PDA

View Full Version : Analogue output Quality



Howard Darwen
2004-11-24, 17:26
an interesting and entertaining (not to mention informative) debate.

am not discounting anything, and have spent a lot on audio kit throughout my
life in order to get as good a sound as i can afford.

the thing i am having trouble understanding though is the following ...

i can see that jitter is going to be present in any digital audio system.
but then you are using the clock to extract 1's and 0's in a digital stream.
if you get this wrong it means you read the bits wrong. at the analogue
level, i'd have thought that would be represented as
spikes/pops/glitches/whatever in the sound, not that it could make a
difference to the overall fidelity of what you are hearing.

the way that the DAC then shapes the analogue signal i can see is a
different matter though ...

cheers.

h.

-----Original Message-----
From: discuss-bounces (AT) lists (DOT) slimdevices.com
[mailto:discuss-bounces (AT) lists (DOT) slimdevices.com]On Behalf Of Neil Hastie
Sent: 24 November 2004 23:39
To: discuss (AT) lists (DOT) slimdevices.com
Subject: [slim] Analogue output Quality


There seems to be some confusion on this thread regarding the
format and use of the SPDIF. The following link may help
explain the operation.

http://www.epanorama.net/documents/audio/spdif.html

The basic problem is that Sony and Philips decided to pass both
data and clock in the same signal stream. To do this a system called
biphase-mark coding (or Manchester coding) is used.
Data is represeted by the logic level of the signal and is trivial to
extract.
The clock is represented by the edges of the signal stream and is a lot
more complex to reconstruct to the accuracy required by high definition
audio. Timing error in the recovered clock is known as jitter. Modifying the
edge speed of the signal pulses in the spdif stream modifies the jitter
seen by the receiver. There are measurable differences in jitter between
optical and wired connections in most systems due to this.

A lot of time and effort has been spent trying to remove jitter from the
recovered clock. As mentioned above it largely comes down to the amount
of money you are willing to pay. PLLs are good, Multilevel PLLs are better,
very large data buffers seem to be best.

One other point := If you have a transport powered by a cheap switching
power supply and containing a Ghz radio transceiver you may find an optical
conection is the best way to minimise the amount of RF noise injected into
your audio system.

Neil

(Using an optical link to feed a DAC with a 4 second data buffer).

Robin Bowes
2004-11-24, 17:58
Howard Darwen wrote:
> i can see that jitter is going to be present in any digital audio system.
> but then you are using the clock to extract 1's and 0's in a digital stream.

The point is that the DAC doesn't use a clock to extract anything - it
just decodes the stream of bits/bytes that are fed into it.

> if you get this wrong it means you read the bits wrong. at the analogue
> level, i'd have thought that would be represented as
> spikes/pops/glitches/whatever in the sound, not that it could make a
> difference to the overall fidelity of what you are hearing.

If the timing is even slightly out then the reconstructed waveform is
not a faithful reproduction of the sampled waveform. Unless it's really
bad, this will not cause anything as obvious as pops/crackles/etc. but
will rather cause more subtle effects described with subjective phrases
like "dead", "lifeless", "poor soundstage", etc.

> the way that the DAC then shapes the analogue signal i can see is a
> different matter though ...

Once the signal is in the analogue domain then conventional electronics
takes over (and that's a whole area for audiophile debate in itself!)

R.
--
http://robinbowes.com