how does a denon dcd 810 work?
Gwoho Lio
that mean? it says in the manual that it uses "pcm" technology.
since it does not say how many bits the dac is, i assume it is only one, but
i do not see how a one bit dac can work because it would have to time
the output to an accuracy of 1/(65536*44100*4) seconds. the 4 is because of
4 times oversampling. i think this is about .1 ns. i dont think it is possable
to time things that well with normal type stuff.
also i am wondering how normal 16 bit dacs can work. wirewound resistors are
usually the most accurate, and they are about .1%. if the resistors in a
dac are .1% resistors, then i can not imagine how the dac can output stuff
with more than -20%log(.1%) = 60 db of dynamic range. most cd players claim
quite a bit more than 60 db dynamic range. they would have to have the
ratios of the resistors accurate to about .001% to get the dynamic range
they claim. i read a 30 year old book, and silicon resistors 30 years
ago could not have ratios even close to .001%. i doubt they have improved
much. even if the resistors are .001% resistors, i think themperature gradients
in the chip would be enough to mess it up anyway: the chips in my cd player
get quite hot. so how do the dacs in cd players work?
gwoho liu.
Rich Fanning
>i have a denon dcd 810. it says it has a "super linear" dac. what does
>that mean? it says in the manual that it uses "pcm" technology.
Marketing being what it is, I can only speculate. Let's assume it is one of
the new 1-bit schemes...
>since it does not say how many bits the dac is, i assume it is only one, but
>i do not see how a one bit dac can work because it would have to time
>the output to an accuracy of 1/(65536*44100*4) seconds. the 4 is because of
>4 times oversampling.
If it is only 4 times oversampling, then it is probably not a 1-bit scheme.
One 1-bit scheme, the new Sony CDP-X77, uses a 45 MHz clock:
44.1 kHz x 16 (bits) x 64 (oversampling) = 45.16 MHz.
Such a clock, if stable, noise- and jitter-free, can very accurately time
pulses. Check out the June Audio. The specs, including linearity and
distortion, are extremely impressive. Since the three major problems (IMHO)
in CD reproduction have been solved by this player (phase problems due to
anti-alias filter, DAC nonlinearity, and jitter) the thing should sound
very good. I'm anticipating listening to one when it's available locally.
I don't trust specs that much: I've been disappointed numerous times in
the past. I don't trust Leonard Feldman's opinions, either. He's usually
about one notch away from "of all the CD players in the last year I've
reviewed, this is certainly one of them."
>also i am wondering how normal 16 bit dacs can work. wirewound resistors are
>usually the most accurate, and they are about .1%. if the resistors in a
>dac are .1% resistors, then i can not imagine how the dac can output stuff
>with more than -20%log(.1%) = 60 db of dynamic range.
One D/A circuit uses a "ladder" network of resistors. I don't remember the
configuration very well (and it would be a pain to draw in ASCII anyway)
but it only needed resistors of value R and 2R. Assuming the ratios of these
resistors could be made close, things work out. Nonlinearity is
measurable even with lab-grade DACs though. Some people have suggested that
we really need 20 bit precision in digital audio for top quality, but if you
have ever looked at the price of 20-bit A/D and D/A converters (which actually
are *accurate* to the full 20 bits), you might reconsider.
Doing 20 bits with a 1-bit scheme would "only" require an increase in the clock
rate of 16x; let's see, a 720 MHz clock... Maybe next year!
>even if the resistors are .001% resistors, i think themperature gradients
>in the chip would be enough to mess it up anyway: the chips in my cd player
>get quite hot. so how do the dacs in cd players work?
Well, as I've said, it isn't necessary for this precision, but temperature
is indeed a factor in the accuracy of these things. One more reason why
solid-state equipment can sound better after it has warmed up for a while.
--
Rich Fanning {sun,ucbcad,uunet}island!rich