PDA

View Full Version : [OT] Audiophiles: Can the audio reproduction oftheCDbeequalled or bettered?



Simon Turner
2004-02-28, 08:04
I'm not too sure if I should have started another thread. I aplogise If i
should have done.
I am having the same discussion in the PnkFishMedia Audiophile forum. As you
might imagine I am meeting an awful lot of resistanc to the idea that a 4k
CD player might just be replaced (or even improved upon) by using a funny
looking 200 quid plastic box, a good quality but inexpensive DAC, wireless
network, a cheap CDROM drive and a bit of free software.

The only argument that I cannot give an adequate response to is the one
about jitter (mainly as i don't understand wtf it is). I've had this reply
(see below) and was hoping that it may make sense to someone on this group
who could perhaps comment on it in relation to the Squeezebox set up:

(Has anyone noticed how this list software keeps mangling the subject
headers? Shame as it dmanages the ability to search through the forum
archive.)

Simon Turner
Brighton UK

"
It's well known that different transports sound different in front of the
same DAC. Jitter is an issue at the point of domain conversion, but when
using an external DAC it has (in most cases) no local clock as a reference.

Instead the incoming SPDIF is usually presented to a line receiver, this
chip (or discrete circuit) is then responsible for buffering creating a
stable clock signal from the incoming data stream.

This is most commonly done with a PLL (Phase Locked Loop), this can be
simply viewed, in mechanical analogy, as a flywheel. It's own 'inertia'
prevents it from changing speed quickly, it can only respond to slow inputs.
It is, in electronic terms, a low pass filter, passing low frequency noise
(read:jitter) but attenuating high frequency jitter.

Jitter is only an issue (in that it only affect the sound) at the point of
domain conversion but jitter is present on the SPDIF data stream created by
the source.

The PLL's function is to filter the jitter present in order to provide a
stable clock for the subsequent DAC. The bandwidth of the PLL (which is a
trade off between the ability to lock on to the incoming signal and the
ability to remove jitter) determines the low pass characteristic - in many
receivers this pole is above the audio band and jitter within the audio
passband passes straight to the DAC.

By this mechanism some of the jitter present on the SPFDIF data stream from
your PC, which is, by definition, a noisy electrical environment, will be
passed to the DAC, unattenuated. I would be VERY surprised if the quality of
the SPDIF produced by a PC was near to that of a well-engineered, but
inexpensive CD player.

Add this to the fact that most PC's (and DAC's / CD players come to that)
use a phono socket as the SPDIF interface (which has an inherent mismatch to
the 75 ohm native impedance of the interface) this further increases jitter
through the cable reflections that WILL be present.

Now, not all DAC's are created equal, some cascade PLL's; the first provides
gross jiiter reduction and wide lock-in range, the second then further
filters the jitter, lowering the pole of the PLL's loop response. Some, such
as the Chord DAC64, use open-loop techniques with large memory buffers, such
that, theoretically at least, jitter present on the interface can never
reach the DAC.

The end result then largely depends upon the DAC, but if the DAC allows
jitter to pass to the DAC chip, the activity within the PC will no doubt
produce activity-related jitter variations, which I have no doubt will be
audible.
"

Daniel Cohen
2004-02-28, 10:36
On 28/2/04 at 3:04 pm +0000, Simon Turner wrote
>I am having the same discussion in the PnkFishMedia Audiophile forum. As you
>might imagine I am meeting an awful lot of resistanc to the idea that a 4k
>CD player might just be replaced (or even improved upon) by using a funny
>looking 200 quid plastic box, a good quality but inexpensive DAC, wireless
>network, a cheap CDROM drive and a bit of free software.

The discussion of the software has, I think, been about software for Windows.

Does anyone know what the equivalent software is (if it exists) for
mac OS X (and, for that matter, for Linux).
--
Daniel Cohen

Andrew W. Donoho
2004-02-28, 11:20
On Feb 28, 2004, at 09:04, Simon Turner wrote:
> The only argument that I cannot give an adequate response to is the one
> about jitter (mainly as i don't understand wtf it is). I've had this
> reply
> (see below) and was hoping that it may make sense to someone on this
> group
> who could perhaps comment on it in relation to the Squeezebox set up:


I'll give it a shot.



> "
> It's well known that different transports sound different in front of
> the
> same DAC. Jitter is an issue at the point of domain conversion, but
> when
> using an external DAC it has (in most cases) no local clock as a
> reference.


S/PDIF uses biphase mark coding, just like ethernet and many other
digital protocols. In specific, this coding allows the clock to be
recovered from the signal. Jitter occurs when the receiver is unable to
sample at the exact same period as the signal was encoded. This could
occur because of an impedence mismatch or other signal degradation.
There are two ways to handle this - use the optical interface or use a
good coax cable with a 75 Ohm impedence and matching connectors. (This
latter is probably impossible due to the RCA connectors used in the
SqueezeBox. Of course, you could possibly overcome this if you make
your cable the length of a wavelength of the harmonic of the clock.
Heck, just go use the optical and get over it.) Quoting from
<http://www.tnt-audio.com/clinica/convertus2_e.html>, jitter is "...
about 20% of the sampling period: normally we are talking about
hundreds of ps (pico second, 10E-12 sec), that is about 0,001% of
44.1kHz sampling period."

Unlike ethernet, because S/PDIF is a realtime, isochronous protocol
without error correction or retransmission, people worry a great deal
about jitter. I guess audiophiles need something to obsess over.


> Instead the incoming SPDIF is usually presented to a line receiver,
> this
> chip (or discrete circuit) is then responsible for buffering creating a
> stable clock signal from the incoming data stream.
>
> This is most commonly done with a PLL (Phase Locked Loop), this can be
> simply viewed, in mechanical analogy, as a flywheel. It's own 'inertia'
> prevents it from changing speed quickly, it can only respond to slow
> inputs.
> It is, in electronic terms, a low pass filter, passing low frequency
> noise
> (read:jitter) but attenuating high frequency jitter.
>
> Jitter is only an issue (in that it only affect the sound) at the
> point of
> domain conversion but jitter is present on the SPDIF data stream
> created by
> the source.
>
> The PLL's function is to filter the jitter present in order to provide
> a
> stable clock for the subsequent DAC. The bandwidth of the PLL (which
> is a
> trade off between the ability to lock on to the incoming signal and the
> ability to remove jitter) determines the low pass characteristic - in
> many
> receivers this pole is above the audio band and jitter within the audio
> passband passes straight to the DAC.
>
> By this mechanism some of the jitter present on the SPFDIF data stream
> from
> your PC, which is, by definition, a noisy electrical environment, will
> be
> passed to the DAC, unattenuated. I would be VERY surprised if the
> quality of
> the SPDIF produced by a PC was near to that of a well-engineered, but
> inexpensive CD player.


This is where the optical from a SqueezeBox should shine compared to
traditional RCA from a computer. The SqueezeBox is a relatively low
power, low complexity and, hence, low noise system. This means that
Dean's team should have been able to make an excellent digital MP3,
AIFF, WAV to S/PDIF converter. In other words, the SqueezeBox should be
the equivalent of that "well-engineered, but inexpensive CD player"
referenced above.


> Add this to the fact that most PC's (and DAC's / CD players come to
> that)
> use a phono socket as the SPDIF interface (which has an inherent
> mismatch to
> the 75 ohm native impedance of the interface) this further increases
> jitter
> through the cable reflections that WILL be present.
>
> Now, not all DAC's are created equal, some cascade PLL's; the first
> provides
> gross jiiter reduction and wide lock-in range, the second then further
> filters the jitter, lowering the pole of the PLL's loop response.
> Some, such
> as the Chord DAC64, use open-loop techniques with large memory
> buffers, such
> that, theoretically at least, jitter present on the interface can never
> reach the DAC.
>
> The end result then largely depends upon the DAC, but if the DAC allows
> jitter to pass to the DAC chip, the activity within the PC will no
> doubt
> produce activity-related jitter variations, which I have no doubt will
> be
> audible.
> "


The SqueezeBox, because it is a dedicated audio rendering system, will
not have "activity-related" jitter variations. This is referring to the
noise a general CPU makes based upon the tasks it is executing. In
other words, a compile of the Linux kernel could induce different noise
in your music versus just listening to a machine that is only playing
through Winamp. This is not an issue to the SqueezeBox until it's
buffers are starved for data. Then you get skipping and pauses.

Andrew





____________________________________
Andrew W. Donoho
awd (AT) DDG (DOT) com, PGP Key ID: 0x81D0F250
+1 (512) 453-6652 (o), +1 (512) 750-7596 (m)

Pat Farrell
2004-02-28, 12:00
At 01:20 PM 2/28/2004, Andrew W. Donoho wrote:
>On Feb 28, 2004, at 09:04, Simon Turner wrote:
>>The only argument that I cannot give an adequate response to is the one
>>about jitter (mainly as i don't understand wtf it is).

Don't feel bad. Most people don't understand what it is.
Even the folks that write about it in magazines.

>>same DAC. Jitter is an issue at the point of domain conversion, but when
>>using an external DAC it has (in most cases) no local clock as a reference.

True, but the data's clock rate is well known, 44.1kHz.
Benchmark has good information about this at
http://www.benchmarkmedia.com/appnotes-d/whyultralock.asp
The only thing that contributes to jitter in a DAC is the quality of
its clock.

> There are two ways to handle this - use the optical interface or use a
good coax
>cable with a 75 Ohm impedence and matching connectors. (This latter is
>probably impossible due to the RCA connectors used in the SqueezeBox.

Andrew, why do you say this? There is a nice RCA SPDIF output on the
SqueezeBox.
You could have a hard time with an overpriced Monster "digital" cable, but
normal high quality 75-Ohm coax works just fine.

One critical part of digital cabling that the vendors ignore is that
lots of the things (capacitance induced roll off of high frequency,
need for "burn in", etc.) that they sell in the analog cable world
are not needed, useful, or justified in a digital signal world.


>correction or retransmission, people worry a great deal about jitter. I
>guess audiophiles need something to obsess over.

This is the key observation.



>>Add this to the fact that most PC's (and DAC's / CD players come to that)
>>use a phono socket as the SPDIF interface (which has an inherent mismatch to
>>the 75 ohm native impedance of the interface) this further increases jitter
>>through the cable reflections that WILL be present.

I don't follow this at all. Sure, if you use audio interconnects, you will have
an impedance mismatch. You are supposed to use 75-ohm video cable.
And if you do, there is no reason for any additional reflections, or any
bad things.


>The SqueezeBox, because it is a dedicated audio rendering system, will not
>have "activity-related" jitter variations.

In the context of this thread, it is even more restricted. When you use an
outboard DAC, the SqueezeBox is an digital stream transfer system.
Jitter is not then a SlimDevices problem, they don't touch the data
in the analog domain where you have an opportunity to have jitter be
"something to obsess over. "

In short, with a good DAC, jitter is not an issue. Unless you want it to
be, in
which case, spend as much as you want, but you probably won't want
a $300 SqueezeBox in the chain anyway.

Pat

Stephen Ryan
2004-02-28, 13:56
On Sat, 2004-02-28 at 12:36, Daniel Cohen wrote:
> On 28/2/04 at 3:04 pm +0000, Simon Turner wrote
> >I am having the same discussion in the PnkFishMedia Audiophile forum. As you
> >might imagine I am meeting an awful lot of resistanc to the idea that a 4k
> >CD player might just be replaced (or even improved upon) by using a funny
> >looking 200 quid plastic box, a good quality but inexpensive DAC, wireless
> >network, a cheap CDROM drive and a bit of free software.
>
> The discussion of the software has, I think, been about software for Windows.
>
> Does anyone know what the equivalent software is (if it exists) for
> mac OS X (and, for that matter, for Linux).

cdparanoia. Readily available on Linux, and possibly available for OS
X, though I don't know how well tested there.

The Finder also offers some similar functionality, in that you can just
drag and drop "files" from the CD to your hard drive, though I don't
know about the error correction capabilities of the Finder under those
circumstances.
--
Stephen Ryan <steve (AT) deepthought (DOT) dartmouth.edu>

Andrew W. Donoho
2004-02-28, 14:10
On Feb 28, 2004, at 13:00, Pat Farrell wrote:
> At 01:20 PM 2/28/2004, Andrew W. Donoho wrote:
> > There are two ways to handle this - use the optical interface or use
> a good coax
> >cable with a 75 Ohm impedence and matching connectors. (This latter
> is
> >probably impossible due to the RCA connectors used in the SqueezeBox.
>
> Andrew, why do you say this? There is a nice RCA SPDIF output on the
> SqueezeBox.
> You could have a hard time with an overpriced Monster "digital"
> cable, but
> normal high quality 75-Ohm coax works just fine.


Pat,

I am far from an audiophile but I do not believe that the RCA jacks
are 75 Ohm impedence connectors. Therefore, they can contribute to
ringing, etc. I am possibly wrong and if so, please inform me so I can
take my $31 S/PDIF cable back to Radio Shack and get a 75 Ohm cable
with RCA jacks.

Taking my audiophile poseur hat off and putting on my physicist/system
architect hat on, this is just not an issue with digital protocols over
decent connectors and cables. I said decent and not over hyped Monster
cables.

Andrew

____________________________________
Andrew W. Donoho
awd (AT) DDG (DOT) com, PGP Key ID: 0x81D0F250
+1 (512) 453-6652 (o), +1 (512) 750-7596 (m)

Pat Farrell
2004-02-28, 14:35
At 04:10 PM 2/28/2004, Andrew W. Donoho wrote:
> I am far from an audiophile but I do not believe that the RCA
> jacks are 75 Ohm impedence connectors.

There are three RCA jacks on a Squeezebox.
Two audio (left and right) and one SPDIF.
The SPDIF standard calls for 75 ohm cable
connected with RCA or BNC connectors.


> Therefore, they can contribute to ringing, etc. I am possibly wrong and
> if so, please inform me so I can take my $31 S/PDIF cable back to Radio
> Shack and get a 75 Ohm cable with RCA jacks.

I'd get a $5 Radio Shack "video cable" that is 75 ohm with RCAs
on each end. Try it. if you can tell the difference, take back the $5 one.
If not, take back the $31 one.

I use 75 ohm coax all over my recording studio, with zero problems.

> Taking my audiophile poseur hat off and putting on my
> physicist/system architect hat on, this is just not an issue with digital
> protocols over decent connectors and cables. I said decent and not over
> hyped Monster cables.

They sure are good at marketing, aren't they? Monster is everywhere.
They are usually good products, just way over hyped and overpriced. YMMV, etc.

Pat

Mark Bennett
2004-02-29, 11:57
Pat Farrell wrote:

> At 01:20 PM 2/28/2004, Andrew W. Donoho wrote:
>
>>> same DAC. Jitter is an issue at the point of domain conversion, but when
>>> using an external DAC it has (in most cases) no local clock as a
>>> reference.
>
>
> True, but the data's clock rate is well known, 44.1kHz.

just to be pedantic :-) the clock rate of a S/PDIF link will be higher
than 44.1kHz. Just the audio stream is

(44.1ksamples/s)*(2 channels)*(16 bits/channel) = 1.4112MHz

By the time you add overhead for clock recovery and sync pulses the
actual frequency of the signal seems to be 4 times as high.

>> correction or retransmission, people worry a great deal about jitter.
>> I guess audiophiles need something to obsess over.
>
> This is the key observation.

ROTFL - couldn't agree more

>> The SqueezeBox, because it is a dedicated audio rendering system, will
>> not have "activity-related" jitter variations.
>
>
> In the context of this thread, it is even more restricted. When you use an
> outboard DAC, the SqueezeBox is an digital stream transfer system.
> Jitter is not then a SlimDevices problem, they don't touch the data
> in the analog domain where you have an opportunity to have jitter be
> "something to obsess over. "

Here I am inclined to be doubtful. It's really not clear to me how
good external DAC's really are at reducing the jitter on the input
signal. They should be pretty good, but I don't know if they really
are, and it's hard to find much objective information on the subject.

If there is excessive jitter on the squeezebox S-PDIF output then,
depending on the DAC, some of this might get through to that actual
D-A conversion stage and affect the analogue signal.

Some DAC's will be better at this than others.

The only exception is something like the Chord DAC64, which contains
a 4 second RAM buffer, which basically allows it to throw away the
incoming clock and construct it's own which can be completely jitter
free.

But then we're talking about $3k worth of DAC on a $250 transport,
which might just be going a bit far. I'll let you know if I can
persuade a dealer to let me try one out.....

Mark Bennett
2004-02-29, 12:05
Andrew W. Donoho wrote:


> The SqueezeBox, because it is a dedicated audio rendering system, will
> not have "activity-related" jitter variations. This is referring to the
> noise a general CPU makes based upon the tasks it is executing. In other
> words, a compile of the Linux kernel could induce different noise in
> your music versus just listening to a machine that is only playing
> through Winamp. This is not an issue to the SqueezeBox until it's
> buffers are starved for data. Then you get skipping and pauses.

Well, yes and no. At the end of the day a squeezebox is a CPU, running
code and instructions. If nothing else there's the TCP/IP stack software
running to handle the network.

So, it is running code, which _could_ affect the power supply noise, and
therefore the jitter. Of course there's orders of magnitude less going
on than in a PC, and probably a similar amount to a modern day CD
Player, so I'm being pedantic again :-).

Don't confuse jitter with buffers running empty. The jitter would have
to be staggeringly large for it to cause the buffer to empty! If you
get skipping it's because there's a network problem, or the server
has got too busy to keep up with streaming the data. This is not jitter
in the real sense of the word.

T
2004-02-29, 12:14
> just to be pedantic :-) the clock rate of a S/PDIF link will be higher
> than 44.1kHz. Just the audio stream is
>
> (44.1ksamples/s)*(2 channels)*(16 bits/channel) = 1.4112MHz

In an SPDIF frame, there is 32 bits of data/channel, not 16 (not all of
which is audio).

Thus, the clock rate is 2.8224 MHz for a 44.1 kHz sample rate.

Tom

Mark Bennett
2004-02-29, 12:20
BTW, if the jitter does turn out to be a problem, anyone
up for fitting a low jitter clock modification and a better
power supply?

I quite like the idea of building a new case to house the
squeezebox and to incorporate a high quality power supply
and clock generator if necessary :-)

That way If I can find a spare case to match my existing
(Cyrus) kit then it would even fit in well visually!

Cheers,
Mark.

Mark Bennett wrote:

>
> Pat Farrell wrote:
>
>> At 01:20 PM 2/28/2004, Andrew W. Donoho wrote:
>>
>>>> same DAC. Jitter is an issue at the point of domain conversion, but
>>>> when
>>>> using an external DAC it has (in most cases) no local clock as a
>>>> reference.
>>
>>
>>
>> True, but the data's clock rate is well known, 44.1kHz.
>
>
> just to be pedantic :-) the clock rate of a S/PDIF link will be higher
> than 44.1kHz. Just the audio stream is
>
> (44.1ksamples/s)*(2 channels)*(16 bits/channel) = 1.4112MHz
>
> By the time you add overhead for clock recovery and sync pulses the
> actual frequency of the signal seems to be 4 times as high.
>
>>> correction or retransmission, people worry a great deal about jitter.
>>> I guess audiophiles need something to obsess over.
>>
>>
>> This is the key observation.
>
>
> ROTFL - couldn't agree more
>
>>> The SqueezeBox, because it is a dedicated audio rendering system,
>>> will not have "activity-related" jitter variations.
>>
>>
>>
>> In the context of this thread, it is even more restricted. When you
>> use an
>> outboard DAC, the SqueezeBox is an digital stream transfer system.
>> Jitter is not then a SlimDevices problem, they don't touch the data
>> in the analog domain where you have an opportunity to have jitter be
>> "something to obsess over. "
>
>
> Here I am inclined to be doubtful. It's really not clear to me how
> good external DAC's really are at reducing the jitter on the input
> signal. They should be pretty good, but I don't know if they really
> are, and it's hard to find much objective information on the subject.
>
> If there is excessive jitter on the squeezebox S-PDIF output then,
> depending on the DAC, some of this might get through to that actual
> D-A conversion stage and affect the analogue signal.
>
> Some DAC's will be better at this than others.
>
> The only exception is something like the Chord DAC64, which contains
> a 4 second RAM buffer, which basically allows it to throw away the
> incoming clock and construct it's own which can be completely jitter free.
>
> But then we're talking about $3k worth of DAC on a $250 transport,
> which might just be going a bit far. I'll let you know if I can
> persuade a dealer to let me try one out.....
>
>
>
>

Mark Bennett
2004-02-29, 12:24
That explains the bit I couldn't track down. The actual signal
is sent by Manchester bi-phase coding, which means that the max
frequency is twice as high as the data transmission rate, hence
the 5.6448MHz over the cable.

T wrote:

>>just to be pedantic :-) the clock rate of a S/PDIF link will be higher
>>than 44.1kHz. Just the audio stream is
>>
>> (44.1ksamples/s)*(2 channels)*(16 bits/channel) = 1.4112MHz
>
>
> In an SPDIF frame, there is 32 bits of data/channel, not 16 (not all of
> which is audio).
>
> Thus, the clock rate is 2.8224 MHz for a 44.1 kHz sample rate.
>
> Tom
>
>

Pat Farrell
2004-03-04, 15:25
I just hooked a new Benchmark DAC-1 to my Sqbx.
Works great. So far, both Toslink and SPdif work.

I haven't had it long enough for a serious comparison
on sonic quality.

If I get really crazy, I'll compare Toslink to SPdif, RCA connections
to balanced XLR, fixed audio from the Sqbx, fixed audio from the DAC,
variable into my Classe, etc. But I might just play some tunes.