This is only a preview of the December 1990 issue of Silicon Chip. You can view 58 of the 120 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
|
The great CD green
pen controversy
Compact discs are supposedly unimprovable,
aren't they? Well, to the average user that is
virtually true. Which is why claims for
improvement after the application of a green
paint or dye to the outside of a compact disc are
intriguing to say the least. Well, we've done the
tests and we reckon it's a load of garbage.
By LEO SIMPSON
Want to improve the sound of your
compact discs? Of course you do.
After all, no matter how good your
hifi system, you will always be keen
to improve the sound quality, particularly if it can be done for only a
small outlay.
14
SILICON CHIP
And so what would your reaction
be if I told you that you could improve the sound of your compact discs
for an outlay of only $10 or perhaps a
little more, say $39.95? If it gave an
audible improvement to a system
costing a thousand or more dollars, it
would be well worthwhile wouldn't
it?
The modification is that you apply
a green dye from a special felt tip
marker pen, to the inside and outside
edge of your compact discs. This is
claimed to give real sonic improvements. In fact, I'll quote directly from
the literature for the "CD Stop Light"
pen: "... will improve sound in all
areas - depth, imaging, better bass,
smoother mids and highs, less graininess and more musicality. When it's
applied to a well recorded Compact
Disc the difference can be detected
immediately. The first dynamic note
will be tighter and crisper than before".
There is a lot more along these lines
from the distributor of this product.
-.·· ·1
This idealised diagram shows the
optical system of a compact disc and
the laser pickup. Invisible infrared
light from the semiconductor laser at
bottom is focussed by a lens system
onto the disc's recorded surface. Light
reflected from the disc passes back
through the lens and is diverted by
the prism and into the photodiode.
Some of the laser light will be
scattered and will reflect from the
disc edges back to the photodiode. But
will that cause any additional errors?
So why is it supposed to work? The
literature for these pens is not absolutely clear on this point, but let me
quote directly from the CD Stop Light
literature again: "A laser reads light,
if it reads any light other than a direct
reflection of its own transmitted light
it will read it as a distortion. Also if
its own transmitted light is reflected
around the disc and it is then read by
the laser it will also be interpreted as
a distortion."
Are you any the wiser? What I interpret this to mean is that some of
the light from the laser in your CD
player is scattered once it passes
through the clear plastic on the underside of the disc. That scattered
light will travel out to the shiny edge
of the disc and be reflected back,
whereupon it will add in a spurious
way to the light reflected from the
aluminium recorded layer of the disc.
In so doing, it will degrade the signal
from the optical detector (which is
not a laser but an infrared detector
diode) and therefore make it more
prone to give errors in the data signal.
If you accept that concept, then you
should have no problems accepting
the next part. The light from a CD
laser is infrared and therefore invisible. It will be absorbed by a green
paint or dye. So if any stray light from
the laser is absorbed by the green dye,
that should make the data signal from
the CD that much cleaner. In other
words, there should be less errors in
the data and that should, ultimately,
mean cleaner sound.
OK, so if this interpretation is correct (and what other interpretation
can there be?), then it should be possible to test for the reduction of errors
off a CD, once it is coated on the
edges with one of these green pens.
How do you test for data errors from
a compact disc? We're not talking here
about steady state distortion of an
analog signal by the way. We are talking about errors in the digital data
stream. Data errors, if they occur, and
are not corrected, will not show up as
steady harmonic distortion. Instead
they will be momentary imperfections
in the signal which will be virtually
impossible for any analog instrument
to measure. So, we need a compact
disc analyser to do the tests.
Compact disc analyser
The only organisation in Australia
that we know of which has a compact
disc analyser is Disctronics Ltd, in
Melbourne. They are the only manufacturer of compact discs in this country. So we approached them with a
view to using their compact disc analyser to do a whole series of tests to
prove conclusively whether or not the
CD Stop Light and other green pens
actually work.
Disctronics Ltd has four compact
disc analysers in their Quality Assurance laboratory but they are not portable and so we could not borrow one.
We would have to go to them. And so,
on 16th October this year, I visited
Disctronics for the whole day. I was
armed with a selection of compact
discs both good and faulty, a number
of green dye pens and various cameras. I also had the full-time assistance of Disctronics' Quality Assurance staff member, Bill Dines, without whom these tests would not have
been possible.
Projected tests
I had planned a number of tests.
The first of these was fairly simple
and involved looking at the signal
waveform from the optical detector
(the infrared detector diode) before it
is processed by the decoder circuitry.
Contrary to what you might think,
this is not a simple "squared up" digital signal. Rather, it is rounded, because the laser light is being reflected
Since the laser pickup is a mechanical system, it will be subject to variations in
the way it tracks the disc surface. This makes it extremely difficult, if not
impossible, to detect any variation caused by the marker pens.
DECEMBER1990
15
the effect of signal hash due to scattering of the laser light by scratches
on the underside of the disc.
The photo of Fig.1 is from an absolutely pristine disc which is also from
a very good master. It has an almost
ideal "eye pattern". Fig.2 is from an
average quality CD with not very
many scratches or surface defects.
Fig.3 is from a CD which is very badly
scratched and virtually unplayable in
parts.
Hash on the digital signal
Our objective tests of the CD stop light pens was done in the quality assurance
laboratory of Disctronics Ltd, in Melbourne. The author is on the right, decked
out in clean room gear, while QA technician Bill Dines is on the left. Between
them is one of the Sony CD analysers.
from pits in the CD metallisation
which are not perfect (ie, they don't
have vertical walls) and because the
laser spot diameter is not extremely
small with respect to the pit diameter.
The signal I looked at is known as
the "eye pattern" and reveals, to the
practised observer, the depth of the
pits and the general quality of the
pressing from the master. Further-
more, because it is an unsynchronised signal (ie, not stationary on the
oscilloscope screen) and one which
is hard to photograph, I looked at the
signal from the lead-in track of a
number of CDs.
Three photographs of eye patterns
from three separate CDs are featured
in this article - see Figs.1, 2 and 3.
Not only do these photos show the
pressing quality but they also show
Badly scratched discs such as this abused example do
cause the laser to produce a lot of errors, as was clearly
shown in our tests. However, even badly scratched discs
like this can be played on most CD machines. A worse
problem is scratches on the label which make the disc
unplayable.
16
SILICON CHIP
Notice how there is a lot of "hash"
on the signal of Fig.3 compared with
the very clean pattern in Fig.1. If
you're not sure what I'm talking about,
the hash is the blurring at the top and
bottom of the signal pattern.
This is a clear demonstration that
laser light is scattered by scratches
and surface defects on a CD and that
the resultant waveform from the optical detector is noisy. Noisy signals
like these are bound to have more
errors than clean signals.
Just for the exercise I compared the
eye patterns of the same discs when
they were treated with the CD pens
but if there was any effect there was
no way you could see a difference in
the patterns. Clean signals stayed
clean; dirty signals stayed dirty.
Measuring error signals
So far, I had confirmed that laser
light can be scattered by surface imperfections on CDs and that this scat-
Applying the CD Stop Light pen is a little easier because
it has a groove in the tip. It's still a messy job, though.
FIG.1
FIG.2
These three "eye" patterns show the signal directly from the laser photodiode
of the CD player. Fig.1 is from a first quality disc with an almost ideal "eye"
pattern. Fig.2 is from an average good quality disc, while Fig,3 is from the very
badly scratched disc shown in a photo on the facing page. These photos clearly
demonstrate that scratches (and presumably internal reflections) can degrade
the signal from the photodetector. But the CD pen did nothing to improve
matters in this regard.
tering caused distortion (ie, noise and
hash) of the recovered digital signal.
So the first part of the claim for the
CD Stop Light pen is undoubtedly
true. Now I had to test whether the
green dye treatment had any effect.
Measuring errors in the data stream
from compact discs can be done in a
number of ways. Disctronics has two
types of CD analyser in their QA lab,
three made by Sony and one made by
Shape, a US company. The Sony
machines, type CDP-5000, can give a
printout of every block error and interpolation on a disc.
And this is where I came to the first
hurdle. The printout for a typical good
disc with an average block error rate
might easily run to 100 to 200 pages,
with 66 lines per page. For a bad disc,
a complete printout would run to
thousands of pages!
Comparing two such printouts,
before and after treatment with the
green pens, was clearly going to be
impossible. I had quite a few discs to
do, so I had to compromise. I decided
that for each disc I would only do an
analysis of the innermost track (track
one) and the outermost track. Because
these were closest to the inner and
outermost edges of the CD, they would
be the most likely to benefit from the
green pen treatment, if it worked.
I wanted to do at least four discs to
begin with. Two of these would be
pristine, first quality discs. Two would
be really rubbishy, with lots of
scratches. After all, if scratches cause
light scattering, then the green pens
should have an even more beneficial
effect. I also wanted to test whether
the pens worked better on moulded
or pressed compact discs.
Those pressed by Disctronics have
a clear centre section you can see
through and a sharp outer edge. By
contrast, those made by PDO (Philips
Dupont Optics) and used for many of
the Philips and European labels have
the metallisation right across to the
centre of the disc and a somewhat
more rounded outer edge. So I had to
test at least four discs.
I started with a very good quality
PDQ disc. As expected, it had a very
low block error rate right through the
disc but right at the end it suddenly
had a bunch of interpolation errors.
Interpolation errors are those that
the compact disc player cannot correct by using the standard CIRC (Cross
Interleave Reed Solomon Code)
method developed by Sony. Instead,
it virtually "guesses" what the data
should be, from the data immediately
before and after the interruption in
the signal.
Interpolation errors are serious
because they represent a definite
departure from the original signal and
therefore they are distortion. In serious cases, as on badly scratched discs,
interpolation errors become audible the disc sounds distorted.
Block errors are corrected
And that brings me to a most important point. All compact discs have
block errors, lots of them. There are
7350 blocks (ie, parcels of serial data)
per second and therefore around 30
million blocks on a disc. The average
block error rate on a really good disc
FIG.3
is likely to be about 20 in every 10second period. Therefore, such a disc
is going to have around 9000 block
errors. And that's a good disc!
For a badly scratched disc, the average block error rate can be 400 to
500 in every 10-second period. This
disc is going to have between one and
two million block errors over the hour
or so of its playing period!
But the good news is that block
errors are not only detected but that
they are all corrected. The resultant
data is exactly the same as ifno block
errors had occured in the system at
all.
Many people have trouble with this
concept of data correction. It means
exactly what it says - data is corrected to precisely the same value as
it would have been if the error had
never occurred.
So good is the CIRC error correction that it can completely correct for
data interruptions of up to 4000 bits that's equivalent to a spot on the disc
of 2.5mm in diameter. Most CD players cannot track over such a large
interruption by the way, but the error
correction can cope with it. However,
the story becomes even more complicated.
Never twice the same
The trouble with measuring and
comparing block error rates read off
discs is that the errors will be different for each disc each time it is played!
Why? Because the laser tracking
mechffnism just cannot track the disc
the same way each time it is played.
It is a servo-mechanical system.
It can no more track a disc identically each time it is played than you
can drive down a road in an identical
way each time. Your car will always
take a slightly different track than
before - and so will the laser. And
therefore the block errors are bound
to be different, even though the block
DECEMBER 1990
17
01:00:13:00
NOISE B.E.R
01:00:13:14
NOISE B.E.R
01:00:13:30
NOISE B . E.R
01:00:13 : 44
NOISE B.E.R
01:00:13:59
NOISE B.E.R
01:00:13:74
NOISE B. E . R
01:00:13:74
NOISE INTER
01:00:14:14
NOISE B . E.R
01:00:1,4:29
NOISE B.E.R
01:00:14:44
NOISE B.E . R
01:00:14:59
NOISE B.E . R
01:00:14:74
NOISE B.E.R
01:00:14:74
NOISE INTER
01:00:15:14
NOISE B.E.R
BLOCK ERROR RATE THRESHOLD
INTER POLATION
THRESHOLD
BLOCK ERROR RATE (MEAN)
BLOCK ERROR RATE (MAX.)
01:00:13:06
01:00:13:21
01:00:13:21
01:00:13:36
01:00:13:51
01:00:13:66
01:00:14:06
01:00:14:21
01 : 00:14:22
01:00:14:36
01:00:14:51
01:00:14:67
01:00:15:06
01:00:15:22
BLOCK
INTER
BLOCK
BLOCK
ERROR RATE
POLATION
ERROR RATE
ERROR RATE
NOISE
NOISE
NOISE
NOISE
NOISE
NOISE
NOISE
NOISE
NOISE
NOISE
NOISE
NOISE
NOISE
NOISE
POLATION
POLATION
B901001300033703710984
B901001314033903200848
B9010013300353039908F7
B901001344032703230869
B901001359033403290891
B9010013740306017408C7
C003060174020040
B901001414031202230925
B9010014290292024508DA
B9010014440289021508BC
B9010014590281020908BD
B9010014740294020908EB
C002940209020063
B901001514027701010866
0070
0071
0072
0073
0074
0075
0076
0077
0078
0079
0080
0081
0082
0083
B901001306031102150806
B901001321033002150840
C00330021502000C
B90100133603290324085E
B901001351033403220882
B90100136603060217085D
B90100140603250217081D
C002930217020070
B9010014220293021708A6
B901001436028401080 89 B
B901001451025700030883
B9010014670276000308B8
B901001506025500030837
B90100152202800003097F
:0200
:0100
:327 . 5
:0479
B.E.R
B.E . R
INTER POLATION
B.E.R
B.E.R
B.E.R
B.E.R
INTER POLATION
B . E.R
B.E . R
B.E.R
B. E.R
B.E.R
B.E . R
THRESHOLD
THRESHOLD
(MEAN)
(MAX.)
0069
0070
0071
0072
0073
0074
0075
0076
0077
0078
0079
0080
0081
0082
: 0200
:0100
:325 . 2
:0479
Shown here are two CD analyser printouts of a scratched compact disc. Both
printouts are for the same compact disc, played twice in succession on the same
machine, without any treatment or even being removed from the machine. Note
how none of the tracking errors are the same! Clearly, our tests showed that not
only does the green dye treatment not work but there is no way that anyone
could claim that it does!
error rates will be much the same.
I had not thought about this before
but it was brought home to me
strongly when I tried to compare two
printouts for a CD, before and after
treatment. I expected to find that
many, if not all, block errors would
be much the same. They're not.
For example, I compared a
scratched disc at time 01:00:02, before and after treatment with the green
pen. In this very short time interval
18
SILICON CHIP
(one hundredth of a second), there
were two block errors, followed by an
interpolation, followed by another
three block errors. But while the block
error rate was identical for both
readouts, none of the block errors or
the interpolation were the same.
Now I can't say this for sure, but
having carefully examined and compared dozens of pages of block error
printouts, I don't think there is much
likelihood of any given block error
occuring twice in succession, for two
successive playings of a disc.
Think about it: each time your car
goes over a pot-hole, it will respond
in a slightly different way each time.
And so, when the laser encounters an
interruption in the data of a CD, the
error is detected (and corrected) in a
slightly different way each time.
Comparing the printouts
After doing a series of printout
comparisons, I concluded that not
only did the green pens not appear to
have any measurable effect on the
block error rate but they had no effect
on the number of interpolations either. Still, I had to pursue the corn-
parisons for at least the four discs I
mentioned above. So for the remainder of the tests done on the Sony CD
analyser, I just took the printout summary. This gives the maximum and
average block error rates for the disc.
As I have just stated, the differences
between the before and after treatment results were negligible.
There was one anomaly though.
You might remember that I mentioned
measuring a very good PDO disc
which had a bunch of interpolations
at the end, printed out on the summary. Since this was early in the process and I had not thought enough
about the difficulties of the test, I
decided to see if the green pen had
any effect on this bunch of interpolations. The trouble was, I did not think
to carefully examine the disc for dust
or fingerprints at that stage.
Messy business
I then applied the green pen to the
outer and inner edges of the disc according to the instructions. Now this
is a tricky process. If you are not absolutely careful and do not have a
steady hand, it is quite easy for the
pen to wander over onto the label or
the playing surface. On this occasion
I made rather a mess of things, so I
cleaned it off thoroughly with isopropyl alcohol and then reapplied the
green pen.
On the subsequent test, the disc
went through with no interpolation
errors at all. Voila! The pen really
worked! Or did it? I then thought that
perhaps there was a fingerprint on
the disc for the first test and when I
had subsequently cleaned the disc,
after smudging it with the pen, I had
removed it.
So I thoroughly removed all the
green dye from the disc, made sure it
was spotlessly clean and then put it
through the same test again. And what
do you think? Not only were the interpolations still absent but the maximum block error rate for the disc was
now slightly lower.
The pen had had no effect at all!
But cleaning the disc thoroughly had
made a big difference to the interpolations.
Block error histograms
Just in case the Sony CD Analyser
was not giving the full picture, I decided to do some discs on the Shape
CD Analyser. This unit uses a Revox
According to the instructions which come with the pens, the green dye must
also be applied to the inner edge of the compact disc and to any grooves that
may also be present in the central area.
CD player linked to a special card in
an IBM PSZ computer. As well as
giving block error summaries, this
gives a histogram of the block errors
on a disk. Two of these histograms are
shown with this article.
They show that even though the
maximum and average block errors
may be similar, the actual histograms,
showing a 10-second moving average
of block errors, are slightly different.
But that is what you would expect,
knowing that the laser cannot track
the disc identically in each case.
We also did some bi-refringence
tests on the discs before and after
treatment. These tests, performed
with red light from a helium-neon
laser, show the stresses and strains in
a compact disc. Since this test is certainly affected by laser light reflections inside the plastic of the disc, we
thought that it might just show the
effect of the green dye. But you
guessed it, there was no measurable
effect.
Conclusions
So what are the conclusions? You
might say that the green pen treat-
ment is a load of old rubbish and
you'd be correct but let's take it a bit
further than that.
I started by accepting the theory
that internal reflections of laser light
might distort the signal coming off
the disc. Certainly, a badly scratched
disc gave a lot more hash on the signal than that from a clean disc. But
digital systems are designed to be
largely immune to superimposed
noise and so should not be affected in
the subsequent decoding.
Even so, I still went along with the
theory, considering the possibility
that, at times, noise on the digital
signal from the disc might just cause
more errors. I did all the comparison
tests and could find no significant
difference in the block errors and no
difference at all in the number of interpolations. As stated above, block
errors are all corrected so even if there
had been a difference in their number, there still would have been no
difference at all in the final sound
quality.
There was no difference in the
number of interpolations, so the
sound quality could not be improved
DECEMBER1990
19
due to this source either. It really is a
load of garbage.
What about listening tests?
Some hifi writers have stated that
they can clearly hear the difference of
11..0CK ERROR HISTOGRAM
70-
rn10-
the green pen. Now we regard this as
straight out self-delusion. How do you
do a fair test? You can't just listen to a
disc, treat it with a pen and then listen to it again. It takes a couple of
minutes to apply the green pen and
16/10/90
ID: BEACH BOYS
LINE: "BAD"
STAMP:BEFORE
RUN: IRK ONE
NI T_HOLD: 1
BURST T_HOLD: 7
NI ERROR:0
BURST ERROR:0
AUG BLER: 48
MAX BLER: 54
30-
*iHHHtPASS*****
i0-
10-
Minutes:
BLOCK ERROR RATE (1 0 sec. ave~ages>
BLOCK ERROR HISTOGRA M
16 / 10/90
I D: BEACH BOYS
LINE: "BAD "
STAMP:AFTER
RUN: IRK. 1
70-
'9--
50--
10-
rn--
NI T_HOLD: 1
BURST T_HOLD: 7
NI ERROR:0
BURST ERROR:0
AUG BLER: 48
MAX BLER: 56
*****PASS*****
20--
10-
Minutes:
I
160
I
7 ~1
then another five minutes to let it dry.
And you have to be absolutely sure
that you have not put any fingerprints
or dust on the disc in the process because that will definitely change
the number of errors.
The time difference between your
two listening tests is likely to be
around 10 minutes. With that sort of
time gap, how can you be sure of
anything?
Nor can you directly compare two
identical discs, one treated and one
untreated, in two identical players.
For a start, no two players are absolutely identical and neither are two
CDs, even if they are from the same
batch. And as we have already said,
the same disc is never played through
in exactly the same way by the laser,
even though the resultant sound quality is the same.
Are you getting the picture? Not
only does this process not work but
the people who are promoting it don't
have the evidence to support their
claims. If the process really did work,
it would be supported by experimental evidence from some official testing organisation.
These pens have no such supporting data, just a lot of uncorroborated
statements. Our advice to people
wanting the best sound quality from
their CDs is this:
(1) Buy a good quality CD player
with a well known brand. Avoid the
cheapies - they're an unknown quantity.
(2) Take great care with your CDs.
Always keep them in their cases when
not in use. When handling them, take
care not to touch the playing surface the less fingerprints and dust you get
on them, the less you have to clean
them.
(3) Keep the discs as clean as possible. Fingerprints cause interpolation
errors - they are worse then scratches
in this respect.
(4) Take even greater care not to
damage the label in any way. Never
write on it with any sort of pen or
pencil. The screen printed label and
the surface immediately under it is
the only protection for the metallisation layer.
BLOCK ERROR Rfi YE ( 10 sec . a ve~ages )
These two block error histograms show the first track of a <:ompact disc
before (top) and after treatment with a CD green dye pen. In this case, the
histogram after treatment is slightly worse but this can easily be caused by
a fingerprint or a slightly different tracking error in the mechanism.
20
SILICON CHIP
Acknowledgement
Our thanks to the staff of Disctronics Ltd for their assistance in the preparation of this article and specifically to Alan Bremner and Bill Dines.
|