Can you hear a 1kHz sine wave distortion of 0.85% from tube preamplifier

[pma]

This test is on audibility of harmonic distortion of the pure sine wave that went through a tube preamplifier, resulting in THD = 0.85%. Two files, pure sine and output from the tube may be downloaded from

https://drive.google.com/file/d/1Npg8YFJAfRvqqN6lalsBjVAkHS2Q1lnd/view?usp=sharing

If you are interested, please download the file, unzip and post your ABX test result. ABX result is mandatory this time for the positive result.

The distortion looks like this

This is my ABX report

foo_abx 2.0.2 report
foobar2000 v1.4.8
2021-09-03 10:10:35

File A: sample1.flac
SHA1: c7fa3903f280279e07fd8a7981017b0634969711
File B: sample2.flac
SHA1: 6f45f66aef302ed42b9199074fbfaca1ab4dd708

Output:
ASIO : Focusrite USB ASIO
Crossfading: NO

10:10:35 : Test started.
10:11:26 : 01/01
10:11:38 : 02/02
10:11:47 : 03/03
10:11:54 : 04/04
10:12:02 : 05/05
10:12:09 : 06/06
10:12:15 : 07/07
10:12:22 : 08/08
10:12:29 : 09/09
10:12:34 : 10/10
10:12:40 : 11/11
10:12:45 : 12/12
10:12:50 : 13/13
10:12:56 : 14/14
10:13:01 : 15/15
10:13:07 : 16/16
10:13:07 : Test finished.

 ----------
Total: 16/16
Probability that you were guessing: 0.0%

 -- signature --
984d946f5a76043c47b8e4c14ef9bebc8b4b6405

 

[abdo123]

This is a bit interesting since 4th harmonic is high-ish and 3rd harmonic is lower. I assume most people will be able to tell the difference.

 

[xaviescacs]

I'm completely new at hearing distortion. I've downloaded a software for ubuntu called abx. The issue is that it doesn't produce a report like this, just a screen with the results and the p-value. I will try with this other software when I have more time.

Here are my (perfect) results:

EDIT: Using AKG 701 headphones, S/N 58920.

This is my first test and was pretty easy! To my ears, used to the piano, the sample with distortion sounds like a fundamental with its first (octave) harmonic louder than usual, almost like an octave interval, where you really hear more prominently the higher pitch. So, to me, this was about choosing from a single tone or a chord really. That's what you call distortion?

Thanks!! That was fun!!

 

[charleski]

I tried this on my crappy desktop computer speakers and was surprised at how easy it was to distinguish the tones. A couple of seconds of training was all it needed to lock into the timbral change.

foo_abx 2.0.6d report
foobar2000 v1.6
2021-09-03 10:28:30

File A: sample1.flac
SHA1: c7fa3903f280279e07fd8a7981017b0634969711
Gain adjustment: -13.91 dB
File B: sample2.flac
SHA1: 6f45f66aef302ed42b9199074fbfaca1ab4dd708
Gain adjustment: -13.87 dB

Output:
Default : Primary Sound Driver
Crossfading: NO

10:28:30 : Test started.
10:29:37 : 01/01
10:29:58 : 02/02
10:30:11 : 02/03
10:30:22 : 03/04
10:30:47 : 04/05
10:30:59 : 05/06
10:31:11 : 06/07
10:31:26 : 07/08
10:31:39 : 08/09
10:32:00 : 09/10
10:32:19 : 10/11
10:32:42 : 11/12
10:32:53 : 12/13
10:33:07 : 13/14
10:33:25 : 14/15
10:33:41 : 15/16
10:33:41 : Test finished.

----------
Total: 15/16
p-value: 0.0003 (0.03%)

 

[danadam]

Interesting thing I noticed, it was fairly easy up to some volume level and it becomes difficult when it is too loud.

Spoiler: abx result

foo_abx 2.0.6d report
foobar2000 v1.5.5
2021-09-03 11:52:22

File A: sample1.flac
SHA1: c7fa3903f280279e07fd8a7981017b0634969711
File B: sample2.flac
SHA1: 6f45f66aef302ed42b9199074fbfaca1ab4dd708

Output:
WASAPI (event) : InternalAudio (Conexant 20672 SmartAudio HD), 24-bit
Crossfading: NO

11:52:22 : Test started.
11:52:36 : 01/01
11:52:49 : 02/02
11:52:59 : 03/03
11:53:10 : 04/04
11:53:25 : 04/05
11:53:37 : 05/06
11:53:43 : 06/07
11:53:51 : 07/08
11:54:02 : 08/09
11:54:09 : 09/10
11:54:20 : 10/11
11:54:33 : 11/12
11:54:33 : Test finished.

----------
Total: 11/12
p-value: 0.0032 (0.32%)

-- signature --
bea2bb006b57a5d12e6d2098f49d473440c89e63

 

[Chyżwar]

I didn't do the ABX test, I believe in my golden ears It's very easy to hear which sample has more harmonic distortion, but it's a single tone. Detecting 0.85% THD in music is extremely difficult.

 

[artburda]

I can't hear the difference. Tested it with different headphones. How can this be so easy to hear for some people?

 

[pma]

A small hint to those who could not hear a difference - try lower volume level. The 2nd harmonic distortion here is audible at low volume, otherwise it is masked by ear intrinsic distortion.

 

[xaviescacs]

A small hint to those who could not hear a difference - try lower volume level. The 2nd harmonic distortion here is audible at low volume, otherwise it is masked by ear intrinsic distortion.

True!! I thought I would hear it anyway but from a certain level it becomes impossible. I had a "normal" volume when I did the test and that helped. Otherwise I would have failed for sure.

My subjective perception though is that the main tone gets louder while the 2nd harmonic don't, and there is a point where I can't hear the 2nd.

 

[artburda]

I know why I can't hear it. Tinnitus. It's masking it. When I listended to a 500 Hz tone with -40dB 2nd harmonic I could hear it, but with 1kHz fundamental and higher up I start to hear the distortion when it's -30dB.

 

[Jimbob54]

Hurrah- I did one right. Lower volume is indeed the key- I started near my normal listening volume and had no chance- I could hear the higher harmonics when lowered it down.

 

[anmpr1]

In the '70s some editors at Stereo Review and Bob Carver ran distortion tests with pure tones. This test has been discussed at ASR, and threads can be found with a search. Tones were easily distinguished at lower distortion levels than with music. I'm doing this from memory, but with music, recognizable distortion was four or five percent before it became noticeable, maybe a little higher depending on the complexity of the recording. In their test, the specific distortion was notch distortion, as I recall.

Today, obsession (and that's what it is for many) over distortion measurements is pretty ridiculous from a practical (ie. auditory) standpoint, although measurements can be a good indication of who is doing the best engineering at any given price point.

On the other hand, if one's idea of great sounding music is pure tones, or if you like to play that silent track from Cage at super loud levels, then SINAD and distortion figures are going to be an important consideration, and should be scrutinized accordingly.

 

[Blumlein 88]

In addition to psycho-acoustic reasons you don't hear this at higher volumes would be the possibility your speaker/phones are distorting more than the test signal.

 

[pma]

In addition to psycho-acoustic reasons you don't hear this at higher volumes would be the possibility your speaker/phones are distorting more than the test signal.

Maybe not, because we are talking about highest sensitivity of human ear to H2 distortion at about 60 dB SPL of the fundamental, so the transducers should not distort much yet. At 80 dB SPL the human ear distortion is quite high and it still is not high level for the transducer distortion.

 

[restorer-john]

 

[wwenze]

Hmm while doing the ABX I feel the 1kHz tone produces a white noise inside my ear instead

 

[restorer-john]

Hmm while doing the ABX I feel the 1kHz tone produces a white noise inside my ear instead

Turn it down. The difference is quite obvious. The clean tone is softer, the distorted tone has an edginess.

 

[KMO]

Today, obsession (and that's what it is for many) over distortion measurements is pretty ridiculous from a practical (ie. auditory) standpoint, although measurements can be a good indication of who is doing the best engineering at any given price point.

I'd be inclined to agree, having just had a real-world experience.

One of my new KEF LS50 Metas had a distortion fault (apparently in the crossover) - no error I could spot in the frequency response, just 2% THD around 2.5kHz.

Took me a few weeks to be certain there was a problem from listening. I'd had a sense something was off, and wasn't fully happy, but assumed it was a room reflection issue. Astrud Gilberto's vocals panned hard right in The Girl From Ipanema turned out to be a killer test though. She sounded harsh.

I'd done a fair bit of REW + UMIK-1 testing, tuning EQ, but not spotted any problems doing that. On further investigation, I then heard something significantly wrong in a pink noise signal - a "crackle" (so no longer just harmonic distortion?), and I then found the distortion error in all previous recorded measurements.

If I hadn't had REW, I wonder if I'd have ever been confident enough that something was wrong to return it.

Given how hard it was to hear that 2%, I'm really doubtful chasing 0.07% versus 0.1% (as in LS50 Meta marketing) is significant...

(This was measured at 80dB at about 30cm. There was also about 0.2% 3rd harmonic - graph just shows 2nd because it dominated.)

But then, as Mark Dodd said in a talk, don't assume you can't hear things. If they do manage to get those numbers down, they must be doing everything right, which would mean clearing up all sorts of potential issues. Even if you couldn't spot a pure distortion test.

 

[maxm272]

Finally an ABX I could pass (failed miserably at the CDP vs. DAC and power amp tests).
Although I didn't quite get the softness vs. edginess difference mentioned above, at least the timbral difference was easily audible with K612 PRO connected to my laptop's headphone jack, but only at lower volume.
Really looking forward to further listening tests. Thank you, @pma !

foo_abx 2.0.6c report
foobar2000 v1.6.6
2021-09-03 19:28:04

File A: sample1.flac
SHA1: c7fa3903f280279e07fd8a7981017b0634969711
Gain adjustment: -13.91 dB
File B: sample2.flac
SHA1: 6f45f66aef302ed42b9199074fbfaca1ab4dd708
Gain adjustment: -13.87 dB

Output:
Default : Primary Sound Driver
Crossfading: NO

19:28:04 : Test started.
19:28:53 : 01/01
19:29:24 : 02/02
19:29:52 : 03/03
19:30:15 : 04/04
19:30:30 : 05/05
19:30:45 : 06/06
19:31:06 : 07/07
19:31:29 : 08/08
19:31:39 : 09/09
19:31:54 : 10/10
19:32:07 : 11/11
19:32:13 : 12/12
19:32:23 : 13/13
19:32:35 : 14/14
19:32:46 : 15/15
19:32:55 : 16/16
19:32:55 : Test finished.

----------
Total: 16/16
p-value: 0 (0%)

-- signature --
5b48032c88160c10f18ec45402a103b4ab4dc5d2

 

[xaviescacs]

softness vs. edginess difference mentioned above

This is quite subjective, isn't it? To me the distorted tone sounds more natural, more like a real musical instrument, because of the harmonics I guess. The single tone is quite artificial to my ears.

To me the key point here is to heard the 2nd harmonic. With a piano this sound can be simulated playing a B4 with the B5 pressed, to let it resonate. Or try to play the B5, which is just a bit lower than 1kHz, and the B6 with at the same time but at lower volume.

I guess though this is a special case because the 2nd harmonic dominates the others, and it's audible by itself. If that weren't the case, I suppose this strategy simply doesn't work and other capabilities come into play.