February 8, 2009 at 4:11 pm #15163Ted FletcherKeymaster
During the development process for the P38EX, I had noticed a subtle ‘thinness’ and ‘edginess’ in the prototype I was working on. It was nothing that I could put my finger on, just a thought that the the sound was not quite right….. the opposite to what I had noticed with the P9 when that was in development; with the P9 it did not seem to matter what you put through it, it sounded better!
With the P38EX I did not worry about it, I just noted it to be looked into before completing the development.
Just before testing the first production ‘EX’s I went back to the problem and looked really carefully at the ‘straight line’ performance of the amplifiers…. The whole thing looked perfect; the ‘sum and difference’ systems were operating correctly with very accurate processing; response, overload, distortion and noise all quite brilliant. Then I noticed a slight anomaly in the high level square-wave response….. there was a tiny ‘kick’ at the leading edge, indicating some sort of phase shift very high up, probably around 50KHz.
It turned out to be exactly that, a phase error due to path-length differences in the ‘sum’ and ‘difference’ paths, showing up as a sharp phase shift well above audible sound!
I corrected the fault and restricted the audio bandwidth to a roll-off at about 40KHz, and suddenly the whole thing sounded more full, cleaner and yet transparent, in fact, just like it is supposed to sound!
The reason I am writing about this is that it is an example of how even professional audio gear can be, and often is, spoiled by inherent design problems that simply do not show up in conventional audio testing. This phase problem would have been invisible if I was relying on digital test equipment…. It was outside the audio band so how could it be important? !!!
So it’s comforting to know that all P38EXs (and everything else for that matter) is checked on old fashioned analogue test equipment where we can look at those places where we can’t hear (!) (but which makes all the difference!)
- You must be logged in to reply to this topic.