www.neumann.com
Forums:
Email:









Forums / Archive / Thread
Re: Interesting...
Author: Posted by dan.dan
Date: 04/15/2004

'Early digital recorders, measured in terms of analog performance, were 'perfect', except that they sounded terrible. I think it was Ed Meintner's work, who identified and reduced jitter, that started to make digital recording sound more musical.'
Okay, I think I see the disconnect between what you and I are saying.
In fact, it is *not* true that 'early digital recorders, measured in terms of analog performance, were 'perfect'.'
Some people postulated that they were perfect incorrectly. The fact that people did that does not change the metaphiscal reality that 'if it cannot be measured, it doesn't exist.'
The problem with spotting the limitations of early digital recorders was never a problem with an ability to measure. It was a problem with the abilities of people to interpret measurements. Those are very different problems.
I fully admit that people do sometimes oversimplify their interpretations of measurements.
Objectively, 'perfect', in the context of a recording medium, means a unity transfer function H(s) = 1. In other words, if x is the source signal that you are going to record, and y is the playback signal, then x = y.
That was never true of early digital recorders. Nor did anyone ever say it was true.
Firstly, digital recording can never have a unity transfer function except in relation to an already bandlimited signal. That fact was known mathematically decades before anyone thought of making a digital recorder.
Secondly, it is impossible to make real world converters that exhibit a unity transfer function even to bandlimited signals because the theory of playback is idealized, while the practice of playback cannot be. In a digital signal, samples represent impulses of zero length in time. To reconstruct the original analog signal from a digital signal, one recreates those impulses and filters them (conceptually) by convolution with a sinc function, which is the impulse response of an ideal lowpass filter. Neither perfect impulses nor ideal lowpass filters exist in the analog domain.
All of these things were known long before digital recording revolutionized audio in the 1980s.
The problem is that zealots glossed over some of the finer implementation details. They quickly concluded that all of these erratic phenomena could be reduced to 'noise', and that this 'noise' could be made arbitrarily small with enough precision in the implementation. By their models, a converter was 'perfect' when its steady state frequency response was flat, and its harmonic distortion was lower than the best analog recorders, and its noise was lower than the best analog recorders.
Early digital recorders met these criteria and still sounded awful. The 'sin' of their makers was that they did not immediately ask the question 'why'. If the *had* asked that question, they would have gone back and done some more measurements, and they *would* have been able to find what they were looking for. The culprit, in this case, was engineering arrogance, not an inability to measure.
The problems with digital were later discovered by scientists and engineers using the math and measurement technology that had already been available for years. It wasn't that it couldn't be measured. It was that people didn't measure the right things.
(Incidentally, the solutions to those problems are conceptually brilliant: Sample the signal at a very high rate, and do the rest of the conversion in a digital world where filters can be made more or less 'ideal', and perfect impulses can be mathematically represented. The drawbacks tend to start disolving from the picture when the oversampling rate is high enough, as long as a good enough clock is used. But I digress...)
So, one way to look at this present situation is as a repeat of that old one. We could say that Neumann engineers have buried their heads in the sand about this P48 stuff. There is a difference, but they are not looking for it.
But I think it would be wise to have the discipline not to do that. You and I don't really know anything about what Neumann has or hasn't studied, except that they have always been known for excellence, and they have a legacy of pioneering in the designs of microphones.
So, we have Neumann engineers saying they've studied it up and down and they can't see a difference anywhere. And we have Erik saying he can easily hear a difference. That's where we stand today.
All we 'objectivists' have said is that we simply cannot assume that what Erik hears is actually the shortcomings of P48 powering. He has made a lot of compelling arguments, and none of those (to our knowledge) have swayed Neumann engineers to go out and investigate the problem. Does Neumann have a history of avoiding research that might improve their products? Why would they do that? All I can conclude is that Neumann has thought about it and really does not consider it to be a big deal.
Neumann could be right, or they could be wrong. None of us knows which is the case. In the mean time, since Erik is passionate about this issue, perhaps he could offer some evidence that might turn Neumann around and get them looking at the issue.
Regards, --Dan  
Posted by Oliver Archut
04/12/2004
 
Posted by arthur
04/13/2004
 
Posted by dan.dan
04/13/2004
 
Posted by arthur
04/14/2004
 
Posted by dan.dan
04/14/2004
 
Posted by arthur
04/15/2004
 
Posted by dan.dan
04/15/2004
 
Posted by arthur
04/15/2004
 
Posted by dan.dan
04/15/2004
 
Posted by Tom
04/13/2004
 
Posted by arthur
04/14/2004
 
Posted by Tom
04/15/2004
 
Posted by Erik
04/13/2004