• 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.

Measurement and Review of FX-Audio DAC-X6 DAC (compared to Topping D30 and Fiio E10K)

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
15,369
Likes
9,360
Location
Seattle Area
#41
The thd+n 1% graphs, and these type of graphs moving forward. Am I projecting or is vrms more familiar to people? Can you add the fiio e10k in there as well or does this test only apply to devices with a line in?
In general, Vpp is a more accurate measure because it doesn't rely on conversion to RMS which may or may not be accurate in the instrument. That's why I picked it. But I don't have a strong feeling about it.

That aside, I went to survey what others do and it is all over the place when it comes to headphone amp measurements. I am going to create a new thread and see if we can land on a specific set of measurements and units.

Anyway, here is that graph (without Exasound) in Vrms.

headphone THD+N % Vrms.png
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
15,369
Likes
9,360
Location
Seattle Area
#42
The FX-Audio seems to be doing a classic rendition of high frequency nonlinear distortion. I see lot of this. A CCIF IM (19+20 KHz) test would seem to be in order. A pair of swept tones a constant 1 KHz apart starting at say 5+6 kHz could also be enlightening. I have a ESS 9018-based USB stick that does this. Again it is not uncommon. Subjectively, this sort of thing might cause audible problems with certain fairly atypical but natural sounds like my infamous "Keys Jangling" sample.
Thanks for the suggestion. I will make that measurement and post the results. I will also try different sampling rate just in case that has something to do with it as suggested. I think all of these detailed tests are done at 48 Khz. I repeated one of them at 44.1 Khz and it made little difference.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
15,369
Likes
9,360
Location
Seattle Area
#43
Can you add the fiio e10k in there as well or does this test only apply to devices with a line in?
Forgot this one. For now, I can only run these tests with units that have S/PDIF input as that allows my audio analyzer to control the signal, unlike USB where a media player would do so asynchronously. Longer term I will get them working with USB too but am not there yet. So for that reason I can't include Fiio yet.

But yes, analog line in also works (i.e. for headphone amplifier measurements as opposed to DACs) although not as good as we will then bring in the distortion of analog signal generator into play.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
15,369
Likes
9,360
Location
Seattle Area
#44
Seriously, even a Fiio E10K drive headphones like the HD600/650 and 250 Ohm Beyerdynamics decently.
They can do that when the music is recorded near or at full amplitude. When testing with my more "audiophile recordings" that are well away from the peaks, I find that the extra gain that these external devices provide to be quite useful.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
5,574
Likes
1,542
Location
Riverview, FL
#45
Am I projecting or is vrms more familiar to people?
Vrms = .7071 Vpk for a sine wave (test tone)

or

Vpk = 1.4142 Vrms going the other way.

(Unless I need to be corrected)
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
5,574
Likes
1,542
Location
Riverview, FL
#46

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
15,369
Likes
9,360
Location
Seattle Area
#47
Vrms = .7071 Vpk for a sine wave (test tone)

or

Vpk = 1.4142 Vrms going the other way.

(Unless I need to be corrected)
That works until distortion and noise become dominant.
 

mindbomb

Active Member
Joined
Nov 11, 2017
Messages
276
Likes
130
#48
vrms is 0.35*vpp, right? I think vrms is more common, and it also makes it slightly easier to get rms power of a headphone amp.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
15,369
Likes
9,360
Location
Seattle Area
#49
vrms is 0.35*vpp, right? I think vrms is more common, and it also makes it slightly easier to get rms power of a headphone amp.
No, I as I mentioned, that ratio depends on the waveform itself. Here is a nice table from Wiki:

upload_2017-12-2_17-27-16.png


In this case I am using sine wave but at low amplitude we are mostly measuring noise and at high amplitude, we have a clipped clipped sine wave of some sorts.

In practice here it probably doesn't matter much but there it is. :)
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
15,369
Likes
9,360
Location
Seattle Area
#50
This is yet another THD+N measurement versus frequency in stereo (pairs of lines for each):

View attachment 9629
Per request, he is comparison of this measurement at 48 Khz vs 44.1 Khz:

upload_2017-12-2_21-9-32.png


Red is 44 Khz and yellow is 48. We see that at 20 Khz things actually get worse for 44.1 Khz. And we still have this odd peak at 5 to 6 Khz.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
15,369
Likes
9,360
Location
Seattle Area
#51

Soniclife

Addicted to Fun and Learning
Joined
Apr 13, 2017
Messages
689
Likes
276
Location
UK
#53
Next let's look at ability to reproduce the lower two bits of 16-bit audio, attempting to approximate a sine wave:

Thanks for doing this work, really interesting stuff.
Regarding these low level sine waves, how quickly do different dacs start producing nice output if you raise the input level? Do they all improve to pretty graphs by -85db, or do some need a lot more? This X6 looks like it will need loads more input, to play nice.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
15,369
Likes
9,360
Location
Seattle Area
#54
Thanks for doing this work, really interesting stuff.
Regarding these low level sine waves, how quickly do different dacs start producing nice output if you raise the input level? Do they all improve to pretty graphs by -85db, or do some need a lot more? This X6 looks like it will need loads more input, to play nice.
My pleasure. I only did one quick test at something like -85 db and there was still noise in the bad DACs. I will repeat and post the results.
 

mindbomb

Active Member
Joined
Nov 11, 2017
Messages
276
Likes
130
#55
speaking of low levels, can you check the volume pot. specifically, how low you can go before channel balance gets really bad?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
15,369
Likes
9,360
Location
Seattle Area
#56
speaking of low levels, can you check the volume pot. specifically, how low you can go before channel balance gets really bad?
Oh sorry, I forgot to do that. I remembered a couple of days ago but forgot it again. :D
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
5,574
Likes
1,542
Location
Riverview, FL
#57
Oh sorry, I forgot to do that. I remembered a couple of days ago but forgot it again.
Maybe you can reveal which units have Digital and which have Analog volume control?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
15,369
Likes
9,360
Location
Seattle Area
#58
Here are the results. I started the frequency response test with volume at max and then manually reduced it a few notches at a time (steps were not the same). Both channels are measured and if they are on top of each other, then matching is very good. This occurs at higher levels but then the matching starts to get lost:

FX-Audio DAC-X6 channel matching.png


There was one deviation at higher volumes but it was small. Matching became good until about 9:00 o'clock position or bottom 1/6 or so at -45 db. As you see there, the two channels -- yellow and green -- have separated already. I went down a couple of notches and everything fell apart. One channel was at -65 and the other at -85 db or so. Seeing how the max volume was at +7 db, useable attenuation is around 50 db.

Subjectively, I listened with my Sennheiser HD600. There, the lowest volume I would want to use on some of my loudest clips was at 11:00 o'clock position. On Sony MDR-V6 headphone, that was around 10:00 o'clock. With the Hifiman HE400i minimum level was also 10:00 o'clock. So from listening point of view with these three headphones the above measured problem is not an issue. Would need a much more efficient headphone to have it be a problem.

Let me know if this was not what you were looking for.
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
4,382
Likes
2,681
#60
I have seen this in much gear that uses pots for volume like preamps. Usually I found if you tested resistance of the total pot for each channel they differed a bit. Maybe a 47 k pot would have 47,100 ohms on one side and 45,900 ohms on the other. I found adding a metal film resistor to trim up both sides to the same total helped with tracking. It didn't make them all perfect, but it often made for dramatic improvements below -20 db. Of course modern surface mount gear sometimes makes that tough to do.
 
Top Bottom