why does a bit-identical copy of an SACD track sound better/different when played from the physical disc than when played from storage media through USB on the same player, using the same signal chain, out the same outputs, into the same downstream gear?

There are only three possible answers:

1. The file is not in fact identical to what's on the disc;
2. The external drive-to-USB-to-internal circuitry signal chain is in some way different than the disc-read-head-to-internal circuitry chain;
3. There is no sonic difference between the two that would survive a proper double-blind test.

No one here has any way of checking item 1, but it seems quite unlikely that this is the culprit.

To everyone's knowledge so far, item 2 is not the culprit either: same clock and no known aspect of the signal path from the USB input port to the main circuit board versus from the disc drive to the main circuit board that could make any difference.

And however mediocre the 105's measurements are, or however mediocre folks think it might be as an SACD player compared to other units, none of those comments and opinions has any bearing on the OP's question because in that case the 105 is equally mediocre with SACDs and DSF files.

So the most likely - not certain, but most likely - culprit is item 3, aka "you didn't do the comparison right."

Finally, I have left out three specific scenarios, which we might call "sub-possibilities" of items 1 and 2 above:

1. When playing content from an external drive via USB input, the 105 for some reason applies digital gain or attenuation that it doesn't apply when playing content from an SACD disc. I view this as incredibly unlikely.

2. The 105's settings as to whether to play SACDs "pure" as bitstream or convert them to PCM before outputting them to analogue might apply separately to SACD vs USB input, so the comparison might be DSD to PCM here. Again, I view this is extremely unlikely since I owned a 105 for several years and my recollection is that the DSD/PCM setting is a global one. If it were an issue, I would expect it to manifest itself as a volume mismatch and/or an issue with clipping, not an actual sound-quality difference

3. @MasterApex 's original post does not actually say whether the DSF files were played (a) from a USB-connected hard drive/thumb drive/SSD, OR, (b) from an external "smart" digital device like a computer or streamer, plugged into the Oppo's "USB DAC" input. The strong implication is that it's the former, and in that case this factor is irrelevant. But if by some chance it was the latter, then the possibility of unmatched volume levels, software-based DSD-to-PCM conversion, or possibly even digital clipping, does come into play. If MasterApex has clarified this aspect of the signal chain elsewhere in the thread, then my apologies and please disregard this part of my comment.

 

[olieb]

Humans are kinda silly.

I agree but I would not call it "silly" in this case.
One needs considerable scientific discipline to avoid falling for bias or jumping to false conclusions.
And there might actually be an obvious cause for audible difference here as the level matching seems to be done only with (too) low resolution/precision. But the OP prefers to ignore that point.

I see three reasons for what the OP calls the observation of hearing an audible difference.
The difference is due to
1. a difference in volume (SPL),
2. bias in the sighted listening that gives rise to "hearing" a difference that is not actually (audibly) there,
3. an actual audible difference (in quality) of the signal from the different sources/source files.
I let Ockham decide.

 

[GXAlan]

2. The external drive-to-USB-to-internal circuitry signal chain is in some way different than the disc-read-head-to-internal circuitry chain;

To everyone's knowledge so far, item 2 is not the culprit either: same clock and no known aspect of the signal path from the USB input port to the main circuit board versus from the disc drive to the main circuit board that could make any difference.

….

If it were an issue, I would expect it to manifest itself as a volume mismatch and/or an issue with clipping, not an actual sound-quality difference

…

the possibility of unmatched volume levels, software-based DSD-to-PCM conversion, or possibly even digital clipping, does come into play.

I think we are too quick vs to rule out #2. I forgot to ask if @MasterApex is playing the DSF files from a PC and the USB input or using the in built app

The first revision of the BDP-105 didn’t take DSD via USB do there would be a DSD to PCM conversion occurring that pathway versus straight from disc. The BDP didn’t ship with multichannel DSF playback from the internal app on day one.

Some AVRs have no problems playing back multichannel DSD from HDMI from a BluRay or SACD player with HDMI but can only play 2 ch DSD using the USB or DLNA pathway. So even though the hardware path is the same, the software isn’t as intuitively the same and for all DSD, there are noise shaping and ultrasonic roll off options.

Maybe SACD playback uses a canned, closed source library for copy protection and DSF playback is done by a different team with different libraries or algorithms. Even among SACDs, there were disagreements how to filter the ultrasonic noise.

I think it’s actually very convincing that the ABX test was borderline, not obvious. I use my own example of being able to ABX two recordings with near 100% accuracy by identifying the piece that gave me the ASMR tingles but on subsequent trials, I was never able to generate any tingles ever

Does DSD recording benefit Japanese traditional instruments?

Is DSD 11.2 MHz "required" for recording taiko drums and Japanese stringed instruments? https://www.phileweb.com/review/column/202303/29/2088.html Original article in Japanese "The powerful and deep ultra-low sound of Japanese taiko drums and the penetrating treble of Noukan and Hichiriki...

audiosciencereview.com

These differences are never night and day. When you ABX two identical files, try to get p = 0.15 or 0.10. You actually get p=0.5 most of the time too.

 

[antcollinet]

….

…



I think we are too quick vs to rule out #2. I forgot to ask if @MasterApex is playing the DSF files from a PC and the USB input or using the in built app

The first revision of the BDP-105 didn’t take DSD via USB do there would be a DSD to PCM conversion occurring that pathway versus straight from disc. The BDP didn’t ship with multichannel DSF playback from the internal app on day one.

Some AVRs have no problems playing back multichannel DSD from HDMI from a BluRay or SACD player with HDMI but can only play 2 ch DSD using the USB or DLNA pathway. So even though the hardware path is the same, the software isn’t as intuitively the same and for all DSD, there are noise shaping and ultrasonic roll off options.

Maybe SACD playback uses a canned, closed source library for copy protection and DSF playback is done by a different team with different libraries or algorithms. Even among SACDs, there were disagreements how to filter the ultrasonic noise.

I think it’s actually very convincing that the ABX test was borderline, not obvious. I use my own example of being able to ABX two recordings with near 100% accuracy by identifying the piece that gave me the ASMR tingles but on subsequent trials, I was never able to generate any tingles ever

Does DSD recording benefit Japanese traditional instruments?

Is DSD 11.2 MHz "required" for recording taiko drums and Japanese stringed instruments? https://www.phileweb.com/review/column/202303/29/2088.html Original article in Japanese "The powerful and deep ultra-low sound of Japanese taiko drums and the penetrating treble of Noukan and Hichiriki...

audiosciencereview.com

These differences are never night and day. When you ABX two identical files, try to get p = 0.15 or 0.10. You actually get p=0.5 most of the time too.

That's a whole bunch of maybes. And for all of those - say conversion to PCM, or noise shaping: With these high resolution files how audible to you really expect them to be? DSD64 far exceeds the capability of human hearing - and any competent conversion should be totally transparent.

At the end of the day we are only going to be able to speculate unless OP is able to actually measure/record the output from the DAC/player and demonstrate audible differences in (for example) deltawave.

 

[elberoth]

In my experience, it all depends on:

- USB input receiver implementation (there was a steep learning curve and early USB input implementations left much to be desired in terms of SQ)
- USB transport quality

On modern CD/SACD players, when fed from a quality USB transport or the inbuilt Ethernet input (which in many cases sounds superior), files often sound superior to physical media (CDs or SACDs).

On older models, whatever transport you use, physical media often sounds better.

 

[tmtomh]

….

…



I think we are too quick vs to rule out #2. I forgot to ask if @MasterApex is playing the DSF files from a PC and the USB input or using the in built app

The first revision of the BDP-105 didn’t take DSD via USB do there would be a DSD to PCM conversion occurring that pathway versus straight from disc. The BDP didn’t ship with multichannel DSF playback from the internal app on day one.

Some AVRs have no problems playing back multichannel DSD from HDMI from a BluRay or SACD player with HDMI but can only play 2 ch DSD using the USB or DLNA pathway. So even though the hardware path is the same, the software isn’t as intuitively the same and for all DSD, there are noise shaping and ultrasonic roll off options.

Maybe SACD playback uses a canned, closed source library for copy protection and DSF playback is done by a different team with different libraries or algorithms. Even among SACDs, there were disagreements how to filter the ultrasonic noise.

I think it’s actually very convincing that the ABX test was borderline, not obvious. I use my own example of being able to ABX two recordings with near 100% accuracy by identifying the piece that gave me the ASMR tingles but on subsequent trials, I was never able to generate any tingles ever

Does DSD recording benefit Japanese traditional instruments?

Is DSD 11.2 MHz "required" for recording taiko drums and Japanese stringed instruments? https://www.phileweb.com/review/column/202303/29/2088.html Original article in Japanese "The powerful and deep ultra-low sound of Japanese taiko drums and the penetrating treble of Noukan and Hichiriki...

audiosciencereview.com

These differences are never night and day. When you ABX two identical files, try to get p = 0.15 or 0.10. You actually get p=0.5 most of the time too.

I'm going to be just a little grumpy for a moment and point out that towards the end of my comment I specifically noted the #2-type issue you mention here: Is the OP using an external drive connected to the USB input, with the Oppo therefore controlling all the processing/handling of the DSF files internally - or is he using an external smart device like a computer or similar and playing the DSF files from there though the Oppo's USB DAC input?

Now that I've gotten that out of my system ... I think you make a great set of points here. Specifically, I'd forgotten that the 105 didn't fully support DSD file playback. Only the later 105D supported it - and it was an often-overlooked difference between the two since the "D" stood for Darbee Edition, which solely had to do with video processing.

However, the original 105 DOES play DSD (and therefore DSF) files directly through its regular USB inputs if you plug an external device/drive into it. Where the 105 does NOT support DSD is if you stream content to it from an external computer/streamer into its USB DAC input.

So if the OP is just plugging a thumb drive or external drive into one of the 105's regular USB-A ports, then it should be a "pure" DSD signal all the way. If he's instead using the USB-B "USB DAC" port and streaming from an external device, my understanding is that it simply wouldn't work at all. But it is possible that the external device is converting the DSD to PCM internally.

Either way, I agree with @antcollinet that DSD sounds the same as properly converted DSD-PCM. That's why I wrote above that I would expect any audible difference from PCM conversion to be in the form of volume/level changes or mismatches. This is because the equivalent of digital 0.0 in DSD defaults to -6.0dB when converted to PCM. So the standard practice is to apply makeup gain of +6.0dB when converting from SACD to PCM.

However - and I don't quite understand how or why this is - some SACDs are not mastered to whatever the rule or standard is supposed to be for the format, so when you convert them to PCM the max level ends up being greater than -6dB. I've ripped and converted SACDs where the PCM conversion without makeup gain is around -4dB or in one case somewhere between -2 and -3dB.

My point is that the defacto volume level of DSD converted to PCM will be lower than the original DSD - and if your software converts DSD to PCM on the fly it will apply a standard makeup gain, which will result in PCM volumes that vary considerably, from a couple dB quieter than the DSD original to a couple dB louder and full of clipping.

The only way to truly volume match when comparing DSD and PCM converted from that DSD is to convert the entire album to a single PCM file, open it in an app like Audacity, and see what level its loudest peak reaches. Then you go back to the DSD rip and reconvert it to individual PCM tracks with the precise makeup gain based on what you found in Audacity.

Even then you're working with an assumption that the DSD file's loudest peak will play at the equivalent of digital 0.0.

So with DSD-PCM volume matching is the issue, not actual sound quality differences (except possibly for clipping-based distortion as a secondary effect of a volume mismatch problem).

 

[GXAlan]

At the end of the day we are only going to be able to speculate unless OP is able to actually measure/record the output from the DAC/player and demonstrate audible differences in (for example) deltawave.

Agreed, but read through my old thread linked about the DSD for Japanese drums. First two pages.

Does DSD recording benefit Japanese traditional instruments?

Is DSD 11.2 MHz "required" for recording taiko drums and Japanese stringed instruments? https://www.phileweb.com/review/column/202303/29/2088.html Original article in Japanese "The powerful and deep ultra-low sound of Japanese taiko drums and the penetrating treble of Noukan and Hichiriki...

www.audiosciencereview.com

The OP could go through the same process with his Oppo 105 as I did with my Sony TAZH1ES, but since he is planning on buying something new, I would argue that my advice is sound advice.

Get the AV10 or HTP-1 to allow Dirac and use a Sony player to convert DSD to 176 kHz PCM. Can play direct without PEQ/RoomEQ and benefit from pretty good SINAD or enable room EQ that decimates to 44.1 kHz, but preserves >16-bit dynamic range during processing.

 

[GXAlan]

So with DSD-PCM volume matching is the issue, not actual sound quality differences (except possibly for clipping-based distortion as a secondary effect of a volume mismatch problem).

All great points but take a look at the thread that I asked anticollinet to read. First post and then page 2 follow up post by me with more testing.

Does DSD recording benefit Japanese traditional instruments?

Is DSD 11.2 MHz "required" for recording taiko drums and Japanese stringed instruments? https://www.phileweb.com/review/column/202303/29/2088.html Original article in Japanese "The powerful and deep ultra-low sound of Japanese taiko drums and the penetrating treble of Noukan and Hichiriki...

www.audiosciencereview.com

I am 100% with you in that DSD can be transcoded to PCM with zero loss in quality. But since proving the #2 scenario was so tricky with the TAZH1ES, it is a lot of work to do that with the Oppo, when you might as well just get a new AVR with a good DAC and use a Sony BD/UHD player that does proper PCM conversion.

 

[tmtomh]

All great points but take a look at the thread that I asked anticollinet to read. First post and then page 2 follow up post by me with more testing.

Does DSD recording benefit Japanese traditional instruments?

Is DSD 11.2 MHz "required" for recording taiko drums and Japanese stringed instruments? https://www.phileweb.com/review/column/202303/29/2088.html Original article in Japanese "The powerful and deep ultra-low sound of Japanese taiko drums and the penetrating treble of Noukan and Hichiriki...

www.audiosciencereview.com

I am 100% with you in that DSD can be transcoded to PCM with zero loss in quality. But since proving the #2 scenario was so tricky with the TAZH1ES, it is a lot of work to do that with the Oppo, when you might as well just get a new AVR with a good DAC and use a Sony BD/UHD player that does proper PCM conversion.

Thanks Alan. I read your first post and additional-testing post in that thread.

From the first post my first reaction would be that when you compare two PCM files that have different sample rates and bit depths, you're never going to get them to match up, even if the higher sample rate is an integer-multiple of the lower one.

I understand you're saying there are signal-correlated peaks that differ between the two, but from what I can see there's no way we can even begin to know why that is, which means we can't even begin to attribute any signal correlated differences to DSD vs PCM - especially in your second post with the added A/D step involved.

As for DSD to PCM conversion, your DSD reference file is being compared to the CD version after being converted to DXD, which is PCM. So it's a PCM to PCM comparison, yes?

The bottom line is that if for some reason it turns out that the OP's "DSF file playback" was actually DSF to PCM in the signal chain, he's going to have a very high mountain to climb in order to establish that the volume was the same as the playback volume of the SACD disc. If he were somehow able to climb that mountain - highly unlikely - then we might want to investigate further RE DSD-PCM conversion.

 

[GXAlan]

From the first post my first reaction would be that when you compare two PCM files that have different sample rates and bit depths, you're never going to get them to match up, even if the higher sample rate is an integer-multiple of the lower one.

Agree, but they match super well meaning that it’s not a gross mixing difference like a vinyl LP mix versus a radio mix vs a studio uncompressed mix.

I understand you're saying there are signal-correlated peaks that differ between the two, but from what I can see there's no way we can even begin to know why that is, which means we can't even begin to attribute any signal correlated differences to DSD vs PCM - especially in your second post with the added A/D step involved.

There are two sets of data. Since deltawave only works with PCM, I took DSD 2.8 to PCM 352.8 kHz. Deltawave then takes CD 44.1 to PCM 352.8.

All done digitally. This is the input files. The pkmetric says its - 90 dB or so.

This makes sense because I show
1) there is not a gross mismatch in mix
2) the CD rolls off the HF while the DSD has HF including ultrasonic noise
3) -90 dB is beyond threshold of hearing anyway

So it's a PCM to PCM comparison, yes?

Yes. For page 1.

But remember, my ABX testing is showing a difference. I don’t believe the ultrasonic content is the difference (maybe it is?)

On page 2, I record the output using my E1DA. This time, it is DSD native vs CD layer native.

We have the volume differences which exceed that predicted by the source files which was not expected.

The bottom line is that if for some reason it turns out that the OP's "DSF file playback" was actually DSF to PCM in the signal chain, he's going to have a very high mountain to climb in order to establish that the volume was the same as the playback volume of the SACD disc.

Right and my experiment shows that the volume difference may be non linear at which point it is IMPOSSIBLE to level match.

(which gets back to the OP. If there is a difference, the answer is in the Oppo 105 not the fundamental of physical media on a disc versus a DSF file extracted from the disc).

And I would put odds that the Oppo 105 is doing something different because someone else had the same comment about a difference, took the trouble to ABX, and instead of p=0.50, it reportedly was closer to 0.10.

 

[antcollinet]

I'm going to be just a little grumpy for a moment and point out that towards the end of my comment I specifically noted the #2-type issue you mention here: Is the OP using an external drive connected to the USB input, with the Oppo therefore controlling all the processing/handling of the DSF files internally - or is he using an external smart device like a computer or similar and playing the DSF files from there though the Oppo's USB DAC input?

Now that I've gotten that out of my system ... I think you make a great set of points here. Specifically, I'd forgotten that the 105 didn't fully support DSD file playback. Only the later 105D supported it - and it was an often-overlooked difference between the two since the "D" stood for Darbee Edition, which solely had to do with video processing.

However, the original 105 DOES play DSD (and therefore DSF) files directly through its regular USB inputs if you plug an external device/drive into it. Where the 105 does NOT support DSD is if you stream content to it from an external computer/streamer into its USB DAC input.

So if the OP is just plugging a thumb drive or external drive into one of the 105's regular USB-A ports, then it should be a "pure" DSD signal all the way. If he's instead using the USB-B "USB DAC" port and streaming from an external device, my understanding is that it simply wouldn't work at all. But it is possible that the external device is converting the DSD to PCM internally.

Either way, I agree with @antcollinet that DSD sounds the same as properly converted DSD-PCM. That's why I wrote above that I would expect any audible difference from PCM conversion to be in the form of volume/level changes or mismatches. This is because the equivalent of digital 0.0 in DSD defaults to -6.0dB when converted to PCM. So the standard practice is to apply makeup gain of +6.0dB when converting from SACD to PCM.

However - and I don't quite understand how or why this is - some SACDs are not mastered to whatever the rule or standard is supposed to be for the format, so when you convert them to PCM the max level ends up being greater than -6dB. I've ripped and converted SACDs where the PCM conversion without makeup gain is around -4dB or in one case somewhere between -2 and -3dB.

My point is that the defacto volume level of DSD converted to PCM will be lower than the original DSD - and if your software converts DSD to PCM on the fly it will apply a standard makeup gain, which will result in PCM volumes that vary considerably, from a couple dB quieter than the DSD original to a couple dB louder and full of clipping.

The only way to truly volume match when comparing DSD and PCM converted from that DSD is to convert the entire album to a single PCM file, open it in an app like Audacity, and see what level its loudest peak reaches. Then you go back to the DSD rip and reconvert it to individual PCM tracks with the precise makeup gain based on what you found in Audacity.

Even then you're working with an assumption that the DSD file's loudest peak will play at the equivalent of digital 0.0.

So with DSD-PCM volume matching is the issue, not actual sound quality differences (except possibly for clipping-based distortion as a secondary effect of a volume mismatch problem).

This is what I love about hanging out here. Every day is a school day at ASR.

 

[pkane]

Since deltawave only works with PCM, I took DSD 2.8 to PCM 352.8 kHz. Deltawave then takes CD 44.1 to PCM 352.8.

As it happens, DeltaWave will read DSD and DSF files and even give you the choice of filter and the PCM rate to downsample to:

 

[GXAlan]

As it happens, DeltaWave will read DSD and DSF files and even give you the choice of filter and the PCM rate to downsample to:

View attachment 384029

To be clear, I used DeltaWave to do the PCM conversion. What I should have said is that DeltaWave does its math on PCM samples and lets you control the strategy (CD at 44.1 and downsample DSD to 44.1) versus (DSD to 352 kHz and CD to 352).

DeltaWave the single most powerful tool for separating pure sighted bias versus “we can spend more time digging”