Information
-
Patent Grant
-
5351231
-
Patent Number
5,351,231
-
Date Filed
Monday, April 13, 199232 years ago
-
Date Issued
Tuesday, September 27, 199429 years ago
-
Inventors
-
Original Assignees
-
Examiners
Agents
- Blakely, Sokoloff, Taylor & Zafman
-
CPC
-
US Classifications
Field of Search
-
International Classifications
-
Abstract
Described in detail herein are the detection and windowing capabilities of the present invention which are critical for reading correct data in normal operations and recovering data during error conditions. Also described herein are the various programmable features of the detection windows of the invention. The flexibility of these features allows the designer to optimize the optical controller for the unique requirements of the particular optical drive and medium in use.
Description
BRIEF SUMMARY OF THE INVENTION
The preferred embodiment of the present invention is embodied in the CL-SM330 Optical Disk ENDEC/ECC, designed to work with the CL-SM331 SCSI Optical Disk Controller (both of which are manufactured and sold by Cirrus Logic, Inc. of Fremont, Calif.) to provide a SCSI (small computer systems interface) optical disk controller solution with minimal hardware. The function of the SM330 component in this optical controller chip set is to perform the encoding and decoding of the raw RLL encoded data written to and read from the disk, and to perform the error detection and correction functions.
One facet of the data decoding that the SM330 performs is to convert the encoded (2,7) RLL data pattern into the serial NRZ data which is transferred to the SM331, and subsequently transferred in bytes to the SCSI host. Before any decoded NRZ data is sent to the SM331 however, the correct sector must be detected, and the SM330 must be synchronized to the incoming data stream from the optical drive. Detection of the correct sector and maintaining synchronization throughout the sector are achieved with the detection and windowing capabilities of the SM330.
Described in detail herein are the detection and windowing capabilities of the SM330, which are critical for reading correct data in normal operations and recovering data during error conditions. Also described herein are the various programmable features of the SM330 detection windows. The flexibility of these features allows the designer to optimize the optical controller for the unique requirements of the particular optical drive and medium in use.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 illustrates the Pre-Formatted Area of a sector.
FIG. 2 shows the four possible cases that can occur with the timing of SMDET with respect to TSMDET.
FIG. 3 and Table 1 describe in detail the method and sizes for the special mark windows, FIG. 3 presenting a flowchart describing the process in which the windows are widened.
DETAILED DESCRIPTION OF THE INVENTION
As described in the Brief Summary of the Invention section, the preferred embodiment of the present invention is embodied in the CL-SM330 Optical Disk ENDEC/ECC, designed to work with the CL-SM331 SCSI Optical Disk Controller (both of which are manufactured and sold by Cirrus Logic, Inc. of Fremont, Calif.) to provide a SCSI (small computer systems interface) optical disk controller solution with minimal hardware. The function of the SM330 component in this optical controller chip set is to perform the encoding and decoding of the raw RLL encoded data written to and read from the disk, and to perform the error detection and correction functions.
One facet of the data decoding that the SM330 performs is to convert the encoded (2,7) RLL data pattern into the serial NRZ data which is transferred to the SM331, and subsequently transferred in bytes to the SCSI host. Before any decoded NRZ data is sent to the SM331 however, the correct sector must be detected, and the SM330 must be synchronized to the incoming data stream from the optical drive. Detection of the correct sector and maintaining synchronization throughout the sector are achieved with the detection and windowing capabilities of the SM330.
Described in detail herein are the detection and windowing capabilities of the SM330, which are critical for reading correct data in normal operations and recovering data during error conditions. Also described herein are the various programmable features of the SM330 detection windows. The flexibility of these features allows the designer to optimize the optical controller for the unique requirements of the particular optical drive and medium in use.
Special Mark Detection Requirements
The most common operations performed by an optical controller are to read or write a sector or group of sectors. Before reading or writing a particular sector, the correct sector ID must be found. The Sector Mark and ID fields are located in the "Pre-Formatted Area" for rewritable, magneto-optic (MO) disks. Once the correct sector has been found, the SM330 must maintain bit synchronization with the sector data field (during a read operation), so that the data is read exactly as it was written. Thus the requirements of the data detection function are to 1) correctly detect the sector to be read, written or erased and 2) maintain bit synchronization while the sector data is being transferred.
Detection Windows
The concept of windows will be used throughout this document, so it is useful to explain the basic concept of windowing at this point. While reading this document, it would also be helpful to refer to Appendix D for the CL-SM330 data sheet, which provides the ANSI/ISO standard track formats for 90 mm and 130 mm media with sector sizes of 512 or 1024 bytes per sector. Also for reference, a set of circuit drawings for one embodiment of the invention is provided as Appendix B, the signal definitions therefor being provided in Appendix A. Appendix C provides the source code for the Window Sequence.
While reading or writing a sector of data, several different fields or events must be detected in the order they appear in the Pre-Formatted Area and data field. For example, the Address Mark detection must occur after the Sector Mark detection. To minimize the possibility of false detection of these fields, a windowing scheme is used. A detection window is a time frame in which the detection of a particular field is enabled. These detection windows (such as the Sector Mark window or Address Mark window) are internal to the SM330 and are based on the REF2F input clock, which is twice the NRZ data rate and equal to the RLL bit rate. The REF2F clock provides an ideal frequency at which the data is read from the disk. The actual data rate will deviate from the ideal frequency due to speed variations of the drive. Thus, the detection windows should be wide enough to account for speed variations of the drive motor and the sector size. Other factors also affect the detection windows for specific events, which will be discussed in later sections.
Sector Mark Detection
The Pre-Formatted Area of a sector is shown in FIG. 1. The first field in the Pre-Formatted Area is the Sector Mark Field. The Sector Mark consists of an embossed pattern of long marks and long spaces which does not occur in a normal data pattern. This unique pattern is designed to enable the drive to identify the start of a sector without the use of a phase-locked loop (PLL). The following SM330 registers are used to control Sector Mark detection:
______________________________________11h Configuration Register 224h Sector Mark Control Register28h Window Control Register2Ah Sector Mark/ALPC Length______________________________________
The first thing the SM330 will look for during a read or write operation is a valid Sector Mark. The SM330 detects a Sector Mark by examining the raw data from the disk on the RAWIN input (pin 94). After power-up or head repositioning, a Sector Mark Reset should be issued by writing to the Sector Mark Control Register (24h). A Sector Mark Reset will disable the Sector Mark window until a valid Sector Mark has been found with the Sector Mark detection circuitry continuously active. After a Sector Mark has been found, Sector Mark windowing is enabled. The Sector Mark windowing can also be completely disabled by setting bit 6 of the Window Control Register (28h), which may be useful in sector size determination or Search ID operations.
The size and position of the Sector Mark window is a function of the medium size (90 or 130 mm), the sector size (512 or 1024) and the speed tolerance of the drive motor (0.5 or 1.0%). These parameters are programmed in Configuration Register 2 (11h). In addition, the positions of the address mark and data sync mark windows with reference to the sector mark window are controlled by bits 7-6 of the Sector Mark Control Register (24h). These bits will indicate whether the windows should be positioned normally, delayed or advanced depending on the speed of the disk being normal, slow or fast respectively. More details about the Sector Mark windowing are provided in a later section.
Sector Mark Thresholds
Sector Mark detection is contingent on the Sector Mark-Mark Threshold and the Sector Mark-Space Threshold, which are programmed in bits 5-0 of the Sector Mark Control Register (24h). These thresholds determine the minimum number of long marks and long spaces that must be correctly detected in qualifying a valid Sector Mark. The mark threshold has a range from zero to five and the space threshold has a range from zero to four. It is recommended that mark and space thresholds of at least four and three respectively be used to insure that false detection does not occur. Lower values for the mark and space thresholds can be used for error recovery.
Additional Sector Mark Controls
The Sector Mark/ALPC Length Register (2Ah) provides additional control over the Sector Mark detection. When Reg. 2Ah, bit 7 is reset, the Sector Mark detection circuit will ignore the first mark bit or first space bit at the boundary of a space to mark or mark to space transition respectively. When bit 7 is set, the Sector Mark detection circuit will ignore the first two bits of the mark or space at each boundary.
Also, Sector Mark Asymmetry Compensation can be enabled (Reg. 2Ah, bit 6 is set). If this mode is enabled, then bit 5 will determine whether the mark lengths are lengthened or shortened by one-half of one REF2F clock period in the Sector Mark detection circuitry.
Pseudo Sector Marks and True Sector Marks
There are two signals related to the Sector Mark detection, SMDET (Sector Mark Detected) and TSMDET (True Sector Mark Detected). The SMDET signal is set when a true Sector Mark detected or when a Pseudo Sector Mark is generated, whichever occurs first. TSMDET is set when a true Sector Mark is detected.
A true Sector Mark is detected (and TSMDET is asserted) when both the Sector Mark-Mark and Sector Mark-Space thresholds have been satisfied within the Sector Mark window. A Pseudo Sector Mark is generated (SMDET is asserted) at the expected position of the Sector Mark detection within the Sector Mark window. FIG. 2 shows the four possible cases that can occur with the timing of SMDET with respect to TSMDET. For each of these four cases, the following assumptions were made:
(i) 512-byte sector size, 0.5% speed tolerance
(ii) The previous Sector Mark has been found.
(iii) The current Sector Mark is been searched for with a Sector Mark window that is eight NRZ clock bytes wide.
The nominal expected position of the Sector Mark is in the center of the Sector Mark window for each case.
CASE 1: The true Sector Mark is not found within the Sector Mark window. A Pseudo Sector Mark is generated and SMDET is asserted at the nominal position in the Sector Mark window. TSMDET is not asserted.
CASE 2: The true Sector Mark is found at the nominal (center) position in the Sector Mark window. A Pseudo Sector Mark is generated at the nominal position in the Sector Mark window and SMDET is asserted. Likewise, TSMDET is also asserted at this time since a true Sector Mark has been found. The exact position of SMDET will vary from the exact position of TSMDET by +/-2 REF2F clocks.
CASE 3: The true Sector Mark is found early in the Sector Mark window. Since the Sector Mark is found before the Pseudo Sector Mark can be generated at the nominal position, both SMDET and TSMDET will be asserted simultaneously. The timing difference between SMDET and TSMDET will vary by +/-2 REF2F clocks. The window timing logic will now adjust to this Sector Mark as the new nominal position for calculating subsequent window positions.
CASE 4: The true Sector Mark in found late in the Sector Mark window. A Pseudo Sector Mark will be generated at the nominal position in the Sector Mark window and SMDET will be asserted. When the true Sector Mark is found later in the window, TSMDET will be asserted. The window timing logic will now adjust to the true Sector Mark as the new nominal position for calculating subsequent window positions.
Another note about the Pseudo Sector Mark: After a Sector Mark Reset, the generation of a Pseudo Sector Mark (along with the Sector Mark windowing) is inhibited until a true Sector Mark has been detected. Also, an optional interrupt can be generated when a true Sector Mark is not detected and a Pseudo Sector Mark is generated. This interrupt is enabled by setting bit 7 of the Media Error Enable Register (21h). When this bit is set, the sector for which the Pseudo Sector Mark was generated will not be read or written and the operation will be aborted.
The SMDET signal is connected to the SMDET input (pin 18) of the CL-SM331. SMDET is used by the Format Sequencer in the SM331 as an indication to branch to the next sequence in the Writable Control Store (WCS). The TSMDET signal can be optionally used by the drive servo system.
Sector Mark Window and Read Gate Positions
After a Sector Mark is detected, the VFO1 field immediately follows. The Device Read Gate signal (DRG, pin 86) should be asserted at this time so that the PLL can achieve initial phase and frequency lock to the repetitive VFO1 pattern. In normal operating conditions DRG should not be asserted while the Sector Mark is being read in order to prevent the PLL from locking to the low frequency pattern of the Sector Mark.
The assertion of DRG with respect to the Sector Mark window is a function of the window position programmed in the Sector Mark Control Register (Reg. 24h, bits 7-6). These bits determine whether the windows are positioned nominally, delayed or advanced. The assertion of DRG with respect to these window positions will be described:
(i) Advanced Window Position (Reg. 24h, bits 7-6=10b). In this case, the disk is assumed to be spinning fast by either 0.5% or 1.0% according to the value programmed in Configuration Register 2. The detection of the Sector Mark is expected to be early in the Sector Mark window, and DRG will be asserted before the nominal position of the Sector Mark window. This timing mode should only be used in recovery operations.
(ii) Nominal Window Position (Reg. 24h, bits 7-6=00b). The operation of this mode is identical to that of the Advanced Window Position. This timing mode should only be used in recovery operations.
(iii) Delayed Window Position (Reg. 24h, bits 7-6=01b). The disk is assumed to be spinning slow by either 0.5% or 1.0% according to the value programmed in Configuration Register 2. After the Sector Mark is found (and TSMDET asserted) within the Sector Mark Window, DRG will be asserted, allowing the PLL to lock to the VFO1 data pattern. If the Sector Mark is not found, then DRG is asserted at the end of the Sector Mark window to ensure that the VFO1 field has been entered. The Delayed Window Position is the recommended operating mode for the SM330.
Address Mark Detection
In addition to the detection of the Sector Mark, the ID Address Marks of a sector must be detected to establish bit and byte synchronization for the ID fields. The detection of the Address Marks is also windowed to prevent false detection. The ID Address Mark is a special one-byte pattern containing a run-length violation of the (2,7) RLL code. Each ID Address Mark is preceded by a VFO sync field and is followed by the ID/CRC field (***see Appendix D of the SM330 data sheet). There are three AM/ID fields in the Pre-Formatted Area of each sector. This is done for redundancy in case one or two of the AM/ID fields is corrupted or cannot be read by the PLL.
The following SM330 registers are used for AM/ID field detection:
______________________________________24h Sector Mark Control Register26h ID Field/Data Sync Control Register28h Window Control Register______________________________________
The Sector Mark Control Register will adjust the position of the AM window (normal, delayed, or advanced) based on the assumed speed of the drive. This feature is also used for the Sector Mark and Data Sync window positions. Bits 5-4 in the ID Field/Data Sync Control Register (26h) determine the ID Voting Threshold. This threshold will determine how many error-free ID fields must be read for the operation to continue without reporting an ID error. It is also possible to disable the AM window altogether by setting bit 5 in the Window Control Register (28h). This will make AM detection active throughout the entire Pre-Formatted Area. The windowing for the AM field will be explained in a later section.
Prior to the beginning of the VFO1 field, the external PLL is locked to the reference clock (REF2F) frequency. At the beginning of the VFO1 field, the PLL will switch its input from the REF2F reference frequency to the actual disk data and lock to the bit pattern in the VFO field. It is possible that defects in any of the VFO fields could put the PLL out of lock and prevent the Address Marks from being detected. This problem could be solved by skipping the first ID field, the first two ID fields or all three of the ID fields by programming bits 1-0 in the Window Control Register (28h). These bits control the assertion of DRG, thus allowing the PLL to skip over a defective ID field.
ID/CRC Field Detection
Before starting a read or write operation, the Target Sector and Target Track registers (14-16h) are set for the first desired sector to be read or written. The ID Found signal will be asserted when the correct sector ID has been found. IDF is asserted at the end of the Pre-Formatted Area after the following conditions have been met:
(i) The minimum number of error-free ID fields have been read as specified by the ID Voting Threshold (Reg. 26h, bits 5-4). An error-free ID field is defined as an ID field in which a valid Address Mark and ID field with good CRC have been found.
(ii) The track and sector information read from each error-free ID field matches the Target Sector and Target Track registers.
The IDF signal is used by the Format Sequencer in the SM331 as a branch condition in the sequencer control store (WCS).
Data Sync Mark Detection
The Data Sync Mark allows the SM330 to synchronize to the sector data field. The detection of the Data Sync Mark is also windowed to minimize the chance for false detection. Thus, we have seen that the Sector Mark detection, Address Mark detection, and now the Data Sync Mark detection are critical elements required for proper synchronization to the disk data. All three of these special marks are windowed to prevent false detection. Whenever a Sector Mark, Address Mark or Data Sync Mark is detected, the window timing logic will adjust to this special mark as the new nominal position for calculating subsequent window positions.
The following SM330 registers are used for Data Sync Mark detection:
______________________________________26h ID Field/Data Sync Control Register28h Window Control Register______________________________________
The Data Sync Mark consists of a special three-byte pattern which indicates the beginning of the user data. The SM330 divides the RLL channel bit pattern into 12 four-bit groups and compares them to the Data Sync Mark pattern in order to detect a valid Data Sync Mark. The Data Sync Mark Threshold bits in the ID Field/Data Sync Control Register (Reg. 26h, bits 3-0) determine the minimum number of four-bit groups that must be correctly matched in order for the Data Sync Mark to be found. A typical value to use is nine. If the Data Sync Mark Threshold is set to zero, then the Skip Data Sync Mark Mode is activated. This mode will be fully discussed in the Resync Mark description. The next section will describe how the Sector Mark, ID Address Mark and Data Sync Mark detection windows are controlled by the SM330 windowing method.
Special Mark Windowing
A brief description of the Sector Mark, ID Address Mark and Data Sync Mark windows has already been given. The registers which control the window size and position have also been described (Regs. 11h, 24h and 28h). To summarize, the window size is a function of form factor (90 mm or 130 mm), sector size and speed tolerance (0.5% or 1.0%). The window position can be delayed, advanced, or left at its nominal position. Alternately, the window can be disabled altogether.
In addition to this initial sizing of the Sector Mark, Address Mark and Data Sync Mark windows, the SM330 has a complex special mark windowing scheme which will widen the Sector Mark, Address Mark or Data Sync Mark windows when the previous special mark field cannot be found. By widening these windows, the SM330 will account for the additional uncertainty in the occurrence of these special marks due to the cumulative effects of drive speed variation.
FIG. 3 and Table 1 describe in detail the method and sizes for the special mark windows. FIG. 3 is a flowchart describing the process in which the windows are widened. The amount that the windows are widened depends on the sector size (512 or 1K) and the speed tolerance (0.5% or 1.0%) parameters, which are initialized in Configuration Register 2 (11h). In FIG. 3, the following definitions and comments apply:
______________________________________SM Sector MarkSMW Sector Mark WindowPSM Pseudo Sector MarkAMW Address Mark WindowID ID/Address FieldDS Data Sync MarkDSW Data Sync Mark Window______________________________________
"ID found" in this flowchart means that at least one good Address Mark and ID with valid CRC was found.
A Pseudo Sector Mark generated will halt operation if Reg. 21h, bit 7 is set.
A valid ID field not found will halt operation if Reg. 21h, bit 5 is set.
A Data Sync Mark not found will halt operation if Reg. 21h, bit 3 is set. The Default, Group 1 and Group 2 window sizes are listed in Table 1.
In Table 1, the following definitions and comments apply:
E=Early window
N=Normal window
L=Late window
(Based on values programmed in Reg. 24h, bits 7-6)
Table 1 is a list of the actual window sizes in NRZ clock bytes that are used for each of the various blocks in the windowing flowchart. The window sizes are divided into four sections according to the selected form factor and sector size. In addition, each section has three size groups: Default window size, Group 1 window size, and Group 2 window size. As shown, the window sizes for the Sector Mark, Address Mark and Data Sync Mark windows also depend on the window position (early, nominal or late). The window position is programmed in bits 7-6 of the Sector Mark Control Register (Reg. 24h) as described earlier.
Example Using the SM330 Windowing Method
Due to the complexity of the windowing, it would be useful to run through an example using the windowing flowchart. Assume that we have a sector size of 512, a speed tolerance of 0.5%, and the "late" window position is selected (Reg. 24h, bits 7-6=01b). This would be a typical configuration for a 90 mm optical drive. A read operation is to be performed. Refer to FIG. 3 and Table 1 throughout this example.
1. The SM330 starts off by trying to find the Sector Mark for the first sector to be transferred. We will assume that the previous sector header has been found. This means that either the Sector Mark or at least one of the ID Fields with valid CRC for the previous sector was found. A detected ID Field implies that its corresponding Address Mark was also correctly detected. The window logic is now synchronized to the timing of the previous sector.
2. The Sector Mark window is set to its default size (eight NRZ bytes) , which is the same as the Group 1 window size. The Sector Mark default window size is relatively large in order to account for the error tolerance that has been accumulated over the previous sector data field. That is, due to the speed variation of the drive, the actual position of the Sector Mark detection will vary from the expected nominal position.
3. If the Sector Mark is found, then the window logic is resynchronized to this event, and the Address Mark window is set to its default size of one NRZ byte. The Address Mark window will go active for each of the three Address Marks in the Pre-Formatted Area. The default Address Mark window size is small because the SM330 is already synchronized from the Sector Mark, and minimal error uncertainty has been introduced since the Sector Mark detection.
4. If the Sector Mark is not found, then a Pseudo Sector Mark is generated, and the Address Mark window is increased to the Group 1 size (38 bytes in this example). This increased window size takes into account the error uncertainty from the previous sector data field. The increased Address Mark window will be active over all three Address Mark/ID Field pairs.
5. When an ID Field with valid CRC has been read, the Address Mark window will reset back to its default size of one NRZ byte, and the window logic will resynchronize its timing to this event. The ID Field does not have to be the correct sector ID that the SM330 is looking for. This means that the windowing logic is independent of the assertion of the IDF (ID found, pin 51) output.
6. If an ID Field with valid CRC cannot be found, this means that the Address Mark was not detected and the window logic is out of sync with the ID data. In this case, the Data Sync Mark window will be increased to its Group 1 window size (10 NRZ bytes). The window sizing logic is not affected by whether or not a valid Data Sync Mark is detected.
7. At this point, we have read through the first sector. If the Sector Mark or any ID Field with valid CRC has been found, then the Sector Mark window will be reset to its default size, and we start back at the top of the flowchart in FIG. 3. If the Sector Mark and none of the IDs have been found, the Sector Mark window will increase to its Group 2 size.
8. Let's assume that the previous Sector Mark and IDs were not found. Now with the Sector Mark window at its Group 2 size, the error uncertainty of two sectors is taken into account. As before, if the Sector Mark is found, then Address Mark window is reset to its default size and the window logic is resynchronized to the new Sector Mark timing.
9. If the Sector Mark is not found with its Group 2 size, then a Pseudo Sector Mark is generated, and the Address Mark window is increased to its Group 2 size (39 in this example). The Address Mark window will be active over all three AM/ID Fields.
10. If any ID Field with valid CRC is read, then the Address Mark window and Data Sync Mark window will be reset to their default sizes. If no valid IDs are found, then the Data Sync Mark window will be increased to its Group 2 size (18 NRZ bytes in this example).
11. At this point, two consecutive Sector Marks have been missed, and two Pseudo Sector Marks have been generated. The detection of a valid Data Sync Mark with the Group 2 sized window will be used to resynchronize the window timing. That is, if a Data Sync Mark is detected, the Sector Mark window will reset to its default size.
12. Suppose that the Data Sync Mark was not found. The Sector Mark, Address Mark and Data Sync Mark window are now at their maximum Group 2 sizes. These windows will remain at their Group 2 sizes until a Sector Mark, ID Field or Data Sync Mark is detected.
Window Modes During Error Recovery
The special mark windowing just described is subject to the various error recovery modes programmed in the SM330 registers. These modes have a direct affect on the paths taken in the window flowchart in FIG. 3. This section will describe how various error recovery modes affect the windowing.
Pseudo Sector Mark Error
Normally, the SM330 will be programmed to abort the current operation if a Pseudo Sector Mark is generated (Reg. 21h, bit 7 is set). Thus, when a Sector Mark is not found, an interrupt will be generated and the operation is aborted. The window sizes will not be increased. During error recovery, this mode could be disabled (Reg. 21h, bit 7 is reset). The generation of a PSM will increase the size of the AM window and the operation will continue.
ID Voting Thresholds
Under normal operating conditions, the ID Voting Threshold (Reg. 26h, bits 5-4) is set as high as possible to insure data integrity (a value of two is recommended). During error recovery, the ID threshold may be decreased in order to find a bad sector ID. The operation of the Address Mark windows is independent of the ID Voting Threshold. For example, if the threshold is set to one, the default Address Mark windows will still be asserted for every ID field. In addition, the Group 1 or Group 2 Address Mark window (see Table 1) will be asserted over all three AM/ID fields.
ID Skip Control
During error recovery, it is possible to skip the first ID, first two IDs, or all three ID Fields (Reg 28h, bits 1-0). In this case, the assertion of the Address Mark windows is disabled by each skipped ID Field. For example, if the first and second ID Fields are skipped, the AM window will not be active until the third ID Field.
Data Sync Mark Window
The windowing for the Data Sync Mark is only enabled when the ID Voting Threshold is met and the sector ID matches that in the ID Target Sector and ID Target Track registers (Regs. 14h-16h). That is, even if valid IDs are found for a given sector, the Data Sync Mark will not be windowed or looked for it the Target ID does not match.
Fatal Data Sync Error
Under normal conditions, the Fatal Data Sync Error interrupt will be enabled (Reg. 21h, bit 3 is set). When the Data Sync Mark is not detected the operation will be aborted and the windowing is reset. During error recovery, this interrupt may be disabled. In this case, when a Data Sync Mark is not detected, the next Sector Mark window will be increased (see FIG. 3 ). For example, if a Sector Mark, ID Fields or Data Sync Marks are not found, then the Sector Mark window is increased to its Group 2 size.
Resetting the Windowing
After a power-up, drive selection, seek, or other break in drive orientation, it is very useful to issue a Sector Mark Reset. This is accomplished by writing to the Sector Mark Control Register (24h). The special mark windowing will be reset (see FIG. 3 ). The Sector Mark window and PSM generation will be disabled until a valid Sector Mark is detected. After a valid Sector Mark is found, the special mark windowing is activated.
Resync Mark Detection
The Resync Mark is a one-byte pattern interspersed at regular intervals throughout the data field. It is used so that the PLL and SM330 can regain bit and byte synchronization to the data stream and limit the propagation of errors. The initial timing for the Resync Mark window is defined by the Data Sync Mark detection and is refined by successive Resync Mark detection.
The following SM330 registers are used for the Resync Mark detection:
______________________________________25h Resync Mark Control Register28h Window Control Register______________________________________
False Resync Protection Mode
Bits 7-6 in the Resync Mark Control Register control the False Resync Protection mode. If bit 7 is reset, then normal Resync Mark detection is enabled. That is, every time a Resync Mark is detected the window logic will reset the position and width of the next Resync Mark window. For purposes of retry or data recovery, the False Resync Protection Mode can be enabled by setting bit 7. In this case, bit 6 will determine the False Resync Protection Mode. See page 57 of the CL-SM330 data sheet for more details on this mode.
Resync Mark Error Threshold
Bits 3-0 of the Resync Mark Control Register determine the Resync Mark Error Threshold. This threshold has a range of zero to 15 and specifies the number of "missing" or "used" Resync Marks that must be exceeded before the Resync Mark Threshold Error bit is set (Media Error Status Register 23h, bit 2). A "missing" Resync Mark is one that cannot be detected within the Resync Mark window. A "used" Resync Mark is one that is detected in the Resync Mark window, but the decoder must be used to reestablish bit synchronization.
Resync Mark Windowing
The Resync Mark window has two basic modes of operation. The first mode of the Resync Mark window controls how initial Resync Mark detection is windowed. The second mode of operation controls how subsequent, or normal Resync Marks are windowed after the initial Resync Mark has been found.
Initial Resync Mark Detection
Initial Resync Mark detection refers to the first Resync Mark that is detected after the Data Sync Mark. The initial Resync Mark window is dependent on whether the previous Sector Mark or one of the ID fields was found, whether the preceding Data Sync Mark was found, and whether the Skip Data Sync Mark Mode is enabled or disabled.
As discussed earlier, the Skip Data Sync Mark Mode is enabled when the Data Sync Mark Threshold is set to zero (Reg. 26h, bits 3-0=0000b). In this case, the Resync Mark Skip Count will also be enabled (Reg. 25h, bits 5-4). Thus, when the Skip Data Sync Mark Mode is enabled, the detection logic will skip the Data Sync Mark and the first N number of Resync Marks as specified in the Resync Mark Skip Count. The Valid Data Sync byte "OD hex" and an appropriate number of dummy data bytes will be transferred and synchronization will be attempted on the next Resync Mark.
If the Skip Data Sync Mark Mode is disabled (when the Data Sync Mark Threshold is nonzero), then the detection logic will look for the Data Sync Mark and all of the Resync Marks.
With this background information, we can now examine the operation of the initial Resync Mark detection more closely. This operation can be viewed as four separate cases:
CASE 1: The Skip Data Sync Mark Mode is disabled. Either the previous Sector Mark or one of the ID fields was found, but the Data Sync Mark was not found.
When the Data Sync Mark is not found, the Resync Mark window is set to a fixed value of 40 REF2F clock periods. If the first Resync Mark is not found, then the controller will abort the current operation and report a Fatal Data Sync Error (Reg. 23h, bit 3 is set).
CASE 2: The Skip Data Sync Mark Mode is disabled. The Sector Mark, ID fields and Data Sync Mark were not found.
When the Data Sync Mark is not found, the Resync Mark window is set to a fixed value of 160 REF2F clock periods. If the first Resync Mark is not found, then the controller will abort the current operation and report a Fatal Data Sync Error (Reg. 23h, bit 3 is set).
CASE 3: The Skip Data Sync Mark Mode is enabled. Either the previous Sector Mark or one of the ID fields was found.
The Data Sync Mark will be skipped. The controller will also skip the first N number of Resync Marks, where N is the value of the Resync Mark Skip Count (Reg. 25h, bits 5-4). The Resync Mark window is set to a fixed value of 40 REF2F clock periods for the next (N+1) Resync Mark. If the N+1 Resync Mark is not found, then the controller will abort the current operation and report a Fatal Data Sync Error (Reg. 23h, bit 3 is set).
CASE 4: The Skip Data Sync Mark Mode is enabled. Neither the Sector Mark or any of the ID fields was found.
The Data Sync Mark will be skipped. The controller will also skip the first N number of Resync Marks, where N is the value of the Resync Mark Skip Count (Reg. 25h, bits 5-4). The Resync Mark window is set to a fixed value of 160 REF2F clock periods for the next (N+1) Resync Mark. If the N+1 Resync Mark is not found, then the controller will abort the current operation and report a Fatal Data Sync Error (Reg. 23h, bit 3 is set).
NOTE: For all of the above cases, if the initial Resync Mark is detected, then the detection circuit will adjust its nominal window position for the next Resync Mark. At this point, normal Resync Mark detection will be used.
Normal Resync Mark Detection
Normal Resync Mark detection is used after the initial Resync Mark has been detected. Remember, the initial Resync Mark is either the first Resync Mark (if the Skip Data Sync Mark Mode is disabled) or the N+1 Resync Mark after N Resync Marks have been skipped (if the Skip Data Sync Mark Mode is enabled).
The Resync Mark detection is subject to the Resync Mark Error Threshold as described earlier. The normal Resync Mark window width is controlled by the Resync Mark Window Control bits in the Window Control Register (Reg. 28h, bits 3-2). These bits control the size and growth of the Resync Mark window. There are four selections for the size and growth of the Resync Mark window: small progressive, large progressive, small fixed, and large fixed. Page 61 of the CL-SM330 data sheet will provide more details.
Example of Resync Mark Detection
The following example of Resync Mark detection will help explain its operation during initial and normal Resync Mark detection.
Example conditions:
1. Skip Data Sync Mark Mode is enabled (Reg. 26h, bits 3-0 =000b)
2. Resync Mark Skip Count is two (Reg. 25h, bits 5-4=10b)
3. Small Progressive Resync Mark Window (Reg. 28h, bits 3-2=00b)
4. Either the previous Sector Mark or one the ID fields was found
The Data Sync Mark and the first two Resync Marks are skipped. The initial Resync Mark window width is set to 40 REF2F clocks. If the initial (third in this case) Resync Mark is not found, then the controller will abort the current operation and report and Fatal Data Sync error. If the initial Resync Mark is detected, the Resync Mark window resets to the nominal width of nine clocks. The Resync Mark window will remain at nine clocks as long as the Resync Marks are detected. If a Resync Mark is missed, the Resync Mark window will increase to 15. If further consecutive Resync Marks are missed, the Resync Mark window will first increase to 19, then 25, and then remain at the maximum width of 25 until a Resync Mark is found. When a Resync Mark is found, the Resync Mark window will reset back to nine clock periods. If the Resync Mark Error Threshold (Reg. 25h, bits 3-0) is exceeded while the data field is being read, the controller will abort the current operation and set the Resync Mark Error Threshold Exceeded bit (Reg. 23h, bit 2).
Other Window Functions
The Pre-Format window, TOF window and ALPC window are additional window functions of the SM330. When these different windows are active, the PREFMT, TOFWIN and ALPCWIN signals are asserted respectively. These window control signals are used by the drive electronics for various purposes . See the SM330 data sheet for more details about these window functions.
While a preferred embodiment of the present invention has been disclosed and described herein, it will be obvious to those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the invention.
TABLE 1______________________________________Special Mark Window SizesSECTORSIZE AND SPEED 512, 1.0%TOLERANCE 512, 0.5% 1024, 0.5% 1024, 1.0%WINDOW POSITION E N L E N L E N L______________________________________DEFAULT SIZESector Mark 8 16 31WindowAddr Mark 1 1 1WindowData Sync 2 2 2WindowGROUP 1 SIZESector Mark 8 16 31WindowAddr Mark 30 30 38 30 30 39 30 39 39WindowData Sync 10 18 32 32 27WidowGROUP 2 SIZESector Mark 16 31 57WindowAddr Mark 30 30 39 30 39 39 30 39 39WindowData Sync 18 32 32 27 58 55 27Window______________________________________ Notes: E = Early window N = Normal window L = Late window (Based on values programmed in Reg. 24h, bits 7-6) ##SPC1##
Claims
- 1. A method of detecting a Data Sync Mark in an ISO optical track format having one or more errors therein comprising the steps of;
- (a) segmenting the expected Data Sync Mark pattern into a sequence of 12 expected four-bit groups;
- (b) comparing each four-bit group of the Data Sync Mark detector data stream with its expected four-bit group value to determine whether they match;
- (c) comparing the total number of the four-bit groups matching their expected four-bit group values with a threshold number of four-bit groups to determine if a Data Sync Mark was detected.
- 2. A method of detecting a Sector Mark when the Sector Mark Marks differ in time length from a predetermined nominal time length by the steps of:
- (a) generating a reference clock;
- (b) sampling the input data to the mark detector on both the rising and falling edges of said reference clock;
- (c) logically ANDing the two samples of step (b) to shorten the detector input when the Sector Mark Marks are shorter in time length than the predetermined nominal time length and logically ORing the two samples of step (b) to lengthen the detector input when the Sector Mark Marks are longer in time length than the predetermined nominal time length.
- 3. A method of detecting a certain defective Sector Mark comprising the steps of:
- (a) receiving a defective Sector Mark data stream; and
- (b) validating all Sector Mark bits except the first bit of each Sector Mark MARK and SPACE.
- 4. A method of improving the reliability of special mark detection comprising the steps of:
- (a) windowing each special mark detector; and,
- (b) expanding each subsequent special mark window upon a failure to detect a special mark, up to the maximum window size.
- 5. A method of detecting a certain defective Sector Mark comprising the steps of:
- (a) receiving a defective Sector Mark data stream; and
- (b) validating all Sector Mark bits except the first and second bits of each Sector Mark MARK and SPACE.
- 6. A method of detecting a valid Address Mark by keeping the PLL in lock comprised of programming a register to skip the first, or the first two, or all three of the ID fields of a Sector Format.
- 7. A method of preventing false detection of an Address Mark comprised of programming a window to adjust the Address Mark time delays based on the speed tolerance of the drive, and the drive form factor, and the sector size.
- 8. A method of preventing false detection of a Data Sync Mark comprised of programming a window to adjust the Data Sync Mark time delays based on the speed tolerance of the drive, the drive form factor, and the sector size.
- 9. A method of preventing false detection of a Sector Mark comprised of programming a window to adjust the Sector Mark time delays based on the speed tolerance of the drive, and the drive form factor, and the sector size.
- 10. A method in accordance with claim 9, wherein said window size is programmably adjustable in accordance with the previous sector data field errors.
- 11. A method of preventing errors when detecting a Sector Mark in a sector format of a disk storage system comprised of widening a programmably adjustable window when the Sector Mark of the previous sector was not found.
- 12. A method of preventing errors when detecting an Address Mark in a sector format of a disk storage system comprised of widening a programmably adjustable window when the Address Mark of the previous sector was not found.
- 13. A method of preventing errors when detecting a Data Sync Mark in a sector format of a disk storage system comprised of widening a programmably adjustable window when the Data Sync Mark of the previous sector was not found.
- 14. A method for detecting Resync Mark with a programmably adjustable window comprising the steps of:
- (a) skipping the first predetermined number of Resync Marks;
- (b) setting the Resync Mark window length to a predetermined number of reference clock cycles of a continuous reference clock signal;
- (c) detecting the initial Resync Mark;
- (d) variably adjusting the size of the Resync Mark window by widening said window size when Resync Mark is missed and shortening said window size when Resync Mark is found.
US Referenced Citations (2)
Number |
Name |
Date |
Kind |
4663752 |
Kakuse et al. |
May 1987 |
|
5083303 |
Yanagida et al. |
Jan 1992 |
|