Chapter 2 Physical Format of Read-Only Discs
| 2.1 Design Concept of the Physical Specification |
2.1.1 DVD design target
The basic design of the DVD began with the goal of media for movies as content. Therefore, a basic goal was for a playback time of about 133 minutes, which is long enough to allow most movies to fit on a single disc.
However, since DVD was intended as a technology to replace LaserDisc (LD), DVD needed to provide at least equivalent video quality. As the result of many rounds of video quality evaluation, and with the assumption that DVD would use variable-rate video compression, it was determined that a data rate of 3.5 Mbps was the minimum requirement. Then, considering audio quality, flexibility for international use, and multimedia capability, it was decided to provide capacity for Dolby AC-3 audio in three languages (384 kbps x 3) and subtitles in four languages (10 kbps x 4), resulting in the design of a specification which required a disc capacity of 4.7 GB.
The difference between the DVD specification and the CD specification is not just the move from a near-infrared laser to a red laser; the difference is that the entire specification is designed to achieve a disc capacity of 4.7 GB, based on the evolution in technology in the ten-plus years since the CD was introduced in 1982. For instance, requirements for parameters such as disc eccentricity (radial run-out) and tilt have become considerably more strict than in the CD specification. This recognizes evolution in disc manufacturing technology, as well as the fact that the recording density has increase proportionally more than the laser wavelength decreased, reducing total system margin.
For example, the standard CD track pitch is 1.6 microns. Reducing this by the ratio of DVD to CD laser wavelength (650/780) would result in a 1.33 micron track pitch. However, DVD actually requires a track pitch of 0.74 microns, meaning that tracks are considerably more packed than one might expect. As the track pitch decreases, crosstalk increases, and the radial tilt margin is severely reduced. In order to achieve the required density the average track pitch variation was tightened to 
0.01 microns. To reduce crosstalk the maximum allowed variation was also tightened to 0.03 microns.
To satisfy this specification, of course, it is necessary for the disc mastering equipment to be sufficiently precise, and the variability in playback device mechanisms and pickups must also be more tightly controlled than in CD players.
The specification known as DVD Book Part 1 describes the physical characteristics of a ROM disc needed to achieve such a system design. That is, the specification describes such things as the required disc mechanical properties, optical properties, and properties of the signal generated upon playback, as well as things like the modulation methods and error correction required to design DVD hardware. The DVD specification also allows dual layer discs and small (
8 cm) discs; these definitions are also contained within this specification.
Note:
This information is based on DVD Specifications for Read-Only Disc: Part 1 PHYSICAL SPECIFICATIONS, version 1.03. Disc and equipment designers should refer to the most recent versions of the relevant specifications. Pioneer makes no warranty concerning the accuracy of information presented in this article, and is not liable for any damages suffered as a result of any inaccuracies contained herein.
Simply put, the difference between a 4.7 GB DVD and a CD is in the recording density. Below are electron microscope photographs of features called "pits" recorded on CD and DVD media.
From these photos, it is easy to see how much smaller the DVD pits are. The problem is how to record and play back the information from these tiny pits.
CD and LD cutting is done with a mastering machine called a Laser Beam Recorder (LBR). These machines use light sources consisting of argon lasers with 457 nm wavelengths, or helium-cadmium lasers with wavelengths of 442 nm. To increase the recording density for DVD discs, LBRs have been developed which use argon or krypton lasers with near-ultraviolet wavelengths of 351 nm.
For playback, DVDs require a more tightly concentrated light than CDs do. The playback beam size is proportional to 
/NA, so DVD players need the numerical aperture (NA) to be large and the wavelength () to be small in order to realize the tiny playback beam necessary. However, a problem arises when the disc is inclined, or tilted, with respect to the light source. A tilted disc surface cause an optical degradation known as coma aberration, which causes the light spot to be distorted and interferes with correct playback. The degree of coma aberration is proportional to d x NA3/, where d is the disc substrate thickness. DVDs use a substrate that is 0.6 mm thick, compared to 1.2 mm used in CDs, thus reducing the effect of this degradation due to disc tilt by a factor of 2.
The figure above shows the relationship of disc tilt to coma aberration.
A 0.6 mm substrate is not physically strong enough, and so it is necessary to bond two 0.6 mm substrates together to form a 1.2 mm substrate. This adds a bonding step in addition to the steps that are required to form the 1.2 mm substrate used in CD media. This extra step tends to increase substrate cost. However, since the substrate is thinner, the cooling time needed in injection molding is shorter. This reduces the cycle time required between injecting the disc material and removing the formed disc. Since a large percentage of the disc cost is the amortization of the manufacturing equipment, the shorter cycle time makes up for the addition of the bonding step, allowing a DVD to be produced for about the same cost as a high-density 1.2 mm substrate.
Further, the DVD specification provides for dual layer discs, which are extremely well-suited to this bonding process, which in turn leads to additional added value.
The next issue that arises is compatibility with CDs. CDs use 1.2 mm substrates while DVDs use 0.6 mm substrates, and the pickup must account for the difference in spherical aberration resulting from this difference in thickness. This can be accomplished using devices which have already been announced, such as dual-focus pickups, dual-lens pickups, or pickups which use liquid crystal devices to provide variable apertures.
Finally, 0.6 mm substrates are at a disadvantage when it comes to surface dirt and damage. Since the pickup beam diameter at the disc surface is only one-half that of the diameter for the 1.2 mm substrate, the DVD pickup is twice as sensitive to surface dirt and damage. This disadvantage is compensated for by using powerful error correction schemes. The CD format provides correction for error bursts up to 2.29 mm long, while the DVD format can correct for error bursts as long as 6.0 mm & more than twice as long. And when it comes to scratches, the CD information layer is covered only by protective lacquer and the printed surface, making the information layer quite vulnerable to scratches on the label side. The DVD, on the other hand, is actually composed of two bonded substrates. Since the information layer is protected by a full 0.6 mm substrate, it is much less vulnerable to label-side scratches than a CD.
The DVD design target is that when the worst-case disc allowed by the specification, considering the economics of production, is played using the worst-case pickup that can be produced in volume economically, the byte error rate after error correction will still be 1 x 10--20, which is good enough to be acceptable for computer applications.
Since the above target is for "after error correction," the error correction capability must be calculated. Considering the tradeoff between error correction capability and the overhead of the added redundancy, the DVD format was set to one ECC block per 32 kB. This requires a byte error rate before correction of 1 x 10-2.
In order to achieve good economy on both the part of the discs and the playback mechanisms, and many experiments were performed. The current disc tilt specification was determined as a result of the efforts on both sides.
As will be explained hereafter, it is difficult to make the error rate a specification of the disc itself. Therefore, a jitter standard is set by the DVD specifications. A simple calculation based on a normal distribution requires that the jitter rate be under 15.4%, and experimental results indicate that jitter must be under 16%, to achieve the required error rate. Since the disc tilt varies within a revolution, it was decided to adopt the design concept that jitter must remain within 16% at the instantaneous peak value of tilt. Since it is actually very difficult to measure the peak value, the concept became to measure the average jitter at under 15%, and the byte error rate at under 5 x 10-3.
The basic concept of system margin is shown in the figure below. In the figure the horizontal axis indicates tilt, while the vertical axis represents jitter.
First, let's consider the best results obtained from many experiments. If there is no tilt, then the jitter value includes components from light source noise, circuit noise, disc noise, standard interference between symbols (inter -symbol interference), and some small amount of crosstalk from the neighboring tracks.
Next, let's find the minimum jitter level due to disc manufacturing variations other than tilt. Experimental results indicate that 8% is a reasonable value, based on the DVD specification.
Next we consider manufacturing variation in the circuitry.
Variation due to the disc and the circuitry have noise-like characteristics, and increase the minimum jitter level, but are thought to have a very small effect on tilt margin. Factors such as offset in the servo circuit, however, both increase the jitter level and decrease tilt margin. The figure shows the components of the reduction in margin, based on experimental results and past volume manufacturing experience. The remaining components are allocated to disc tilt and pickup tilt (including aberrations), resulting in the current specification.
The pits on CDs are typically about /6 deep. This enables the use of both a push-pull tracking error scheme, which works best at a pit depth of /8, and three-beam and differential phase tracking methods, which work best at a pit depth of /4. In the DVD specification, however, the top priority was to increase recording density, and so the specification was developed assuming the /4 pit depth necessary to obtain optimal signal quality. A push-pull signal is too small to be usable at a /4 pit depth, but the push-pull scheme has problems with offset due to lens shift and disc tilt anyway, so it was decided that support for that scheme was not important.
The differential phase tracking method, however, met all the requirements for compatibility, including handling dual layer discs and different track pitches. Further, this method was felt to be well suited to future recording density increases, and thus it was chosen as the standard tracking method for DVD. The differential phase tracking method does have the weakness of producing an offset in regions where there is strong correlation in bit pattern with the neighboringtracks , but this can be avoided by scramblingthe signal.
Dual layer discs
The DVD specification provides for dual layer discs, in a such a manner that either layer can be played without the need to turn over the disc. This gives rise to problems such as increased spherical aberration due to different substrate thickness when reproducing a signal from the different layers, and a decrease in signal-to-noise ratio due to reflected light from the surface not being played (inter-layer crosstalk). Track density is reduced by about 10% in dual layer DVD discs as a means of increasing margin.
The two layers need to be far enough apart that inter-layer crosstalk is small for standard pickups, but close enough that spherical aberration doesn't become fatal. With this in mind experiments were performed, and the inter-layer distance was specified to be 55 15 microns. The substrate thickness specification for dual layer discs was able to be made thin because the thinner the substrate, the easier it is to control coma aberration, thus giving plenty of margin.
| 2.2 Features of the DVD Physical Specification |
2.2.1 Standard evaluation specifications
The table below shows a comparison of the basic specifications used in evaluating specification compliance for CD and DVD discs. As shown in the table, the NA value is considerably larger for DVD. Since these are standard evaluation specifications, the limits on variability are quite tight. Further, there are specifications for items which were not specified for CDs, such as servo characteristics and transfer characteristics of playback system elements like laser diodes and waveform equalizers. Since the recording density is higher in DVD, the shortest mark is shifted toward the high range of the MTF low-pass optical system transfer function. The DVD has an MTF of 68%, compared with 50% for a CD system. As a result, the system requires waveform equalization for demodulation. Because the waveform equalization function changes the error rate and jitter value, the playback system transfer characteristics are set, including the waveform equalization function of a standard test device. As a result, it is possible to indirectlyspecify the performance of the LBR master recording device. The DVD specification also defines standard servo characteristics. These are closely related to the disc mechanical characteristics, which will be described later. For CDs, disc surface deviation and radial deviation are specified as accelerations. However, it is difficult to measure acceleration and produce repeatable measurement values. This issue was investigated during the work on the ISO 3.5" disc. The result was to specify standard servo characteristics and to specify disc mechanical properties with the residual error after servo compression. The key feature of this method is the ability to generate repeatability of data. The DVD specification uses the same method. In specifying servo characteristics, open-loop specifications were avoided, as they are subject to wide variation; instead, closed-loop parameters were specified. Please refer to the DVD specification for playback system transfer characteristics and servo characteristics.
Further, the detector size is specified for measurement of dual layer discs. This is done with the intent of limiting the amount of inter-layer crosstalk. Care must be taken, as use of a larger detector will introduce obstacles to other measurements, particularly to reflectivity measurements, due to inter-layer crosstalk. As a side note, detectors usually utilize PIN photo diodes, and electrons excited by light entering from other than the detector portion can result in a considerable DC offset. Therefore, during measurement the detector should be shielded from stray light, or grounded to bleed off the stray electrons.
| Comparison of standard evaluation specifications | ||
|---|---|---|
| DVD | CD | |
| wavelength | 6505 nm |
78010 nm |
| NA | 0.600.01 |
0.450.01 |
| polarization | circular | - |
| light intensity at the rim (of the pupil of the abjective lens) |
RAD:60-70% TAN:90% or more |
50% or less |
| surface aberration | 0.033 RMS or less |
0.07 or less |
| laser diode noise | -134 dB/Hz or less | - |
| measuring scanning velosity | 3.49 0.03 m/s (single layer) 3.84 0.03 m/s (dual layer) |
- |
| disc clamping force | 2.00.5 N |
1-2N |
| circuit characteristics | standard servo, equalizer, PLL, CLV servo, slicer characteristics |
- |
2.2.2 Jitter specification
The DVD specification includes a standard for jitter, which is not found in the CD specification. The logical format specifies disc mechanical and optical characteristics, but specifications for sources of degradation, like inter-symbol interference introduced in the disc cutting step, are not included in the current CD specification. There are also items which cannot be expressed by parameters listed in the current CD specification, such as degradation introduced by the mastering machine or unevenness in pit replication.
These effects can't be ignored in DVDs with their higher recording density, and thus it is necessary to specify such factors. There was an error rate specification in the CD specification, but this is impossible to measure unless there are defects or degradation due to the playback device.
The disc specification for Pioneer's 
Karaoke System specified an error rate using a tilted pickup, but this was a difficult measurement to make, and certainly not efficient. In the DVD specification it was decided to add a jitter specification, as jitter is a parameter where degradation can be measured numerically. Jitter is measured in the absence of tilt, which is an ideal disc specification, but it isn't practical to compensate for tilt at all points. Since the effect of jitter due to the varying component of surface tilt is small, it was decided to measure across one full revolution and measure the average value of tilt variation. As a result, it is only necessary to compensate for the average radial tilt when taking measurements.
2.2.3 Tilt specification
The disc tilt limit is 0.8
in the radial direction, and 0.3in the tangential direction. The specification for the radial direction is larger in consideration of the fact that it's easy for the disc to curve into a bowl shape.
Note that the tilt angle defined by the specification is not just the physical angle of inclination, but rather the angle between incident and reflected light (), measured optically.
Of course, the specification defines characteristics of discs when shipped from the factory; but it also requires guaranteed disc characteristics after being subject to conditions in the marketplace. However, there are a wide variety of environmental conditions in the marketplace, making that very difficult to specify. Therefore, an Informative Annex to the DVD specification describes the minimum environmental tests. Care must be taken, for example, not to use adhesives that will degrade at the specified high temperatures.
As a side note, there have been reports of degradation due to disc tilt resulting from the use of improper cases and packing.
2.2.4 Reflectivity specification
For CDs, the reflectivity specification is for a disc with a reflective surface only, with no information recorded on the disc; practically speaking, this is very difficult to test. The DVD specification takes into account the player design, and specifies reflectivity in terms of the maximum playback signal level I14HThis is very easy to measure. In this case there are effects from disc birefringence, so the specification defines values for both polarizing and non-polarizing optical systems.
There are actually two sets of reflectivity specifications, as reflectivity differs between single layer and dual layer discs. For the polarizing optical system, the specified values are 45-85% for single layer discs, and 18-30% for dual layer discs. However, note that there are specifications for I14Hvariations across the surface and around a revolution, for the purpose of limiting variation in servo gain.
Further note that the minimum I14Hreflectivity value for DVD-RAM discs is about 10%. For more detailed information, please refer to the DVD-RAM specification. Ordinarily, the detection of whether or not a disc is loaded in a tray is done by attempting to focus on the disc. This means that the focussing mechanism must handle discs with a reflectivity of only 10%.
2.2.5 Tracking specification
The DVD specification adds an item for a crosstalk signal which expresses the contrast when cutting across a track, and which is used in pulling in of tracking and disc access. Since the track pitch is narrow and crosstalk noise is high, the crosstalk signal is specified to be measured after running through a 30 kHz low-pass filter.
2.2.6 Other disc parameter specifications
The table below shows the difference in some other parameters between the CD and DVD specifications. When playing video from a DVD, the disc spins at a rate which provides a linear speed of 3.49 m/s. (The linear speed of a CD is 1.2 to 1.4 m/s.) If disc warpage and eccentricity is the same for a DVD disc as for a CD, it would require a high bandwidth actuator to allow the pickup to follow the disc movement. This would raise the further complication of increased heat generation. This must be avoided to enable portable applications, and so the DVD specifications for disc eccentricity and surface deviation were made more strict than for CDs. Note that for dual layer discs the layer closest to the pickup is aligned during the clamping process. Therefore, the maximum eccentricity value is large, considering the de-center that can occur during bonding with the other bonded layer.
| DVD | CD | |
|---|---|---|
| track pitch | 0.74 0.01  m (average)0.74 0.03 m (instantaneous) |
1.60.1 m |
| eccentricity (radial runout) |
100m peak to peak |
140m peak to peak |
| warpage (surface deviation) |
0.3mm |
0.5mm |
Since the DVD medium is a bonded disc, there must be a specification for the maximum allowable de-center in the bond. Since the specification includes variation in mass, a specification item for dynamic balance was created.
Further, so that the back side of the tapered cone used during clamping doesn't contact the inside diameter of the disc, and so that no problems occur if a DVD is accidentally loaded into a CD player, the inside diameter is specified to not be less than 15.0 mm when viewed through both surfaces.
To maintain bonding strength, the maximum value of the depression on the inside of the clamping area was changed to 0.1 mm, from the CD value of 0.2 mm. And, considering LD and DVD compatible players, the maximum value of the stack ring thickness variability area protrusion on the outside of the clamping area was changed to 0.25 mm, from the CD value of 0.4 mm.
The program start radius of a CD is 25 mm. This was reduced to 24 mm for DVD to increase the disc capacity. The lead-in start radius is 22.6 mm. The maximum radius of the program region is 58 mm, followed by a lead-out of at least 0.5 mm width. Further, in order to insure that there is some region with program recorded, the minimum value of the outer radius of the information region is required to be 35 mm.
| 2.3 The DVD Data Format |
2.3.1 ID, IED, and EDC
The figure below shows the process used in encoding.
First, two bytes of error correction code are added to a four-byte ID. This is done to enable fast access by making it easy to read the ID using only the ID's error correction code, without having to calculate and check the error correction code which is later added to the entire data block. To this ID is added six bytes of control data, 2048 bytes of main data, and a four-byte EDC code. This EDC code is used for checks such as determining whether scrambling has been performed correctly, and whether error correction has occurred after the error correction code is calculated and checked.
2.3.2 Scrambling
After the EDC is appended, the data is scrambled. This is done to randomize the data and prevent problems like interference from a repeating pattern in the neighboring track, or a repeating pattern in the data resulting in a large DC component which affects the data slicing or servo. Note that the data used in the scrambling process will not have any fixed pattern, but will be comprised of 16 values, based on four bits in the ID, in a manner chosen to also be effective in recording. The initial value is taken from the four bits beginning with the fifth bit from the end of the ID, to provide the same scrambling to 16 sectors of data. Therefore, the scramble pattern makes one complete cycle in 16 x 16 = 256 sectors. This scrambling is done for the purpose of randomizing the data, which is particularly important for differential phase tracking. In differential phase tracking, a proper error signal cannot be generated if the pit arrangement in the adjacent track is in some particular pattern. At the inner circumference of the disc there are about 29 sectors around one complete revolution. Since the same scrambling continues for 16 sectors, the necessary condition has been met at the inside circumference. At the outer circumference there are about 70 sectors in one complete revolution. This is less than a full cycle of 256 sectors, so again the necessary condition has been met. The conditions will still be met a blue laser is used to increase recording density by a factor of 1.5. ECC encoding is done on the scrambled 16 sectors, using product codes.
2.3.3 ECC (Error Correction Code) block and interleaving
The figure above shows the data in block structure after ECC has been added, with 10 bytes of Reed-Solomon check code (182, 172, 11) added to each row of 172 bytes, and 16 bytes bits of Reed-Solomon check code (208, 192, 17) added to each column of 192 bytes .
After the ECC has been added, each of the bottom 16 rows is interleaved with the data so that there are 12 rows of data followed by one row of parity check code, as shown in the figure at right. This block of 13 rows of 182 bytes each comprises one recording frame, before the addition of modulation and synchronization signals.
2.3.4 Sync Code
Each row of a recording frame is divided into two equal parts, and a 32-bit synchronization (SYNC) Code is added to each group of 91 bytes(91 x 16 channel bits= 1456 bits). The pattern of this SYNC Code field is
AAA*******0001000000000000010001
The latter part of this field is a combination of 14T and 4T. The Tmax in the data is 11T, so adding 3T to make a pattern of 14T in the SYNC field means that even if 11T becomes 12T due to an edge shift, and if 14T becomes 13T due to an edge shift, it will still be possible to distinguish them. After the 14T comes a fixed 4T, and with the previous having a gap of at least 4T, it is prevented from having symbol interference with the 14T. The AAA portion is chosen to be either 000, 001, or 100, depending on the relationship with the previous word (defined by the condition and the (d, k) limitation). The seven bits indicated by ******* are used in combination with the three AAA bits to form one of 32 different patterns, and assigned one of two different codes with different edge transition numbers for eight types of SYNC Codes ranging from SY0 through SY7. Utilizing the two codes with different edge transition numbers for each SYNC Code allows look-ahead DC control. (DC control is done by choosing the the one of the two types of SYNC Code which will result in a smaller DSVuntil the point where DC control is next performed.)
The figure at right shows the combination of the SYNC Codes for the 13 rows in a sector. Each sector begins with SY0, and each row is uniquely identifiable by the pattern of cyclically repeating SY1 through SY4 and SY5 through SY7 codes. Error correction codes are generated over 16 sectors. The ID information following the SY0 at the beginning of the block is read and recognized as an address which is divisible by 16. SY0, that is to say, the beginning of the sector, plays an important role in decoding the data.
Since the individual rows are uniquely identifiable in the sector structure, several rows can be read and the location of a coming SY0 can be calculated from the periodicity of the rows' SYNC Codes. This makes it possible to interpolate and read the next ID, even if for some reason the SY0 code is unreadable. Key features of this sector block structure are the large error correction code block and the ability to determine the sector head even if the actual pattern is unreadable.
The SYNC Codes were defined to realize in just 32 bits the features described above, namely a 14T length SYNC pattern that is 3T longer than the Tmax in the data region, DC control, and the identification of the sector start.
The synchronization frequency of a standard test unit is based on the 27 MHz clock of the video system, and is divided down by (512 x 3) to 17.578125 kHz. Since the channel clock is one SYNC frame interval, or (91 + 2) x 16 = 1488 bits, the channel clock frequency becomes 27 MHz x 1488 / 512 / 3 = 26.15625 MHz.
The ID information added to each sector is comprised of four bytes. The lower-order three bytes contain the sector number. The upper byte provides a bitmap of information which the drive requires in real time, namely the sector format (ROM or RAM), tracking method (pit tracking or groove tracking), reflectivity (greater or less than 40%), disc region (lead-in, lead-out, or middle data), and layer information (layer 0, layer 1, other). Note that the layer information is contained in the lowest-order bits of the byte, putting it in a position to be considered as the upper bits of the sector number.
2.3.5 Lead-in, middle, and lead-out regions
The CD's table of contents information is contained in the lead-in area, and contains a table of information used to access the rest of the disc. The DVD specification, however, adopts the concept that all information about the data content is contained within a file system, and exists within the data itself. Therefore, the information written in the lead-in (middle region) is information necessary only for the drive itself, such as disc compatibility and drive control information. (There are also regions for manufacturer information or related to copying.)
This information is called control data, and is written in the 17.5 to 105 tracks located inside the data region start radius of 24 mm. One block of the control data is an ECC block (16 sectors), and is written across 192 blocks, or in other words, is repeated 192 times. The first sector of the control data contains physical format information. The physical format information describes what type of disc it is, conforming to what revision of the specification. It also describes the disc size, the maximum transfer rate with consideration for portable players, number of layers, track path type, whether the disc is all ROM or partial ROM, recording linear density, track density, and start and end sector numbers.
Starting 16 tracks inside of the control tracks and covering two blocks of length is recorded a reference code used for equalizer calibration. Inside of this, ROM discs also contain a region of all-zero data extending inward to a radius of 22.6 mm.
Chapter 2 Physical Format of Read-Only Discs