• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

A Thread Dedicated to Cirrus Logic CS43131

@CedarX on QCC5125 the second microcontroller (M2 if I remember correctly) and DSP are in user (OEM) space along with profile adjustment EQ not toy 5 band QC GEK but 10 PEQ one trough ADB tool. It's better to ask FiiO directly what they cooked and not really wanting to know how they did it. Old QCC5125 has a plethora of other abilities like spartial sound which no OEM ever actually implemented and it's now old and dirty cheap. Interesting device, more plane color schemes and more of them would have been more than welcome. Tho it is far reach from efficient running (using 3~4x more power than similar unbalanced BT dongles) who cares you can get and higher capacity battery after all.
 
I bought a Snowsky Retro Nano (I paid $59.50). Snowsky is a (new) sub-brand of FiiO, they also have an entry-level DAP (~$50), and an on-ear headphones (~$18).

The Retro Nano is a Bluetooth/USB battery-powered dongle. It has both 3.5 and 4.4mm outputs and I believe it is closely related to FiiO own’s BTR13: it uses the same dual CS43131 DAC/Amp (pfff..... I'm not off-topic this time! ;)), and a QCC5125 as USB bridge & BT receiver. The complete spec, including FiiO APx500 measurements, are available at: https://www.fiio.com/retronano_parameters
Here it is, with the BTR13 and Qudelix 5K on its right:
View attachment 423053

It’s a cool device! :) I like the retro “cassette” look (it's an animation!).
It has a removeable 10440 (450mAh) battery—easy to replace when dead, around ~$3 a piece. But you can also increase the autonomy by carrying a spare (charged!) battery(ies). A battery container is included. Oh, and it also works with no battery, powered from the USB port.
View attachment 423056

It is not yet (?) supported by the FiiO Control WebApp, but the FiiO Control app (tested with iOS) enables quite a range of settings, including a 10-band PEQ (up to 3 custom "sticky" PEQ settings—they are saved onboard—plus the usual Jazz, Pop, Rock, etc.). It’s not quite as comprehensive or "advanced" as the Q5K app, but it makes for an easier experience overall: for example, all the main settings are accessible from the on-screen menu, so most of the time, you don’t need the app at all.

Ok, now my real CS43131-related question… The app offers these two THD compensation sliders:
View attachment 423057

The BTR13 does not have these... I have seen comparable features on ESS DACs: for example, the E1DA ESS-based dongles expose similar H2/H3 comp. settings in the app, and the corresponding registers are documented in the ESS datasheets. They are intended to "compensate" (some of) the THD generated in the I/V stage, potentially the amp. stage, and even the Headphones or IEM... On E1DA dongles, you can also use them the other way around to purposedly add euphonic THD (I suspect some OEMs use this trick to create their own “house sound”).

But this is the first time I see a THD compensation on a Cirrus Logic DAC. I could not find any mention of THD-comp. registers in CS4313 datasheet either...
Is this an undocumented feature of the CS43131? Or, is this something FiiO implemented in the QCC5125 Kalimba DSP core (I believe it's relatively easy to do in a DSP)?

Here you go again ;) Very nice!
I'm not really into bluetooth devices but surely the removable\serviceable battery is a great selling point that could make it potentially interesting for me in some scenarios, thank you for sharing this.
Conversely i find the cassette player look pretty meh, but ok, who cares, aesthetics is something i don't consider at all (in the end, my Kuang Pai Plyer 3 too have a lame retro boombox look :facepalm: )
harmonics compensation is probably something done in the dsp together with PEQ that is very welcome.
Fiio measurements are good, 97 dB sinad @ 32 ohm load is on par with the JM20.

Oh, and before somebody asks: are these H2/H3 comp. audible? Maybe… So far, I have refused to drag my immediate family & friends in this hobby, meaning I can’t really perform a valid controlled blind test on my own!
Never say never...
 
I had a BHD pro (43131x2), and now I ordered a JCally JM28 (single 43198). Did I do anything wrong?. I wonder if double chip means a better sound quality in any way.
 
I had a BHD pro (43131x2), and now I ordered a JCally JM28 (single 43198). Did I do anything wrong?. I wonder if double chip means a better sound quality in any way.
Nothing wrong, number of chips doesn't relate to sound quality, both will sound the same. Anyway CS43198 is basically a CS43131 without the integrated headphone driver stage, so it has to use an external op-amp for this. Unless this op-amp is very poor quality or very badly implemented, they will be both transparent and indistinguishable.
Jm 28 power specs are much lower than CS43131 dongles, so they must have used a not very powerful op-amp, for most iems it will be perfectly fine but if you have more demanding headphones it could be somewhat limited.
 
Last edited:
If you use DAC's as indifference biasing it will improve performance additional 2~3 dB and you have two integrated amplifiers to get 4V or close to balanced out. Let's face it no one of OEM's so far didn't even implement capacitor rails to get it working in H class mode but it's getting increasingly popular to put a dozen of them in 2.0 stereo DAC.
 
If you use DAC's as indifference biasing it will improve performance additional 2~3 dB and you have two integrated amplifiers to get 4V or close to balanced out. Let's face it no one of OEM's so far didn't even implement capacitor rails to get it working in H class mode but it's getting increasingly popular to put a dozen of them in 2.0 stereo DAC.
The FiiO KA15 and TTGK TT39518F01-Pro module have a Class AB/H switch (both are dual CS43198). Are you saying that they are not actually working in Class H?
 
The FiiO KA15 and TTGK TT39518F01-Pro module have a Class AB/H switch (both are dual CS43198). Are you saying that they are not actually working in Class H?
It can be only D or H, it's H when you iron switch buffer V so it's rock solid and future more V of batteries varies with state. It's done on the SoC subsystems on your phone or rails on MB but I doubt they paid much attention to it for audio handled even not a very cheap one. The USB audio protocol 2.0 is old the 3.0 is out more than three years and no one of them even thought of implementing it even power save and delivery is much improved with it.
 
I found the following in the datasheet:
"2 Wideband Flatness Mode The CS43131 specifically optimizes for wideband flatness playback, which is enabled by PCM_WBF_EN bit. This mode should only be enabled when PCM plays back at 192 kHz. When operating in other speeds or modes, this bit should be properly disabled. If wideband flatness mode enabled, the filter spec complies to Table 3-10. There is no option for filter roll-off or phase response in this mode. NOS filter mode should be disabled for proper operation. Note that wideband flatness mode can only be enabled or disabled when PDN_HP is set."

Did I understand correctly that the input sampling rate should be 192khz when the NOS mode is activated?

When I measured the FR of my SMSL DL100, I noticed that as soon as I oversample the signal with the Neutro Player, it doesn't matter which filter I select on the DL100.
The FR and IR are then dominated by the OS activation in the software.
Why is that?
 
Did I understand correctly that the input sampling rate should be 192khz when the NOS mode is activated?
Datasheets--when they are published--tend to be a little vague on the functionality details and can even be wrong sometime (ESS is notorious for bogus datasheets...). You understanding may be correct, but that's not how I read the CS43131 datasheet: the ‘Wideband Flatness Mode’ is one possible operating mode of the CS43131. The datasheet doesn't say what this mode consists of (no noise shaping?). The default is 'disabled', but it can be turned on in the 'PCM Filter Option' register (see 7.5.2 in the datasheet). The point 4.2.2 you referenced says that the 'Wideband Flatness Mode' should only be used with 192 kHz input (and above?), that there is no other filter option available when this mode is activated (they'll be ignored), and that the NOS filter option should be disabled.
I don’t read anywhere that the input sampling rate must be 192 kHz when the NOS mode is activated...

As for the DL100, I believe only SMSL can answer what they programed into the XMOS (XU308?) controller. For example, the firmware may automatically enable this 'Wideband Flatness Mode' whenever the input signal is 192 kHz or above. Then on the CS43131 side, any filter will be disabled, which would match what you've observed. The datasheet does not say that 'Wideband Flatness Mode' must be enabled when the input signal is 192 kHz or above, but it appears to be a valid way to program the CS43131 behavior.
 
It's not really a high/low gain.

Instead, the JM20 will measure the headphone impedance at startup and limit output voltage accordingly.

The aim is to prevent clipping regardless of headphone impedance and volume.

Here's the result:
View attachment 378762

You can trick/bypass this mechanism by plugging an "empty" extension cable into the dongle, then the dongle into your phone/PC, then your headphone into the extension cable.

That way, the impedance sensing is defeated and the JM20 can output more voltage/power, even hard-clipping the built-in Amp in some circumstances.

Hardware volume can be adjusted on Windows using the default volume control:
View attachment 378766

If a compatible DAC is recognized, then Windows will automatically switch to controlling it directly.

Hardware volume control and impedance sense auto gain are completely separate mechanisms on the JM20.
Is this the necessary procedure?
Step 1: Connect the extension cable
Step 2: Plug it into the computer
Step 3: Connect the headphones
Is this mechanism exclusive to the CS43131? What about other chips? For example, the CS43198?
 
Is this mechanism exclusive to the CS43131? What about other chips? For example, the CS43198?
I didn't dig further into datasheets but at first glance it seems it only works on 43131:

CS43131
• Headphone detection
— Headphone DC and AC impedance measurement
— Headphone plug-in detection
— Popguard® technology eliminates pop noise

CS43198
• Headphone detection
— Headphone plug-in detection
— Popguard® technology eliminates pop noise

Note that it is not always implemented on all CS43131 dongles, many of them disabled it and let you set gain mode manually via hardware buttons or a dedicated app.
 
I do feel DSD is best on cs43131 (dawn pro/fosi ds2) compared to ESS/AKM. I had ak4493 (topping dx3) which i dont remember much but id say it was about the same or marginally better as 9038q2m (khadas t2p) that i currently have and used extensively.

On cirrus it just feels ever slightly smoother while also having the same or even better detail retrieval and dynamics. Could be subjective tho.

What is a fact is the DSD volume matching to PCM is properly implemented on Cirrus while AKM/ESS just always boosts +6db and its super annoying.


In PCM i believe i liked AKM the most, followed by Cirrus, and ESS. Super subjective opinion, was long ago i had AKM. Mostly a feels thing since all these devices mentioned perform very well. I imagine its probably filter preference that might change depending on gear and its barely perceptible.

Id recommend dawn pro over fosi ds2, better build quality and usb bridge/firmware. Small differences, biggest issue with ds2 are fade ins/outs, lower quality jacks and loose pcb. Dawn pro isnt perfect but definitely not worse and sound quality is either same or hair better.
 

1742061151833.png

1742061209388.png

XDUOO DM01 CS43131 Desktop DAC & Headphone Amplifier​

 
Last edited:
oh my goodness... more than CYBERPUNK it seems a kitchen oven for your Barbie... well, so she can listen to music while cooking... :facepalm:
 
Back
Top Bottom