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

[MaxwellsEq]

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.

Absolutely - if you disable DRE the clicks depart, BUT the overall DAC performance now measures less well. That's why some consider DRE a "cheat" to make these chips look better when seen through an AP test set, than they actually are.

 

[robbiekhan]

If it measures less so, does it ultimately matter though if the delta between pre/post DRE are inaudible though?

 

[MaxwellsEq]

If it measures less so, does it ultimately matter though if the delta between pre/post DRE are inaudible though?

No, it doesn't matter. If DRE is switched off, the DAC will sound fine. But, it will measure quite a bit worse than an equivalent ESS or AKM which do not have a DRE issue.

What's upsetting some people is that objective reviews of DACs with CS chips get excellent scores because AP tests look top class. But if you disable DRE the objective tests are closer to early 2000s CS DACs.

BUT an early 2000 CS DAC is still audibly better than we can hear!

 

[CedarX]

What's upsetting some people is that objective reviews of DACs with CS chips get excellent scores because AP tests look top class. But if you disable DRE the objective tests are closer to early 2000s CS DACs.

I hope their new CS4308P is better, but Cirrus product page is not very encouraging:

“The CS4308P uses a proprietary analog FIR architecture to reduce out-of-band noise and minimize the external component requirements. This provides a dynamic range of 123 dB and THD+N below -114 dB, all within a 10 mW/ch power budget.”

Is Cirrus devising a new trick to achieve good DR measurements and not much else? What are the tradeoffs?

 

[robbiekhan]

So by the sounds of it then it seems CS ended up chasing the numbers rather than just sticking to what works and has always worked and is still beyond human hearing - And then forgot to give the memo to OEMs for ages until later OEMs figured it out and put FW updates in place or had the option to disable by default.

Personally speaking, I have not had any noise or click issues with any of the major 3 DAC chip makers, the only issue that did exist was resolved by installing the DAC USB driver and enabling the streaming mode to Always On which stops the initial part of any audio being cut off, though this function too impacts most DACs connected via USB to a computer as well and it's up to the OEM to release a driver to toggle that Streaming option. There are still too many Chinese brands that do not offer a driver leaving people having to put up with the issue.

Mac and Linux users have no hope as the drivers that do get supplied are for Windows only lol. The only solution here is to run a script that keeps a constant silent audio signal going to the DAC which stops it from internally sleeping and thus mitigating the initial part of sounds being cut off.

 

[Sebby]

I ran the tests and the Tempotec Sonata BHD Pro and the iBasso DC07 PRO have this issue. The FiiO KA11 (which should be Cirrus) doesn’t seem to have this problem. On all of these devices, it’s impossibile to disable the DRE

 

[IsaacOscar]

The FiiO KA11 (which should be Cirrus) doesn’t seem to have this problem.

Yes, it has a CS43131. .

 

[tccalvin]

I hope their new CS4308P is better

Me too. Universal Audio is using this new flagship chip in the Volt 876. AFAIK they have been careful to avoid the CS431XX family of chips unlike other brands such as Apple, Audient and Focusrite. It would be great to put that interface through its paces. Also, no mention of DRE on the CS4308P Data Sheet.

The FiiO KA11 (which should be Cirrus) doesn’t seem to have this problem.

It probably does, but it might be one of the "non-hump" DACs, which have fewer (and therefore less detectable) clicks. To know for sure, someone needs to record the KA11's output.

 

[Sebby]

Me too. Universal Audio is using this new flagship chip in the Volt 876. AFAIK they have been careful to avoid the CS431XX family of chips unlike other brands such as Apple, Audient and Focusrite. It would be great to put that interface through its paces. Also, no mention of DRE on the CS4308P Data Sheet.



It probably does, but it might be one of the "non-hump" DACs, which have fewer (and therefore less detectable) clicks. To know for sure, someone needs to record the KA11's output.

The KA11 is a budget device with good power. If it doesn’t have strange distortions, that’s fine, and even if they were slightly audible compared to others but still present, you could forgive it. On a device like the DC07 PRO, which I paid €210 for, it’s unacceptable. Luckily the seller agreed to refund me after I explained the issue, and I’ve promised myself I won’t buy another Cirrus DAC. There are plenty of other options out there at any price point.

What makes me angry is that between Tempotec and iBasso, I’ve always heard these flaws, and not only on the famous Hans Zimmer track that I’ve always used as a killer track (one in my killer playlist). I always blamed something else because "it measures well, so it can’t be the DAC or its amplification stage" and I convinced myself it was placebo.
In the end I’ve always used the Topping G5 much more, even though it was very inconvenient in some contexts, because I never fully liked those two Cirrus devices, but I wouldn’t accept the idea of seeming like a subjectivist who hears things that don’t exist.

The fact that the measurements came out and showed what specific issue should occur… damn, I was right! So I wasn’t being an idiot! This doesn’t mean becoming a subjectivist or anything like that, but in my opinion objectivism should be used more maturely and more in depth measurements should be included in the future, because a DAC judged "incredible!!" turned out to be crap.
If the goal is the highest possible audio quality, these DACs are objectively bad because the problems are clearly audible. If more thorough measurements aren’t added, there’s no point in reading flawed technical reviews.

Sorry for the rant. I know there are more serious things in life, I agree, but doing things halfway makes no sense.
This issue was pointed out by another member in a different discussion and he was attacked in a cowardly and shameful way, like members of the worst kind of subjectivist or fundamentalist forum. I hope some people calm down and understand that what emerged here is first and foremost a flaw in the measurements that can be fixed without any problem. That’s part of how science works.

 

[rcstevensonaz]

[7/23/25] ... disabling DRE on the ...

This is an awesome thread. But would you mind updating the leading post to with the full name of whatever DRE is an acronym for as part of the very first use?

For example: [7/23/25] ... disabling Department of Real Estate* (DRE) on the ...

* When I do a google search on DRE, that is the only meaningful definition that was returned (along with references to Dr. Dre). But somehow, I suspect DRE in this context means something different than California's Department of Real Estate.

Edit: As a general convention, the full name should always be provided the first time an acronym is used. I realize that it is defined in the text further down the body, but it would be useful to add the definition for people who come across the thread and just read the very valuable change log to get a sense of the overall post.

 

[tccalvin]

because a DAC judged "incredible!!" turned out to be crap.

I wouldn't go as far as to call these DACs "crap", but I still don't like that they are being presented as flawless, state of the art devices when they clearly do have some minor flaws.

Take the SMSL DL100, for instance. If I hadn't known about the DRE thing, I probably would have chosen it over its better-performing but similarly-priced ESS sibling, the DL200. After all, based on the review, I might have concluded that its performance is just as great, and I like its I/O more.

I'm not unreasonable. When the DL100 review came out, most of us didn't know about DRE artifacts. Amir's time is also limited and he provides reviews for free, so I don't expect him to go and edit all CS DAC reviews only to mention these problems.

The real problem is that when some members asked him to take the CS DAC issue into account going forward, he just flat out refused. In the case of the DL100, we're talking about a 200€ device. In this price range, there are better alternatives. Yet, if another such CS device ended up on his test bench, he would present it as SoTA despite the evidence the ASR forum members have gathered, just because it would clear all the usual tests he does.

Given these premises, one could understandably begin thinking that there are some shady reasons why Amir doesn't want to put some products in a bad light. I personally don't believe in such a conspiracy, but I did lose some trust in the objectivity of the main reviews. If he doesn't have the time to be thorough, then why keep doing reviews? No one is forcing him.

 

[SoySalsaPicante]

This is an awesome thread. But would you mind updating the leading post to with the full name of whatever DRE is an acronym for as part of the very first use?

For example: [7/23/25] ... disabling Department of Real Estate* (DRE) on the ...

* When I do a google search on DRE, that is the only meaningful definition that was returned (along with references to Dr. Dre). But somehow, I suspect DRE in this context means something different than California's Department of Real Estate.
Edit: As a general convention, the full name should always be provided the first time an acronym is used. I realize that it is defined in the text further down the body, but it would be useful to add the definition for people who come across the thread and just read the very valuable change log to get a sense of the overall post.

It's Dynamic Range Enhancement. I don't think anyone really wants to type all of that out every time.

 

[rcstevensonaz]

It's Dynamic Range Enhancement. I don't think anyone really wants to type all of that out every time.

You only type it the first time it is used in the change log section, not every time; especially since a quick Google search brings up nothing.

And yes, my request may be overkill so feel free to ignore it. Just posted out of frustration since I was not seeing it defined and wasn't finding help on Google. Eventually, I did notice it in the main body... so also free free to attribute the issue to my not spending enough time to scroll down below the change log and below the picture to read the text in the main body of the post.

 

[CedarX]

Just posted out of frustration since I was not seeing it defined and wasn't finding help on Google.

See post #116: “It is most likely caused by the DRE (dynamic range enhancement) function in the DAC chip which could be controlled in firmware.”

 

[danadam]

See post #116:

He knows:

I realize that it is defined in the text further down the body,

He only asks to do it on the first use in the first post.

 

[CedarX]

I hope some people calm down and understand that what emerged here is first and foremost a flaw in the measurements that can be fixed without any problem. That’s part of how science works.

Which measurements: 1/ Cirrus Logic measurements & verifications of their own design & feature? 2/ The OEMs who selected these chips and did not verify what they were buying and their implementation? or 3/ Amir' measurements, which AFAIK have always been a "standard set of measurements", not intended to catch everything?

Given these premises, one could understandably begin thinking that there are some shady reasons why Amir doesn't want to put some products in a bad light. I personally don't believe in such a conspiracy, but I did lose some trust in the objectivity of the main reviews. If he doesn't have the time to be thorough, then why keep doing reviews? No one is forcing him.

Also, no mention of DRE on the CS4308P Data Sheet.

So, the new CS4308P may not use DRE... But I did loose some trust in Cirrus Logic: what new measurement should Amir add to catch the next Cirrus "cheat"?
I'm not trying to defend Amir, but I don't think exploring all the possible corner conditions (which DRE is) it is that simple...

 

[tccalvin]

what new measurement should Amir add to catch the next Cirrus "cheat"?

I don't believe he should try to catch anything. But I do think that measurements should adapt to new evidence.

For example, if a new drug is developed, standard tests should be performed to determine its safety. But if there is evidence that some side effects of the drug slipped past the measurements, then the test suite should probably be updated in order to catch them. Or at the very least, patients should be made aware that the drug has some correlation with those side effects, even if their origin hasn't yet been studied.

Pretty standard stuff, really.

 

[Somafunk]

* When I do a google search on DRE, that is the only meaningful definition that was returned (along with references to Dr. Dre)

Should’ve searched for “dre dac chips” , a useful example of correct search criteria, if you didn’t find it on google using your search terms then perhaps you should take time to understand that when searching for information it is often very relevant to include as much detail in the search bar as you need to find the correct resource/result

 

[oleg87]

but I don't think exploring all the possible corner conditions (which DRE is) it is that simple...

Perhaps not, but for the simple full-scale output DAC use case, I think continuously sweeping the standard multitone over amplitude and looking at what's going on in the noise floor would be a good start for catching meaningfully audible gremlins like this.

Interestingly, I can't seem to reproduce the aberrant behavior of the Apple 3.5mm dongle with my M3 Macbook Air, even though it sounded like some people were seeing the same problems with their Macs. Wonder if Apple has patched it?
edit: guess the distortion is still there in the CMaj test, but very faint compared to the dongle.

 

[rcstevensonaz]

Should’ve searched for “dre dac chips” , a useful example of correct search criteria, if you didn’t find it on google using your search terms then perhaps you should take time to understand that when searching for information it is often very relevant to include as much detail in the search bar as you need to find the correct resource/result

Gee, thanks for the subtle ad hominem dig there, my friend. Do you really think I am so stupid that I can't figure out how to take on the added effort to add additional terms to guide a Google search? But anyway, thanks for the tutorial to help out this old befuddled man who is baffled by the newfangled technology known as "Google search".

But I wonder if you missed the point:

1) The reason every style guide calls for full spelling of term on first use of an acronym — when it is not already well understood by the audience (e.g., "DAC") — is because the burden should not fall on the reader to have to go and do research. The style guides puts that on the author.

2) If the burden of determining an acronym's meaning is being left to the reader, then that reader should be able to search using the acronym to obtain the answer.

Now, @Somafunk, you would have scored at a strong point if you used a different example. For example, you could have noted that simply selecting DRE along with the surrounding word(s) — e.g., "disabling DRE on the Cirrus" — would have brought up a useful response. And you would be spot on correct. The reason that example would score a point (unlike the one you provided) is because it uses the text that was already written; and thus would be largely consistent with #2 above.

But style guidelines, as noted in #1 above, are still the best way to write in a way that readers can comprehend.

P.s., we should get back to the issue of distortion on Cirrus Logic chips. I wasn't intending a simple wording suggestion to the OP to turn into a long sub-thread on style guides and whether the burden should fall on the reader to figure it out, or on the author to provide the definition on first use. If anyone wants, add your last word and I'll leave that as the final word.