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

[jkim]

The Topping D30Pro has a filter setting of 5, Non-Oversampling. Does that mean there are no more problems with this setting?

Yes Most likely. All DACs supporting NOS on my test bench behave the same way.

 

[jkim]

Isn’t this a bad design and/or bad implementation from Cirrus Logic and their OEMs?

Right, this must definitely not be an end product they intended.

The solution is to switch NOS on, which turns DRE off, and to upsample your tracks upstream of the device. In other words, “dumbing down” the CS chip is the fix…

Yes, that is a proper view. The concept of DRE itself is nice, but Cirrus Logic's implementation or their choice of default parameter values did not pan out well.

 

[MCH]

The Topping D30Pro has a filter setting of 5, Non-Oversampling. Does that mean there are no more problems with this setting?

I am very curious to know if the D30pro shows any of these issues at all. Unlike some of their direct competitors, topping tends to do things a bit differently and not just follow the datasheets. I will be surprised if they didn't notice these issues. Unfortunately johnyang doesn't post much anymore.

 

[Swin]

As this is mostly about portable devices, are there a good upsampler app for Android?

 

[HeebieGenie]

Yes. With NOS, there's no Cirrus hump. Upsampling will help lower adverse NOS effects. 192 kHz should be adequate.


Can you find that review for us?

Does this apply to the CS43198 as well?

 

[jkim]

I am very curious to know if the D30pro shows any of these issues at all. Unlike some of their direct competitors, topping tends to do things a bit differently and not just follow the datasheets. I will be surprised if they didn't notice these issues. Unfortunately johnyang doesn't post much anymore.

I doubt Topping knew this. This issue is tricky enough.

 

[jkim]

Does this apply to the CS43198 as well?

Yes.

 

[mc.god]

Only if it supports selection of NOS. It seems that some DACs, like Hiby FC5, do not have that option.

KP Player 3 supports NOS

I took the test and can confirm that in NOS mode the clicks on C-Maj at native 44k become inaudible, while they are clearly there cycling through all other 4 filters.

Oddly, if i set Pipewire to resample at 192k instead of native rate, the file produces a couple of strong pops during the second and third iteration of the signal, while the clicks are still not present in NOS mode. If i set a 96k it pops only at the second iteration, while setting at 384k there are pops in all the three iterations. These loud pops are there no matter which filter is selected, but maybe it's just some misconfiguration in pipewire that I have no intention to investigate now since usually it runs at native rate and all is ok.

 

[VyseLegendaire]

Thanks a lot for the discussion on this. I recently bought Black Pearl and was so happy with the sound quality, only to find out that there is this hidden issue - which admittedly I didn't actively notice before hearing the specific test tracks - which as people who value our music listening time take very seriously either way.

Observations from my experiments:

- Turning on NOS filter did eliminate the continuous background clicking, thankfully

- In the below listed programs I did test out the tracks and some music using the NOS with system audio at 44 and 48Hz, and did noticeably hear 'static/fuzz/white noise' like an old record player on top of songs which didn't have it before with my usual rates (32-Bits, 48 or 96K)

- last night before reading jkim's recent posts explaining this pictorally, I tested out upsampling to the max settings (32-bit, 384K) and did notice that the 'white noise' I mentioned was noticeably less or gone. Keep in mind its MUCH easier to hear it on a quiet, high-res piece with solo piano recorded live in a hall or recording studio (rather than direct line recording) than with any other type of music due to the captured ambience and 'headroom' in the recording itself.

-but I also noticed some applications will or will not play the "beginning 3 pops" depending on audio settings within the program itself.

For instance, I opened C maj. test and Song of the Sea in 3 programs:
Mpv.net (latest revision)
MPC-HC
Foobar2000
VLC

-After upsampling in Windows 10 to the max available from the Black Pearl - 2 Channel, 32-bit, 384000 Hz - all clicking sounds were removed, and treble NOS noise nearly 100% reduced, except in MPC-HC I could still hear the '3 beginning-pops' minus the continuous clicking. I'm not sure of this, but It seems to be analogous to the 3-pops measured in Method 2 that Jkim used in the original tests to discern click/DRE behaviour in 'non-continuous click' devices such as Jcally Max 20, etc.

So if you did follow the above steps, were able to remove the continuous distortion but still heard the '3 starting pops' sounds, I found this setting which removed that within MPC-HC:

Options --> Internal Filters --> Audio Switcher --> [Off] Enable built in audio switcher filter [requires restart]

-After doing this change, now every program listed above appears to be totally free of the pop/clicking sounds that I can tell.

Its important to run the test tracks on whichever audio player you're using in order to determine whether that program's own conditions are causing the particular behavior or not.
At least this seems like proof that the CS chip will react differently with the same audio files, DAC filters, and systemwide audio settings if the audio preferences within that program are hijacking or diverting playback somehow.

If someone runs into the same issue with another program with another CS chip test instance, feel free to share this and maybe someone can help you find an app-specific setting causing the chips to behave differently within that instance.

 

[HeebieGenie]

With some helpful guidance from @jkim, here's how my Moondrop Dawn Pro dongle (CS43131) performs with a dual-tone (41 Hz/89 Hz) test signal.

View attachment 456435

I'm pretty sure the Dawn Pro has a NOS mode.

When the LED indicator blinks twice, it goes into a "Default" mode where you see that no filter is selected in the app. So this dongle DAC can avoid the Cirrus Hump.

 

[jkim]

Major update on the effect of NOS (non-oversampling) mode on the Cirrus hump distortion

A complete set of measurements was performed and added to this post. Now we have a much better understanding of the relationship between DRE and Cirrus hump distortion.

 

[pangcukaipang]

currently i could only give the android app right now. since the windows version is a python script and sorta kinda made windows defender unhappy and deleted it.(idk why, maybe it doesnt like seeing code that interacts with usb devices). ill try to recover or remake it.

Content not found | LimeWire

Content not found on LimeWire

limewire.com

maybe ill just post them in github if i ever stopped procrastinating
edit: heres the source code for the android app: https://github.com/cheesyserg/blackPearlControl-Android

Thanks, it's working.

---

I wish there's a dedicated thread for TRN Black Pearl. There's some question regarding the Walk Play app for Black Pearl. The app layout for “DAC working mode” on Walk Play android app is different from the web app. If I select Class AB on Walk Play app on my android then switch to my laptop and using it on the web app, it switched to Class H and vice versa. I don't know which one is the right one, lol.

But, when I tried @KDRN android app, it didn't switch. So, when I select Class AB on @KDRN app on my android then switch to my laptop and using it on the web app, it stays on Class AB.

 

[pangcukaipang]

I also have TempoTec Variations V1 DAP, which also uses Dual CS43131, on the way. I'm curious about its performance regarding this Cirrus Hump issue.

 

[Theriverlethe]

I guess I won't order a JM20 for my desktop.

 

[VyseLegendaire]

Thanks, it's working.

---

I wish there's a dedicated thread for TRN Black Pearl. There's some question regarding the Walk Play app for Black Pearl. The app layout for “DAC working mode” on Walk Play android app is different from the web app. If I select Class AB on Walk Play app on my android then switch to my laptop and using it on the web app, it switched to Class H and vice versa. I don't know which one is the right one, lol.

But, when I tried @KDRN android app, it didn't switch. So, when I select Class AB on @KDRN app on my android then switch to my laptop and using it on the web app, it stays on Class AB.

In my case, I was totally unable to get the Android app to work beyond the login screen at all since I'm using an old Pixel 3XL with no capability to run these DACs that require above USB 1.0 backward compat. (which Black Pearl lacks), so I've use the Android app to create an account, then the web app only to log in. It seems settings remain as they should once disconnected/reconnected several times if you use the web app only. In my case of course it means I'm only using the Black Pearl for desktop use.

Since the BP almost certainly has an on-board EProm memorizing the settings, its probably a good idea to configure on PC and just ignore the Android app altogether. I don't know how upsampling to 192 or 384k would be done an Android though, probably via your playback engine (USB Audio Player Pro?)

 

[MBO]

Major update on the effect of NOS (non-oversampling) mode on the Cirrus hump distortion

A complete set of measurements was performed and added to this post. Now we have a much better understanding of the relationship between DRE and Cirrus hump distortion.

What do you think? Maybe it's time to contact Cirrus officially? Maybe Amir will write a serious letter? This could greatly help Cirrus to take these errors into account. Maybe they will make recommendations to manufacturers to avoid the problem?

 

[VyseLegendaire]

Thanks, it's working.

---

I wish there's a dedicated thread for TRN Black Pearl. There's some question regarding the Walk Play app for Black Pearl. The app layout for “DAC working mode” on Walk Play android app is different from the web app. If I select Class AB on Walk Play app on my android then switch to my laptop and using it on the web app, it switched to Class H and vice versa. I don't know which one is the right one, lol.

But, when I tried @KDRN android app, it didn't switch. So, when I select Class AB on @KDRN app on my android then switch to my laptop and using it on the web app, it stays on Class AB.

In my case, I was totally unable to get the Android app to work beyond the login screen at all since I'm using an old Pixel 3XL with no capability to run these DACs that require above USB 1.0 backward compat. (which Black Pearl lacks), so I've use the Android app to create an account, then the web app only to log in. It seems settings remain as they should once disconnected/reconnected several times if you use the web app only.

me regardless of the

What do you think? Maybe it's time to contact Cirrus officially? Maybe Amir will write a serious letter? This could greatly help Cirrus to take these errors into account. Maybe they will make recommendations to manufacturers to avoid the problem?

Likely Cirrus says that the large sell-through and popularity of their chips is proof of the correctness of the decision to implement the DRE algorithm. That and most listeners won't be able to notice the problem under normal circumstances so who cares, Apple has bought into it so how can it be bad anyway?

 

[MBO]

In my case, I was totally unable to get the Android app to work beyond the login screen at all since I'm using an old Pixel 3XL with no capability to run these DACs that require above USB 1.0 backward compat. (which Black Pearl lacks), so I've use the Android app to create an account, then the web app only to log in. It seems settings remain as they should once disconnected/reconnected several times if you use the web app only.




Likely Cirrus says that the large sell-through and popularity of their chips is proof of the correctness of the decision to implement the DRE algorithm. That and most listeners won't be able to notice the problem under normal circumstances so who cares, Apple has bought into it so how can it be bad anyway?

They may say so, but they'll still take note next time.

 

[Rotiv]

Excellent work, congratulations.
I have TANCHJIM Space and through the app I selected non-oversampling (NOS). In UAPP, settings, USB Audio, authorize "increase to highest rate," rate relative to the DAC's capabilities (TANCHJIM Space 384khz)
Will this do the job?

 

[Theriverlethe]

Has anyone done an ABX with the "good" units, such as the JM20?