Review and Measurements of E1DA 9038S BAL Portable DAC & Amp

[IVX]

Some comments on the DSD performance. Archimago always uses the same set of DSD files generated at -3dbfs, I used Foobar2000 DSD processor to online generate DoP DSD64-128 at 0dbfs. My THD+N result is slightly worse than PCM 32/96 -124db(PCM -125.5db, I guess a worse result caused by the fact that THD compensation is not active in DSD). Also, I noticed that Foobar2000 has noticeably less HF noise(30-100kHz) vs Archimago's files, so to get a reliable reading of THD+N with these files I used my passive LPF. Foobar2000 allows THD+N reading directly by Cosmos APU without the LPF.

 

[IVX]

a short video about 9039S as a sine source:

 

[Sokel]

Some comments on the DSD performance. Archimago always uses the same set of DSD files generated at -3dbfs, I used Foobar2000 DSD processor to online generate DoP DSD64-128 at 0dbfs. My THD+N result is slightly worse than PCM 32/96 -124db(PCM -125.5db, I guess a worse result caused by the fact that THD compensation is not active in DSD). Also, I noticed that Foobar2000 has noticeably less HF noise(30-100kHz) vs Archimago's files, so to get a reliable reading of THD+N with these files I used my passive LPF. Foobar2000 allows THD+N reading directly by Cosmos APU without the LPF.

Archimago's DSD files were always a point of conflict,some people has pointed that to him.
Luckily we now have Multitone Analyzer where we can measure any DSD signal possible (including DoP and Native and from DSD64 to DSD1024 )

I find Multitone much cleaner down low with my ancient PC who can add kind of tiny (down to -130db ) spikes to the chart when played through foobar,depending on it's mood.
With Multitone is always clean as a whistle.

 

[IVX]

a short video about 9039S as a sine source:

It is interesting that adding a 32b dither improves the THD for a couple of db more.

 

[pkane]

It is interesting that adding a 32b dither improves the THD for a couple of db more.

How can dither affect THD? Noise floor, I understand. But 2dB of THD?

 

[IVX]

yeah, hard to say.

 

[CedarX]

yeah, hard to say.

Could it be related to this?

https://precisionmechatronicslab.com/wp-content/uploads/2017/01/J17c-Preprint2.pdf

 

[IVX]

well, I guess no, 24-32 bits isn't large dither and isn't high-frequency at all because it is applied before the interpolator.

 

[IVX]

It seems not dither is a matter but REW gen itself. Usually, I use another sine gen but today I run REw's one and got THD -146db on the same sample! Completely vanished H7 for instance.

 

[pkane]

It seems not dither is a matter but REW gen itself. Usually, I use another sine gen but today I run REw's one and got THD -146db on the same sample! Completely vanished H7 for instance.

THD compensation is a great tool for measurements, but what does it look like at different frequencies, say 100Hz? Too bad the notch is only at 1k and 10k. Maybe a future version will have a lower setting added, say at 100Hz?

 

[IVX]

I just tested 9039S 1000; 100; and 50Hz with AP SYS2522 and I got -119.3db; -118.5db and -117.3db. SYS2522 loopback has 118.4db; 116.7db and 115.3db.

 

[pkane]

AP SYS2522

One can now measure well below the capability of that AP Just need that one more notch frequency at 100Hz for complete happiness