A recent Listening Test generated much interest and debate on both the technical aspects of the devices and material used as well as the validity of ABX testing.
I thought it would be informative (read fun 🙂) to come at this from an even more extreme angle i.e. how 'bad' or 'how different' does something have to be to get a reliable ABX result.
As a first step I put these two files together. Both are opamps that might be considered 'opposites' in the performance stakes. The circuit is based on Pavels buffer, it is fed from battery supplies and has an approximate 6k loading on the output when all coupled up.
One distinguishing feature of this test is that one of the devices can be made much worse via a simple component change. This change is 'reverse compatible' with the better opamp, in other words the circuit can be kept the same for both and so the question of differing circuits for the test is cancelled out.
I'm afraid the equipment I have available (PC's soundcards etc) does not match that used by Pavel, however I hope something useful might come out of all this.
So the question is, can these two devices be reliably differentiated via their audible traits ?
You don't have to say which you prefer, just whether or not you can identify a difference.
ABX Test File
I thought it would be informative (read fun 🙂) to come at this from an even more extreme angle i.e. how 'bad' or 'how different' does something have to be to get a reliable ABX result.
As a first step I put these two files together. Both are opamps that might be considered 'opposites' in the performance stakes. The circuit is based on Pavels buffer, it is fed from battery supplies and has an approximate 6k loading on the output when all coupled up.
One distinguishing feature of this test is that one of the devices can be made much worse via a simple component change. This change is 'reverse compatible' with the better opamp, in other words the circuit can be kept the same for both and so the question of differing circuits for the test is cancelled out.
I'm afraid the equipment I have available (PC's soundcards etc) does not match that used by Pavel, however I hope something useful might come out of all this.
So the question is, can these two devices be reliably differentiated via their audible traits ?
You don't have to say which you prefer, just whether or not you can identify a difference.
ABX Test File
Hi Karl,
thank you for the test. Yes I was able to get positive result repeatedly, with no failures. You can verify my result
foobar2000 ABX Log Signature Tool
thank you for the test. Yes I was able to get positive result repeatedly, with no failures. You can verify my result
Code:
foo_abx 2.0.2 report
foobar2000 v1.3.7
2017-07-15 16:04:29
File A: OP1T.wav
SHA1: 9b19126ad72d403b615f2b994e1d90169a4046fc
File B: OP2T.wav
SHA1: 269805e99c4dcb0775e9c7958e1d462ea44dea06
Output:
WASAPI (event) : SPDIF-Out (USB Sound Blaster HD), 24-bit
Crossfading: NO
16:04:29 : Test started.
16:04:54 : 01/01
16:05:00 : 02/02
16:05:05 : 03/03
16:05:10 : 04/04
16:05:14 : 05/05
16:05:19 : 06/06
16:05:24 : 07/07
16:05:30 : 08/08
16:05:34 : 09/09
16:05:40 : 10/10
16:05:44 : 11/11
16:05:52 : 12/12
16:05:52 : Test finished.
----------
Total: 12/12
Probability that you were guessing: 0.0%
-- signature --
10a173c3062c95945cea77cb111a5cefce0ff566foobar2000 ABX Log Signature Tool
Attachments
Top result Pavel 
I hope we get some interest in thi and I also hope to continue this by performing another ABX of the same set up, but this time changing things a little more.
I hope we get some interest in thi and I also hope to continue this by performing another ABX of the same set up, but this time changing things a little more.
ODAC + AVR Marantz SR4500 + KEF Q100
With foobar2000 + ASIO4all and J River Media 22.* + Kernel Streaming is a very very very easy test because the sound is more different.
I like more the second file, more clean sound.
With foobar2000 + ASIO4all and J River Media 22.* + Kernel Streaming is a very very very easy test because the sound is more different.
I like more the second file, more clean sound.
Thanks Maty
All will be revealed at some point, and then depending on the interest I might take things a little further.
And anyone with Foobar results, please feel free to post them.
All will be revealed at some point, and then depending on the interest I might take things a little further.
And anyone with Foobar results, please feel free to post them.
Top result Pavel
I hope we get some interest in this
Me too, and I also hope we shall get some real results from the others as well. The real result means the valid ABX test report. The real result is NOT a hand waving and saying how easy it was.
This test is ideal for the ABX method. It shows both weak and strong points of the method.
I am out for the weekend but will certainly try and share my results, so please don't move any goal posts.
If you like tests, new:
Archimago's Musings: INTERNET BLIND TEST: MQA Core Decoding vs. Standard Hi-Res Audio Saturday, 15 July 2017
Archimago's Musings: INTERNET BLIND TEST: MQA Core Decoding vs. Standard Hi-Res Audio Saturday, 15 July 2017
Thanks Maty, there is a lot to read there 🙂
OK folks, I am continuing with this experiment and will shortly post a new thread called ABX Test no 2 which essentially picks up where this one finishes off but this time we are taking things down a few notches in the quality department (for one of the op amps). The circuit configuration is the same except for one change that pulls one of the opamps closer to 'terrible'.
This is all being done very much on the fly (and all for fun) but what I am hoping to achieve is a point at which we can reliably pick out a true difference in perceived audio quality.
Edit... ABX Test No 2 is here:
http://www.diyaudio.com/forums/everything-else/310060-abx-test-2-a.html#post5132825
OK folks, I am continuing with this experiment and will shortly post a new thread called ABX Test no 2 which essentially picks up where this one finishes off but this time we are taking things down a few notches in the quality department (for one of the op amps). The circuit configuration is the same except for one change that pulls one of the opamps closer to 'terrible'.
This is all being done very much on the fly (and all for fun) but what I am hoping to achieve is a point at which we can reliably pick out a true difference in perceived audio quality.
Edit... ABX Test No 2 is here:
http://www.diyaudio.com/forums/everything-else/310060-abx-test-2-a.html#post5132825
As there has been a drop off in interest in all these threads I shall sum up what you were listening to.
This test together with ABX Test no 2 were of a unity gain buffer based on Pavels original circuit.
The devices used were the TLE2072 and the LM358 (the dual version of the LM324 quad opamp). The key was as follows:
ABX Test 1
OP1 = LM358
OP2 = TLE2072
ABX Test 2
1st OpAmp = LM358
2nd OpAmp = TLE2072
The LM358 was biased into Class A via a resistor returned to the negative rail. I was surprised how well the LM358 acquitted itself.
This led me to do a further test (different music) which is here:
ABX Test No 3
The above is just for completeness and for anyone wishing to try the files out.
There are four files.
ABX Test 3
Sine = LM358
Cosine = TLE2072
Tangent = TLE2072
Cotangent = LM358
Sine/Cosine are the full test track. You can clearly hear the distortion of the LM358 on the quieter intro.
Tangent/Cotangent are the louder portion of the track. The distortion curiously is much harder to pick up here.
The FFT shows the LM358 distortion. This was unbiased (no pull down resistor) to show just how bad the opamp is. In the final test the pull down was 33k, so not very much.
Everyone should be able to get a 100% Foobar result from Sine vs Cosine.
This test together with ABX Test no 2 were of a unity gain buffer based on Pavels original circuit.
The devices used were the TLE2072 and the LM358 (the dual version of the LM324 quad opamp). The key was as follows:
ABX Test 1
OP1 = LM358
OP2 = TLE2072
ABX Test 2
1st OpAmp = LM358
2nd OpAmp = TLE2072
The LM358 was biased into Class A via a resistor returned to the negative rail. I was surprised how well the LM358 acquitted itself.
This led me to do a further test (different music) which is here:
ABX Test No 3
The above is just for completeness and for anyone wishing to try the files out.
There are four files.
ABX Test 3
Sine = LM358
Cosine = TLE2072
Tangent = TLE2072
Cotangent = LM358
Sine/Cosine are the full test track. You can clearly hear the distortion of the LM358 on the quieter intro.
Tangent/Cotangent are the louder portion of the track. The distortion curiously is much harder to pick up here.
The FFT shows the LM358 distortion. This was unbiased (no pull down resistor) to show just how bad the opamp is. In the final test the pull down was 33k, so not very much.
Everyone should be able to get a 100% Foobar result from Sine vs Cosine.
Attachments
- Status
- Not open for further replies.