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

[Amperage]

"This DAC contained in our product has a huge flaw that is in the hardware itself and we cannot fix it, sorry guys"

Well, depending on the market's consumer protection laws, that could very well warrant a refund - for consumers, and even the OEM if it's Cirrus at fault.

 

[dmitrykos]

Calling it "Lo-Fi" is too harsh. I think the end result is audibly fine. I would still prefer the devices with mitigated distortion from less aggressive DRE, though, like the JCally JM20 or JM20 Max.

Maybe I am a bit harsh with my assertion in relation to CS but ESS and AKM chips show excellent, flawless performance in this respect.

By the way, while doing measurements with my Beats test file you demonstrated results from device with Cirrus Humps and without. In case of JM20, although it does not have Cirrus Hump detected, the behavior of the noise shaping has distinctive feature - it fades out very quickly in comparison with Stargate II. Based on your Spectrogram screenshots I would say, around 3 times faster. So DRE is still On but probably another register setting increased speed of noise shaping fade-out and thus probably it is possible catch Cirrus Hump effect on those devices but with different dual-tone signal frequencies (or it was a workaround against Cirrus Hump).

After all, the design must have been somewhat complex. I believe the designer may have thought the end product would be an optimal tradeoff for the chip's proposed applications.

Nevertheless, they really need to remove "pro audio" from Applications from the Specification based on revealed performance deficiencies. It is interesting to see whether distortion exist if DRE is Off, if it still exists then the design of the chip is really flawed. If with DRE = Off there is no distortion then it can be used but who needs 50 kHz audible range and 94-100 dB DNR in modern devices?..

 

[MCH]

I am probably building a DIY cs43131 headphone amp for one of my projects, if I find schematics to add an output op-amp, then might have the chance to test what settings make what.
Thanks for gathering all this information, will be useful.

About to abandon the idea. The plan was to use a Meizu hifi non pro I have as donor and capture the i2c communication with the bridge before the transplant, but:
- I have checked the Meizu carefully and the cs43131 inside is the solder ball variant: 42 pins in a 3x2.7 mm footprint . On top of that, seems that the PCB has a conformal coating, that could make things even more difficult.
- The usb bridge is CS46L41, also solder ball package, what means that the I2c tracks between the two are internal and cannot be tapped.

Maybe I try to get a new cs43131 in qfn package if I see it cheap in Ali, but for the current project I will probably use something completely different, what a bummer, this was interesting to tinker with....

 

[jkim]

50+ kHz range is still important, although not audible, if user is using USB DAC for tasks not directly related to generic music listening, for example for some experiments.

Nevertheless, they really need to remove "pro audio" from Applications from the Specification based on revealed performance deficiencies.

By the way, I wonder in what "pro audio applications" clean response beyond 50 kHz would be important. Would you clarify?

 

[dmitrykos]

By the way, I wonder in what "pro audio applications" clean response beyond 50 kHz would be important. Would you clarify?

If we are talking about applications then it can be useful for different measurements and scientific experiments where you need clean spectrum above 50 kHz. For music listening of course we are limited by frequency response of the headset/speakers and human hearing and 50 kHz is sufficient. On the other hand, if comparing implementations by different IC vendors then Pro becomes that device which does its work better, just like athlete in competition - you are Pro when you do your discipline better. For example competitor ESS's ES2919Q demonstrates clean analog spectrum, has no distortions, can reach 121 dB DNR without DRE, has similar chip size, it does its work better overall then it becomes Pro over similar CS DAC ICs.

 

[jkim]

If we are talking about applications then it can be useful for different measurements and scientific experiments where you need clean spectrum above 50 kHz.

Right, for some academic purposes, for instance, a study on bats' hearing.

On the other hand, if comparing implementations by different IC vendors then Pro becomes that device which does its work better, just like athlete in competition - you are Pro when you do your discipline better. For example competitor ESS's ES2919Q demonstrates clean analog spectrum, has no distortions, can reach 121 dB DNR without DRE, has similar chip size, it does its work better overall then it becomes Pro over similar CS DAC ICs.

I agree that if you compare the CS43131 to the ES9219, then the latter is a better SoC. But I wonder why there are greater variations in THD performance (not including noise) of ES9219-based DACs whereas most CS431xx-based DACs perform very similarly. What would be the primary design factor in harmonic distortion performance of ES9219-based DACs? Your Neutron HiFi v1 has the lowest harmonic distortions I've seen among the DACs adopting ES9219.

That is, we are comparing this:

To this:

From a measurement point of view, this is a huge difference.

EDIT. By the way I recently measured both revisions of the Qudelix 5K with ES9218P and ES9219C. They measured the same and matched Amir's results.

 

[mc.god]

Right, for some academic purposes, for instance, a study on bats' hearing.


I agree that if you compare the CS43131 to the ES9219, then the latter is a better SoC. But I wonder why there are greater variations in THD performance (not including noise) of ES9219-based DACs whereas most CS431xx-based DACs perform very similarly. What would be the primary design factor in harmonic distortion performance of ES9219-based DACs? Your Neutron HiFi v1 has the lowest harmonic distortions I've seen among the DACs adopting ES9219.

That is, we are comparing this:

To this:

From a measurement point of view, this is a huge difference.

EDIT. By the way I recently measured both revisions of the Qudelix 5K with ES9218P and ES9219C. They measured the same and matched Amir's results.

Could ES9219 performance possibly be more dependent on the used usb bridge?

 

[dmitrykos]

@jkim I replied to your question in DAC V1's thread to avoid polluting CS thread:

Neutron HiFi DAC V1 Review

ASIO driver is not provided but you can use FlexASIO (https://github.com/dechamps/FlexASIO) configured to use Exclusive WASAPI. WASAPI backend is provided by PortAudio (https://github.com/PortAudio/portaudio/) and it is currently very stable and allows low-latency operation. Could you clarify...

www.audiosciencereview.com

Could ES9219 performance possibly be more dependent on the used usb bridge?

There is no direct effect but can be influenced by several factors:

• DAC chip operates in Slave mode and MCU/bridge has basic clock source and is additionally causing some phase jitter with its own PLL
• MCU/bridge has issues with passing valid audio data to DAC chip over I2S/SAI bus (less likely but still can happen due to firmware bugs)
• Firmware of MCU/bridge can't handle USB Asynchronous transfers well due to its limited/simple algorithm causing periodic dropouts, pops & clicks (the weakest part of bridges except XMOS known from Neutron Player's development of its USB driver) - it is the trickiest failure, on one device all can be well and on another (different SoC, OS) you get periodic interruptions

So itself MCU/bridge can't provide Quality for audio output (like affect frequency response or deliver other qualities of audio signal) but it can affect it with its misbehavior and flaws of firmware. Bridge and DAC's input are both digital, so if data bits are flowing without an issue then quality of analog signal generation is not affected and main actors for audio quality are DAC chip itself and analog output circuit.

 

[testp]

i would add E1DA 9038D6K (unbalanced) as competent alternative next to (balanced) E1DA 9039S, archimago has done great measurements for it and found it to be excellent as well..

 

[testp]

This was has been actually great read and study!

maybe worthy of pump to main page.. @amirm

 

[kr0mka]

@jkim - have you tried installing a firmware update via the Walkplay app on the Stargate II? Maybe they fixed the hump this way, there's also an option in the walkplay app to switch the 43131 into class H or class AB hp amp mode, worth checking if either of these behave different.

 

[jkim]

@jkim - have you tried installing a firmware update via the Walkplay app on the Stargate II? Maybe they fixed the hump this way, there's also an option in the walkplay app to switch the 43131 into class H or class AB hp amp mode, worth checking if either of these behave different.

Of course, I tried everything. See the note I put in the review:

Note. It is notable that this distortion behavior does not depend on the device's playback sample rate, impedance load, or any firmware settings like digital filters, Class H/AB modes, etc. (if available)---it always occurs regardless of such conditions.

I will definitely update the information and measurements if a new firmware release addresses the problem. But I suspect that the issue is due to the chip's built-in logic. Below is what I wrote in the MacBook measurement thread:

Cirrus Logic's primary design goal must have been low power consumption. The analog gain part of the chips must not have been low-noise circuits due to thermal noise.

DRE must have been a great solution in this case. My hunch is that Cirrus Logic designers did not expect this multitone-induced distortion when designing the chip's DRE operation logic. DRE must involve determining the amplitude of incoming signals to decide how large digital headroom exists. I guess the logic is using a very tight threshold for this, which causes the signal, when amplified digitally, to be borderline saturated. The problem might be easily solved by slightly relaxing such a threshold of signal amplitude in the DRE operation logic.

 

[AcousticTheory]

I have two DACs from SMSL, the RAW-MDA1 based on ESS 9039Q2M, and the DO200 based on 12x CS43131. Both DACs have exemplary measured performance in the top tier of DACs (>110dB SINAD) through Amir's tests, and both DACs sound good, but I have noticed a very subtle difference between the two DACs in the treble range, almost a slight warmishness of the DO200 where the RAW-MDA1 seems more lit up, though I hesitate to say 'more bright'. It's so subtle that I am fine with chalking it up to the different implementation of filters, etc., and neither device is 'broken' sounding, but it's just enough of a difference in the sonic presentation that I can say I think I notice it. That's my subjective opinion, so you can ignore it if you want. But I wonder, based on this analysis of simpler DACs, whether the 12x CS43131 implementation of the DO200 experiences the same measured distortion effect as what was noticed in these smaller and simpler DACs that was tested in this thread.

 

[MBO]

I also have a Topping D70Pro Octo with 8x cs43198, I am very interested in the behavior of this chip in this device

 

[jkim]

I have two DACs from SMSL, the RAW-MDA1 based on ESS 9039Q2M, and the DO200 based on 12x CS43131. Both DACs have exemplary measured performance in the top tier of DACs (>110dB SINAD) through Amir's tests, and both DACs sound good, but I have noticed a very subtle difference between the two DACs in the treble range, almost a slight warmishness of the DO200 where the RAW-MDA1 seems more lit up, though I hesitate to say 'more bright'. It's so subtle that I am fine with chalking it up to the different implementation of filters, etc., and neither device is 'broken' sounding, but it's just enough of a difference in the sonic presentation that I can say I think I notice it. That's my subjective opinion, so you can ignore it if you want. But I wonder, based on this analysis of simpler DACs, whether the 12x CS43131 implementation of the DO200 experiences the same measured distortion effect as what was noticed in these smaller and simpler DACs that was tested in this thread.

I also have a Topping D70Pro Octo with 8x cs43198, I am very interested in the behavior of this chip in this device

It would be interesting to see if these desktop devices with CS431xx exhibit the same artifacts.

As for the SMSL DO200 Pro

The usual shift of linearity commonly seen in any CS431xx-based devices indicates DRE (dynamic range enhancement) in action. Although this does not necessarily mean the presence of a Cirrus hump, I am quite sure some distortion (at least when the signal level is increasing, as shown in Part II of my review) can be observed. In fact, any device based on CS431xx, including the Topping D70 Pro Octo, should at best produce some artifacts like the case of JCally JM20 described in Part II.

 

[cyril_meysson]

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.

 

[jkim]

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

Thanks for the test. The list has been updated!

 

[HeebieGenie]

The FiiO K11 seems to be very popular. I am wondering if it has this same issue.

Also, could the same DAC, such as the Moondrop Dawn Pro or any other have the issue in one unit, but not another?

 

[jkim]

The FiiO K11 seems to be very popular. I am wondering if it has this same issue.

Also, could the same DAC, such as the Moondrop Dawn Pro or any other have the issue in one unit, but not another?

We do not know if the FiiO K11 has the same issue until it's actually measured. But see what I wrote in the conclusion:

It is really bothersome that any device employing CS43131 or CS43198, whether it is a headphone dongle, desktop DAC/HP amp, digital audio player, or a music streamer, should be suspected to potentially produce this distortion.

And if there's distortion, it will occur to all units of the model as the issue is due to the DAC chip's implementation.

 

[HissingFree]

In practical terms, would you say devices not affected by Cirrus hump in the first scenario, like Jcally JM20, are much better than the ones with Cirrus hump? what I mean by this is if it's better despite still having the issue in part II?
That is to say, in the case these issues could be audible in some real-world scenarios, is it likely that JM20 without the cirrus hump ends up with all or most of these scenarios not audible because you are less likely to be in the scenario of part II compared to the scenario in part I?