Your right... how on earth did that happen. I've obviously screwed up either the input to the test circuit or the output going into the laptop on "orange".
OK... if anyone who has contributed to this thread so far (i.e. they have posted and contributed to the discussion) and would now either have voted "orange" as their favourite or they voted orange and now don't think it their favourite please pm me.
And these are the actual devices in the quick 741 vs OPA134. Offered these as a choice for a new preamp or to replace some chips in some audio gear you wouldn't think twice... would you 
The OPA's are really hard to photograph for some reason.
The OPA's are really hard to photograph for some reason.Attachments
Ah ha! So that's it. I kept hearing a change, or a sort of flip. It seemed more obvious coming to Orange than from it. Funny I didn't notice the channel flip. Guess that invalidates my vote. Will have to listen to the file corrected.
I will tell you without voting that the best opamp is the one I choose to use.
Can't argue with one
A channel phase flip that went mostly unnoticed !.
That causes me to question the abilities of the playback systems, and to really question the ability of the listeners, to the point of invalidating any proper results.
600 ohms is a too heavy loading for most opamps.
Is this loading intended to exaggerate sonic differences, or is it a wrongly representative choice ?.
Dan.
That causes me to question the abilities of the playback systems, and to really question the ability of the listeners, to the point of invalidating any proper results.
600 ohms is a too heavy loading for most opamps.
Is this loading intended to exaggerate sonic differences, or is it a wrongly representative choice ?.
Dan.
I did want to load the opamps... well, this was the thinking behind it,
http://www.diyaudio.com/forums/ever...otout-where-you-get-decide-3.html#post3697484
Maybe 10 selections was a bit much with hindsight, hence the quicky B vs C mini challenge
Are we getting near to a grand unveiling do you think ? Maybe tomorrow or Wednesday...
The oldest stuff I have are some 301's and two 748's. The Quad 405 used the 301.
Anyhow, the results have certainly shown up one or two surprises but I'll not say any more just yet. One thing seems pretty obvious though and that is that the difference isn't "night and day" between most devices, let alone between the really old stuff like the 741 and later offerings.
I was hoping to at least pick out the 741, but after trying an ABX test in foobar2k, I couldn't reliably tell any of them apart (there should have been a "can't tell" option!).
In the interests of science, I then tried to "cheat" by comparing FFTs of them all, and noticed that one is 0.5dB louder than the others, which is presumably the original (I know I'm not the first to notice, but I deliberately didn't read the thread until after I had done this). This is enough difference that I'm sure those with more sensitive ears than mine could hear it. You can see this in the first attached screenshot. There are smaller differences between the others, with the lowest one probably being the 741. I won't say which trace is which file, to avoid biasing anyone.
In the second screenshot I normalized all the files to the same value (-18dB RMS). Now the traces nearly overlap exactly. I expect that if I had taken the time to align the files so they started at exactly the same point, and were exactly the same length, and there was no clipping, then the difference would be even less.
In the interests of science, I then tried to "cheat" by comparing FFTs of them all, and noticed that one is 0.5dB louder than the others, which is presumably the original (I know I'm not the first to notice, but I deliberately didn't read the thread until after I had done this). This is enough difference that I'm sure those with more sensitive ears than mine could hear it. You can see this in the first attached screenshot. There are smaller differences between the others, with the lowest one probably being the 741. I won't say which trace is which file, to avoid biasing anyone.
In the second screenshot I normalized all the files to the same value (-18dB RMS). Now the traces nearly overlap exactly. I expect that if I had taken the time to align the files so they started at exactly the same point, and were exactly the same length, and there was no clipping, then the difference would be even less.
Attachments
And here's me thinking it was deliberate! That's why I posted the crytic comment earlier on - thought Mooly was trying one on ...
Picked it up when looking at the overall waveforms in Audacity ...