• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Measurements Of SMSL M8 DAC

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
5,208
Likes
3,942
#81
Ok thank you. So I am likely looking for an explenation in the wrong area.
Sorry for taking so long to get that.

Any clue as to why I'm hearing subtle differences between two good DACs then? :) And possibly between DAC chips?
But perhaps I've asked enough questions for today..
@JustIntonation
I was reminded of this today on another thread. Try listening to pink noise thru both DACs. You can hear rather unambiguously FR differences with pink noise. Don't listen too loud. Now FR differences aren't the only thing that would effect the sound of pink noise, but it is a good signal.
 

andreasmaaan

Major Contributor
Patreon Donor
Joined
Jun 19, 2018
Messages
2,776
Likes
1,849
#82
@JustIntonation you could also try level matching the two DACs (if you have the ability to do this) and then have someone else switch between the two while you keep your headphones on and face the other way to see if you still feel you can reliably pick any difference blind.

Not to suggest there is no difference. But the effects of background knowledge can be profound and this might be a way to at least rule this possibility out.

But of course don't do this unless you can be satisfied that the levels are matched correctly.
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
5,208
Likes
3,942
#83
@JustIntonation you could also try level matching the two DACs (if you have the ability to do this) and then have someone else switch between the two while you keep your headphones on and face the other way to see if you still feel you can reliably pick any difference blind.

Not to suggest there is no difference. But the effects of background knowledge can be profound and this might be a way to at least rule this possibility out.

But of course don't do this unless you can be satisfied that the levels are matched correctly.
Good point by andreasmaaan. Level matching is important even in sighted listening comparisons. If you do it by ear you'll maybe get within a .5db. Unfortunately if the difference is more than .1 db your ear hears a quality difference, and not a volume difference. Even a cheap multimeter measuring the output with say a 400 hz (or 1 khz) test signal should do the trick. You need the output voltages to be closer than a 2% difference (preferably 1%).
 

JustIntonation

Senior Member
Joined
Jun 20, 2018
Messages
333
Likes
129
#84
I already sent the M8A v3 back. Can't do anymore listening.

I didn't think about pink noise. I have been using pink noise to detect effects of EQ on headphones, works very well! Can easily set the best way to EQ out resonant driver peaks that way. Very audible.

Level was possibly matched well enough. I used both DACs at full output which is stated as 2Vrms for both. (The measurements in this thread for the M8 show 1.95Vrms for the left channel and Anedio states 2.0 Vrms for unbalanced outs with max 0.1dB volume difference between channels which seems believable as their other specs measure as stated so that would be a 0.2dB difference if my M8A v3 were to have the same output as the M8 in this thread).

I remain convinced that the differences I heard are real. And they are infact the differences I thought I heard in not direct A/B comparison, only smaller than I thought in direct A/B on my nearfields.
I do see frequency response differences in the measurements of ES9038Q2M DACs vs the D2. The D2 measures ruler flat with only a small 0.15 drop off at 20KHz. The ES Q2M DACs tend to show a more wiggly line with larger frequency deviations (+- 0.3dB for instance which is 0.6dB total vs the 0.15dB of the D2 which happens only at 20KHz not below so can be seen as about 0dB in practice). I tried to counter this by upsampling the music to 176.4 KHz in Foobar2000 which hopefully gave less deviation from flat for the M8A v3. Though perhaps it is still there and this alone could have been enough to explain the percieved difference? (That is, if the M8A v3 measures as flat as the M8, there are no specs given for this for the M8A v3 by SMSL)

As for IMD, the D2 shows about -116dB IMD with a hard test of 0dBFS equal volume 19+20KHz tones. This seems to be better than any DAC measured on this site, even the OPPO 205.
The M8A v3 possibly does slightly worse than the D50 in this regard if we add up a similar bump to D50 to the M8 measurements as the M8A v3 uses the ES9038Q2M which all seem to show a +15dB or so extra bump in IMD in mid levels.
Its possible that the M8A v3 gives for instance 30dB more IMD at mid levels than the D2 (which is about where I heard things like hi-hats in my listening test, which were done at roughly -24dB set in Foobar giving possibly about -24dB to -36dB for most of the musical signal). So quite possible there's -60dB IMD for M8A v3 vs -90dB IMD for D2 in my listening, possibly even more for high frequency signals. -60dB translates to about 0.1%, perhaps audible for IMD? Audible as a quality difference, subtle forwardness of the highs in listening? Maybe.
Maybe it's a combination of minor frequency response differences, IMD, jitter, etc. The D2 measures significantly better in many of these aspects.
But then still.. I don't consider the following proven to myself but I can't shake the idea either that DAC chips have something of a subtle inherent sound where the ESS mobile chips I heard had something similar and the ESS full chips had something similar among them as well different from the mobile chips.. Perhaps I'm wrong about this but perhaps not, this would perhaps point in the direction of frequency response though as the full chips seem to display a more even frequency response than the mobile versions. However, I don't even hear differences when selecting different filters.. hmm.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
17,832
Likes
16,020
Location
Seattle Area
#85
As for IMD, the D2 shows about -116dB IMD with a hard test of 0dBFS equal volume 19+20KHz tones. This seems to be better than any DAC measured on this site, even the OPPO 205.
My IMD measurements use SMPTE tones which is sharply different than the ones you used above. So the results can't be compared.
 

JustIntonation

Senior Member
Joined
Jun 20, 2018
Messages
333
Likes
129
#86
My IMD measurements use SMPTE tones which is sharply different than the ones you used above. So the results can't be compared.
Ok. Was I wrong in thinking that a 19+20KHz 0dBFS IMD test is at least as hard for a DAC as an SMPTE IMD test? I thought to read between the lines that it was about slew rate and you said a two tone max level test is about as stressful as it gets and showed a picture of what appears to be 19+20KHz tones :)

edit: Nevermind. Looked up some results of both CCIF and SMPTE IMD measurements and sometimes one is higher sometimes the other. I see they indeed can't be compared the way tried to do.
 
Last edited:

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
5,208
Likes
3,942
#87
Ok. Was I wrong in thinking that a 19+20KHz 0dBFS IMD test is at least as hard for a DAC as an SMPTE IMD test? I thought to read between the lines that it was about slew rate and you said a two tone max level test is about as stressful as it gets and showed a picture of what appears to be 19+20KHz tones :)

edit: Nevermind. Looked up some results of both CCIF and SMPTE IMD measurements and sometimes one is higher sometimes the other. I see they indeed can't be compared the way tried to do.

The issue is the tone being used as a reference, 7 khz, starts 12 db lower in level. The predominate tone is at 60 hz. So even if they were directly comparable (which they aren't), you'd get several to 12 db worse results in the SMPTE version vs the one you used. I personally value the one you used as more meaningful, but the two variants can't be directly compared.
 

JustIntonation

Senior Member
Joined
Jun 20, 2018
Messages
333
Likes
129
#88
Well.. in another twist. My listening test between DACs was very compromised I just found out.
Kind of ashamed about it, but only fair to post it here.
I still had Sonarworks Reference 4 running as an ASIO plugin in Foobar2000. Still had it in the chain, bypassed of course, because I had just been testing some EQ on my headphones (don't like the Reference curve but it was one of the things I was comparing with). And I had an upsampler to 176.4 KHz running as I reported. I was just listening to some music and I don't remember why but just changed the order of the resamper to come after Reference 4 instead of before it as before. And it changed the sound!! Immediately removed Reference 4 completely from Foobar2000. And boy, what a great sound! I had just moved my nearfields 2 days ago to a temp position on small stands on my desk (long story) and when testing DACs I already though hmm this doesn't sound that great thinking it was due to the position of my nearfields. Was I stupid, it was Reference 4 in bypass mode messing up my sound (no I'm not talking about the lower level from Reference 4 in bypass mode, but sound quality being totally degraded).
I don't know yet if this is something Foobar2000 specific in combination with Reference 4 as an ASIO plugin, or 176.4 KHz specific for Reference 4 in general.
Anyhow, too bad I can't test against the M8A v3 anymore. Perhaps I was not all wrong with my earlier impressions that the difference is quite large between it and the D2, an opinion I formed on my main speakers but in less carefull A/B, but without Reference in the chain.
In any case, I did hear a subtle difference even with Reference bypassed but in the chain, but both DACs were at that point outputting pretty much garbage..
Again, ashamed I didn't notice just how bad it was right away and thought it was something else..
But very happy I now have absolutely fantastic sound on my nearfields as well! :)
I may test Reference bypassed in a DAW later on with different sample rates to see if the huge bug is there as well, though I may not as there is currently no software around I hate more..

edit: It also messed up my recent headphone EQ testing btw.. Up to the point where I thought I was mistaken about my DT1990 and already filled in a return request and ordered a HD800. I can now blame this to Reference being in the chain. Perhaps a good thing to come from this is I now get to do a good comparison between a HD800 and DT1990, in good quality this time.

edit 2: Have been testing the HD800 past 2 days and the Beyerdynamic DT1990 is simply better in almost every way. That is post EQ for both, pre EQ the HD800 is more listenable. And Sonarworks has in my opinion very bad EQ curves for both. Better go with the EQ curves and measurements by rtings.com (I use them without the bass bump).
 
Last edited:

andreasmaaan

Major Contributor
Patreon Donor
Joined
Jun 19, 2018
Messages
2,776
Likes
1,849
#89
Well.. in another twist. My listening test between DACs was very compromised I just found out.
Kind of ashamed about it, but only fair to post it here.
I still had Sonarworks Reference 4 running as an ASIO plugin in Foobar2000. Still had it in the chain, bypassed of course, because I had just been testing some EQ on my headphones (don't like the Reference curve but it was one of the things I was comparing with). And I had an upsampler to 176.4 KHz running as I reported. I was just listening to some music and I don't remember why but just changed the order of the resamper to come after Reference 4 instead of before it as before. And it changed the sound!! Immediately removed Reference 4 completely from Foobar2000. And boy, what a great sound! I had just moved my nearfields 2 days ago to a temp position on small stands on my desk (long story) and when testing DACs I already though hmm this doesn't sound that great thinking it was due to the position of my nearfields. Was I stupid, it was Reference 4 in bypass mode messing up my sound (no I'm not talking about the lower level from Reference 4 in bypass mode, but sound quality being totally degraded).
I don't know yet if this is something Foobar2000 specific in combination with Reference 4 as an ASIO plugin, or 176.4 KHz specific for Reference 4 in general.
Anyhow, too bad I can't test against the M8A v3 anymore. Perhaps I was not all wrong with my earlier impressions that the difference is quite large between it and the D2, an opinion I formed on my main speakers but in less carefull A/B, but without Reference in the chain.
In any case, I did hear a subtle difference even with Reference bypassed but in the chain, but both DACs were at that point outputting pretty much garbage..
Again, ashamed I didn't notice just how bad it was right away and thought it was something else..
But very happy I now have absolutely fantastic sound on my nearfields as well! :)
I may test Reference bypassed in a DAW later on with different sample rates to see if the huge bug is there as well, though I may not as there is currently no software around I hate more..
Nice to hear you have better sound now :) And good on you for letting us know what other factors were affecting your comparison.
 

JohnYang1997

Senior Member
Joined
Dec 28, 2018
Messages
382
Likes
114
#90
May I ask the lowest harmonic distortion values you can get from qa401 using rew with 1M fft length? And a screenshot?
I have focusrite forte at -20db from headphone output it has very very low distortion. Individual harmonics are below 0.00003%.
And i recently got motu 828es which has ak5574 and es9016. And the result is similar.
 
Top Bottom