SMSL DO100 Review (Stereo DAC)

[Sokel]

When does that peak occur?

It's not one,what you see is a sum of about a minute.Approximately at 2.46, 2.51, 2.57, 3.04, 3.12, 3.18, 3.25 and there's more after.

 

[staticV3]

If @amirm doesn't have time to check this, I have both the M2 and UltraLite-mk5 here and can measure their filters (outputting white noise into RME ADI-2).

I just got done measuring the 9038Q2M's PCM filters actually: Graph

Maybe my method could be useful to you.

 

[PeteL]

Not today, it is announced however: https://www.digitaltrends.com/home-theater/qualcomm-s3-s4-snapdragon-sound-lossless-audio-bt-le/
That said, IDK why BT and not wi-fi, maybe because wi-fi is not proprietary?

Did you mean why wifi can have lossless and not BT? Wifi will use more power, and unlike Bluetooth whih is simply point to point streaming wifi is IP based therefore you have a whole set of network protocols to conform too, with packet verifications traffic control, etc. your streaming device will need some Internet based interface to stream AOIP, it's a more complicated venture and it's partly why Network streamers are still quite expensive, you need some computing that's why the cheapest ways to do wifi rihght now is something like a Raspberry Pie because it's a mini computer that already have all that, not just an integrated circuit chip.

 

[respice finem]

Did you mean why wifi can have lossless and not BT? Wifi will use more power, and unlike Bluetooth whih is simply point to point streaming wifi is IP based therefore you have a whole set of network protocols to conform too, with packet verifications traffic control, etc. your streaming device will need some Internet based interface to stream AOIP, it's a more complicated venture and it's partly why Network streamers are still quite expensive, you need some computing that's why the cheapest ways to do wifi rihght now is something like a Raspberry Pie because it's a mini computer that already have all that, not just an integrated circuit chip.

OK what do I know about BT after all... But can BT stream 1080p for instance? Cause my cheap phone via WIFI can... So it seems at least bandwidth capable.

 

[PeteL]

OK what do I know about BT after all... But can it stream 1080p for instance? Cause my cheap phone via WIFI can...

Bluetooth don't stream video no.

 

[mdsimon2]

If @amirm doesn't have time to check this, I have both the M2 and UltraLite-mk5 here and can measure their filters (outputting white noise into RME ADI-2).

Please do. See below for my measurements of a MOTU M4 and an Ultralite Mk5 which look very different from the ideal filter measurements previously posted. This is on a Mac but I also tried Linux and got the same results. Playback sample rate is 44.1 kHz, capture device is a Cosmos ADC running at 192 kHz. Stimulus is -10 dBFS periodic white noise.

Ultralite Mk5

M4

Michael

 

[AdamG]

Lucky that the display stopped at “C” as it was one letter from “D” and the resulting acronym “DOA” would not bode well for product placement.

 

[manisandher]

See below for my measurements of a MOTU M4 and an Ultralite Mk5 which look very different from the ideal filter measurements previously posted.

Interesting. I'll check my M2 and UltraLite (but it'll have to wait until the weekend).

Mani.

 

[PeteL]

Please do. See below for my measurements of a MOTU M4 and an Ultralite Mk5 which look very different from the ideal filter measurements previously posted. This is on a Mac but I also tried Linux and got the same results. Playback sample rate is 44.1 kHz, capture device is a Cosmos ADC running at 192 kHz. Stimulus is -10 dBFS periodic white noise.

Ultralite Mk5
View attachment 208656

M4
View attachment 208657

Michael

The M4 filter is pretty much text book perfection.... But for a 48 KHz sample rate... Are you absolutely certain that nothing weird happened in the chain and some 48k resampling somehow was triggered? Not doubting your method but its so damn bang on. Especially since ADC is running at 192. any way you could retry this with ADC at 176.4? Sometimes computer os does some of those oddities.

 

[BeerBear]

So, you think Amir measured the UltraLite and M2 incorrectly, but the M4 correctly?

I strongly suspect that Amir's measurements for the UltraLite, M2 and M4 are all correct. For whatever reason, MOTU seem have gone with a 'slower' filter for the M4.

I don't know, it's rather unlikely that he messed up both measurement. It's possible that MOTU used different settings, not just for the M2 and M4, but also for different versions/firmwares of the same model. The mdsimon2's measurements above hint at that possibility too.

One thing that sticks out about the M2 is the high round trip latency, as measured by @Julian Krause in his review. The latency is normal up to until about 64 samples, when it starts to drift further and further away from the norm. At 256 samples it's already 10ms higher than typical. Maybe this is caused by a higher number of filter taps, which also improves the filter response (just guessing here, though).
If you're doing filter measurements perhaps also measure the RTL, just to see if there's a correlation.

 

[mdsimon2]

The M4 filter is pretty much text book perfection.... But for a 48 KHz sample rate... Are you absolutely certain that nothing weird happened in the chain and some 48k resampling somehow was triggered? Not doubting your method but its so damn bang on. Especially since ADC is running at 192. any way you could retry this with ADC at 176.4? Sometimes computer os does some of those oddities.

A few thoughts on the validity of my measurements...

1) I always appreciate peer review which is the primary reason I asked @manisandher to replicate the measurements as he has the same DAC (Ultralite Mk5) but a different ADC
2) As mentioned I used two different OS, mac and linux (ALSA), so OS related resampling seems unlikely as the results were identical
3) I can say that 1 kHz FFTs do not indicate any resampling for either DAC, when ALSA or Core Audio are resampling the results are evident on a 1 kHz FFT
4) My results for the M4 exactly match Amir's results for the M4 although to be fair it was his measurements of the M2 and Ultralite Mk5 which were in question, not the M4
5) My M4 is a first generation with an ES9016 DAC which has rather limited filter options compared to newer ESS options, the filter response appears to exactly match @staticV3's measurements of the ES9038Q2M for the linear phase fast roll off filter which is one of the options for the ES9016
6) The Ultralite Mk5 results seem to exactly match the minimum phase fast roll off filter as measured by @staticV3 for the ES9038Q2M, from impulse response measurements of the Ultralite Mk5 the results seem to be minimum phase

Michael

 

[Steamrolly]

If it had a good headphone output, it'd be perfect!

So what you want is the 2022 Sabaj Audio A10d which is basically a DO100 with headphone jacks and handles. Whether the headphone output is good or not is yet to be seen.

Sabaj Audio Official-Sabaj音频官网

Since the establishment of our company, we have been committed to an enterprise integrating audio product design, R&D, production, market operation and service.There are mainly three categories: decoder/headphone amplifier/power amplifier. Our company is composed of a group of members who love...

www.sabaj.com.cn

 

[PeteL]

A few thoughts on the validity of my measurements...

1) I always appreciate peer review which is the primary reason I asked @manisandher to replicate the measurements as he has the same DAC (Ultralite Mk5) but a different ADC
2) As mentioned I used two different OS, mac and linux (ALSA), so OS related resampling seems unlikely as the results were identical
3) I can say that 1 kHz FFTs do not indicate any resampling for either DAC, when ALSA or Core Audio are resampling the results are evident on a 1 kHz FFT
4) My results for the M4 exactly match Amir's results for the M4 although to be fair it was his measurements of the M2 and Ultralite Mk5 which were in question, not the M4
5) My M4 is a first generation with an ES9016 DAC which has rather limited filter options compared to newer ESS options, the filter response appears to exactly match @staticV3's measurements of the ES9038Q2M for the linear phase fast roll off filter which is one of the options for the ES9016
6) The Ultralite Mk5 results seem to exactly match the minimum phase fast roll off filter as measured by @staticV3 for the ES9038Q2M, from impulse response measurements of the Ultralite Mk5 the results seem to be minimum phase

Michael

Fair enough, yeah I see now it’s M4 and not M2,my bad so yeah same exact thing. I did not know there was a Mk5 controversy when reading trough its was the M2 that was in question. So what do you think could be the problem with The Mk5 measurments Amir Got? Firmware different between your unit and his perhaps?

 

[mdsimon2]

Fair enough, yeah I see now it’s M4 and not M2,my bad so yeah same exact thing. I did not know there was a Mk5 controversy when reading trough its was the M2 that was in question. So what do you think could be the problem with The Mk5 measurments Amir Got? Firmware different between your unit and his perhaps?

I have upgraded the firmware on my Mk5 which solved the high sample rate issues observed in Amir's review, it is certainly possible that this could have resulted in other unexpected changes.

The reason I measured my Mk5 is @manisandher posted in this thread (post #61) that he wondered why other manufacturers could not achieve the same results as MOTU did with the Mk5. This then prompted the comments by @Tks and others regarding the validity of the M2 measurements (which look identical to the Mk5 measurements).

At this point I agree with others that @amirm's measurements of the M2 and Mk5 were likely inaccurate.

Michael

 

[PeteL]

I have upgraded the firmware on my Mk5 which solved the high sample rate issues observed in Amir's review, it is certainly possible that this could have resulted in other unexpected changes.

The reason I measured my Mk5 is @manisandher posted in this thread (post #61) that he wondered why other manufacturers could not achieve the same results as MOTU did with the Mk5. This then prompted the comments by @Tks and others regarding the validity of the M2 measurements (which look identical to the Mk5 measurements).

At this point I agree with others that @amirm's measurements of the M2 and Mk5 were likely inaccurate.

Michael

My guess is that would be the firmware differences, it’s totally plausible. Bottom line, ESS has a programmable filter, but sometimes in dev people would go in tunnel vision, May have worked hard in creating this perfect filter not even considering other issues it created. Then reverted back to stock filters that work fine.. There is no reason to think that Amir’s measurment rig would all of a sudden start giving false results with the same process for now a couple hundred DACs. Audio Precision is the gold standard, it just works. I really think it’s the firmware.

 

[Garrincha]

So what you want is the 2022 Sabaj Audio A10d which is basically a DO100 with headphone jacks and handles. Whether the headphone output is good or not is yet to be seen.

Sabaj Audio Official-Sabaj音频官网

Since the establishment of our company, we have been committed to an enterprise integrating audio product design, R&D, production, market operation and service.There are mainly three categories: decoder/headphone amplifier/power amplifier. Our company is composed of a group of members who love...

www.sabaj.com.cn

Looks like a really nice stylish, tiny device, maybe the headphone amp could be a bit more powerful. But I just ordered the Topping Dx3Pro+, so next time.

 

[staticV3]

The Ultralite Mk5 results seem to exactly match the minimum phase fast roll off filter as measured by @staticV3

And the Motu M4's result seems to match Linear phase fast roll-off
Which is kind of an odd choice for studio equipment where you want low latency

 

[PyramidElectric]

And the Motu M4's result seems to match Linear phase fast roll-off
Which is kind of an odd choice for studio equipment where you want low latency

So 'Linear phase fast roll-off' obviously doesn't fully filter 22-24k, but since the signal shouldn't have any signal above 22k anyway, what exactly is getting through here? What are the consequences?

 

[staticV3]

What are the consequences?

Aliasing will occur

 

[PyramidElectric]

Aliasing will occur

Are the A-D filters not meant to filter the 20-22k stuff during recording? Or are you talking about worst case stuff when that's not done properly and then you basically get these aliased signals in the 22-24k range? (I need to read up on this I think)