Decent speaker cables at last! (soft troll)
 

On Sat, 26 Jul 2003 13:37:03 +0100, Jim Lesurf
wrote:

IIUC S/PDIF is a form of byphase modulation which essentially xors the data
with a clock to aid signal recovery using a fresh clock at the receiver.
The bandwidth should not, in principle, be critical, but may matter of the
recovery method is not well implimented.

From memory, biphase mark (as used in S/PDIF) is a form of frequency
modulation with zeros represented by the clock frequency and ones by
twice the clock frequency. (Much like the FSK used in 300 baud modems
- remember them?) It is not bandwidth efficient, but makes clock
extraction at the far end very easy. It's very similar to Manchester
encoding used for 10base Ethernet.

So both your VHF tuner and S/PDIF links are FM - one with analogue
signals and the other with digital.


--
Chris Isbell
Southampton
UK

Decent speaker cables at last! (soft troll)
 

In message , Chris Isbell
writes
On Sat, 26 Jul 2003 13:37:03 +0100, Jim Lesurf
wrote:

IIUC S/PDIF is a form of byphase modulation which essentially xors the data
with a clock to aid signal recovery using a fresh clock at the receiver.
The bandwidth should not, in principle, be critical, but may matter of the
recovery method is not well implimented.

From memory, biphase mark (as used in S/PDIF) is a form of frequency
modulation with zeros represented by the clock frequency and ones by
twice the clock frequency. (Much like the FSK used in 300 baud modems
- remember them?) It is not bandwidth efficient, but makes clock
extraction at the far end very easy. It's very similar to Manchester
encoding used for 10base Ethernet.

So both your VHF tuner and S/PDIF links are FM - one with analogue
signals and the other with digital.



Although you sometimes see the various types of Bi-phase modulation
described as 'frequency modulation', this is really not the case, and
the books that describe it as 'FM' are targeted at a fairly unlearned
market. These digital modulation schemes like biphase-M, biphase-S,
WAL-1 and WAL-2 are, as others have pointed out, controlled digital
phase inversions of the clock according to a particular algorithm. They
are a way of encoding the clock in with the data to ease recovery, and
in some cases, to provide a means of compensating for the HF roll-off
and dispersion caused to the signal by the characteristics of the cable.
The easiest way to see the effect of cable degradation of the signal is
still the 'text book' eye diagram.

The clock shouldn't be though of as a carrier.
--
Chris Morriss

Decent speaker cables at last! (soft troll)
 

On Sat, 26 Jul 2003 21:17:51 +0100, Chris Morriss
wrote:

Although you sometimes see the various types of Bi-phase modulation
described as 'frequency modulation', this is really not the case, and
the books that describe it as 'FM' are targeted at a fairly unlearned
market.

I not sure that I agree. Consider a bit stream of all zeros. This will
result in only the clock frequency appearing on the output. Likewise a
stream of all ones will only produce double the clock frequency. How
is this different from frequency modulation?

In other words, a zero results in no transition in the middle of the
clock period whilst a one does.

(I have just been working on a project implementing HDLC with bi-phase
mark encoding in firmware. All good clean fun! We had to get it right
because the receiver was a Zilog communication controller which had
the algorithms built into its guts and therefore outside of our
control.)


--
Chris Isbell
Southampton
UK

Decent speaker cables at last! (soft troll)
 

On Sun, 27 Jul 2003 08:12:42 +0100, Chris Morriss
wrote:

I'm not saying it can't be described as synchronous FM, but that it's
misleading. Your description only applies to biphase-M (or S). The
commoner WAL-1 (Biphase-L) can't really be described as FM in that the
coding algorithm is that a 1 has a 01 transition in the middle of the
bit period, while a 0 has a 10 transition. (With other transitions at
the bit boundaries to sort everything out or course.)

I completely agree that my description only applies to biphase-mark
and you point about it being misleading is well taken. However, I
personally found it easier to understand why there are more
'efficient' coding methods after realising that biphase-mark is
essentially FM (or FSK). (But then I have been told on a number of
occasions that I have a warped mind. :-)

The similarities between the digital S/PDIF and analogue FM radio are
interesting, notwithstanding the considerable differences. This ties
in quite nicely with the recent discussions here on the blurred
boundaries between digital and analogue systems. (I especially like
Jim Lesurf's analysis of an analogue record player as a digital system
that is on his Web pages.)


--
Chris Isbell
Southampton
UK

Decent speaker cables at last! (soft troll)
 

In article , Chris Isbell
wrote:
On Sun, 27 Jul 2003 08:12:42 +0100, Chris Morriss
wrote:

I'm not saying it can't be described as synchronous FM, but that it's
misleading. Your description only applies to biphase-M (or S). The
commoner WAL-1 (Biphase-L) can't really be described as FM in that the
coding algorithm is that a 1 has a 01 transition in the middle of the
bit period, while a 0 has a 10 transition. (With other transitions at
the bit boundaries to sort everything out or course.)

I completely agree that my description only applies to biphase-mark and
you point about it being misleading is well taken.

Actually, I found your description quite neat as an explanation of S/PDIF
as it makes it very easy to see how the clock becomes relatively easy to
recover.

However, I personally found it easier to understand why there are more
'efficient' coding methods after realising that biphase-mark is
essentially FM (or FSK). (But then I have been told on a number of
occasions that I have a warped mind. :-)

My mind must be of a similar distorted shape to yours. ;-

Mind you, I spent *years* trying to explain to military types why an FM
system that had smooth modulation did not *have* 'hop slots'. (And found
it even harder to get them to understand that things like this are bad news
if you want to reduce the signature of transmissions. :-) )

The similarities between the digital S/PDIF and analogue FM radio are
interesting, notwithstanding the considerable differences.

i have also found it interesting that you can argue that the 'Zenith'
stereo system used for FM radio is either frequency division *or* time
division, whichever suits the points you're trying to make. My own view of
many of these matters is that the 'standard' explanation is just the one
that most people find most convenient on most occasions. Yet alternative
explanations can sometimes be useful for illuminating specific points.
(Another example was the approach Don took to amp loads and feedback loops
a while ago.)

This ties in quite nicely with the recent discussions here on the
blurred boundaries between digital and analogue systems. (I especially
like Jim Lesurf's analysis of an analogue record player as a digital
system that is on his Web pages.)

OK, the cheque is in the post... :-)

Unfortunately, I could not do a precise calculation due to not knowing some
specific values - e.g. the typical size/shape of the 'PVC' sic molecules
used for LPs.

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.st-and.demon.co.uk/AudioMisc/index.html
Armstrong Audio http://www.st-and.demon.co.uk/Audio/armstrong.html
Barbirolli Soc. http://www.st-and.demon.co.uk/JBSoc/JBSoc.html

Decent speaker cables at last! (soft troll)
 

In message , Chris Isbell
writes
On Sun, 27 Jul 2003 08:12:42 +0100, Chris Morriss
wrote:

I'm not saying it can't be described as synchronous FM, but that it's
misleading. Your description only applies to biphase-M (or S). The
commoner WAL-1 (Biphase-L) can't really be described as FM in that the
coding algorithm is that a 1 has a 01 transition in the middle of the
bit period, while a 0 has a 10 transition. (With other transitions at
the bit boundaries to sort everything out or course.)

I completely agree that my description only applies to biphase-mark
and you point about it being misleading is well taken. However, I
personally found it easier to understand why there are more
'efficient' coding methods after realising that biphase-mark is
essentially FM (or FSK). (But then I have been told on a number of
occasions that I have a warped mind. :-)

The similarities between the digital S/PDIF and analogue FM radio are
interesting, notwithstanding the considerable differences. This ties
in quite nicely with the recent discussions here on the blurred
boundaries between digital and analogue systems. (I especially like
Jim Lesurf's analysis of an analogue record player as a digital system
that is on his Web pages.)



Thanks, I'll have a look at that!
--
Chris Morriss

Decent speaker cables at last! (soft troll)
 

Chris Morriss in uk.rec.audio:

In message , Jim H
writes


Digital signals genarally do not use 'square' waves, their signal is
added to a carrier wave. [1]


Wrong.
10 and 100 baseT Ethernet use digital signals. The leading edge has
digital pre-emphasis, but it's still done digitally.

I said generally. Yes, ethernet is baseband.

I think your missing the point to say a plot of pd against which produces
'square' signal that it IS digital, whereas if it looks curvy it IS
analogue.

It is often said that a modulated signal is analogue because a pd/time plot
shows a series of continuous values. However, if a phase/time plot were
made instead (for Phase Modulated signal) the values would be best
expressed as square line. It could be said therefore that since it occupies
discrete space, the phase of this signal is digital, whereas the pd is
continuous and therefore analogue.

The point I'm making is that digital and analogue shouldn't be thought of
in such absolute terms. They are ideas, a signal may be thought to be both
simultaneously if it helps.

The main diference is in the data, not how it is put on a wire. Dgital data
may only have discrete values, mhich may be perfectly copied. This is
imposible for analogue data, however a signal most comonly thought of as
analogue may carry digital data with the same perfection.

--
Jim H
3.1415...4999999 and so on... Richard Feynman

Decent speaker cables at last! (soft troll)
 

In article , Jim H
wrote:


It is often said that a modulated signal is analogue because a pd/time
plot shows a series of continuous values. However, if a phase/time plot
were made instead (for Phase Modulated signal) the values would be best
expressed as square line. It could be said therefore that since it
occupies discrete space, the phase of this signal is digital, whereas
the pd is continuous and therefore analogue.

I am not quite sure I know what you mean by "square line". Do you mean the
path of the phase on a polar vector plot would show a square? If so, are
you not assuming an infinite bandwidth for the modulated signal? Not clear
what you mean...

Of course, this opens up discussion of what we mean by 'phase' and even
'frequency'. ;-

The point I'm making is that digital and analogue shouldn't be thought
of in such absolute terms. They are ideas, a signal may be thought to
be both simultaneously if it helps.

Need to be careful with the term 'signal' as well. :-)

The actual variations of field or poential or current or whatever physical
property that is being used to convey the 'signal' are, I would say,
neither analog or digital. The chosen scheme for creating and understading
the meaning of the 'signal' (i.e. the resulting information-bearing
pattern) may be analog or digital, though, depending upon the choices made.

The main diference is in the data, not how it is put on a wire.

Agreed.

Dgital data may only have discrete values, mhich may be perfectly
copied.

Not quite 'perfect' as there is always a finite chance of errors produced
by noise, etc. We can reduce this chance to a very low level, but not
guarantee to exclude it entirely. The advantage of digital is that it
provides some level of error immunity, and provides relatively easy methods
for error detection and correction.

Slainte,

Jim



This is imposible for analogue data, however a signal most
comonly thought of as analogue may carry digital data with the same
perfection.

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.st-and.demon.co.uk/AudioMisc/index.html
Armstrong Audio http://www.st-and.demon.co.uk/Audio/armstrong.html
Barbirolli Soc. http://www.st-and.demon.co.uk/JBSoc/JBSoc.html

Decent speaker cables at last! (soft troll)
 

Jim Lesurf in uk.rec.audio:

In article , Jim H
wrote:

It is often said that a modulated signal is analogue because a
pd/time plot shows a series of continuous values. However, if a
phase/time plot were made instead (for Phase Modulated signal) the
values would be best expressed as square line. It could be said
therefore that since it occupies discrete space, the phase of this
signal is digital, whereas the pd is continuous and therefore
analogue.

I am not quite sure I know what you mean by "square line". Do you mean
the path of the phase on a polar vector plot would show a square? If
so, are you not assuming an infinite bandwidth for the modulated
signal? Not clear what you mean...

I don't like the term square line, maybe there is a better word for it?
For square I imagine cartesean axis showing a line which (ideally) jumps
quickly between values it stays at for some amount of time, in which all
lines are perpendicular to one axis.

Away from the particulars, the real point I was trying to make is that
however digital data is encoded, a plot can be made which shows the data
as descrete values, otherwise the digital data could not be retrieved at
the other end.

I don't like, for example, for someone to describe the output from a
computer modem as analogue, given that at some level it contains possibly
perfect representatitions of descrete values. I consider 'contains
possibly perfect representatitions of descrete values, to be a pretty
good definition of digital.

Of course, this opens up discussion of what we mean by 'phase' and
even 'frequency'. ;-

The point I'm making is that digital and analogue shouldn't be
thought of in such absolute terms. They are ideas, a signal may be
thought to be both simultaneously if it helps.

Need to be careful with the term 'signal' as well. :-)

Yes, agreed. medium may have been a beter word here. (although that
implys physical distribution)

The actual variations of field or poential or current or whatever
physical property that is being used to convey the 'signal' are, I
would say, neither analog or digital. The chosen scheme for creating
and understading the meaning of the 'signal' (i.e. the resulting
information-bearing pattern) may be analog or digital, though,
depending upon the choices made.

Agreed. I don't really think there is such thing as digital whilst the
signal is in the wire, other than if we think of it that way. Therefore
if it helps our understanding I don't think it is a contradiction to
think of a single wave to be simultaneously digital and analogue, or that
one property of the wave is digital whilst another is analogue.

The main diference is in the data, not how it is put on a wire.

Agreed.

Dgital data may only have discrete values, mhich may be perfectly
copied.

Not quite 'perfect' as there is always a finite chance of errors
produced by noise, etc. We can reduce this chance to a very low level,
but not guarantee to exclude it entirely. The advantage of digital is
that it provides some level of error immunity, and provides relatively
easy methods for error detection and correction.

The important word was 'may'! There is a possiblity that a digital piece
of data sent will be exactly what arrives. This is in contrast to an
analogue medium which can never be copied exactly. I agree that no
checking can ever rule out the possibility an error was made. So whilst
digital data may be perfectly copied, we never really know if it was.

This is imposible for analogue data, however a signal most
comonly thought of as analogue may carry digital data with the same
perfection.


--
Jim H
3.1415...4999999 and so on... Richard Feynman

Decent speaker cables at last! (soft troll)
 

In article , Jim H
wrote:
Jim Lesurf in uk.rec.audio:

In article , Jim H
wrote:

It is often said that a modulated signal is analogue because a
pd/time plot shows a series of continuous values. However, if a
phase/time plot were made instead (for Phase Modulated signal) the
values would be best expressed as square line. It could be said
therefore that since it occupies discrete space, the phase of this
signal is digital, whereas the pd is continuous and therefore
analogue.

I am not quite sure I know what you mean by "square line". Do you mean
the path of the phase on a polar vector plot would show a square? If
so, are you not assuming an infinite bandwidth for the modulated
signal? Not clear what you mean...

I don't like the term square line, maybe there is a better word for it?
For square I imagine cartesean axis showing a line which (ideally) jumps
quickly between values it stays at for some amount of time, in which
all lines are perpendicular to one axis.

I'd simply say the path drawn out on such a display was a square. The
signal may dwell for relatively long periods at the 'corners' of the
square, though.

Away from the particulars, the real point I was trying to make is that
however digital data is encoded, a plot can be made which shows the data
as descrete values, otherwise the digital data could not be retrieved
at the other end.

From what you say I recognise one of the standard display methods which is
used to show multiple-state AM/PM modulations. Typically giving an 'array'
of blobs on the screen with some fuzzyness or tracking inbetween if you are
using a simple 'analogue' display that shows the state on a time averaged
continuous basis. (As opposed to a processed display which would simply
group the states into a pre-defined set of locations and suppress the
low-level effects of noise, finite bandwidth, etc.)

I don't like, for example, for someone to describe the output from a
computer modem as analogue, given that at some level it contains
possibly perfect representatitions of descrete values. I consider
'contains possibly perfect representatitions of descrete values, to be
a pretty good definition of digital.

You need to distinguish between the actual time-voltage variations and the
way they are *used* to convey the information according to a specified
coding scheme. So far as the phone lines are concerned, the signal is just
like speech, and is a continuously varying voltage/current. However the
method used to impose information is 'digital'.


Of course, this opens up discussion of what we mean by 'phase' and
even 'frequency'. ;-

The point I'm making is that digital and analogue shouldn't be
thought of in such absolute terms. They are ideas, a signal may be
thought to be both simultaneously if it helps.

Need to be careful with the term 'signal' as well. :-)

Yes, agreed. medium may have been a beter word here. (although that
implys physical distribution)

There are a set of standard definition in Information Theory for these
things - although, despight having written a book about this, I often
muddle them myself! Hence I tend to try and distinguish between the
physical quantity used to carry or store the info with the coding method
employed to associate specific information with specific patterns of
variation of this quantity.


Not quite 'perfect' as there is always a finite chance of errors
produced by noise, etc. We can reduce this chance to a very low level,
but not guarantee to exclude it entirely. The advantage of digital is
that it provides some level of error immunity, and provides relatively
easy methods for error detection and correction.

The important word was 'may'! There is a possiblity that a digital piece
of data sent will be exactly what arrives.

Indeed, and with well designed and used systems, the probability of this is
quite high. :-) The snag is that - in information theory terms - we can
never be absolutely certain this *has* just happened, or will occur on the
next try.


This is in contrast to an analogue medium which can never be copied
exactly.

Need to be careful with the words 'exactly' and 'medium' here. :-)

The recovered infomation with analogue is always subject to noise, etc.
Hence the output can never be a 'perfect' representation of what was
intended. Each time we play the LP the noise will be different, and so we
can expect a slightly different result *if* true random noise is the
problem as opposed to systematic imperfections of the LP surface producing
the same 'noise' each time.

With digital systems the behaviour is slightly different as 'most of the
time' channel noise does not alter the output. However for a finite but
small amount of the time there will still be unintended changes due to the
finite risk of uncorrected errors, etc.

Digital systems - properly used - are much better than analogue in avoiding
effects due to noise added during transmission/storage. But they are not
perfect in the strict sense, just potentially very good if done well. You
still have to add some dither prior to sampling to avoid other problems, so
the output will still have a finite noise level, but this is systematic

Slainte,

Jim

--
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Audio Misc http://www.st-and.demon.co.uk/AudioMisc/index.html
Armstrong Audio http://www.st-and.demon.co.uk/Audio/armstrong.html
Barbirolli Soc. http://www.st-and.demon.co.uk/JBSoc/JBSoc.html