• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

A Thread Dedicated to Cirrus Logic CS43131

Saidera

Senior Member
Joined
Jul 18, 2020
Messages
323
Likes
217
Location
Adelaide, South Australia, Australia
The USB-C Dongle Market

Apparently CS43130 was a Wolfson chip.

It was Hidizs / Tempotec / Meizu that first drew my attention to CS43131.

Later in 2020 products with dual CS43131 such as iBasso DC03 began appearing.

We’ve had these pages:

https://www.audiosciencereview.com/...pple-vs-google-usb-c-headphone-adapters.5541/

https://www.audiosciencereview.com/...dizs-s8-usb-c-headphone-adapter-review.10823/

https://www.audiosciencereview.com/...sonata-hd-pro-review-headphone-adapter.22625/

https://www.audiosciencereview.com/...a-better-budget-option-than-the-hd-pro.16175/

https://www.audiosciencereview.com/...zu-hifi-usb-c-headphone-adapter-review.10947/


I have repeatedly read that Meizu Hifi will be affected by the quality of the 5V USB rail. On the other hand, IVX has recently again proposed Meizu Hifi as the better choice over Tempotec SHDP. This led to me getting the Meizu Hifi.

Meizu Hifi cannot record DSD, but at least it can record only 16/48 PCM mono which Tempotec SHDP cannot. Tempotec SHDP has Savitech’s ASIO driver, and so can sometimes playback DSD256. Meizu did not provide ASIO drivers. Meizu Hifi is like a longer version of Apple’s TPU dongle, except black. Meizu Hifi’s circuitry is encased in metal though. If I had to choose between Meizu and Apple, I’d choose Meizu. I value DSD and 44.1kHz multiples of PCM. I still can’t stand 24/48. But Apple can record at 24/48 mono.

It is now common knowledge that these first appeared for smartphone jack replacement, but subsequently, thanks to USB-C to USB A adaptors, could be used in PCs to replace Realtek ‘crab’ sound. Ever since Sony VAIO’s strong advertising of DSD and their Sigmatel/IDT (VAIO 2005-08) CXD9872 implementations I have had an infatuation with codec chips. Notably the CXD9872 was stated by Sony to be their ‘own’ codec chip, when in reality, buyers noticed that the foundations were from Sigmatel. Later Sony switched to ALC889DSD by Realtek, which could allow MIC IN DSD recording. Some desktops paired ALC889DSD with the Sony S-Master amp. I assume that ALC889DSD’s own amp wouldn’t be used in that case.

CXD9872 and ALC889DSD were marketed as 'Sound Reality' chip by Sony - they used to list S/N ratios on documents pertaining to walkmans, S-Master amps and VAIO Sound Reality (as chip direct output, not individual implementations). Now they don't.

That's my infatuation with codec chips and their useless and sometimes misleading specs.

Now:

I wanted to ask, some dongles only use CS46L41 (not CS46L06 of Apple)+SA9302L. Meizu has CS46L41 but still uses CS43131…why? Isn’t CS43131 superfluous? It’s not like the DAC and amp are split between chips. What’s CS46L41 doing – is it acting as a USB bridge?

Now we’re seeing PCM 32/768 on UA2. So:

I wanted to ask, why is UA2 able to use ESS ES9038Q2M DAC & Ricore RT6863 amplifier to achieve PCM 32/768 + DSD512 when the same ES9038Q2M is used in E1DA’s DAC but no PCM 32/768 is supported?

As for pricing, dual CS43131 dongles are pricey and might not be that much different, given that CS43131 is easy to implement but has little room for improvement, unlike higher ESS AKM chips. Dual CS43131 is usually for balanced output, but to be fair, they seem to be doing the same anyway for unbalanced. Just for placebo.

A rough division of price ranges for best cost performance ratio dongles:

Apple’s $9USD -$15 AUD

or $13 USD on taobao or $24 AUD on aliexpress like Meizu HIFI

or on taobao $35-$40-ish with TempoTec SHDP…


There is the question: in a volume matched comparison of various dongles in a professional studio with studio monitors (speakers) will the sound be subjectively different between dongles. If so it would it be due to implementation? We do know that CS43131 is unlikely to be vastly different across different products.

On speakers, it’s usually a comparison between terrible motherboards and actual DACs which highlight vast differences. When we get to the territory of Apple’s dongle and above, I presume it’s safe to say that subjectively perceivable differences are minimal? Is it worth testing? Is that not what quite a few reviewers (who appear to seriously spend their time listening on varied earphones etc and writing reviews) are aiming to do? Sometimes it’s possible to hear how Apple’s dongle is somewhat different, less ‘present’ as if the thickness of sound has been reduced. But maybe not. I have no standing to speak since even volume matching is beyond my ability.

Finally, Cirrus Logic has supported DSD playback, but no matter what kind of comparison you do, there is just no difference between its PCM and DSD. It’s as though DSD support is included just because the market has released DSD in the past – a mere compatibility and consumer preference issue. Many DACs originating from Japan have clearly presented a different sound when comparing PCM and DSD playback e.g. AKM, but it could be down to implementation.

So basically I have complaints about the DSD aspect of CS43131, but it is like a dream come true to have DACs with native DSD at $16-24 AUD (Meizu on taobao). This was not the case in 2018! So I wanted to express some gratitude to CS43131 for enabling DSD to be obtained on any USB C device and capable of taking anywhere.

Meizu Hifi does not support DSD DoP on some apps, since it comes out as white noise. No amount of tweaking appears to change this.
It works on HQPlayer, but not JRiver or foobar DSD Processor. Oh well.

After comparing Meizu and Meizu Pro teardown pictures from 52audio, I found that Meizu Pro has components on the bottom side as well. The only difference is the opamp. DoP support limits remain the same.

I know that Meizu Pro suffers phone antenna interference as noise. I wonder why.
 
OP
Saidera

Saidera

Senior Member
Joined
Jul 18, 2020
Messages
323
Likes
217
Location
Adelaide, South Australia, Australia
Totally from the layperson's perspective, I like the way CS doesn't heat up like AKM, and how it's slightly cheaper and easier to implement than ESS chips, supposedly. I like how CS's DSD implementation is straightforward, closer to the textbook DSD Processor way. ESS makes it complex and anyway, none of this is disclosed. AKM's DSD might be better, since it's Japanese. Ridiculous statements here, please excuse me.

CS46L41 is acting as a USB bridge. Rhetorical question.

I still wanted to ask, why is UA2 able to use ESS ES9038Q2M DAC & Ricore RT6863 amplifier to achieve PCM 32/768 + DSD512 when the same ES9038Q2M is used in E1DA’s DAC but no PCM 32/768 is supported?
 

Jet Black

Member
Joined
Jan 7, 2021
Messages
73
Likes
19
Totally from the layperson's perspective, I like the way CS doesn't heat up like AKM, and how it's slightly cheaper and easier to implement than ESS chips, supposedly. I like how CS's DSD implementation is straightforward, closer to the textbook DSD Processor way. ESS makes it complex and anyway, none of this is disclosed. AKM's DSD might be better, since it's Japanese. Ridiculous statements here, please excuse me.

CS46L41 is acting as a USB bridge. Rhetorical question.

I still wanted to ask, why is UA2 able to use ESS ES9038Q2M DAC & Ricore RT6863 amplifier to achieve PCM 32/768 + DSD512 when the same ES9038Q2M is used in E1DA’s DAC but no PCM 32/768 is supported?
Chinese tends too over exaggerate them specs for sales probably
 
OP
Saidera

Saidera

Senior Member
Joined
Jul 18, 2020
Messages
323
Likes
217
Location
Adelaide, South Australia, Australia
Don't think so? maybe?
 
OP
Saidera

Saidera

Senior Member
Joined
Jul 18, 2020
Messages
323
Likes
217
Location
Adelaide, South Australia, Australia
Some extensive info on CS43131 in its datasheet! Hard to understand for me though.

One can infer that Meizu’s CS43131, made in Korea, is CS43131 42 ball WLCSP package (balls underneath it and its vertical sides are longer than the horizontal side – a rectangle) rather than pads attached to pins in the 40 pin QFN –CWZR ver. Made in Taiwan and used in SHDP.
 
OP
Saidera

Saidera

Senior Member
Joined
Jul 18, 2020
Messages
323
Likes
217
Location
Adelaide, South Australia, Australia
To clarify, you are saying that UA2 is able to use ESS ES9038Q2M DAC & Ricore RT6863 amplifier to achieve PCM 32/768 + DSD512 when the same ES9038Q2M is used in E1DA’s DAC without PCM 32/768 support because of the different usb bridges they use.

Many thanks!
 
OP
Saidera

Saidera

Senior Member
Joined
Jul 18, 2020
Messages
323
Likes
217
Location
Adelaide, South Australia, Australia
In regards to DSD, it says there is no need for an ‘intermediate decimation stage’ or interpolation filter for volume control etc although it says DSD gets special pins just for itself, it appears it is shared with PCM streams, or maybe not. At max output, 32 bit can outperform DSD for THD+N values. But the cause for this might be due to how Cirrus recorded the DSD source for the test.

DSD does end up passing through a multibit delta sigma modulator before the DAC and filter.

Cirrus uses a Class H amp like an AB amp. Then there’s phase modulation and mixing PCM with DSD. Then power up and down sequences.

Then filter response graphs for DSD. Though there are plenty of impulse responses, which have microsecond units ranging from many thousands to one hundred only, DSD falls in the category of one hundred microseconds.

Maximum operating temp 70 degrees.

Datasheet is from Oct 2017.
 
OP
Saidera

Saidera

Senior Member
Joined
Jul 18, 2020
Messages
323
Likes
217
Location
Adelaide, South Australia, Australia
On Cirrus’ DSD pathway

While Cirrus is a name attached to Apple, and acquired Wolfson, its history with DSD is quite long, starting from the early 2000s. The chip CS4398 has been used in TEAC devices and especially KORG’s DS-DAC-100 measured on ASR.

There appear to be two types of DSD chips in Cirrus’ lineup. One is the older type with both DSD Processor and direct DSD pathways. When it is programmed to be direct, it does not use DSD Processor for the volume control and 50 kHz filter. It does not use the multibit delta sigma modulator. It goes directly to the MUX/DAC or switched capacitor array. But those chips also lacked the DoP function it seems. After Wolfson was acquired, the CS431xx way of doing DSD has been via DSD Processor only. There is no direct DSD pathway, which on the surface does not sound like an improvement. However, if the DSD Processor was not used, volume control and 50 kHz filter would have to be done in some other way. It would not be possible to put all the components necessary into a finger-sized DAC adapter for the phone. So most likely for the sake of portability and ease of use, the direct DSD pathway was abolished. Although it is not mentioned in the datasheets, CS431xx apparently uses a DAC with 512 1 bit DACs or something like that. It could be fundamentally different to such older chips as CS4398. What isn’t so clear is which pathway is better? DSD or PCM pathway?
 

JohnYang1997

Master Contributor
Technical Expert
Manufacturer
Joined
Dec 28, 2018
Messages
7,005
Likes
15,833
Location
China
Cs43198/cs43131 has direct DSD mode. It is a little bit better than processing DSD in harmonic distortion but halved in level so worse in SNR and overall THD+N.
Cs43198/cs43131 are multibit delta sigma dacs with very good performance. If there's a way to reduce noise at full level it will be even better.
 
OP
Saidera

Saidera

Senior Member
Joined
Jul 18, 2020
Messages
323
Likes
217
Location
Adelaide, South Australia, Australia
Cs43198/cs43131 has direct DSD mode. It is a little bit better than processing DSD in harmonic distortion but halved in level so worse in SNR and overall THD+N.
Cs43198/cs43131 are multibit delta sigma dacs with very good performance. If there's a way to reduce noise at full level it will be even better.

Thank you. It is an absolute privilege to have you answer my queries. Respectfully, I think I am referring to direct DSD in a different sense as shown in the diagrams below:
https://www.cirrus.com/images/product_diagrams/cs4398blkdiag-782ad57e36.svg
https://www.cirrus.com/images/product_diagrams/cs43131blkdiag-a38055d2c0.svg
https://www.cirrus.com/images/product_diagrams/cs43198blkdiag-0f9ab80757.svg

But I also looked at this: https://au.mouser.com/images/marketingid/2019/microsites/128513667/2019-03-08_16-04-11.png

I am confused, but perhaps CS43198 has direct DSD mode as you have said.

Edit: After checking CS43198_F1.pdf, I can confirm that CS43198 HAS DSD Direct, whilst cs43131 does NOT. I am shocked at the difference. But as you say. Harmonic distortion of DSD is better in Direct mode, but it's halved in level and thus worse in SNR and overall THD+N. And so Direct mode is not practical or rational as a choice.
 

JohnYang1997

Master Contributor
Technical Expert
Manufacturer
Joined
Dec 28, 2018
Messages
7,005
Likes
15,833
Location
China
Screenshot_20210626-105001__01.jpg
 
OP
Saidera

Saidera

Senior Member
Joined
Jul 18, 2020
Messages
323
Likes
217
Location
Adelaide, South Australia, Australia
What isn’t so clear is which pathway is better? DSD or PCM pathway?

I must have come across the answer somewhere, but I can’t find it. What advantage or disadvantage does DSD Processor path have over the ordinary PCM path in Cirrus chips? Direct DSD mode measures slightly worse than the DSD Processor path, which in turn measures worse than the ordinary PCM path, although it’s usually at full level output. There is also the possible limitation of the DSD source test file. Based on this datasheet info alone, and the historically unimpressive measurement results of DSD DACs on ASR the conclusion would be that there is only a disadvantage. But still the bypassing of the decimation stage and so on seems alluring.

Regarding CS43198, I only learned of its existence after A&K’s first dongle came out, using dual CS43198. Obviously they would probably not use Direct DSD at all. But it surprises me now, why would I have been shocked to discover that CS43131 doesn’t have Direct DSD? Clearly it would be a headphone and hearing risk if DSD were to be put through a DAC so untouched. Why would I have been shocked to discover that CS43198 DOES have Direct DSD? It’s obvious that it should, given that external components can take care of the resulting ‘low quality’ but untouched analogue signal. Now why on earth am I so curious about Direct DSD mode, given that it is almost like the widely disliked NOS mode, a detraction on the goal of higher fidelity playback of sound?
 
OP
Saidera

Saidera

Senior Member
Joined
Jul 18, 2020
Messages
323
Likes
217
Location
Adelaide, South Australia, Australia
DSD is better.
 
OP
Saidera

Saidera

Senior Member
Joined
Jul 18, 2020
Messages
323
Likes
217
Location
Adelaide, South Australia, Australia
Yeah PCM, because if a user doesn't trust the decimation etc why should the user trust 'DSD Processor' and multibit dsm? Less is more simply does not apply in any case, seeing how Direct DSD ends up as terrible quality despite minimal interference and manipulation of the source.

But how can science defeat marketing?
 

Wes

Major Contributor
Forum Donor
Joined
Dec 5, 2019
Messages
3,851
Likes
3,568
my FPGAs bleed when I read chip SQ threads
 
OP
Saidera

Saidera

Senior Member
Joined
Jul 18, 2020
Messages
323
Likes
217
Location
Adelaide, South Australia, Australia
my FPGAs bleed when I read chip SQ threads
I was always a chip person since VAIO DSD CXD9872 came into my existence. For years it was the only laptop without noise plus could play and record DSD64. And today it is still! If you bar the gaming laptops with ALC1220.

FPGA is great but for casual users we need chips and satisfaction.

I want to break open the black box, to understand what is going on in those microprocessors, semiconductor designs, even if the end result sound is no different!
 
Top Bottom