Fosi Audio DS2 Portable DAC & Amp Review

[Eddie]

Jcally JM20 with the same Savitech doesn't behave in this way, when you start playing after it's gone into muting status the very first samples are not truncated\faded but instead you can hear a very brief bakground pink noise. What is preferable it's to you.

Then it seems to be a firmware issue rather than a hardware issue.
In the meantime, I contacted Fosi about this behavior. So far they don't seem to understand the problem.

 

[alexanderino]

I just didn't had the courage to push the plug into the socket until the final 'click'. It takes some kind of force, and I was a bit afraid of damaging the little device.

The same is true when using the Sony MDR-EX15AP earphones (with mic) and the 3.5mm output.

 

[AnalogSteph]

I mentioned that I had problems to do the sweep. Here is what is there and also what happens: enormous out-of-band junk, which at certain levels (around -25 dBFS) makes the special dual high performance input of the APx fail. The noise floor then is shown wrong, the notch filter fails as well, measurement values are all way off:

Do you reckon this is what happened here as well? The device in question is an inexpensive CS4272-based job with out-of-band noise to match.

My working theory so far has been input stage overload (slew rate limiting) at high gain. The jumps have to be where the gain switching occurs, it seems very unlikely that they'd be generated by the DUT.

 

[GeekyBastard]

Jcally JM20 with the same Savitech doesn't behave in this way (Sonata BHD does), when you start playing after it's gone into muting status the very first samples are not truncated\faded but instead you can hear a very brief background pink noise. What is preferable it's to you.

The JM20 uses SA9227, DS2 & Sonata BHD uses SA9312, I don't think it's comparable.
After reading SA9227's documents, I think you're right, it only supports up to DSD128, while JM20 supports up to DSD256, so it could be using a 9312 chip.

 

[ObjectiveSubjectivist]

To say it's compact is an understatement.

If you are looking for something smaller and lighter check Moondrop Dawn Pro

 

[ObjectiveSubjectivist]

Unfortunately I have a very shaky connection with the 4.4 mm input of the Fosi. If I insert the plug all the way in, I get no sound. If I pull the plug out again a little, I get the R-channel, the then L-channel - then at some point both. This could indicate that the stop of the socket is not correct.

Can anyone confirm this? It doesn't have to be a fault with the 4.4 socket of the Fosi, it could of course also be due to my XLR adapter. This may be what you get when the plug/adapter is larger than the device to which it is connected to.

Unfortunately, I don't have another device on which I could test my adapter.

View attachment 396517

It seems not to be fully inserted.

 

[mc.god]

The JM20 uses SA9227, DS2 & Sonata BHD uses SA9312, I don't think it's comparable.

I also supposed JM20 using SA9227 since got it working with "SA9227" drivers from HiFime and not being able to find specific information about the JM20.
But then I discovered that the same driver is fine with Sonata BHD, probably 9312 is just a refresh on the same architecture of 9227 (Bravo HD), so I'm now not sure which chip the JM20 is using, and unless you have precise information in regard I'll be assuming it is the newer 9312.

 

[GeekyBastard]

I also supposed JM20 using SA9227 since got it working with "SA9227" drivers from HiFime and not being able to find specific information about the JM20.
But then I discovered that the same driver is fine with Sonata BHD, probably 9312 is just a refresh on the same architecture of 9227 (Bravo HD), so I'm now not sure which chip the JM20 is using, and unless you have precise information in regard I'll be assuming it is the newer 9312.

After reading SA9227's documents, I think you're right, it only supports up to DSD128, while JM20 supports up to DSD256, so it could be using a 9312 chip.

But we need to sign an NDA for Savitech to provide us with firmware tools, I think that's impossible to complete.

 

[Kuma]

Received this a couple of days ago. Can confirm it’s a direct plug-in to an iPad running iOS18. The provided cable works fine. The Wall has never sounded better from a tiny dongle. What a great time of audio we’re living in.

 

[AndiGee]

The unit continues to give me random sound dropouts and fading truncations at the beginning of music tracks.

While the fades are already described in this thread, I'm wondering if anyone is also encountering those random dropouts. So far, those are only happening on Windows 11. Android and Windows 10 seem to be fine, I will also test OSX and report later.

I'm using foobar and Tidal player. It does not matter if I use exclusive mode, or the Windows standard audio output.

 

[roladyzator]

The unit continues to give me random sound dropouts and fading truncations at the beginning of music tracks.

While the fades are already described in this thread, I'm wondering if anyone is also encountering those random dropouts. So far, those are only happening on Windows 11. Android and Windows 10 seem to be fine, I will also test OSX and report later.

I'm using foobar and Tidal player. It does not matter if I use exclusive mode, or the Windows standard audio output.

I am getting random dropouts with the Fosi DS2 on my Windows 10 work laptop (Lenovo L14 Gen2), and my Samsung Note 10 Lite phone.
No dropouts on my Lenovo T460S laptop, that's also running on Windows 10.

On my phone, I use UAPP. I tried playing with sample rates, buffer sizes and so on.
I thought that there was a difference (less problems with larger buffer sizes and higher sample rates), but still got the problem every now and then.
Every now and then means that there can be no dropout in 15 minutes, then few dropouts within half a minute.
It may or may not have something to do with what I'm doing on my PC / phone.

Can you run tools like DPCLatencyChecker or LatencyMon and see the difference between your Windows 10 and Windows 11?
I don't have admin privileges on my work laptop to use those tools, but possibly some driver is causing this (LatencyMon would show you which).
As a corporate notebook, it has Cynet, which operates in the kernel space and may affect things like audio streaming.

EDIT: I'm not getting any dropouts with my Creative GC7 or Samsung USB-C Dongle.
I used to have such dropouts on Android only with a Shanling M0 until some firmware revision which specifically addressed USB audio stability.

 

[roladyzator]

To add to the above, there is even the Fidelizer tool, which does disable a lot of services in Windows and leave only audio-related ones.
The more hardcore presets even shuffle all non-audio stuff to specific CPU cores.

This used to be free, now the more extreme measures are paid.

Obviously any claims of improved audio quality besides removing stuttering / droputs (if there was any) are purely subjective and no objective measurements were ever provided.
He did try to use a loopback using Virtual Audio Cable (not the bit-perfect "Hi-Fi Cable" mind you).
Fell free to read if you you want to have a bit of laugh

Measuring digital audio qualities of bit-perfect playback with Diffmaker’s correlation depth | Fidelizer.Audio

It's been a challenge to measure digital audio's qualities and most of the time audiophiles don't know any measurement outside RMAA's analog metrics and got failed evaluations as you can see below: This was done through pure software environment using VB-Audio Virtual Cable to make...

fidelizer-audio.com

The guy has expanded and is now selling "audiophile" rebranded network switches and the like.

 

[AndiGee]

Every now and then means that there can be no dropout in 15 minutes, then few dropouts within half a minute.
It may or may not have something to do with what I'm doing on my PC / phone.

Exactly those.

Thanks for your advices, I'm really not sure if I declare investigating faulty audio devices to be my new hobby - but hey, what else can we do on a stormy, rainy afternoon? I ordered directly from Fosi - cost free returns seem to be something unheard of in China, otherwise the little guy would be on his way home by now.

 

[Labjr]

I have one of these on order direct from China. I wonder if I should try to cancel before they ship?

 

[ptrg]

Anyone use DS2 with Google Pixel 8? Any issues? Thank you.

I've just received mine. With Pixel 8, it works like a charm.
As 7jm wrote, the volume control of DS2 is independent of Android's one.

btw, hi everybody! (As usual, longtime lurker, finally signed up.)

 

[Hasan Aydin]

@amirm @MC_RME
What about the CS43131 click issue as described here: https://reference-audio-analyzer.pro/en/demo-cs131.php#gsc.tab=0

 

[GeekyBastard]

@amirm @MC_RME
What about the CS43131 click issue as described here: https://reference-audio-analyzer.pro/en/demo-cs131.php#gsc.tab=0

Interesting read, also wonders about if those settings would affect the SQ.

 

[GeekyBastard]

Then it seems to be a firmware issue rather than a hardware issue.
In the meantime, I contacted Fosi about this behavior. So far they don't seem to understand the problem.

Guys, I think we might be blaming the wrong guy this whole time…

After carefully reading every page of the CS43131 chip's document, this part caught my eyes:

These settings might actually be the problem, especially the soft ramp and auto mute function.

From what I understand, it takes time for the DAC to get out of the muted state, so it could cause the first half second of the audio to be lost, the auto mute function is the most suspected.

@Fosi Audio Maybe ask your engineers about this, and try to solve this annoyance, it could be as simple as changing literally several bits in your firmware.

 

[mc.god]

Guys, I think we might be blaming the wrong guy this whole time…

After carefully reading every page of the CS43131 chip's document, this part caught my eyes:

View attachment 398062

These settings might actually be the problem, especially the soft ramp and auto mute function.

From what I understand, it takes time for the DAC to get out of the muted state, so it could cause the first half second of the audio to be lost, the auto mute function is the most suspected.

@Fosi Audio Maybe ask your engineers about this, and try to solve this annoyance, it could be as simple as changing literally several bits in your firmware.

It can be possible, I don't have necessary knowledge to interprete how the implementation of those functions would translate into real world user experience, all i can say is that other than Jm20 (that produces pink noise when unmuting) and Sonata BHD (that cuts the very first audio bits) I also own a Ugreen HiFi Audio Pro with the CS43131 controlled via a CS41L46 used as bridge and it doesn't produce any strange behavior when unmuting or in any other circumstances.
Note that CS41L46 has the downside of working in SYNC USB mode instead of ASYNC, resulting into much worse jitter measures like this:

What we can desume from all of this, honestly I don't know

 

[Hasan Aydin]

Interesting read, also wonders about if those settings would affect the SQ.

It obviously can affect SQ. A piece from a song was used as an example (this can also be downloaded). The clicks occur depending on the structure of the signal and DAC Volume. I will try it out later with my Sonata BHD Pro.

These clicks were also confirmed for the SMSL M300 in the last 4 comments here and a few pages further: https://www.audiosciencereview.com/forum/index.php?threads/new-s-m-s-l-m300-se-120.44485/page-6


I don't want any random clicks in the signal that weren't there before. Especially when I'm doing audio editing...