Well, I'll defend Alan (one L - last name Lofft - two F's) here... I read his articles in the trade mag Sound & Vision Canada long before I knew about Axiom. The fact he was standing behind what Ian was doing was a big decision in my trying out the Axiom brand sound unheard.
As a long time reader of his, he's never once written anything I didn't believe or didn't have proven to me in time.
On the other side of the coin, I was playing with digital audio when IBM was still selling PCjrs with a Charlie Chaplin lookalike - built a basic audio digitizer out of a 0820ADC in the Commodore 64 days, there's really nothing to analog-digital or digital-analog ICs - ADCs are actually a bit more finicky than DACs. As for a certain chipset being light years ahead of another brands... one only has to cut through marketing hype and remember what the circuit does. A basic 16-bit DAC takes a digital number between 0 and 65535 and translates it into a voltage (say for simplicity 0-1v).
There are error-diffusion and oversampling DACs as well. Oversampling DACs simply take adjacent samples (or a series of samples) and extrapolate intermediate values between the actual sampled values. Technically, in the purest sense, this is distortion, but the manufacturers mean well - and it works - to a point. Put it this way - you see your wife at 3 - she's happy, at 6 - she's happy, at 9 - she throws a pan at you. Using interpolation, you'd say she was happy all the way from 3 til 6 (when really, she might have cried at a chick flick and gotten mad when she stubbed her toe, then realized she stubbed her toe on her birthday present from you, which made her happy again - all of which your interpolation missed), then you'd say she was 33% angry at 7, 66% angry at 8 on her way to becoming 100% angry at 9. None of which may even be remotely true - you're just connecting the dots between known samples. Not the worst technology in the world, but not exactly something to write home about either.
The error-diffusion type ICs use various mathematical equations (like 1 bit delta-sigma) to pass along the remainder of each sample to the next sample, this happens at the sub-sample level (less than a 1/65535 of total amplitude in a sample by definition!) so we're talking about a very minimal level of audio "sweetening". If you can hear the difference between samples at say binary 1111111111111111 and 1111111111111110, then call the Guinness people.
The bigger question SACD-P owners should be asking is "why am I even allowing them to screw me this way?"... we finally get a digital datapipe between components and receivers in TOSlink and SPDIF - then they panic about perfect digital copies and revert back to 50+ year old analog technology? Next they'll bring back the cassette tape so there will be generational loss with each copy.
Bren R.
