On the Distortion of Cirrus Logic CS431xx-Based Devices: A Comparative Review

[jkim]

Just try the procedure except the "-15dB step" since I do not know how to set it, on iBasso DC07 Pro (4xCS43131), with Fast filter, I do not hear any clicking noise along with the low-frequency sound.

So, you used the volume control on the iBasso. Right? What headphones/IEMs did you use? Which connection (3.5mm or 4.4mm) did you use?

Here's my suggestion. Use relatively sensitive IEMs connected to the 3.5mm output. Use player software that supports volume control on dB scale. Set it to -15 to -14.5 dB.. Set the operating system's volume as well as the iBasso's volume to maximum.

 

[jkim]

Some thoughts after some time thinking and experimenting with this;

A. its a mistake I think to assume DAC/AMP makers did not know about the CS distortion issue. I don't have sources but its a known phenomenon I have read about in places for years. Likely they just believed it would go ignored or considered it too minute compared to other chip characteristics considerations (basically, there is a near universal concensus that the CS43131 is 'the chip' for marketing purposes)

And its nearly 100% for sure that the behavior of some CS devices minimizing the click sounds to the beginning of the audio clip is also an engineering goal and not a coincidence.

I understand what you mean although my thought is the opposite. These engineers must have simply designed these products using their standard design flow. Everything is our pure conjecture for now, though.

B. Why does it matter to engineer this away if we can just lock the device to or manually choose NOS filter?

Noise floor and Dynamic Range are audible characteristics in music, and it shows. Nearly all the audio A-B tests I've done give me better listening results with the standard filters compared to NOS, taking into account the potential background distortion.

In other words, the activation of the DRE feature actively improves signal clarity/ reduces background noise, and boosts dynamism in any given music sample. An example would be a high-res piano solo music where with NOS, audible recording background noise is audible increased (with 364k upsampling active), or any given rock song featuring clearly strum guitar chords - they will lack initial attack and articulation and/or total power in comparison to standard shape filters. These examples seem to apply nearly universally with slightly different interpretations for any type of music. Basically, noise goes down and music signal sounds stronger with DRE active.

If you're talking about the TRN Black Pearl, I doubt you can hear the dynamic range / noise performance difference b/w standard filters and NOS. Both are well beyond the audible threshold, as tested here. Unless you use a very, very (unrealistically) sensitive IEM, I bet you won't hear a difference. In fact, actual audio recordings, even hi-res, can barely achieve a 20-bit noise floor---consider microphone noise which can only be lowered by post-processing. The Black Pearl's noise performance in its NOS mode is more than adequate and will be masked by audio content' real, effective resolution.

A potentially bigger issue with NOS would've been the effect of no oversampling, which should be taken care of by the playback side when NOS is used on the DAC side.

The CS431xx's analog, thermal noise level is still decent. It does not need DRE for high quality audio playback. Chips in cheaper dongles have substantially higher noise levels. Still, I would avoid DRE in those chips, too. Because DRE seems to cause all kinds of peculiar distortions in exchange for lowered noise. See this and this.

 

[roladyzator]

@jkim Did you reach out to FiiO support about this?

I've had some luck with them when I complained about loud pops over 4.4mm jack when connected to active speakers, every time sound was stopped for few seconds.
They have addressed this in 1.0.4 FW (well, it's still there, you just have to wait few minutes now, which is why I have a tone generator at 20 kHz and -100 dBFS in the backround when using speakers - it never shuts down the output).

 

[Gitfiddle]

I saw that the moondrop dawn pro was on the naughty list here and decided to give mine a try with your C major file. I set it to max volume and using voicemeeter banana (windows 11, wasapi exclusive, no EQ) I dropped the db to -15.

Listening with truthear gate IEM's I did not hear any audible distortion on any of the moondrop dawn pro filters. I tested this straight from the device's 3.5mm out. I expected to hear distortion so I then added a headphone amp in the thick of it and cranked up the volume. Still nothing.

Perhaps later I'll do more and test at different DB levels.

I did force distortion myself with some EQ settings, and doing that I get the clicks through each of the entire C major phases. So I know what to listen for.

 

[roladyzator]

I was able to hear it with my KA15 with the CMajor file using the following setup:

Headphones used: Beyerdynamic MMX 100.

1. Volume maxed out on KA15, set to high gain.
2. Device volume set to -15 dB in Windows. (If you use Sound Control panel, you can switch to dB view when adjusting device volume).
3. Playing the file in a browser tab.

The issue manifested as audible "clicks" accompanying the bass notes.
Switching to NOS filter on the KA15 got rid of those.

 

[Ahmonge]

I repeated the test playng Original_CMaj file with Topping D30 Pro in preamp mode, at -15dB and -22dB volume levels, deafult F2 Fast roll off minimum (default) filter and I've got the same result than in DAC mode: just two clicks, in the middle of he second and third periods of bass sound.

 

[jkim]

I saw that the moondrop dawn pro was on the naughty list here and decided to give mine a try with your C major file. I set it to max volume and using voicemeeter banana (windows 11, wasapi exclusive, no EQ) I dropped the db to -15.

Listening with truthear gate IEM's I did not hear any audible distortion on any of the moondrop dawn pro filters. I tested this straight from the device's 3.5mm out. I expected to hear distortion so I then added a headphone amp in the thick of it and cranked up the volume. Still nothing.

Perhaps later I'll do more and test at different DB levels.

I did force distortion myself with some EQ settings, and doing that I get the clicks through each of the entire C major phases. So I know what to listen for.

You should hear it. The Moondrop Dawn Pro was tested by @cyril_meysson and reported here.

 

[Gitfiddle]

You should hear it. The Moondrop Dawn Pro was tested by @cyril_meysson and reported here.

Yes that is precisely why I expected to hear it, I saw those measurements. I'll do more testing later today.

 

[Gitfiddle]

So to follow up, I got in a room with ambient noise at <30 DB and tested again. Results went as expected this time when switching between filter 1 and no filter.

Distortion happens with the C major test when I take Voicemeeter Banana out of the equation on the moondrop dawn pro; except when no filter is selected. Very easily audible this time with the first filter selected.

With voicemeeter in the loop the distortion does exist, but I have to turn up the volume to levels I would never want to listen at. Again not having a filter selected solves the distortion as far as I can tell at all volumes tested.

Very neat results. I can go back to using this device as I always have I guess. No filters and WASAPI exclusive.

 

[jkim]

No filters and WASAPI exclusive.

WASAP exclusive selected in what software you mean?

 

[Gitfiddle]

WASAP exclusive selected in what software you mean?

Voicemeeter banana. I use it for EQ as well.

 

[markanini]

Voicemeeter banana. I use it for EQ as well.

Last time I tried it prevented windows automatic resampling to work normally, causing resampling artifacts. Disabling it should really be the default for any critical audio evaluations.

 

[Gitfiddle]

Last time I tried it prevented windows automatic resampling to work normally, causing resampling artifacts. Disabling it should really be the default for any critical audio evaluations.

The entire point is to remove windows from the equation altogether. VM has its own built in resampling. Using the test samples on the thread below I heard nothing out of the ordinary. On my setup everything seems okay.

T

'Resampling artifacts'

Jul 31, 2020

Is there a test file that is specifically designed to reveal resampling artifacts? I want to know if my Android phone resampling from 44.1 kHz to 48 kHz is added any distortion or not.

• T3RIAD
• Replies: 4
• Forum: General Audio Discussions

• 

 

[testp]

The entire point is to remove windows from the equation altogether. VM has its own built in resampling. Using the test samples on the thread below I heard nothing out of the ordinary. On my setup everything seems okay.

T

'Resampling artifacts'

Jul 31, 2020

Is there a test file that is specifically designed to reveal resampling artifacts? I want to know if my Android phone resampling from 44.1 kHz to 48 kHz is added any distortion or not.

• T3RIAD
• Replies: 4
• Forum: General Audio Discussions

• 

could it be banana is lowering the signal giving lower sinad, so no clicks are heard anymore?

 

[Gitfiddle]

could it be banana is lowering the signal giving lower sinad, so no clicks are heard anymore?

I don't think I have an audio interface to test this kind of stuff, so I'll say a definite maybe. It would be neat to see if voicemeeter does have problems with sampling, I guess I could make the switch to EQ apo, I've simply never needed to. Here is the thread explaining why I want to bypass windows in the first place by using WASAPI exclusive.

'Ending the Windows Audio Quality Debate'

Jan 17, 2021

A debate has been long raging in the audio community that Windows isn't fit for use in a high resolution audio system. Sceptics counterargued that bits are bits and Windows audio degradations are a figment of an over-active audiophile imagination.

I decided to investigate this as I kept hearing issues with Windows, even after following common practice to (controversially) supposedly make it clean. I found some surprises along the way.

Sorry for the long post but my motivation is to share how to get effectively perfect audio from Windows, to save folks money and headaches, and to help...

• DDF
• apo audio upsampling caudiolimier digital filter intersample over pc audio remote windows windows 10
• Replies: 848
• Forum: Home Music Servers, Computers and Streamers

• 

 

[jkim]

I don't think I have an audio interface to test this kind of stuff, so I'll say a definite maybe. It would be neat to see if voicemeeter does have problems with sampling, I guess I could make the switch to EQ apo, I've simply never needed to. Here is the thread explaining why I want to bypass windows in the first place by using WASAPI exclusive.

'Ending the Windows Audio Quality Debate'

Jan 17, 2021

A debate has been long raging in the audio community that Windows isn't fit for use in a high resolution audio system. Sceptics counterargued that bits are bits and Windows audio degradations are a figment of an over-active audiophile imagination.

I decided to investigate this as I kept hearing issues with Windows, even after following common practice to (controversially) supposedly make it clean. I found some surprises along the way.

Sorry for the long post but my motivation is to share how to get effectively perfect audio from Windows, to save folks money and headaches, and to help...

• DDF
• apo audio upsampling caudiolimier digital filter intersample over pc audio remote windows windows 10
• Replies: 848
• Forum: Home Music Servers, Computers and Streamers

• 

The point of that article is that we CAN rely on Windows audio processing including its resampling. I also tested the quality of Windows upsampling to that of the PGGB-RT Real Time Upsampling add-on for foobar2000. There was a negligible difference.

 

[Gitfiddle]

The point of that article is that we CAN rely on Windows audio processing including its resampling. I also tested the quality of Windows upsampling to that of PGGB-RT Real Time Upsampling add-on for foobar2000. There was a negligible difference.

as the article suggests: "Install Equalizer APO and use it to disable original APOs and set EAPO's preamp gain at ~ -4 dB to avoid upsample overs, filtering induced peaks, and the Windows CAudioLimiter." I just do this a different way by using voicemeeter banana to set it as WASAPI exclusive. That is all.

I never had a problem with windows sampling and I never meant to imply that. I simply never had any audible problem with voice meeter's sampling either.

 

[markanini]

The entire point is to remove windows from the equation altogether. VM has its own built in resampling. Using the test samples on the thread below I heard nothing out of the ordinary. On my setup everything seems okay.

T

'Resampling artifacts'

Jul 31, 2020

Is there a test file that is specifically designed to reveal resampling artifacts? I want to know if my Android phone resampling from 44.1 kHz to 48 kHz is added any distortion or not.

• T3RIAD
• Replies: 4
• Forum: General Audio Discussions

• 

Your setup, Voicemeeter into a VM, is novel and any credible claims of quality improvements, or even quality parity, need be to be validated. It's very easy to Introduce unforeseen quality degradations when working off assumptions. The performance of Windows resampler on the other hand has been well studied and serves users even with high standards, it's expected from the use of a textbook resampler implementation.

Are you asking others to believe that you have control over how Voicemeeter signals sample rate changes to both the OS and the host audio device? Voicemeeter is closed source software so unless you have exclusive insight it's not very likely.

 

[jkim]

as the article suggests: "Install Equalizer APO and use it to disable original APOs and set EAPO's preamp gain at ~ -4 dB to avoid upsample overs, filtering induced peaks, and the Windows CAudioLimiter." I just do this a different way by using voicemeeter banana to set it as WASAPI exclusive. That is all.

I never had a problem with windows sampling and I never meant to imply that. I simply never had any audible problem with voice meeter's sampling either.

Your setup, Voicemeeter into a VM, is novel and any credible claims of quality improvements, or even quality parity, need be to be validated. Simply assuming that the quality must be better is a sure way to introduce unforeseen quality degradations. The performance of Windows resampler on the other hand has been well studied and serves users even with high standards, it's expected from the use of a textbook resampler implementation.

Are you asking others to believe that you have control over how Voicemeeter signals sample rate changes to both the OS and the host audio device? Voicemeeter is closed source software so unless you have exclusive insight it's not very likely.

When I get a chance, I will run a quick test of the Voicemeeter's upsampling effect as it provides a 30-day trial.

 

[VyseLegendaire]

I understand what you mean although my thought is the opposite. These engineers must have simply designed these products using their standard design flow. Everything is our pure conjecture for now, though.


If you're talking about the TRN Black Pearl, I doubt you can hear the dynamic range / noise performance difference b/w standard filters and NOS. Both are well beyond the audible threshold, as tested here. Unless you use a very, very (unrealistically) sensitive IEM, I bet you won't hear a difference. In fact, actual audio recordings, even hi-res, can barely achieve a 20-bit noise floor---consider microphone noise which can only be lowered by post-processing. The Black Pearl's noise performance in its NOS mode is more than adequate and will be masked by audio content' real, effective resolution.

A potentially bigger issue with NOS would've been the effect of no oversampling, which should be taken care of by the playback side when NOS is used on the DAC side.

The CS431xx's analog, thermal noise level is still decent. It does not need DRE for high quality audio playback. Chips in cheaper dongles have substantially higher noise levels. Still, I would avoid DRE in those chips, too. Because DRE seems to cause all kinds of peculiar distortions in exchange for lowered noise. See this and this.

You're probably right, but tbh I can't think of why my ears felt the differences. I know that the filters used for DACs impact the perceived sound, at least theoretically (On Black Pearl Walk Play app interface, the digital filter graph choice pages tries to describe the impact on the sound of each graph type). But I couldn't see how those could change my sense of the tonality in such a clear way, including altered background noise and generally increased power of the music. Or maybe they do. Would this possibly indicate that NOS mode is contributing to a 'higher fidelity' sound signature than any of the shaping effects can - as they actually altering the perceived musical content due to their shaping effect?

These are only questions and I'm rather new to this entire field having just started this year.