Fosi Audio DS2 Portable DAC & Amp Review

[Jim Creek]

Amir standard set of measurements is a proxy for “good engineering”. It doesn’t mean it covers everything… As @jkim showed in their measurements, the CS hump is relatively easy to detect if you know what to look for, but it’s true audibility is debatable.
It doesn’t invalidate Amir’s assessment at all. The disappointing part IMO are the OEMs: they should know better and pay umpattention to these “details”.
Not sure about Fosi…

Amir does a great job. My question is why didn’t this so called distortion hump show up in his distortion measurements?

 

[staticV3]

Amir does a great job. My question is why didn’t this so called distortion hump show up in his distortion measurements?

Amir's APx555B has high inherent noise in tests where the internal Notch filter cannot be used, like IMD tests for example.

This tends to hide humps and the like.

The early ESS hump was highly elevated and so still shows up in APx555 measurements, but less obvious humps like the one on Cirrus DACs are easily masked by the noise.

Higher FFT size would be required to compensate, which exponentially increases data acquisition time, which Amir likely won't invest.

 

[torgeirs]

Higher FFT size would be required to compensate, which exponentially increases data acquisition time, which Amir likely won't invest.

The cs hump is a noise hump i my opinion. I think it should show up in amirs measurements.
My understanding is that the hump is the rise of noise in the last desibels

 

[torgeirs]

Is this just the measurement of the AP internal noise?

 

[torgeirs]

Yes mc_rme measurements has the hump

 

[torgeirs]

Tilt my graph 30 degrees to get the same slope

 

[staticV3]

The cs hump is a noise hump i my opinion. I think it should show up in amirs measurements.
My understanding is that the hump is the rise of noise in the last desibels
View attachment 461298

Two different phenomena. There's an unintentional IMD (=distortion) hump, as well as a purpose-built DRE mechanism in place (signal amplitude modulates noise floor).

Is this just the measurement of the AP internal noise?

No. It's a somewhat flawed measurement setup with 16-bit truncation.

Yes mc_rme measurements has the hump

Again: not the hump that we're talking about.



You can read more about it here:

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

TL;DR / Changelog [5/20/25] Eleven DAC/HP amp combo devices adopting either the Cirrus Logic CS43131 or CS43198 were tested, and seven of them were found produce the distortion dubbed a "Cirrus hump." Based on measurements from multiple sources (including this review), a list of...

www.audiosciencereview.com

 

[torgeirs]

Ok, so a camel and not a dromedar
Almost 0.1 %, thats is almost a artifact!
Thanks for the link. And confirming that Amirs graph are flawed in this measurement

 

[Xombul]

A big thank you to Amir for the review.

The DS2 is one of our products as we make our entry into the high-performance DAC market. The DAC market is filled with many outstanding companies with which everyone is familiar. We will humbly learn from these companies as we continue to grow, and we hope to follow in their footsteps by bringing Fosi Audio’s unique touch to DACs for audiophiles worldwide. In addition to portable DACs, we’ll soon be introducing a desktop DAC as well.

To show our appreciation for the support from ASR members, we’re offering a special discount code for the DS2. You can use this code to get 20% off your DS2 purchase!
ASRDS220

link:
https://fosiaudio.com/products/fosi-audio-ds2-2024-dac-headphone-amplifier
https://www.amazon.com/dp/B0CTHN2QB3
https://www.amazon.de/dp/B0BJ66KJ2B
https://www.amazon.com.au/dp/B0BJ66KJ2B
https://www.ebay.com/itm/296311456320

Could you please send the link for Amazon France thank you.

 

[torgeirs]

Amir does a great job. My question is why didn’t this so called distortion hump show up in his distortion measurements?

Looks like the fosi DS2 is not affected by that hump. https://docs.google.com/spreadsheets/d/1fYP8cqPMtZigR1VdPIHaax5nyGDaHZpKeS1w3AKd5RI

 

[jkim]

Amir does a great job. My question is why didn’t this so called distortion hump show up in his distortion measurements?

Amir's APx555B has high inherent noise in tests where the internal Notch filter cannot be used, like IMD tests for example.

This tends to hide humps and the like.

The early ESS hump was highly elevated and so still shows up in APx555 measurements, but less obvious humps like the one on Cirrus DACs are easily masked by the noise.

Higher FFT size would be required to compensate, which exponentially increases data acquisition time, which Amir likely won't invest.

View attachment 461270

Looks like the fosi DS2 is not affected by that hump. https://docs.google.com/spreadsheets/d/1fYP8cqPMtZigR1VdPIHaax5nyGDaHZpKeS1w3AKd5RI

It seems that you guys missed what I posted yesterday to this thread.

Amir's tests do not show the Cirrus hump simply because his standard package of measurements does not include a stepped sweep of multitone tests (needed to show the hump). The APx555B's noise performance is sufficient to show this artifact because the Cirrus hump distortion is much higher than the "ESS hump" (which became even milder in recent iterations of ESS chips).

EDIT. Also see Part II of my review. The DS2 (or JCally JM20 or JM20 Max) is not completely free of this distortion phenomenon. The 'crunch' or 'clicking' is still audible in some test clips.

 

[jkim]

By the way, I added to my Cirrus hump distortion thread a post about some listening tests and my take on the audibility of the distortion:

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

In the list of devices ONIX Alpha XI1 not producing hump, but we don't know firmware version, which one of this versions 1.0.3, 1.1.3, 1.2.3. And can someone please give me the DAC measurements of Onix Alpha XI1. Can't find any, there is even no PCB pic)

audiosciencereview.com

 

[Jim Creek]

It seems that you guys missed what I posted yesterday to this thread.

Amir's tests do not show the Cirrus hump simply because his standard package of measurements does not include a stepped sweep of multitone tests (needed to show the hump). The APx555B's noise performance is sufficient to show this artifact because the Cirrus hump distortion is much higher than the "ESS hump" (which became even milder in recent iterations of ESS chips).

EDIT. Also see Part II of my review. The DS2 (or JCally JM20 or JM20 Max) is not completely free of this distortion phenomenon. The 'crunch' or 'clicking' is still audible in some test clips.

Question #2: Why are some reviewers claiming there is no hump in certain CS431xx devices like the Fosi DS2? https://docs.google.com/spreadsheets/d/1fYP8cqPMtZigR1VdPIHaax5nyGDaHZpKeS1w3AKd5RI/edit?gid=0#gid=0

So who’s measurements are correct and who’s measurements are incorrect? I’m confused!

I’ve been listening to my jazz collection with the DS2 and AKG371 headphones. I hear nothing but transparency, no artifacts. My ears are good, not great, but good.

 

[torgeirs]

I measured my FOSI DS2 myself now with one of the methods described in the links.
Maybe it is tainted with the hump 30 dB lower than worst case, but at those low levels it is no problem for me, at least

 

[torgeirs]

So who’s measurements are correct and who’s measurements are incorrect? I’m confused!

They are all correct just measuring different things. JKIMs measurements are very thorough to visualize this particular artifact.
Seems like some of the DACS are severely affected and others much more less. Think for DS2 the signal has to be very unfortunate to the artifact to be heard.

 

[torgeirs]

Started a 32 tone stepped response now, but they take a lot of time to run


Some raised noise in low frequency, but not bad at all

 

[staticV3]

Amir's tests do not show the Cirrus hump simply because his standard package of measurements does not include a stepped sweep of multitone tests (needed to show the hump)

The hump shows up in traditional CCIF IMD as well, as you can see in the graph I posted, given that your ADC has low enough noise.

 

[jkim]

Question #2: Why are some reviewers claiming there is no hump in certain CS431xx devices like the Fosi DS2? https://docs.google.com/spreadsheets/d/1fYP8cqPMtZigR1VdPIHaax5nyGDaHZpKeS1w3AKd5RI/edit?gid=0#gid=0

So who’s measurements are correct and who’s measurements are incorrect? I’m confused!

I’ve been listening to my jazz collection with the DS2 and AKG371 headphones. I hear nothing but transparency, no artifacts. My ears are good, not great, but good.

They are all correct just measuring different things. JKIMs measurements are very thorough to visualize this particular artifact.
Seems like some of the DACS are severely affected and others much more less. Think for DS2 the signal has to be very unfortunate to the artifact to be heard.

@torgeirs is correct. The measurements had no problem no matter who measured the devices. It is just about which aspect and which device is measured.

@Jim Creek Please read my review, especially Part II. There are two kinds of devices: One is with the full "Cirrus hump" distortion and the other's distortion is detected only when a certain signal's onset. The DS2 is the latter kind. Also, carefully follow the instructions given in this post. You will be able to hear the distortion in the C Major test file at the onset of each iteration of the signal.

 

[jkim]

I measured my FOSI DS2 myself now with one of the methods described in the links. View attachment 461345
Maybe it is tainted with the hump 30 dB lower than worst case, but at those low levels it is no problem for me, at least

What you showed is NOT the Cirrus hump. The Fosi DS2 should NOT exhibit a Cirrus hump when measured using a stepped test with steady signals, as described in Part II of my review. The JCally JM20 or JM20 Max's TDFD-Bass IMD measurements are just like what you showed above. These DAC's TDFD-Bass IMD levels (when measured with steady test signals) are normal and low enough to compete with the best DACs.

 

[jkim]

The hump shows up in traditional CCIF IMD as well, as you can see in the graph I posted, given that your ADC has low enough noise.

You are comparing apples to oranges. The AP measurements of SMPTE IMD, as defined by AP, includes noise whereas the REW measurements you showed do not include noise. And AP's definition of SMPTE (DIN) appears to be different from TD+N in REW, too.

And the Cirrus hump distortion does NOT occur in SMPTE IMD tests. The rise of SMPTE IMD in your graph is low enough and on par with recent ESS DACs'. This SMPTE IMD rise has a very different cause.