Audio CD has been popular for thirty years. Among them, the playback
equipment has undergone many technical updates, and the sound quality has also been greatly improved. But so far, the phenomenon
that audio CD sounds are less natural than LP records seems to be common.
On the other hand, different levels of players
have different sound quality. Even if the same CD is played twice before and after the same player, there may be two different
sound effects. For digital equipments with strong consistency, such a situation is a bit weird.
The above problems
all originated from the jitter* of the data on the CD. These jitter interfere the player's Servo system, causing Read data (data stream) with a large amount of jitter. The data therefore has errors that ultimately affect
the sound quality of the CD playback.
* The physical structure deviation occurs in the pits arranged on the CD. When
the data represented by these pits is read by the player, it is reflected as the jitter of the electrical signal in the horizontal
axis (time) direction.
Below, we will analyze the problems encountered in CD playback from a technical level. Use an
objective perspective to understand and use audio CDs and playback devices more effectively. For other digital music formats,
if the data is still using a plastic disk as a storage carrier, the mechanism for generating data errors is the same.
CD requires multiple processes. Some production processes can cause some structural deviations in the product. These deviations
cause jitter in the data read by the player and affect the correctness of the EFM decoded data. When the error rate of the
decoded data reaches a certain level and exceeds the error correction capability of the player and cannot be completely corrected;
or the error data is some verification data, it will affect the sound quality of the CD playback. The degree of deviation
depends on the production process and the production equipment. Therefore, CDs of the same brand and different origins may
have different sound qualities. In the video below, you can see the difference in data stream jitter of some of the CDs we
use every day.CD Jitter Test Part 1 、 CD Jitter Test Part 2.
CD disk production, in the process of extrusion, because the cooling time is short, the physical properties of the plastic will inevitably
cause the edge of the pit to have a slope (the slope is random, resulting in physical structure deviation) . This sloping
edge becomes a gray zone between 0 or 1. The length of the cooling time changes the width of the gray zone and controls the
degree of data jitter.
The data on the CD is not returning to zero (NRZ) Serial format. To accurately transmit serial data, at low speeds, the transmitted data is required to have a short
synchronization period (eg RS232). A longer synchronization period (with more digits in the synchronization period) will result in more data being corrupted
due to data stream jitter. Therefore, serial data of a long synchronization period (such as I²C) is transmitted at a high speed. Another data transmission line must be used to synchronously transmit the bit clock signal.
The player transmits CD data at a rate of 4.22 MBit per second (4.3218 MHz). Because the player's CD disk data transfer
mechanism can only be a single data line. Therefore, in the case of high speed, long synchronization period (588 digits),
the player still needs to generate a bit clock signal at the data receiving end. There is jitter in the data stream, and the
bit clock must also follow its jitter. This following jitter is very complicated, and it is easy to cause the bit clock to
be misaligned in the time domain, which causes the digit to be in error during the state discrimination.
quality problem of audio CD playback lies in this bit clock used to distinguish the digital state of the data stream.
After the data on the CD is transferred to the player, the bit clock must be used to identify
the status of each digit (0 or 1). The player's bit clock is generated by a voltage controlled oscillator of the phase locked
loop. The frequency of the bit clock needs to undergo double offset correction: first, the bit clock frequency is corrected
by the sync signal of the data stream, and the correction information of the data read speed is provided to the servo
system; secondly, The state transition edge of the data stream provides information to correct the phase offset of
the bit clock. The jitter of the data stream will affect the above corrections.
is a feedback process and requires a response time. After combining the time domain factors, the feedback process is to extract
information from past facts, supply subsequent events as a comparison and repair gaps. The feedback information is taken from
the difference between one object and another reference object.
- This information is valid when the subsequent
object is in the same direction of deviation as the previous object;
- For some irregularly biased objects, this information
is not only invalid, but also incorrectly corrects subsequent accurate objects due to the information provided by the previous
erroneous object, and vice versa;
- When the subsequent object is opposite of deviation direction to the previous object,
this information will produce a maximum magnitude and erroneously modify the subsequent object.
structure deviation of the CD disk causes irregular jitter in the data stream, so the time domain deviation between the digital
bit and the bit clock occurs in the second and third cases. If the modified bit clock, its time-domain deviation from the
digit is still within the allowable range, this modification will not cause data errors. When the irregular jitter increases
to a certain extent and exceeds the range allowed by the time domain deviation, a data error will occur.
The CD player
uses a servo motor mechanism to transmit data on the disc. During this time, the server needs to correct the deviation of
the transmission speed. Since the process of changing the speed of the motor requires a response time, the data reading speed
of the player will therefore have a periodic variation of the lower frequency (relative to the frequency of the synchronizing
signal). That is, the data stream will have a regular periodic jitter.
bit clock signal generated by the phase-locked loop voltage controlled oscillator itself has a large amount of jitter.
is jitter in the data stream, and the bit clock must also follow its jitter.
The bit clock's own jitter (randomness),
the periodic jitter of the data stream (low frequency), and the digital irregular jitter (high frequency), the multi-level
intermodulation produced by them increases the deviation magnitude of time domain between the bit clock and the digits. and
cause the bit clock to follow the data stream jitter more complex. The data error rate will be further increased accordingly.
When the jitter of the data stream increases to a certain extent, in addition to more data
to be wrong, normal data transmission may also be destroyed. And the phase-locked loop may also exceed its lockable range.
Therefore, the servo system needs to set a threshold for the jitter of the data stream. If the amount of jitter reaches this
threshold, the servo system will use another control mechanism to make an intervention correction to maintain the uninterrupted
transmission of data. At this point, the amount of erroneous data will be maximized. This threshold set by the servo system
is the tolerance of the player for data stream jitter.
In industrial production, if
the product is not allowed to be deviation, there is no finished product. Therefore, the player still needs to have some tolerance
for the data stream jitter that occurs during the process of reading data. The low-level player can set a wider tolerance
range (+/- 7T per synchronization period, T = clock period), allowing playback of CDs with a larger amount of jitter. But
during the CD playback, the data will have more errors.
Some high-level players have relatively better jitter-following
features that allow for more accurate tracking of data on the CD. Therefore, tolerance can be set to a lower jitter threshold
(+/- 3T), and CD playback can have a smaller number of errors. However, when the data stream has a large amount of jitter,
the balance of the deviation and correction in the servo control is also more likely to be destroyed. Low-level players
can play all CDs of different quality, but some high-level players can't. This data stream jitter tolerance is one of the
When the CD is playing, the player's phase-locked loop and servo control will
enter a biased and corrected cycle state. The sensitivity of picking deviations, the amount of correction applied, the response
time, etc., determine the accuracy with which the servo system tracks the data stream. Based on the accuracy of the servo
system of each level of the player are differs, the data error rate caused by the data jitter of the CD disk also differs.
decoding. The data recorded on the CD is double-encoded, which is cross-interlaced Solomon
code and 8 to 14 modulation code. Audio CD playback, the first is to perform EFM decoding. EFM decoding is a 14-bit to 8-bit
data conversion using a look-up table. During the decoding process, the EFM decoder has no error correction capability in
the range of 3T to 11T. Therefore, when the data stream is jittered and thus causes data errors, the EFM decoder also provides
erroneous data to the CIRC decoder. In this way, the following two situations will occur:
This error data is an audio data. This error may be fixed during the CIRC decoding process.
error data is a checksum data. The checksum data will be used to check if the audio data is wrong, and if so, correct it.
Checksum data errors will cause serious problems. CIRC decoding has a strong ability
to repair, but the checksum data must be correct. If there is an error in the checksum data, it will modify some of the audio
data that is already correct. We can evaluate the probability of this data error appearing in the second case from the data
format of the audio CD.
Among the audio CD formats, there are 33 data per frame. This includes 1 control display data,
24 audio data and 8 checksum data. If every data has the same chance of getting an error. Then, there is an error in one frame
of data, and the chance that the error belongs to the checksum data will be 24.2% (8/33).
In actual use, the chance
of checksum data error will be higher than other data. Because of the CD data arrangement, the checksum data is arranged in
the middle and the last position away from the synchronization signal. The corrected bit clock, the cumulative deviation amount
will increase with time. The time domain deviation between the bit clock and the digits increases with time. Therefore, the
chances of error in checksum data will be much higher than other data.
Data arrangement sequence for each frame of
|Syn 1byte ||Ctrl, disp 1byte ||right
12bytes ||Check 4bytes ||left 12bytes ||Check 4bytes|
Syn --- Synchronization
Ctrl, disp --- Control and display
right--- 12 bytes of right channel audio
Check --- Checksum
left --- 12
bytes of left channel audio
As long as the correct checksum data is obtained, CIRC
can output the correct audio data. If the EFM decoder fails to provide the correct checksum data, it will not effectively
use the strong repair capability of CIRC, and its high reliability will cause the decoder to output some wrong audio data.
This situation seriously affects the sound quality of the CD.
Audio CDs use digital
methods to store music information. We know from the work of various digital devices that there is a high degree of accuracy,
and it is easy to think that CD players should also be like this. When we realized that the player used an electric motor
to drive the CD, it needed to read 4,321,800 digits per second in physical length of only 1.2m. It can be understood that
this is of course only in the ideal state, and can be easily destroyed by the structural quality of the data carrier and the
data transmission method.
At the beginning, the audio CD was released, and people expected
that LP would gradually wither away. But 30 years later, LP is still being taken care of by many music lovers, and seems to
have a re-emergence trend. LP sound can have today's quality and depends on decades of technological development. Compared
with the early days, the sound quality of today's audio CD has been greatly improved. However, if we want to further develop
the technical advantages of audio CD, we still need to continue to work hard.
Created by Chen
Last revision date: Aug-2019