Looper's Delight Archive Top (Search)
Date Index
Thread Index
Author Index
Looper's Delight Home
Mailing List Info

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Date Index][Thread Index][Author Index]

Re: Re: Quest for a mobile recorder - how manufacturers do designs...



Stefan,

> For D/A it doesn't make a big difference, but for A/D it does. Its about 
> the filters you need before the converter. at 44.1kHz they need to be 
> extremely steep, and create ripple. 

One of the problems is to create analogue filters properly embedded in a 
(CMOS) semiconductor process. Resistors in these processes are not very 
tightly tolerated - manufacturers usually specify +100/-50%, although the 
actual value is more like +/-10%. Capacitors, on the other hand, can be 
tolerated very tightly (which is necessary for DRAM technology, among 
other things), but here, it's a problem of real estate to create big 
values ("big" meaning anything above the lower exp-11 range).
Finally, there's a workaround to the resistor tolerance issue: use a DMOS 
biased by a stabilized voltage source. To make those immune to temperature 
influence, you need to operate them in the inversion region, which means 
low resistance values.

So you got the tradeoff:
use a R -> bad tolerance
use a DMOS in non-inverted region -> strong temperature influence
use an inverted DMOS -> high demands for C real estate
or finally, use an external R -> more expensive, demands on PCB process 
due to parasitic capacitances.

The sum of all this is, as you mentioned, you can stay out of trouble by 
simply going for a Nyquist of 96kHz and tune your filters for the 80kHz 
range , because then if you're off by 10%, it doesn't hurt in either 
direction...

> Though I guess 96 kHz is usually 
> good enough, I can hear a difference to 192 kHz as well. The result is 
> much cleaner and more transparent. 

Frankly, I find it hard to believe that, taking into account how old your 
ears are. What is the gross value of your listening equipment? Did you use 
one and the same converter with the same material in the same testbench in 
a statistically relevant double blind?

> If you down sample afterwards, you 
> don't loose as much, as it is possible to build ideal filters in the 
> digital domain. But I don't know of any device which would do that 
> directly...

Both hardware (e.g. Weiss) and software (e.g. Apogee) solutions are 
available for that off the shelf.

> High end fundamentalists would rather buy four times the processing 
> power and keep everything at 192 kHz...

Actually, no, they won't. Extensive listening test suggest that (possibly 
due to problems with clock bleed in many converter designs), 96kHz is 
often perceived as "better".

> The price, size and weight of the R-44 is aimed at that market...

Again, no. While a lot of fundamentalists may still disagree here, telling 
me that it's basically another kind of noise source, the region to start 
looking for better stuff would be with a converter pair at the price of a 
R-44 (e.g. RME ADI-2).

                Rainer