• Welcome to ASR. 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!

Teardown and (maybe) mod of Meizu Hifi Non-Pro CS43131 dongle DAC

MCH

Major Contributor
Joined
Apr 10, 2021
Messages
3,171
Likes
2,814
Some background:
These last weeks i have been thinking on building a headphone amplifier as part of a larger project (USB interface). I thought that i could use the CS43131 inside my Meizu Hifi Non Pro. But i wanted to do it with output opamps to have more power available, as this is going to be a desktop interface. However, i could not find any schematics of any of the CS43131 dongles with opamps and it is well beyond my knowledge to design one from scratch.
In the meantime, @jkim posted his excellent thread on the analysis of distortion issues some of the CS43131 dongles show. There is no consensus as what is causing the issues, but they are not present in the same degree in all the dongles. Having full control of the i2c programming of the DAC chip, I thought that my DIY amp would offer the chance to play with the different settings and maybe see what is going on with the different implementations of the chip.
Before moving forward with the PCB design, I needed to know which of the versions of the chip (solder ball or qfn) the Meizu non pro has, so peeked through its little window.... and no surprises, what i could see there inside was very similar to the Meizu Pro.
The Meizu Pro is, to my knowledge, one of the only modern dongles for which a detailed teardown exists. The bad news is that both the USB bridge (CS46L41) and DAC have a solder ball package, so the i2c communication between them is through inner layer tracks, and cannot be tapped.... Additionally, this package of CS43131 is tiny (3x2.7 mm) and has 42 pins, and I don't think i am capable of reballing and soldering such a part with any significant chance of success. On top of everything, i could see what seemed to be a conformal coating on the PCB: I have access to all kinds of solvents and i am confident i could get rid of it, but it might not be a clean easy process.
All this made me almost abandon the project, but i though.... what if the Non Pro is just a Pro, but without the opamp populated?? This would solve all my problems at once: i would have the schematic i was looking for + i could use the existing PCB without having to solder any microscopic multi pin IC.

With all this, the curiosity was too strong and this is what i found:

The dongle has an aluminium case, and the natural way of removing it, is to pull in the direction of the cable. I tried this but it didn't work, and when i noticed i was destroying the USB connector, i abandoned this route and used the dremel to cut open the case. At this point, i wanted the dongle to keep working, so i cut the sides carefully. Aluminium is relatively easy to cut. This was an easy process overall:

1748694481651.png


Once done, i could peel off the case, and the first good surprise was already visible. Unpopulated footprints!:

1748694624902.png


There is a second metallic shield inside, exactly like the Pro version, that is easy to peel.

The other side is where the bridge and DAC are, covered by the transparent window, that is a sort of methacrylate like hard plastic:

1748694892155.png


DAC, USB bridge, SPI flash memory, crystal, and pasives. This side is exactly the same as the Pro, even the PCB marking is the same. I believe the PCB is 6 layers, and yes, conformal coating covering both main chips. I wonder if it is to prevent them to fall apart because of the vibration and shocks a pocket device will have to stand:

1748695029837.png


The analog side is also the same than the Pro, but without the OPA1622, LM27762 and that other IC marked "9F HD" on the lower right corner that i don't know yet what is, a regulator maybe? if anyone has a guess please let me know. The larger rectangle is for a metallic shield encapsulating the output stage in the Pro version, and the best yet is that there is no coating here, one could solder directly on the pads:

1748695359178.png


Pro (from online teardown):

1748695843842.png


Looking closely, some of the resistors that divide the PCB in two are missing in one or the other version. The ones in the non Pro that are not populated in the Pro are 0 Ohm, what tells me that they are just the connectors to bypass or not the opamps.

Well, the good news is that the dongle still works in this state. I will now have to reverse engineer the circuit. As I won't have access to the output pins of the DAC, i will have to do it backwards, starting from the headphone terminals. What i would not be able to know is the value of the resistors around the OPA1622 and the LM27762.

Help needed: if someone has an idea of how the schematic could be and what the 9F HD IC is, i would appreciate very much the information. I will publish here the schematic as i read it, it does not seem too complicated, i think it is doable, but any help will facilitate things to me, I am a complete noob in this sort of circuits. Thanks!
 
Last edited:
Checking the connectivity, I suspect that the unknown IC could be a switch that routes the signals through the opamp or not. I don't remember if the Meizu Pro has two modes like high power / low power mode?
Edit: could also be a mute control
 
I think i know what the unknown IC is. After a lot of search, i think it could be an Onsemi FSA2275A or the better performing FSA2276. The markings don't match but the package and pinout yes. Maybe the one in the Meizu is a drop in copy, it is an expensive part.

These parts are audio switches intended to select between two sources, this is the block diagram:

1748777036511.png


However, here i think it is doing a simple mute control, as only one of the two audio sources carries signal (confirmed with an osciloscope).
C_EXT is a capacitor that you can use to tune the switch on time. SEL selects the source and is probably unused in the Meizu and Mute controls the audio output. The specs for the FSA2276 match the -115 dB THD+N of the CS43131.

I don't know why in the Meizu non Pro is not populated, maybe only the presence of opamp makes audible pops?

PS: while searching for this, i came across other parts used for headphone input impedance detection, as i thought that could be a potential purpose of this IC. It called my attention that they also measure.... moisture level! (?)
 
Interesting, this is basically what it is:

1748803219718.png


Unexpectedly, the switch is actually a sort of selector between feeding the headphones directly from the DAC outputs OR from the OPA1622 outputs.
This is how it works:
In the absence of signal, or heaphones, the EN pins of LM27762 and OPA1622 are low and the devices are off. The MUTE pin of the switch is also low, what means that the switch is closed and the headphones are not disconnected but connected directly to the DAC outputs. They also remain physically connected to the outputs of the OPA1622, which is off.
Now when a headphone or a dummy open cable is connected and an audio signal starts, then LM12262 and OPA1622 EN pins go high, both devices turn on, and the MUTE pin of the switch goes also high, what means the switch is open and the headphones are not connected anymore to the DAC outputs but only to the OPA1622 outputs.

I read in TI forum that OPA1622 does not like very low impedances, like 8Ohms or below, i wonder if that is the reason of all this mechanism: when a very low impedance is detected, the OPA1622 is kept off and the headphone stays connected only to the DAC output. It is strange, because my truthear blue still triggered the OPA1622 circuit, and i think it is like 10 Ohms.... Hm, i wonder if i am missing something, all this trigger thing seems too complicated for what it does....
 
Back
Top Bottom