Class D AIYIMA A07 amplifier (ABX) listening test

[pma]

I was curious if inserting AIYIMA A07 TPA3255 class D amplifier in the D/A – A/D loop would be audible. I tested the amp in another thread and found that the frequency response depends, at higher frequencies, on speaker complex impedance, due to amplifier's output LC filter - that is outside amplifier feedback. So I prepared this test. There are two test files, G1_mono and G2_mono. One of them is recorded through AIYIMA amplifier loaded with my complex dummy load (see the test thread, please). The other one is a direct loopback through a signal cable. In fact both files were recorded simultaneously, as a stereo file and then were split into 2 mono files. So they are perfectly time aligned and also level matched. There was the same mono file driving both channels of the stereo file. The maximum power of the class D channel was 6W. The link to the test files is below:

https://www.dropbox.com/s/ppij8yuvpkmx0g1/Test_classD.zip?dl=0

If you would like to participate in the test, please download the test file (you do not need to have a dropbox account. Click somewhere outside log-in window and then click to download the quintet_test.zip file and then unzip the downloaded file. Do no try to play the files directly from dropbox, it is impossible.)

Please feel free to write your listening impressions, however in case you would like to be serious, I would ask you to make an ABX test in Foobar 2000

http://www.foobar2000.org/components/view/foo_abx

and then to post the protocol of the test, which we are able to verify by a signature verification tool.

There is a negligible non-linear distortion with the class D amplifier inserted. The biggest difference is in the frequency response above 8kHz (again see the test thread) and the reason of this test is to find if it might be audible on music. The music sample with considerable content of high frequencies was chosen. The following plot shows the differences between the two test files in a frequency domain.

Have fun with the test!

 

[pma]

Not too many results yet , so I'll start. It is not easy at all ...

foo_abx 2.0.2 report
foobar2000 v1.4.8
2021-01-14 19:20:15

File A: G1_mono.flac
SHA1: 2371edfa3f03a70afbd015fb34510a09b263a213
File B: G2_mono.flac
SHA1: 95f157488acbe1f237cc4c106dd477286a3b58f5

Output:
ASIO : DUO-CAPTURE EX
Crossfading: NO

19:20:15 : Test started.
19:20:48 : 00/01
19:21:00 : 01/02
19:21:15 : 02/03
19:21:40 : 03/04
19:21:56 : 04/05
19:22:21 : 05/06
19:22:32 : 05/07
19:22:48 : 06/08
19:22:59 : 06/09
19:23:14 : 07/10
19:23:14 : Test finished.

----------
Total: 7/10
Probability that you were guessing: 17.2%

-- signature --
3c500e1824584f3ebae2bcb52196e8117a01957d

Immediately following

foo_abx 2.0.2 report
foobar2000 v1.4.8
2021-01-14 19:28:31

File A: G1_mono.flac
SHA1: 2371edfa3f03a70afbd015fb34510a09b263a213
File B: G2_mono.flac
SHA1: 95f157488acbe1f237cc4c106dd477286a3b58f5

Output:
ASIO : DUO-CAPTURE EX
Crossfading: NO

19:28:31 : Test started.
19:28:48 : 00/01
19:29:03 : 01/02
19:29:22 : 02/03
19:29:35 : 03/04
19:29:45 : 04/05
19:30:04 : 04/06
19:30:23 : 05/07
19:30:37 : 06/08
19:30:56 : 07/09
19:31:14 : 08/10
19:31:14 : Test finished.

----------
Total: 8/10
Probability that you were guessing: 5.5%

-- signature --
7225a1b10a90c09bd7c0a65d20053a4f8dfebc86

So 7/10 and 8/10 sequence makes 15/20, maybe it is just telling something.

the tested part is "... but you do not seem to understand" (sibilants, as always .....)

 

[Blumlein 88]

I didn't see this when you first posted. Saw it referenced in the other thread. I'll see if can hear it sometime in the next couple days. Thanks for doing this.

 

[pma]

Not many results . So we may start with something easier, like a test with white noise. The same method as above, one of the files is a direct loopback and the other one is recorded through AIYIMA A07, now loaded with JBL Cotrol1 Pro speaker.

http://pmacura.cz/testwhite_classD.zip

foo_abx 2.0.2 report
foobar2000 v1.4.8
2021-01-27 15:10:31

File A: whiteorig_recL.wav
SHA1: 3950f936521047a580e161773382ef7a66c1b0cd
File B: whiteorig_recR.wav
SHA1: b66993cc0460cc74a9a00b34a0cba979764fcd2b

Used DSPs:
Resampler (PPHS), Downmix channels to mono

Output:
ASIO : DUO-CAPTURE EX
Crossfading: NO

15:10:31 : Test started.
15:10:44 : 01/01
15:10:52 : 02/02
15:10:59 : 03/03
15:11:12 : 04/04
15:11:22 : 04/05
15:11:31 : 04/06
15:11:37 : 05/07
15:11:47 : 06/08
15:11:53 : 07/09
15:12:01 : 08/10
15:12:01 : Test finished.

 ----------
Total: 8/10
Probability that you were guessing: 5.5%

 -- signature --
e510b09c8b3332b0e79ee971f60075fe0d33366b

 

[pma]

Please also find just created DeltaWave plots with @pkane software.

BTW the DeltaWave result of AIYIMA A07 is much worse than the result of Hypex UcD180HG in a similar test

Blind test - can you tell the class D amp from original loop in ABX

This is the test of the looback file vs. the file recorded through a class D amplifier. The files for download are here: https://www.dropbox.com/s/z13zpe0k08opfma/Mutter.zip?dl=0 If you are interested, please download the zip file, unzip it and listen. Dropbox account is not needed, just click...

www.audiosciencereview.com

AIYIMA A07 is definitely worse amplifier than UcD180HG with more bright sound with most speakers. Maybe this feature attracts its users.

 

[grogi.giant]

foo_abx 2.1 report
foobar2000 v1.6.16
2023-02-12 23:01:10

File A: whiteorig_recL.wav
SHA1: 3950f936521047a580e161773382ef7a66c1b0cd
File B: whiteorig_recR.wav
SHA1: b66993cc0460cc74a9a00b34a0cba979764fcd2b

Output:
Default : Primary Sound Driver
Crossfading: NO

23:01:10 : Test started.
23:01:32 : 01/01
23:01:42 : 02/02
23:02:05 : 03/03
23:02:11 : 03/04
23:02:18 : 04/05
23:02:23 : 05/06
23:02:29 : 06/07
23:02:36 : 07/08
23:02:36 : Test finished.

----------
Total: 7/8
p-value: 0.0352 (3.52%)

-- signature --
f60f7010a64f4268e463d0fe34faafff189cc868

Although with the first test files I am completely random and can't tell them apart.