Robert,
I know you are very keen on scientific rigour rather than enjoying the journey but to dismiss this test is insane - it is what 95% of HiFi purchasing is based on and therefore has to represent one way to measure the systems performance. I can't believe that if you were designing a pre-amplifier you wouldn't listen to it against a reference rather than / as well as recording its outputs.
Besides a line out recording would remove the power amp which is likely to be most affected by any changes in the mains imo.
cheers
Jason
Jason, I'm not dismissing this as insane - I'm simply saying that here we have two files and the only difference is supposed to be the mains block effect.
However it isn't the only difference. We have a non constant noise source on each file meaning that the files
must be different. In other words you have completely eliminated the option of null result because it cannot occur. You guarantee a positive result. I would disagree with those who say ambient noise is not significant. It contributes significantly more to the file than many things us audiophiles consider as significant. We worry about PSU effects on for example dacs and pre amps where there is barely any measurable effect - sometimes no effect. So to dismiss something that is clearly audible and recordable is wrong.
On how we listen and compare when designing equipment, I would hope nobody does single quick A/B and says job done. There would be many, many listening sessions which changes things considerably. But you make a good point and where simple, flat, low distortion circuits are concerned, I would say that PJ Walker and Baxandall had it right. You don't actually need to listen to it because it follows rules, has a known transfer function and the result in predictable. That will have some scowling but a good case can and has been made over the years. You listen to say ' does that work, does it sound ok and have we avoided a balls-up?' Great, build them!
line out does eliminate any effect of the power amp circuit itself, but it does include speaker vibration, through feedback, and it does include modulation of the mains line by the amp as this is still drawing current and driving the speakers.
One other thing strikes me here as another variable. This test was performed with valve equipment being turned on and off (presumably as the mains connections were changed over). Make a recording, switch off valve equipment, switch on and record again - and a difference is highly likely for various reasons including the internal valve components contracting and expanding. In fact the noise from this is often easily heard as i'm sure valve users will testify. So you'd need a fairly long wait after power-up.
Of course lots of kit was switched on and off on the day - some feel this is very significant, including I think Steven. It hasn't been mentioned as a factor.
So I'd say this was a good attempt to record something and I genuinely meant what I said about Steven in particular agreeing to some attempt to capture the results. I just think as it stands the object of the test hasn't been realised.
One last thought for the day. I agree we want the amp circuit effect recorded if possible but but this is very difficult in practice.
You could take a line via the speaker terminals via an L Pad but you'd need to be driving a dummy load. With speakers connected you have back EMF and a signal going back to the amp.
A dummy load is a reasonable solution as at least you are loading the amplifier and drawing current.
Do this and make the changes in mains wiring resistance extreme enough and we should see something. It at least gives a feel for the nature of the problem, though given the lengths we evidently have to go in order to find this damned problem, I would personally file under insignificant and move on. Edge of audibility, indeed.