Is Bit Perfect DSD possible through Foobar2000 to Topping D50iii

[drewdawg999]

A component that was downloaded and installed last month listed SACD but it acted the same as this later component in that playing a DSD128 would cause a "352.8khz not supported" error.

I'm thinking you have a wrong setting somewhere, others have gotten it to work. I got my DO100 to work in native DSD, but it doesn't make a difference on an ESS DAC, no Direct DSD path. So the DSD and PCM will sound the same.

DSD when properly decoded doesn't have volume control, and can't have EQ applied. That would defeat the purpose of the PEQ feature, which is one of this DAC's main selling points. I believe Topping couldn't care less about proper DSD decoding, since it's pretty rare that anyone actually cares. On the E70 Velvet they chose not to implement DSD Direct, even though it's on the AKM chip.

 

[AUDACC]

I'm thinking you have a wrong setting somewhere, others have gotten it to work. I got my DO100 to work in native DSD, but it doesn't make a difference on an ESS DAC, no Direct DSD path. So the DSD and PCM will sound the same.

DSD when properly decoded doesn't have volume control, and can't have EQ applied. That would defeat the purpose of the PEQ feature, which is one of this DAC's main selling points. I believe Topping couldn't care less about proper DSD decoding, since it's pretty rare that anyone actually cares. On the E70 Velvet they chose not to implement DSD Direct, even though it's on the AKM chip.

Since I am using both the PEQ and volume control on the D50iii, feeding native DSD to the D50 will not result in bit perfect D to A. A couple of Grok inquiries yielded results that confirm your comments:

”When PEQ is turned off, the volume control on the Topping D50 III is applied digitally inside the dual ES9039Q2M DAC chips, after the native DSD bitstream has been received from the XMOS XU316 and jitter-cleaning stage.

Exact point in the signal path (PEQ off + native DSD)

1. USB → XMOS XU316 receives raw native DSD (no processing since PEQ is off).
2. XMOS passes the untouched DSD bitstream via I²S to the jitter-cleaning stage.
3. Jitter cleaner reclocks the signal (timing improvement only — no data change).
4. Clean DSD bitstream is fed to the ES9039Q2M DAC chips in native DSD mode.
5. Inside the ES9039Q2M chips: The digital volume control is applied directly in the DSD domain (or via the chip’s high-quality digital attenuator that preserves the 1-bit nature as much as possible).
6. The attenuated DSD signal then goes through the chip’s internal DSD-to-analog conversion path → analog output stage (op-amps) → RCA/TRS outputs.

The volume control does not sit in the XMOS (where PEQ lives). It is handled by the ESS DAC chips themselves, which is why the D50 III can offer a smooth, stepped digital volume (0.5 dB steps) even in preamp mode.

Does it change the encoding method?

• No forced PCM conversion for native DSD when volume is adjusted.
• ESS Sabre chips (including the ES9039Q2M) are designed with dedicated hardware that can attenuate DSD data without converting it back to multi-bit PCM. The 1-bit DSD stream remains in its native format through the volume stage and into the modulator/filter path.

This is consistent with how modern ESS DACs handle DSD volume control (confirmed across multiple Topping D50-series and similar ESS-based designs).

Does it add noise or distortion?

• Very little in practice — the volume control in the ES9039Q2M is one of the highest-quality digital attenuators available.
• At 0 dB (maximum volume): Bit-perfect or extremely close to it for DSD.
• When you lower the volume (e.g., -10 dB or more): There is a small theoretical loss of resolution because you're attenuating a 1-bit signal. In real-world measurements on similar ESS implementations, this shows up as:
  • Extremely minor increases in noise floor or very low-level distortion.
  • Still excellent overall performance (far better than most software volume controls or older DACs).
• Many users and reviewers note that the D50 III’s internal volume sounds transparent even at moderate attenuation levels. For the absolute cleanest native DSD path, the recommendation is still to set the DAC volume to maximum (0 dB) and control volume in your amplifier or preamp instead.

Practical advice for your DSD128 playback

• For the purest native DSD (lowest possible added noise/distortion):
  • Keep PEQ off.
  • Set D50 III volume to 0 dB (maximum) in preamp mode or switch to fixed DAC mode if available.
  • Control listening volume with your downstream amp or active speakers.
• If you need to use the D50 III’s volume control regularly, the added artifacts are negligible for most listening and still deliver excellent measured performance.

This setup keeps the DSD stream as native as possible all the way to the analog conversion stage.”

 

[101geo]

I can do it with Foobar feeding a D10B but it was a pain to set up. Regret it now as there was no difference at all with multiple sessions and listeners on any album I have played. My new system/player will have everything at 48Khz.

 

[Scytales]

Since I am using both the PEQ and volume control on the D50iii, feeding native DSD to the D50 will not result in bit perfect D to A. A couple of Grok inquiries yielded results that confirm your comments:

”When PEQ is turned off, ...”

The text Grok has generated is just a melting pot of uneducated guesses or statements the machine gathered on the Internet. It shows no real technical knowledge about the matters it tackles, which isn't surprising from a so-called artificial 'intelligence'. This text content is as informed as some of the famous Star Trek technobables.

That being said, the issue you encountered has nothing to do with signal processing or conversion to analogue. It is that your digital file player and DAC doesn't do a proper handshake to let DSD be transmitted from the player to the DAC, whereas the DAC is specified to be able to receive and play DSD data. It seems to me that this is a problem that deserves to be addressed because, as is, your DAC doesn't meet one of its specified features.

If you receive an answer from Topping, it could be of interest for others.

 

[AUDACC]

The text Grok has generated is just a melting pot of uneducated guesses or statements the machine gathered on the Internet. It shows no real technical knowledge about the matters it tackles, which isn't surprising from a so-called artificial 'intelligence'. This text content is as informed as some of the famous Star Trek technobables.

That being said, the issue you encountered has nothing to do with signal processing or conversion to analogue. It is that your digital file player and DAC doesn't do a proper handshake to let DSD be transmitted from the player to the DAC, whereas the DAC is specified to be able to receive and play DSD data. It seems to me that this is a problem that deserves to be addressed because, as is, your DAC doesn't meet one of its specified features.

If you receive an answer from Topping, it could be of interest for others.

You are correct this free version of Grok doesn't understand what its saying. It's just repeating the words it retrieved from its sources. But it is worth noting that some of its sources are chip manufacturers data sheets and applications notes. So I won't completely dismiss Grok's reply, but move forward with an open mind.

Here is my inquiry to Topping customer service:
"I have a Topping D50iii purchased through Amazon and received September 18, 2025. I has operated very well when playing PCM music files through the Foobar2000 player. However, an error occurs when trying to play DSD files through the Topping ASIO driver in native DSD format. How must Foobar2000 ASIO output components be configured to play DSD64, 128, 256, 512 files through the D50iii?"

and their answer:
"Thank you for contacting TOPPING support. We are happy to assist.
Please download the topping driver. After decompression, you will find driver installation instructions and Foobar player installation tutorials, which you can refer to.
https://www.toppingaudio.com/download/v6-14-0-driver "

I had downloaded and installed this driver, v6.14.0, before starting this thread. So I just now went to Windows>control panel>device manager and I see that the driver for the topping USB DAC is shown to be Microsoft driver v10.0.26100.1. I am going to investigate why this is the case. Maybe I am looing in the wrong place for the topping D50iii driver. Any suggestions?

 

[Scytales]

I ceased to use Windows operating systems for almost a decade now, so I am not in the best position to help, but may I suggest you point the Topping after-sale service to this thread or share with them the full details about your software configuration and the "D50 doesn’t accept the 352.8khz" error message you are confronted with, because it seems to me that your inquiry omitted to mention this important details whereas they may help Topping to pin point the issue.

 

[Sokel]

Some random points:

ESS can absolutely play DSD, even the crappy Tone1 can play DSD512 as shown at my previous posts (tested with any method available, including ESS controller who shows a ton of info)
Foobar-ASIO+DSD-SACD (better the 1.6.xx version, not the new 2.0.xx for SACD) can output native DSD at any rate, tested on various DAC, both ESS and AKM.

BUT.
Playing DSD means that any DSP (and that involves volume control as well even if some have a way around it for 5dB attenuation or so) should either be shut-off or shuts off of its own as the ASIO driver is the only way to go.
Global EQ tools like EAPO and the shorts don't work.

So caution about it for systems depended on that.
The right way, is a DAC playing at full output (0dB) followed by a pre or some other attenuating way.

The right way for foobar:

The above components, the right output chosen is ASIO+DSD Topping (or whatever ASIO own driver) and SACD component set-up as my previous photo.

It will show at driver's CP at the correct rate, Foobar's VC will be disabled (as it should and it must be at full 0dB) and no gimmicks as VUs and stuff will work.

 

[Sokel]

You are correct this free version of Grok doesn't understand what its saying. It's just repeating the words it retrieved from its sources. But it is worth noting that some of its sources are chip manufacturers data sheets and applications notes. So I won't completely dismiss Grok's reply, but move forward with an open mind.

Here is my inquiry to Topping customer service:
"I have a Topping D50iii purchased through Amazon and received September 18, 2025. I has operated very well when playing PCM music files through the Foobar2000 player. However, an error occurs when trying to play DSD files through the Topping ASIO driver in native DSD format. How must Foobar2000 ASIO output components be configured to play DSD64, 128, 256, 512 files through the D50iii?"

and their answer:
"Thank you for contacting TOPPING support. We are happy to assist.
Please download the topping driver. After decompression, you will find driver installation instructions and Foobar player installation tutorials, which you can refer to.
https://www.toppingaudio.com/download/v6-14-0-driver "

I had downloaded and installed this driver, v6.14.0, before starting this thread. So I just now went to Windows>control panel>device manager and I see that the driver for the topping USB DAC is shown to be Microsoft driver v10.0.26100.1. I am going to investigate why this is the case. Maybe I am looing in the wrong place for the topping D50iii driver. Any suggestions?

Go to device manager, hit it to show the hidden devices and unistall everything about D50, including the driver.
Plug out D50, reboot and install Tooping's CORRECT (watch the exact naming! ) ASIO driver.

Plug D50 and it should now show its correct ASIO driver installed (with Topping as publisher) .

 

[drewdawg999]

I Googled and its AI spit out this statement:

ESS Sabre DACs (e.g., ES9039PRO, ES9028Q2M) utilize patented Hyperstream IV/QUAD modulator architecture to process DSD (up to DSD1024 or DSD-22.6MHz). These chips internally convert DSD to a 6-bit or 32-bit multipath digital domain, rather than pure "Direct DSD," allowing DSD volume control and digital filtering.

ESS has been hush hush about how they handle DSD, but this Hyperstream IV modulation is undesired. As PEQ and volume control are being used by OP, I would suggest just accepting PCM conversion and enjoying the music. This DAC is not capable of bit perfect DSD performance.

 

[Sokel]

I Googled and its AI spit out this statement:

ESS Sabre DACs (e.g., ES9039PRO, ES9028Q2M) utilize patented Hyperstream IV/QUAD modulator architecture to process DSD (up to DSD1024 or DSD-22.6MHz). These chips internally convert DSD to a 6-bit or 32-bit multipath digital domain, rather than pure "Direct DSD," allowing DSD volume control and digital filtering.

ESS has been hush hush about how they handle DSD, but this Hyperstream IV modulation is undesired. As PEQ and volume control are being used by OP, I would suggest just accepting PCM conversion and enjoying the music. This DAC is not capable of bit perfect DSD performance.

In case of one of ESS DAC (ES9028Q2M) at some implementation of it (haven't tested lots, but the flawed "humped" ones are interesting) DSD is way better electrically as it gets rid of the hump:

PCM vs DSD

Did a little comparison and I was wondering if any friend with better gear here has tested this. Results you're about to see is probably narrowed to the specific DAC but it's interesting to see how it handles things differently. I will not try to decipher the results other that DSD seems much...

www.audiosciencereview.com

What's also interesting is the VC performance (also shown to an extend), either on the MP side or the DAC side, as the performance varies a lot between them.

I wouldn't worry about either format though, if done right.
There's other stuff (like gain-staging for example or RC) that have a lot more impact at SQ.

 

[Scytales]

I Googled and its AI spit out this statement:

ESS Sabre DACs (e.g., ES9039PRO, ES9028Q2M) utilize patented Hyperstream IV/QUAD modulator architecture to process DSD (up to DSD1024 or DSD-22.6MHz). These chips internally convert DSD to a 6-bit or 32-bit multipath digital domain, rather than pure "Direct DSD," allowing DSD volume control and digital filtering.

ESS has been hush hush about how they handle DSD, but this Hyperstream IV modulation is undesired. As PEQ and volume control are being used by OP, I would suggest just accepting PCM conversion and enjoying the music. This DAC is not capable of bit perfect DSD performance.

What it is you seek by asking an AI? This quote from Google, which to my mind in as uninformed as Grok's (ESS's Hyperstream IV modulation undesired, really?) has no relevance to the interfacing problem you are confronted with.

 

[drewdawg999]

What it is you seek by asking an AI? This quote from Google, which to my mind in as uninformed as Grok's (ESS's Hyperstream IV modulation undesired, really?) has no relevance to the interfacing problem you are confronted with.

I didn't directly ask AI but now when you Google anything it's the first thing that pops up. It's pretty well known that ESS doesn't support bit perfect DSD, which is what the OP wants. Other websites measure for these things and the list of DACs that support this feature is small. This DAC is not on the list.

 

[antcollinet]

My impression is that every time there is a conversion a little noise and/or distortion is added to the data depending on the quality of the converter. I don't know what I will hear since I haven't had the chance to listen to any comparisons yet. Do you think multiple data format conversions can have an effect?

Not audibly. Any imperfections in (correctly done) conversions are way way below the level of audibility.

 

[AUDACC]

I'm thinking you have a wrong setting somewhere, others have gotten it to work. I got my DO100 to work in native DSD, but it doesn't make a difference on an ESS DAC, no Direct DSD path. So the DSD and PCM will sound the same.

DSD when properly decoded doesn't have volume control, and can't have EQ applied. That would defeat the purpose of the PEQ feature, which is one of this DAC's main selling points. I believe Topping couldn't care less about proper DSD decoding, since it's pretty rare that anyone actually cares. On the E70 Velvet they chose not to implement DSD Direct, even though it's on the AKM chip.

Since I am using the volume control and PEQ on the D50iii, there is apparently no hope for hearing the supposed advantages of DSD. Attempting to play a DSD file through the DSD input function on my D50iii still generates the "352.8khz not supported" error with the fb2k component versions recommended by Topping support installed. Testing the DSD input function with the D50iii volume control disabled might blow a tweeter. It is becoming apparent that to hear what DSD has to offer, its time to buy a D200.

 

[drewdawg999]

Since I am using the volume control and PEQ on the D50iii, there is apparently no hope for hearing the supposed advantages of DSD. Attempting to play a DSD file through the DSD input function on my D50iii still generates the "352.8khz not supported" error with the fb2k component versions recommended by Topping support installed. Testing the DSD input function with the D50iii volume control disabled might blow a tweeter. It is becoming apparent that to hear what DSD has to offer, its time to buy a D200.

Have you tried it with PEQ disabled?
The D200 is excellent and does what you are looking for without fuss. I posted some stuff about bit perfect DSD in the review thread here:

SMSL D200 DAC Review

True. One day you'll get it. I head you. If I set the volume on my headphones to exactly the same level at 1 kHz, all DACs will sound exactly the same. Any differences are simply a matter of imagination, mood, and expectation bias. Thank you.

www.audiosciencereview.com

Well it's up to you whether it's worth it to have a DAC just for DSD, depending on your library. It's somehow weaseled its way into my heart despite the gargantuan file sizes, but for some reason I put up with it. Do you need balanced outputs? If not, the D1 is cheaper and no frills.

 

[Scytales]

I am a bit lost, now.

I understand the desire to get a "bit perfect" interface to transmit DSD data without alteration, because it let you do whatever you want with the data-stream in the receiving end without being stuck with what has been done at the sending end.

But the DAC? What is exactly a "bit perfect" DAC for DSD? A DAC that convert the 1s and 0s directly in squarish pulse signal of opposite direction and that's it?

If so, to my knowledge, there is actually very few DAC chips that do that: maybe some old and sometimes obsolete Cirrus Logic, Niigita Semitsu and Wolfson designs from the early 2000s such as CS4391 and CS4392 and of course the venerable Philips TDA1547 which was used once in the first Marantz SACD player (but I suspect even some of those might over-sample the DSD64 data with a zero order hold function to match the actual rate at which their respective DAC output stages operate, which is usually higher than the 64x rate of DSD data).

Even Sony and Philips didn't dispense with some discrete time signal processing on DSD data in their respective very first SACD players in order to improve performance of out of band noise filtering.

As far as the DAC listed as "bit perfect" for DSD on the audiophilestyle forum refered to in the above thread about the SMSL D200 are concerned, I do not know discrete implementations, but all the listed DAC devices that make use of any AKM, Burr Brown (Texas Instruments) from PCM1738 and higher and Rohm DAC chips perform digital computation on the DSD data in order to implement a low-pass filter at their actual DAC stages. For instance, Burr Brown described this mode of operation in a patent application. Are those DACs considered "DSD bit perfect" anyway?

The only off the shelf DAC chips of my knowledge that can converts 1s and 0s as is (notwithstanding probable over-sampling with a zero order hold function) are Cirrus Logic CS4398 and CS43198, which have an optional "direct path" that let the DSD data stream drive their multilevel DAC output stages directly, in which case said output stages will obviously produce only two voltage levels.

 

[AUDACC]

Have you tried it with PEQ disabled?
The D200 is excellent and does what you are looking for without fuss. I posted some stuff about bit perfect DSD in the review thread here:

SMSL D200 DAC Review

True. One day you'll get it. I head you. If I set the volume on my headphones to exactly the same level at 1 kHz, all DACs will sound exactly the same. Any differences are simply a matter of imagination, mood, and expectation bias. Thank you.

www.audiosciencereview.com

Well it's up to you whether it's worth it to have a DAC just for DSD, depending on your library. It's somehow weaseled its way into my heart despite the gargantuan file sizes, but for some reason I put up with it. Do you need balanced outputs? If not, the D1 is cheaper and no frills.

Indeed, disabling the PEQ might allow the D50 to accept DSD native files. I’ll try disabling the PEQ.

The next impediment to hearing the true DSD sound is the D50 volume control. My smallish (10x10ft) listening room requires a low volume setting for the signal going to my Buckeye amp. The volume control comparison in the D200 review, https://www.audiosciencereview.com/forum/index.php?threads/smsl-d200-dac-review.65764/post-2464447 , shows the D50’s volume control gives up 2-3 bits resolution at the -20db to -30db volume settings I use to save my hearing. The solution might be to use the Topping DAC with no PEQ at -0db volume with a ~ 50 watt inexpensive transparent amp with a built-in quality volume control. Any suggestions based on experience would be appreciated.

 

[drewdawg999]

Okay I'm not so technical, but I believe the confusion stems from the definition of "bit perfect". As far as I know, the DACs on the list on Audiophilestyle all have the analogue reconstruction filter. But that's the only filter applied, only one allowed for bit perfect DSD. The list has been confirmed by measurements, the user Miska has an analyzer and has posted results.

Rohm chips do have the direct DSD path, it's the only way they decode DSD which is nice, saves on confusion and fuss. To my ears there's an audible difference between PCM and Direct DSD. Whether or not it's more analogue or more desirable comes down to personal preference.

 

[drewdawg999]

Indeed, disabling the PEQ might allow the D50 to accept DSD native files. I’ll try disabling the PEQ.

Yes please try and report back the results.

The next impediment to hearing the true DSD sound is the D50 volume control. My smallish (10x10ft) listening room requires a low volume setting for the signal going to my Buckeye amp. The volume control comparison in the D200 review, https://www.audiosciencereview.com/forum/index.php?threads/smsl-d200-dac-review.65764/post-2464447 , shows the D50’s volume control gives up 2-3 bits resolution at the -20db to -30db volume settings I use to save my hearing. The solution might be to use the Topping DAC with no PEQ at -0db volume with a ~ 50 watt inexpensive transparent amp with a built-in quality volume control. Any suggestions based on experience would be appreciated.

Or you could insert a preamp. I have a Topping Pre90 and love it. But even then, this DAC using an ESS chip, will process DSD through the same path as PCM. It won't have the true DSD sound even if you get it to accept a native DSD signal. I don't think it's worth the hassle with this DAC, you should enjoy the PEQ. When I A-B on my DO100, there's no difference between PCM and DSD.

The D200 is what you want though for DSD, with the added bonus of high quality analogue volume control. If you have the budget for it, it's great. But the D50 III is great for PCM, plus outstanding feature in PEQ. My suggestion is to just enjoy what you have, the difference is really very subtle.

 

[AUDACC]

"Yes please try and report back the results."

Disabled the PEQ, but still the same "352.8khz not supported" error. I'll just stick with the D50 with PCM until more D200 reviews come in on ASR that might justify getting one to play with for myself. The online file converters offer the ability to convert the DSD128 files to 24bit 96khz. Playing that will be an interesting test.