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

[SoySalsaPicante]

Yup, for the last 5 years or so all my music has been tainted!

Only for people using equipment with a DAC from these cheaters (audible or not). Plenty of alternatives in the market, no necessity to give these cheaters any more money.

I’m not telling anyone what they should or shouldn’t do with their money, but this is how I see it with anyone trying to trick people (CL DRE, MQA, etc.).

 

[IsaacOscar]

Only for people using equipment with a DAC from these cheaters (audible or not). Plenty of alternatives in the market, no necessity to give these cheaters any more money.

The problem being that when I bought my DAC, I had no idea that cirrus was a "cheater". At the time, the chips seemed to be wildly regarded as a high quality.

I’m not telling anyone what they should or shouldn’t do with their money, but this is how I see it with anyone trying to trick people (CL DRE, MQA, etc.).

I have since bought another DAC with a cirus chip/DRE (the Fiio melody), but no clicks are audible on it as Fiio updated the firmware after this bug was discovered. So I'm not really sure it's trying to "trick" people as opposed to an honest mistake (they didn't know there could be a problem, or just thought no one would actually play any problematic audio data)

 

[tccalvin]

Oh and I hear the clicks now when playing it and the Song of the Sisters

I think the reason you (and I) hear them with these tests is that the frequencies contained in the files are quite low.

Since the clicks span the entire audible band and beyond (they are like sudden increases in noise level), it's quite hard to detect them when playing "full frequency" content because they will be almost always masked by other instruments. But play something like a bass drone, and all those higher frequencies stick out very clearly.

An analogy: Imagine a tv screen that exhibits little "white dot" artifacts. When the scene is bright, you don't notice them. When the scene is pitch black, they are extremely visible.

 

[SoySalsaPicante]

The problem being that when I bought my DAC, I had no idea that cirrus was a "cheater". At the time, the chips seemed to be wildly regarded as a high quality.

I have since bought another DAC with a cirus chip/DRE (the Fiio melody), but no clicks are audible on it as Fiio updated the firmware after this bug was discovered. So I'm not really sure it's trying to "trick" people as opposed to an honest mistake (they didn't know there could be a problem, or just thought no one would actually play any problematic audio data)

This artifact doesn’t exist out of coincidence; it was designed specifically for improving dynamic range measurement results (at the expense of output accuracy). This is the very definition of cheating to me, even if it isn’t generally audible and even if the vast majority don’t care or bother to look at the measurements—they went out of their way specifically to create a filter whose sole purpose is to improve stats at the expense of accuracy. As I said, if you don’t hear it and don’t care about the moral implication (enough), that’s totally fine—we’re all here to enjoy music, after all.

I personally refuse to condone/fund this sort of behavior.

 

[IsaacOscar]

This artifact doesn’t exist out of coincidence; it was designed specifically for improving dynamic range measurement results (at the expense of output accuracy).

But what's wrong with wanting to improve dynamic range?

Anyway we don't know if they where aware of the accuracy problem, and perhaps they warned OEMs about it?
I guess I just don't personally go around assuming people have ill intentions...
There was a very minor issue with some highly technical thing, that once noticed, has been fixed by OEMs.
So all in all, I don't see them doing anything wrong here (of course they might have done all this intentionally, hoping that no one would blame them for it, but I've seen no evidence of that).

Of course it's probably bettet to buy chips from manufacturers that have never had such problems. But if the device with the cirrus chip is cheaper, it makes sense to buy it...

 

[robbiekhan]

Having read a bit more on it,. I am in the boat of /don't care/ since it's not audible it seems, it's interesting to be aware of though, cheating is still cheating at the end of the day. Thankfully on the Luxsin X8 the performance and quality appears to be extremely high as it is, and it's the only DAC where 8 of the CS43198 are being used in parallel so perhaps that has something to do with it along with custom implementation and an in-house DSP handling processing.

Will be intrigued to see what the usual outlets think about it when they do their tests too from an objective pov.

 

[SoySalsaPicante]

Having read a bit more on it,. I am in the boat of /don't care/ since it's not audible it seems, it's interesting to be aware of though, cheating is still cheating at the end of the day. Thankfully on the Luxsin X8 the performance and quality appears to be extremely high as it is, and it's the only DAC where 8 of the CS43198 are being used in parallel so perhaps that has something to do with it along with custom implementation and an in-house DSP handling processing.

Will be intrigued to see what the usual outlets think about it when they do their tests too from an objective pov.

Topping D70 Pro Octo uses eight 43198 chips and SMSL DO200 Pro uses 12 43131 chips.

 

[CedarX]

This artifact doesn’t exist out of coincidence; it was designed specifically for improving dynamic range measurement results (at the expense of output accuracy). This is the very definition of cheating to me, even if it isn’t generally audible and even if the vast majority don’t care or bother to look at the measurements—they went out of their way specifically to create a filter whose sole purpose is to improve stats at the expensive of accuracy. As I said, if you don’t hear it and don’t care about the moral implication (enough), that’s totally fine—we’re all here to enjoy music, after all.

That the feature was specifically designed to improve measurements may be exaggerated. DRE is a legit. solution to improve dynamic range. Cirrus' mistake, intentional or not, was to skew the default settings toward "best measurements" (they were greedy) without warning their OEMS about the potential collateral issues...

To me the OEMs are the most guilty: if ASR members (and others) were able to spot the problem with amateur equipment, how come these companies didn't see it? And why some of these companies are still ignoring the issue: that's how much they care about sound quality?

 

[SoySalsaPicante]

... that's how much they care about sound quality?

They care so much as necessary to bag sales of their equipment and no more than that.

 

[robbiekhan]

Topping D70 Pro Octo uses eight 43198 chips and SMSL DO200 Pro uses 12 43131 chips.

Interesting, not heard of this model!

Did a deep research check in perplexity which found:

Yes, the D70 Pro Octo is affected by the Cirrus Logic DRE behaviour, but Topping has added a way to disable it in firmware.

What the issue is​

• Cirrus Logic CS431xx chips (including the CS43198 used in the D70 Pro Octo) have a Dynamic Range Enhancement (DRE) function that can introduce additional distortion and low‑level “artifacts” at certain signal levels.
• This has been measured as a characteristic distortion “hump” on many CS431xx‑based devices when DRE is active.

How it applies to the D70 Pro Octo​

• The D70 Pro Octo uses eight CS43198 DAC chips, so in principle it shares the same DRE‑related behaviour as other CS431xx designs when DRE is enabled.
• Community reports note that Topping released a firmware update adding a DRE filter toggle specifically for the D70 Pro Octo, acknowledging the issue and allowing users to turn DRE off.

Practical takeaway​

• Out of the box or on early firmware, the D70 Pro Octo can show the Cirrus DRE distortion behaviour like other CS431xx DACs.
• On current firmware with the DRE/DRE‑filter option turned off, the problematic DRE‑induced artifacts can be largely or completely mitigated in measurements, and are unlikely to be audible in normal use.

If you own one, update to the latest firmware and look in the filter/DRE settings; use the mode that explicitly disables DRE for the cleanest performance.

So based on other comments above, it's down to OEMs to add in the settings to disable DRE, seems many have even out of the factory (Luxsin) whilst Topping added it in later FW.

So realistically then, is it still an issue or is it a case of knowing that CS and/or OEMs in the past chose to ignore the issue and just left DRE on as standard and this is what people are picking up on even though now it's either off by default or can be turned off?

 

[CedarX]

So based on other comments above, it's down to OEMs to add in the settings to disable DRE, seems many have even out of the factory (Luxsin) whilst Topping added it in later FW.

So realistically then, is it still an issue or is it a case of knowing that CS and/or OEMs in the past chose to ignore the issue and just left DRE on as standard and this is what people are picking up on even though now it's either off by default or can be turned off?

It appears that Cirrus has never been very straightforward about the DRE feature and corresponding settings. On the other hand, some OEMs have known for a long time that the CS431xx dynamic range measurements were fake or, at a minimum, too good to be true. They generally chose to not use these chips.

Topping has acknowledged the problem on the D70 Pro Octo and offered a fix. Great! But I would have expected Topping to detect the issue from the beginning: that's what I find particularly disappointing...

 

[IsaacOscar]

On the other hand, some OEMs have known for a long time that the CS431xx dynamic range measurements were fake or, at a minimum, too good to be true.

Exactly how are they fake? Is it that DRE only increases the dynamic range sometimes?

 

[tccalvin]

Interesting, not heard of this model!

Did a deep research check in perplexity which found:



So based on other comments above, it's down to OEMs to add in the settings to disable DRE, seems many have even out of the factory (Luxsin) whilst Topping added it in later FW.

So realistically then, is it still an issue or is it a case of knowing that CS and/or OEMs in the past chose to ignore the issue and just left DRE on as standard and this is what people are picking up on even though now it's either off by default or can be turned off?

The vast majority of CS DACs still suffer from the DRE issue since it's enabled by default on the chips.

Some manufacturers have provided patches for some of their products.

I have an SMSL DS100 and it looks like it will never see a patch since the company has yet to even acknowledge the problem. I've briefly owned an M300SE as well that exhibited all the same artifacts before eventually settling on a DL200 that uses a different chip (ES9039Q2M) and has a clean output.

My take is that it doesn't make any sense to purchase a CS-loaded DAC unless you're looking for that "Apple dongle" kind of form factor and/or price range. It seems like other more capable chips like the ESS ones need a bigger enclosure and are slightly more expensive. Any more money spent on these CS DACs is wasted since you'll be getting no audible benefits, whereas you will get an audible advantage (however tiny) by going for an ESS or AKM chip.

Another exception might be something like this FiiO JM21 budget DAP that I'm eyeing since it did receive a DRE patch and has other advantages like a custom version of Android that bypasses the mandatory SRC. All in all, a very niche use-case.

 

[robbiekhan]

It appears that Cirrus has never been very straightforward about the DRE feature and corresponding settings. On the other hand, some OEMs have known for a long time that the CS431xx dynamic range measurements were fake or, at a minimum, too good to be true. They generally chose to not use these chips.

Topping has acknowledged the problem on the D70 Pro Octo and offered a fix. Great! But I would have expected Topping to detect the issue from the beginning: that's what I find particularly disappointing...

I lost a bit of faith with Topping even though two of my stack are Topping (A90 D and PA7), just that when I had the DX5 II and witnessed the bugs from day 1 I realised that Topping aren't really in the game of getting even the basics right for a new product launch until people start beta testing after release and then the FW updates start to come weeks/months later.

It's one reason why I'm less interested in new Topping DACs now whereas the analogue stuff like HPA/power and pre-amps are all fine due to their simplicity.

 

[Ahmonge]

Topping has acknowledged the problem on the D70 Pro Octo and offered a fix. Great!

But not for the D30 Pro, who carries four CS43198 chips.

 

[CedarX]

Exactly how are they fake? Is it that DRE only increases the dynamic range sometimes?

What I was trying to say is that the dynamic range measurements show performances that are not representative of the actual dynamic range when playing music tracks (hence "fake"). OTOH, performances with DRE deactivated are more than "good enough" for real music, if not SOTA... So the artifacts caused by DRE look like a bad trade-off.

 

[Sebby]

I’ve just read the discussion and you’ve confirmed what I’ve always heard without being able to explain it.
If I connect the iBasso DC07 PRO and play "Song of the Sister", which is a track I’ve been using for testing for a long time, using it normally that distortion is terribly audible and annoying.
With the Topping G5 everything is perfectly clean again

 

[thcdru2k]

If a CS431xx device does not allow DRE to be disabled, that is a legitimate concern.
If it does allow DRE to be disabled, the issue becomes largely academic for real world listening.

I agree that Cirrus should have been more transparent and that OEMs should have caught this earlier. But I would not conclude that all CS431xx based DACs are sonically compromised once configured correctly.

 

[MaxwellsEq]

Having read a bit more on it,. I am in the boat of /don't care/ since it's not audible it seems

But it is clearly audible! This makes it considerably worse than a DAC with distortion at only -80dB.

You and I can't hear the difference between a DAC with -80dB distortion and one with -105dB distortion (all other measurements being the same). But everyone can hear these clicks as the DRE threshold is crossed.

 

[robbiekhan]

That's on a DAC with DRE on though or one that has no option for it. I'm going by the X8's implementation which has DRE turned off from factory in the menu and has no audible click as far as I can hear.