Conceptio › Archive › ECMA International

ECMA Internationalopen access

ECMA-396 — Test method for the estimation of lifetime of optical disks for long-term data storage (June 2017)

ECMA International · ECMA International

ECMA International · Standards · License: Open Access

data disks ecma ecmainternational estimation long method optical

ecma, standard, ecma international, specification, ecma-396, ecma 396, 396, test, method, for, the, estimation, lifetime, optical, disks, long-term, data, storage

ECMA-396 4th Edition / June 2017

Test Method for the Estimation of Lifetime of Optical Disks for Long-term Data Storage

Reference number ECMA-123:2009

Contents

Page

Scope ...................................................................................................................................................... 1

Normative references ............................................................................................................................ 2

Terms and definitions ........................................................................................................................... 3

Abbreviated terms ................................................................................................................................. 4

Conformance ......................................................................................................................................... 5

6 6.1 6.2 6.3

Conventions and notations .................................................................................................................. 5 Representation of numbers .................................................................................................................. 5 Variables ................................................................................................................................................. 5 Names ..................................................................................................................................................... 5

7 7.1 7.1.1 7.1.2 7.1.3 7.1.4 7.1.5 7.2 7.3 7.3.1 7.3.2 7.4 7.4.1 7.4.2 7.4.3 7.5 7.5.1 7.5.2 7.5.3

Measurements ....................................................................................................................................... 5 Summary ................................................................................................................................................ 5 Stress incubation and measuring ........................................................................................................ 5 Assumptions .......................................................................................................................................... 5 Data error................................................................................................................................................ 6 Data quality ............................................................................................................................................ 7 Regression ............................................................................................................................................. 7 Test specimen ........................................................................................................................................ 7 Recording conditions ............................................................................................................................ 8 General ................................................................................................................................................... 8 Recording test environment ................................................................................................................. 8 Playback conditions .............................................................................................................................. 8 Playback tester ...................................................................................................................................... 8 Playback test environment ................................................................................................................... 8 Calibration .............................................................................................................................................. 9 Disk testing locations ........................................................................................................................... 9 General ................................................................................................................................................... 9 Rigorous testing location ..................................................................................................................... 9 Basic testing location ........................................................................................................................... 9

8 8.1 8.2 8.2.1 8.2.2 8.2.3 8.2.4 8.3 8.4 8.5

Accelerated stress test ....................................................................................................................... 10 General ................................................................................................................................................. 10 Stress conditions ................................................................................................................................ 10 General ................................................................................................................................................. 10 Temperature ......................................................................................................................................... 11 Relative humidity ................................................................................................................................. 11 Incubation and ramp profiles ............................................................................................................. 12 Measuring-time intervals .................................................................................................................... 13 Design of stress conditions ............................................................................................................... 13 Disk orientation ................................................................................................................................... 13

9 9.1 9.2 9.2.1 9.2.2 9.3 9.4

Lifetime estimation .............................................................................................................................. 13 Time-to-failure ...................................................................................................................................... 13 Accelerated-aging test methods ........................................................................................................ 14 Eyring acceleration model (Eyring method) ..................................................................................... 14 Arrhenius accelerated model (Arrhenius method) .......................................................................... 14 Data analysis and judgment of effectiveness................................................................................... 14 Result of estimated disk life ............................................................................................................... 15

Annex A (normative) Outline of Disk-life estimation method and data-analysis steps ............................. 17 A.1 Data analysis for Disk-life estimation ................................................................................................ 17

A.1.1 A.1.2 A.1.3 A.1.4 A.2 A.2.1 A.2.2 A.2.3 A.2.4 A.2.5 A.2.6

General ................................................................................................................................................. 17 Lognormal model and point estimation of lnBˆ and lnBˆ ............................................................ 17 5

Interval estimation for optical disks ................................................................................................. 18 Estimation of β and σ using least-squares method ...................................................................... 18 Data analysis steps for lifetime estimation ...................................................................................... 20 Judgment of effectiveness of test data and time-to-failure determination................................... 20 Judgment of complete data ............................................................................................................... 21 Condition for lifetime-estimation effectiveness .............................................................................. 22 Life-time estimation when there are missing times-to-failure (Informative) ................................ 22 Lifetime-estimation calculation method (Maximum-likelihood method with least-squares method) ................................................................................................................................................ 23 Lifetime-estimation calculation method (acceleration-factor method) ......................................... 23

Annex B (normative) Disk-life estimation for Controlled storage-condition (Eyring method) ................. 25 B.1 General ................................................................................................................................................. 25 B.2 Data analysis and lifetime estimation using least-squares method .............................................. 25 B.3 Data analysis and lifetime estimation using conventional acceleration-factor method (Step 4-7) .............................................................................................................................................. 31 Annex C (normative) Disk-life estimation for Harsh storage-condition (Arrhenius method) ................... 37 C.1 Stress conditions and data-analysis steps for Arrhenius method ................................................ 37 C.2 Data analysis ....................................................................................................................................... 38 Annex D (normative) Alternative non destructive stress-condition .......................................................... 43 Annex E (informative) Interval Estimation for B5 Life using Maximum Likelihood ................................... 45 E.1 Lower confidence bound ................................................................................................................... 45 E.2 Maximum-likelihood method ............................................................................................................. 45 E.3 Calculation method of Fisher information matrix and variance .................................................... 47 E.4 Example of variance calculation ....................................................................................................... 49 Annex F (informative) RSER measurement of BD disks .............................................................................. 51

Introduction Markets and industry have developed a common understanding that the property referred to as the lifetime of data recorded to optical disks plays an increasingly important role in many applications. Disparate standardized test methodologies exist for Magneto-Optical disks vs recordable compact disks and DVD systems. The first edition of ECMA-396 provided a common methodology, applicable for various purposes that included lifetime testing of then-available writable CD and DVD optical disks. ISO/IEC JTC 1/SC 23/JWG 1, which was a Joint working group comprising ISO/TC 42, ISO/TC 171/SC 1 and ISO/IEC JTC 1/SC 23, initiated work on this subject and developed initial drafts with assistance from Ecma International TC31. After the issuance of the first edition of ECMA-396, ISO/IEC standards for the physical formats of BD Recordable and Rewritable disks were published. Accordingly, ISO/IEC JTC 1/SC 23/JWG 1 and TC31 started work again to include testing of writable BD optical disks in the second and now in this third edition of the Standard. Please note that the 2nd Edition of this standard is an Ecma International only version. In the 3 rd Edition of ECMA-396 – which is a result of synchronization with ISO/IEC 16963 Edition 2 - further additions for lifetime estimation had been incorporated. The 4th Edition of ECMA-396 is synchronized with the ISO/IEC 16963 3rd edition.

This Ecma Standard was developed by Technical Committee 31 and was adopted by the General Assembly of June 2017.

iii

"COPYRIGHT NOTICE © 2017 Ecma International This document may be copied, published and distributed to others, and certain derivative works of it may be prepared, copied, published, and distributed, in whole or in part, p rovided that the above copyright notice and this Copyright License and Disclaimer are included on all such copies and derivative works. The only derivative works that are permissible under this Copyright License and Disclaimer are: (i)

works which incorporate all or portion of this document for the purpose of providing commentary or explanation (such as an annotated version of the document),

(ii) works which incorporate all or portion of this document for the purpose of incorporating features that provide accessibility, (iii) translations of this document into languages other than English and into different formats and (iv) works by making use of this specification in standard conformant products by implementing (e.g. by copy and paste wholly or partly) the functionality therein. However, the content of this document itself may not be modified in any way, including by removing the copyright notice or references to Ecma International, except as required to translate it into languages other than English or into a different format. The official version of an Ecma International document is the English language version on the Ecma International website. In the event of discrepancies between a translated version and the official version, the official version shall govern. The limited permissions granted above are perpetual and will not be revoked by Ecma International or its successors or assigns. This document and the information contained herein is provided on an "AS IS" basis and ECMA INTERNATIONAL DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY OWNERSHIP RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE."

Test Method for the Estimation of Lifetime of Optical Disks for Long-term Data Storage

Scope

This Ecma Standard specifies an accelerated-aging test method for estimating the lifetime of the retrievability of information stored on recordable or rewritable optical disks. The method is based on the theoretical assumption that the lifetime of data recorded on an optical disk has a lognormal distribution. Detailed testing is specified for the following formats: DVD-R/RW/RAM disks, +R/+RW disks, CD-R/RW disks and BD Recordable / Rewritable disks. The testing may be applied to additional optical-disk formats, with substitution of the appropriate specifications, and may also be updated by committee in the future as required. This Ecma Standard includes: -

stress conditions 

Basic and Rigorous stress-conditions for testing and subsequent analysis using both the Eyring and Arrhenius methods.

ambient storage conditions in which the lifetime of data stored on optical disk is estimated 

A Controlled storage-condition, 25 °C and 50 % RH, representing full-time air conditioning. The Eyring method is used to estimate the lifetime under this storage condition.



A Harsh storage-condition, 30 °C and 80 % RH, representing the most severe conditions in which users handle and store optical disks. The Arrhenius method is used to estimate the lifetime under this storage condition.

a description of the evaluation system

procedures for specimen preparation and data acquisition

definitions and methods used in testing specific disk types

analysis of test results to determine the lifetime of stored data

a format for reporting the estimated lifetime of stored data

The methodology includes only the effects of temperature and relative humidity. It does not attempt to model degradation due to complex failure-mechanism kinetics, nor does it test for exposure to light, corrosive gases, contaminants, handling, or variations in playback subsystems. Disks exposed to these additional sources of stress or higher levels of temperature and relative humidity are expected to experience shorter usable lifetime.

Normative references

The following referenced documents are indispensable for the application of this document. For dated references, only the edition cited applies. For undated references, the latest edition of the referenced document (including any amendments) applies. ECMA-330 120 mm (4,7 Gbytes per side) and 80 mm (1,46 Gbytes per side) DVD Rewritable Disk (DVDRAM), 3rd edition (ISO/IEC 17592:2004) ECMA-337 120 mm and 80 mm – Optical Disk using +RW Format – Capacity: 4,7 and 1,46 Gbytes per side (Recording speed up to 4X), 4th edition (ISO/IEC 17341:2009) ECMA-338 80 mm (1,46 Gbytes per side) and 120 mm (4,70 Gbytes per side) DVD Re-recordable Disk (DVD-RW) (ISO/IEC 17342:2004) ECMA-349 120 mm and 80 mm Optical Disk using +R Format – Capacity: 4,7 and 1,46 Gbytes per Side (Recording speed up to 16X), 4th edition (ISO/IEC 17344:2009) ECMA-359 80 mm (1,46 Gbytes per side) and 120 mm (4,70 Gbytes per side) DVD Recordable Disk (DVDR) (ISO/IEC 23912:2005) ECMA-364 120 mm and 80 mm Optical Disk using +R DL Format – Capacity: 8,55 and 2,66 Gbytes per Side (Recording speed up to 16X), 3rd edition (ISO/IEC 25434:2008) ECMA-371 120 mm and 80 mm Optical Disk using +RW HS Format – Capacity: 4,7 and 1,46 Gbytes per Side (Recording speed 8X) 2nd edition (ISO/IEC 26925:2009) ECMA-374 120 mm and 80 mm Optical Disk using +RW DL Format – Capacity: 8,55 and 2,66 Gbytes per Side (Recording speed 2,4x) 2nd edition (ISO/IEC 29642:2009) ECMA-382 120 mm (8,54 Gbytes per side) and 80 mm (2,66 Gbytes per side) DVD Recordable Disk for Dual Layer (DVD-R for DL) (ISO/IEC 12862:2009) ECMA-384 120 mm (8,54 Gbytes per side) and 80 mm (2,66 Gbytes per side) DVD re-recordable disk for dual layer (DVD-RW for DL) (ISO/IEC 13170: 2009) ECMA-394

Recordable Compact Disc Systems CD-R Multi-Speed

ECMA-395

Recordable Compact Disc Systems CD-RW Ultra-Speed

ISO/IEC 30190 Information technology – Digitally recorded media for information interchange and storage – 120 mm Single Layer (25,0 Gbytes per disk) and Dual Layer (50,0 Gbytes per disk) BD Recordable disk ISO/IEC 30191 Information technology – Digitally recorded media for information interchange and storage – 120 mm Triple Layer (100,0 Gbytes per disk) and Quadruple Layer (128,0 Gbytes per disk) BD Recordable disk ISO/IEC 30192 Information technology – Digitally recorded media for information interchange and storage – 120 mm Single Layer (25,0 Gbytes per disk) and Dual Layer (50,0 Gbytes per disk) BD Rewritable disk ISO/IEC 30193 Information technology – Digitally recorded media for information interchange and storage – 120 mm Triple Layer (100,0 Gbytes per disk) BD Rewritable disk

Terms and definitions

For the purposes of this document, the following terms and definitions apply. 3.1 Arrhenius method accelerated aging model based on the effects of temperature only 3.2 baseline analysis of an initial test (e.g., initial data errors) after recording and before exposure to any stress condition, i.e. measurement at stress time t = 0 hours 3.3 Basic stress-condition accelerated-aging conditions for estimating the lifetime of data stored on optical disks with a reasonable amount of time and labour 3.4 B5 Life 5 percentile of the lifetime distribution (i.e. 5 % failure time) or 95 % survival lifetime 3.5 (B5 Life)L 95 % lower confidence bound of B5 Life 3.6 B50 Life 50 percentile of the lifetime distribution (i.e. 50 % failure time) or 50 % survival lifetime 3.7 Controlled storage-condition well-controlled storage-conditions with full-time air conditioning (25 °C and 50 % RH), which may extend the lifetime of data stored on optical disks 3.8 Eyring method accelerated-aging model based on the combined effects of temperature and relative humidity 3.9 data error data error measured on a sample disk before error correction is applied 3.10 Harsh storage-condition most-severe conditions in which users handle and store the optical disks (30 °C and 80 % RH) under which the lifetime of data stored on optical disks can be reduced 3.11 incubation process of enclosing and maintaining controlled test-sample environments 3.12 LDC Block ECC Block of BDs using Long-Distance Code [SOURCE: ISO/IEC 30190:2016 13.6 ]

3.13 Maximum Data Error greatest level of data error measured anywhere in one of the relevant areas on the disk: NOTE For BD Recordable SL/DL disks, BD Recordable TL/QL disks, BD Rewritable SL/DL disks and BD Rewritable TL disks, this is the Maximum RSER; for DVD-R/RW disks and +R/+RW disks, this is the Maximum PI Sum 8; for DVDRAM disks, this is the Maximum BER and for CD-R/RW disks, this is the Maximum C1 Ave 10.

3.14 retrievability ability to recover physically-recorded information as recorded 3.15 RH variable for relative humidity used with % 3.16 Rigorous stress-condition accelerated-aging conditions for estimating the lifetime of data stored on optical disks with higher confidence 3.17 shelf life maximum time an unrecorded disk can be stored under specific conditions and still meet the performance requirements specified 3.18 shelf time time of an unrecorded disk spent on the shelf 3.19 stress temperature and relative humidity variables to which the sample is exposed during the incubation subintervals 3.20 system combination of hardware, software, storage medium and documentation used to record, retrieve and reproduce information 3.21 Temp variable for Celsius temperature used with °C

Abbreviated terms

BER

Byte Error Rate

BLER

BLock Error Rate

Dual Layer

ECC

Error-Correction Code

LDC

Long-Distance Code

Parity (of the) Inner (code)

Quadruple Layer

RSER

Random Symbol Error Rate

SER

Symbol Error Rate

Single Layer

Triple Layer

Conformance

A disk tested by this methodology shall conform to all normative references specific to that disk format

Conventions and notations

6.1

Representation of numbers

A measured value is rounded off to the least significant digit of the corresponding specified value. For instance, it follows that a specified value of 1,26 with a positive tolerance of + 0,01 and a negative tolerance of - 0,02 allows a range of measured values from ,235 to 1,2756.2 Variables

6.2

Variables

A variable with "^" above the character denotes that its value is obtained by estimation.

6.3

Names

The names of entities having explicitly-defined meanings for the purpose of this document are capitalized.

Measurements

7.1

Summary

7.1.1

Stress incubation and measuring

A group of disks shall be measured at four stress conditions for Basic stress-condition testing, or five stress conditions for Rigorous stress-condition testing, for analysis by the Eyring method. For analysis by the Arrhenius method, three stress conditions shall be used for Basic stress-condition testing and four stress conditions shall be used for the Rigorous stress-condition testing. Each total incubation time is divided into several incubation sub-interval time periods. The purpose of the subintervals is to provide sufficient data points to enable proper fitting of the data to an exponential curve during analysis. Each disk in each group of disks has its initial data errors measured before exposure to a stress condition. After each incubation sub-interval, each disk shall be measured for its data errors again. A control disk used for monitoring the measurement equipment may also be measured after each incubation sub-interval. 7.1.2

Assumptions

This Ecma Standard is based on the following assumptions for applicability to the optical disks to be tested: -

the life-distribution of the disks is appropriately modeled by a statistical distribution,

the Eyring method can be used to model aging with both stresses involved (temperature and relative humidity),

the dominant failure mechanism acting when disks are in use under normal conditions will be the same as that acting under the stress conditions,

compatibility of a disk and drive combination can assure the initial recording quality, and will not otherwise affect the resulting lifetime estimation,

a hardware and software system needed to read the disk will be available at the time retrieval of the information is attempted,

the recorded format will be recognizable and interpretable by the reading software.

7.1.3

Data error

7.1.3.1

General

Data errors shall be measured at disk locations defined in 7.5. For each format, the Maximum Data Error used to estimate the time-to-failure shall be determined as follows: BD Recordable SL/DL disks, BD Recordable TL/QL disks, BD Rewritable SL/DL disks and BD Rewritable TL disks defined in ISO/IEC 30190, ISO/IEC 30191, ISO/IEC 30192 and ISO/IEC 30193, respectively : Maximum Random SER (Max RSER), DVD-R disks defined in ECMA-359 and ECMA-382 (ISO/IEC 23912 and ISO/IEC 12862), DVD-RW disks defined in ECMA-338 and ECMA-384 (ISO/IEC 17342 and ISO/IEC 13170), +R disks defined in ECMA-364 and ECMA-349 (ISO/IEC 25434 and ISO/IEC 17344), and +RW disks defined in ECMA-337, ECMA-371 and ECMA-374 (ISO/IEC 17341, ISO/IEC 26925 and ISO/IEC 29642) : Maximum PI Sum 8 (Max PI Sum 8), DVD-RAM disks defined in ECMA-330:

Maximum Byte Error Rate (Max BER),

CD-R/RW disks defined in ECMA-394 and ECMA-395 respectively: Maximum C1 Ave 10 (Max C1 Ave 10). 7.1.3.2

RSER

Per ISO/IEC 30190, ISO/IEC 30191, ISO/IEC 30192 and ISO/IEC 30193, a Random Symbol Error Rate (RSER) is defined as the SER where all erroneous bytes contained in burst errors of length ≥ 40 bytes are not counted, neither in the numerator nor in the denominator of the SER calculation:  Ea  Eb  N

i 1

N  75 392 

 Eb i 1

where, Ea i = number of all erroneous bytes in LDC Block i,

Eb i = number of all erroneous bytes ≥ 40 bytes in LDC Block i, N = number of LDC Blocks. RSER shall be averaged over any 10 000 consecutive LDC Blocks with the condition that all Blocks are recorded either in a continuously-written sequence, or in a discontinuously-written sequence excluding disk defects. A burst error is defined as a sequence of bytes where there are not more than two correct bytes between any two erroneous bytes. For determining burst errors, the bytes shall be ordered in the same sequence as they were recorded on the disk. The length of a burst error is defined as the total number of bytes counting from the first erroneous byte that is preceded by at least three correct bytes to the last erroneous byte that is followed by at least three correct bytes.

The number of erroneous bytes in a burst is defined as the actual number of bytes in that burst that are not correct (see example in Figure 1). The maximum value of the RSER measured over the area specified in 7.5 (Max RSER) shall not exceed 10-3.

Figure 1 — Example of burst error 7.1.3.3

PI Sum 8

Per ISO/IEC 16448 or ISO/IEC 16449, a row in an ECC block that has at least 1 byte in error constitutes a PI error. PI Sum 8 is measured over any 8 consecutive ECC blocks. The maximum number of PI errors, also called Max PI Sum 8, before error correction, measured over the area specified in 7.5 shall not exceed 280. 7.1.3.4

BER

The number of erroneous symbols shall be measured in any consecutive 32 ECC blocks in the first pass of the decoder before correction. The BER is the number of erroneous symbols divided by the total number of symbols included in the 32 consecutive ECC blocks. The maximum value of the BER measured over the area specified in 7.5 (Max BER) shall not exceed 10-3. 7.1.3.5

C1 Ave 10

ISO/IEC 10149 specifies that the BLER averaged over any 10 seconds shall be less than 3×10-2. At the standard (1X) data transfer rate, the total number of blocks per second entering the C1-decoder is 7 350. Thus, the number of C1 errors per second before error correction which is averaged over any 10 seconds is called C1 Ave 10. The maximum value measured over the area specified in 7.5 (Max C1 Ave 10) shall not exceed 220. 7.1.4

Data quality

Data quality is checked by plotting the median rank of the estimated time to failure values with a best-fit line for each stress condition. The lines are then checked for reasonable parallelism. 7.1.5

Regression

The log predicted time-to-failure values shall be calculated using linear regression. Multiple linear-regression is used for the Eyring method and linear regression is used for the Arrhenius method.

7.2

Test specimen

The sample disks shall represent the construction, materials, manufacturing process, quality and variation of the final process output.

Consideration shall be made for shelf life. Longer shelf time of optical disks before recording and testing can impact test results. Shelf time shall be a representative of normal usage. NOTE In case the support of disk manufacturer is available, it is recommended to use the disks gathered from as many production lots as possible.

7.3

Recording conditions

7.3.1

General

Before disks are entered into accelerated-aging tests, they shall be recorded as optimally as is practicable according to the descriptions given in the related standard. OPC (Optimum Power Control) during the writing process shall serve as the method to achieve minimum data errors. It is generally assumed that optimallyrecorded disks will yield the longest estimated-lifetime. Disks are deemed acceptable for entry into the aging tests when their data errors and all other disk parameters are found to be within their respective standard’s specification limits. The choice of recording hardware is at the discretion of the recording party. It may be based either on a commercial drive or a specialty recording tester. It shall be capable of producing recordings that meet all specifications. The recording speed used for testing shall be reported. NOTE speed.

It is expected that the lifetime of data on a disk can be affected by recording conditions including recording

7.3.2

Recording test environment

When performing recordings, the air immediately surrounding the disk shall have the following properties: Temperature:

23 °C to 35 °C

Relative humidity:

45 % to 55 %

Atmospheric pressure:

60 kPa to 106 kPa

No condensation on the disk shall occur. Before testing, the disk shall be conditioned in this environment for 48 hrs minimum. It is recommended that, before testing, the entrance surface be cleaned according to the instructions of the manufacturer of the disk.

7.4

Playback conditions

7.4.1

Playback tester

Specimen disks shall be read as described in the relevant format standards identified in Clause 3. 7.4.2

Playback test environment

When measuring the data errors, the air immediately surrounding the disk shall have the following properties: Temperature: Relative humidity: Atmospheric pressure:

23 °C to 35 °C, 45 % to 55 %, 60 kPa to 106 kPa.

Unless otherwise stated, all tests and measurements shall be made in this test environment.

7.4.3

Calibration

The test equipment should be calibrated as needed or prescribed by its manufacturer using calibration disks approved by said manufacturer before disk testing. A control disk should be maintained at ambient conditions, and its data error should be measured at the same time the stressed disks are measured, both initially and after each stress sub-interval. The mean and standard deviation of the control disk shall be established by collecting at least five measurements. Should any individual data error differ from the mean by more than three times the standard deviation, the problem shall be corrected and all data collected since the last valid control point shall be remeasured.

7.5 7.5.1

Disk testing locations General

Disk testing locations for Data Error measurement shall be Rigorous testing location or Basic testing location. Rigorous testing location should be applied combined with Rigorous stress-condition testing. Basic testing location should be applied combined with Basic stress-condition testing. A combination of Rigorous testing location and Basic stress-condition testing and a combination of Basic testing location and Rigorous stresscondition testing are also applicable (see 8.2.1). 7.5.2

Rigorous testing location

Rigorous testing location is all data areas on a disk to be tested. 7.5.3

Basic testing location

Basic testing location is a minimum of three bands spaced evenly across the inner, middle and outer radius regions on the disk as indicated in Table 1. The total testing area shall represent a minimum of 5 % of the disk capacity. For BD disks, each of the three test bands in each layer shall have more than 10 000 LDC Blocks. For DVD disks and +R/+RW disks, each of the three test bands in each layer shall have more than 750 ECC blocks for 80 mm disks, or 2 400 ECC blocks for 120 mm disks. For CD disks, each of the three test bands shall have more than 5 900 sectors. Table 1 — Nominal radii of three test bands BD Recordable disk/BD Rewritable disk (SL/DL/TL/Q) (inner radius)

DVD-R/DVD-RW/+R/+RW disk(SL/DL) (Inner radius)

(Unit; mm)

DVD-RAM disk

CD-R/RW disk (inner radius)

120 mm

80 mm

120 mm

80 mm

120 mm

Band 1

25,0

24,1 to 25,0

25,0

Band 2

40,0

30,0

40,0

29,8 to 38,8

39,4 to 40,4

40,0

Band 3

55,0

35,0

55,0

34,6 to 35,6

54,9 to 55,8

55,0

NOTE For Multi-layer disks it is recommended that additional test band(s) at the outer diameter covering data in the transition(s) between layers in the disk be included in the test.

Accelerated stress test

8.1

General

Accelerated stress testing is used in order to estimate the lifetime of the optical disk. All information needed for this testing is provided in this document.

8.2 8.2.1

Stress conditions General

Stress conditions for this test method are increases in temperature and/or relative humidity. The stress conditions are intended to accelerate the chemical reaction rate from what would occur normally at ambient storage or usage conditions. The chemical reaction is expected to cause degradation in some desired material property that eventually leads to disk failure. Regarding the use of the Eyring method, five stress conditions shall be used for Rigorous stress-condition testing and the minimum number of specimens that shall be used for those stress conditions are shown in Table 2. The four stress conditions that shall be used for Basic stress-condition testing and the minimum numbers of specimens are shown in Table 3. Additional specimens and conditions may be used, if desired for improved precision. The total incubation time for each stress condition shall be greater than or equal to the minimum total incubation time. The minimum total incubation-time for the Rigorous stress-condition is defined in Table 2. The minimum total incubation-time for the Basic stress-condition is defined in Table 3. If all the data errors of specimens for a certain stress condition far exceed the failure criteria (see 9.1) before the minimum total incubation-time and the continuation of testing is judged as irrelevant then the testing for that stress condition may be stopped. The incubation sub-interval time shall be smaller than or equal to the maximum incubation sub-interval time. The maximum incubation sub-interval time for the Rigorous stress-condition is defined in Table 2. The maximum incubation sub-interval time for the Basic stress-condition is defined in Table 3. The number of incubation sub-intervals depends on the total incubation time and the incubation sub-interval time. For example the total time for each stress condition given in Table 2 and Table 3 is divided into five and four equal incubation sub-intervals respectively in the case of a combination of the maximum incubation subinterval time and the minimum total incubation-time. It is recommended to set the number of incubation subintervals to greater than or equal to 4, considering the case that a specimen reaches the failure criteria (see 9.1) before the minimum total incubation-time. Regarding use of the Arrhenius method, stress conditions are given in Table C.1 and Table C.2 in Annex C. The temperature and relative humidity during each incubation sub-interval shall be controlled as given in Table 4 and shown in Figure 2.

Table 2 — Rigorous stress-condition for use with Eyring method

Test specimen group

Test stress condition (incubation)

Number of specimens

Maximum incubation sub-interval time

Minimum total incubationtime

Intermediate relative humidity

Minimum equilibration duration time

Temp °C 85

RH % 80

300

1 500

RH % 30

400

2 000

600

3 000

600

3 000

800

4 000

h 7

Table 3 — Basic stress-condition for use with Eyring method Test specimen group

Test stress condition (incubation)

Number of specimens

Maximum incubation sub-interval time

Minimum total incubationtime

Intermediate relative humidity

Minimum equilibration duration time

RH %

Temp °C

RH %

250

1 000

250

1 000

500

2 000

625

2 500

NOTE Total incubation time and incubation sub-interval time should be determined from the aging characteristic of the disks under test. In the situation where only one condition or less reaches the failure criteria during the minimum total incubation time it is recommended that the test should be extended for all conditions until at least two conditions reach the failure criteria.

8.2.2

Temperature

The temperature levels chosen for this test plan are based on the following: There shall be no change of phase of moisture within the test system over the test-temperature range. This restricts the temperature to greater than 0 °C and less than 100 °C. The temperature shall not be so high that plastic deformation occurs anywhere within the disk structure. In case a stress condition would be destructive for a disk to be tested see Annex D for alternative stressconditions. The typical substrate material used for optical disks is polycarbonate (glass-transition temperature is around 150 °C). The glass-transition temperature of other layers can be lower. Experience with high-temperature testing of BD disks, DVD disks, +R/+RW disks, and CD disks indicates that an upper limit of 85 °C is practical for most applications. 8.2.3

Relative humidity

Experience indicates that 80 % of relative humidity is the generally-accepted upper limit for control within most accelerated test cells.

8.2.4

Incubation and ramp profiles

The relative-humidity transition (ramp) profile is intended to avoid moisture condensation on the substrate, minimize substantial moisture gradients in the substrate and end at ramp-down completion with the substrate equilibrated at the ambient condition. This is accomplished by varying the moisture content of the chamber only at the stress-incubation temperature, and allowing sufficient time for equilibration during the ramp down based on the diffusion coefficient of water in polycarbonate. Table 4 —Temperature and relative humidity transition (ramp) profiles for each incubation sub-interval Temperature

Relative humidity

Duration

°C

hours

Start

at Tamb

at RHamb

—

Temperature, Relative-humidity ramp

to Tinc

to RHint

1,5 ± 0,5

Relative-humidity ramp

at Tinc

to RHinc

1,5 ± 0,5

Incubation

at Tinc

at RHinc

See Table 2 or Table 3

Relative humidity ramp

at Tinc

to RHint

1,5 ± 0,5

Equilibration

at Tinc

at RHint

See Table 2 or Table 3

Temperature, Relative- humidity ramp

to Tamb

to RHamb

1,5 ± 0,5

End

at Tamb

at RHamb

—

Process step

amb= room-ambient temperature or relative humidity (Tamb or RHamb) inc= stress-incubation temperature or relative humidity (Tinc or RHinc) int= intermediate relative-humidity (RHint) that at Tinc supports the same equilibrium moisture absorption in polycarbonate as that supported at Tamb and RHamb Figure 2 — Graph of typical transition (ramp) profile for each incubation sub-interval

8.3

Measuring-time intervals

For data collection, RSER (BD Recordable SL/DL disk, BD Recordable TL/QL disk, BD Rewritable SL/DL disk, BD Rewritable TL disk), PI Sum 8 (DVD-R, DVD-RW, +R, +RW disk), BER (DVD-RAM disk), or C1 Ave 10 (CD-R, CD-RW disk) shall be measured on each disk : 1) before disk exposure to any stress condition to determine its baseline measurement and 2) after each incubation sub-interval. The length of time for intervals is dependent on the severity of the stress conditions. In case all the data errors of specimens do not reach the failure criteria (see 9.1) within the minimum total incubation time, testing at a particular stress condition may have to be stopped (see A.2.1 for guidance).

8.4

Design of stress conditions

A separate group of specimens shall be used for each stress condition. Table 2, for the Rigorous stress-condition, and Table 3, for the Basic stress-condition, specify the temperatures, relative-humidity values, maximum Incubation sub-intervals, minimum total incubation time, and minimum number of specimens for each stress condition. All temperatures shall be maintained within ± 2 °C of the target temperature; all relative-humidity values shall be maintained within ± 3 % RH of the target relative humidity. The intermediate relative-humidity values in Table 2 and Table 3 are calculated assuming 25 °C and 50 % RH ambient conditions. If the ambient is different, the intermediate relative humidity to be used is calculated using the equation:

RHint 

0,24  0,003 7  Tamb  RH amb 0,24  0,003 7  Tinc

where, Tamb and Tinc are the ambient and incubation temperature in units of °C, RHamb

is the ambient relative humidity,

RHint

is the intermediate relative humidity.

The stress conditions in Table 2, Table 3 and Table 4 offer sufficient combinations of temperature and relative humidity to satisfy the mathematical requirements of the Eyring method.

8.5

Disk orientation

The disks subjected to this test method shall be maintained during incubation in a vertical position with a minimum of 2 mm separation between disks to allow air flow between disks and to minimize deposition of debris, which could negatively influence the data-error measurements, on the disk surface.

Lifetime estimation

9.1

Time-to-failure

Ideally, all disks subjected to stress conditions should have their times-to-failure calculated at the stress conditions they have been subjected to. The time-to-failure of a disk is determined by the test data including the Maximum Data Error (see A.2.1). In case any times-to-failures are not available for a stress condition, however, see A.2.3. Failure criteria are: Max RSER exceeding 10-3 for BD Recordable SL/DL disks, BD Recordable TL/QL disks, BD Rewritable SL/DL disks and BD Rewritable TL disks (see Annex F), Max PI Sum 8 exceeding 280 for

DVD-R/RW disks and +R/+RW disks, Max BER exceeding 10-3 for DVD-RAM disks and Max C1 Ave 10 exceeding 220 for CD-R/RW disks. It is assumed that the data errors on a disk are the result of material degradation. The chemical changes are generally expected to cause test data to have a distribution that follows an exponential function over time. Therefore, test values of: PI Sum 8, BER, C1 Ave 10 or RSER as functions of time are expected to exhibit an exponential distribution. The best function fitting an error trend can be found by regression of the test data against time, for example, with a least-squares fit. The time-to-failure per disk type can be calculated using the error-trend function and the failure criteria. But if a determination of time-to-failure is judged not to be effective then that case should be treated as a missing time-to-failure (see A.2.1).

9.2

Accelerated-aging test methods

9.2.1

Eyring acceleration model (Eyring method)

Using the Eyring model, the following equation is derived from the laws of thermodynamics and can be used to handle the two critical stresses of temperature and relative humidity. t  AT a e H / kT e (BC / T )RH

where t is the time to failure, A is the pre-exponential time constant, a is the pre-exponential temperature factor, T ΔH is the activation energy per molecule, k is the Boltzmann's constant (1,380 7 × 10-23 J/molecule degree K), T is the temperature (in Kelvin), B, C are the RH exponential constants, In this Ecma Standard T (in Kelvin) is set as T = 273,15+Temp (°C). For the temperature range used in this test method, “a” and “C” shall be set to zero. The Eyring-model equation then reduces to the following equation: t  Ae H / kT e BRH

or, ln(t )  ln( A )  9.2.2

H  B  RH . kT

Arrhenius accelerated model (Arrhenius method)

The Arrhenius method uses only temperature stress for accelerated aging. The time-to-failure is assumed to be governed by the following Arrhenius-model equation: t  Ae H / kT ,

ln(t )  ln( A ) 

9.3

H . kT

Data analysis and judgment of effectiveness

Data analysis and a method for judging the effectiveness of the data are contained in the following Annexes:

Annex A: Outline of Disk-life estimation method and data-analysis steps, Annex B: Disk-life estimation for the Controlled storage-condition (Eyring method), Annex C: Disk-life estimation for the Harsh storage-condition (Arrhenius method), Annex E: Interval estimation for B5 Life using maximum likelihood.

9.4

Result of estimated disk life

An estimated lifetime based on the data analysis shall be reported as follows. (a) Number and title of this standard. (b) Ambient storage-condition for the lifetime estimation: 25 °C / 50 % RH (Controlled storage-condition) or 30 °C / 80 % RH (Harsh storage-condition). (c) Disk testing location: Rigorous testing location or Basic testing location (see 7.5). (d) Stress and testing condition: Rigorous stress-condition testing or Basic stress-condition testing and whether or not the alternative condition was used. (e) The recording speed used for testing shall be reported (see 7.3). (f) Time-to-failure data Complete data or data with the substitutes of missing times-to-failure. (g) Sample information Number of samples tested under each stress condition. (h) Estimation method and the estimated data Maximum-likelihood method with the least squares method/acceleration-factor method and the estimated log standard deviation (i) B50 Life, B5 Life and 95 % lower confidence bound of B5 Life (= (B5 Life)L) for the maximum-likelihood method with least squares method. B50 Life, B5 Life and the point estimates of the 5 percentile with variation (= B5V Life) for the accelerationfactor method.

Annex A (normative) Outline of Disk-life estimation method and data-analysis steps

A.1

Data analysis for Disk-life estimation

A.1.1

General

Data analysis for lifetime estimation is based on the following assumptions. -

The lifetime of data recorded on an optical disk has a lognormal distribution.

The Eyring method is used for the Controlled storage condition (25 °C, 50 % RH) (see Annex B).

The Arrhenius method is used for the Harsh storage condition (30 °C, 80 % RH) (see Annex C).

The maximum-likelihood method (see Annex E) is applied for a precise analysis and a precise interval estimation. Thus the lifetime estimation in this Ecma Standard is specified based on the maximum-likelihood method estimation. The calculation for the maximum-likelihood method is complicated and it is not so easy to adopt. If the lifetime data is complete and its distribution is lognormal, then the estimated lifetime can also be calculated using the least-squares method and the calculated results will be the same as that of the maximum-likelihood method. Thus for the complete data case the least-squares method, which is relatively easy to calculate, is adopted as the practical calculation method for estimating the population. For the case that the lifetime data is not complete and there are missing times-to-failure, the estimation method is shown in A.2.4 as an informative sub clause. The acceleration-factor method has been widely used for the lifetime estimation of DVD disks. Those who need the evaluation with relation to the past data can refer to the acceleration method, explained in A.2.6 and B.3. There can be the case of the multi-layer disk that the Maximum Data Error occurs in different layers after each incubation sub-interval time according to the acceleration condition. In that case at first it is recommended to confirm that there is not any abnormal Maximum Data Error value. In such a case the time-to-failure of multilayer disk should be estimated for each layer and the time-to-failure of the disk should be the minimum one among the layers.

A.1.2

Lognormal model and point estimation of lnBˆ 5 and lnBˆ 50

As time-to-failure t is distributed with lognormal distribution LN( μ, σ 2 ) , log lifetime ( y  ln t ) follows a normal distribution N( μ, σ 2 ) , where μ and σ 2 are the expected values of y and variance, respectively. μ can be expressed as a function of x as follows, y  μ (x )  σ  z

 β0  β1x1  β2 x 2σ  z NOTE

x is a vector with two dimensions (x1, x2).

where z denotes a percentile of N(0,1), β0 = ln A, β1 = ΔH /k, β2 = B (for the definition of A and B see 9.2.1), x1 represents the variable related to the temperature as x1 = 1/T and x2 represents the variable related to the relative humidity as x2 = RH . The p percentile of the lifetime distribution, or BP Life, is widely used in reliability engineering. The point estimation of lnBp is described as

lnBˆp  βˆ0  βˆ1x1  βˆ2 x2  zp / 100σˆ . Then the point estimates of the 5 percentile and 50 percentile of the lifetime distribution are given by:

lnBˆ5  βˆ0  βˆ1x10  βˆ2 x20 1,64σˆ , lnBˆ50  βˆ0  βˆ1x10  βˆ2 x20 . where, x10 , x20 denotes the Controlled storage-condition ( 25 °C and 50 % RH ). （x10 = 1/(273,15 + 25) , x20 = 50 ） The purpose of the lifetime estimation is to estimate the lifetime of the population. Thus   is the unbiased 2

NOTE variance.

A.1.3

Interval estimation for optical disks

For interval estimation of lnB̂p for an optical disk, one may consider only the lower bound.

(100  α ) % lower confidence bound of log lifetime lnB̂p is given by the following equation: ( lnBˆ p )L  lnBˆ p  zα / 100 var(ln Bˆ p ) ,

where, var ( lnBˆ p ) denotes the variance of lnB̂p (see Annex E).

A.1.4

Estimation of β and σ using least-squares method

The multiple linear-regression model for the ij th specimen is described as follows.

y ij  β0  β1x1j  β2 x2 j  εij

(i  1 to n j ) ( j  1 to J ) ,

where, ε ij denotes errors, nj denotes the number of specimens in each group and J denotes the total number of groups. The estimate ŷ j is given as

yˆ j  βˆ0  βˆ1x1j  βˆ 2 x2 j , where, x1j = 1/(273,15 + Tj (in °C) x2j = RHj. Also, the sum of squared residual errors Se is computed as J

Se    ( y ij  yˆ j )2 . j  1 i 1

If the lifetime data is complete and the distribution is lognormal then the estimated regression coefficients obtained by the least-squares method are the same as that of the maximum-likelihood method and they can be used for the estimation. The following shows the way to utilize the calculation results obtained by the leastsquares method. The estimated regression coefficients of ŷ j can be obtained by applying the least-squares method to Se . The estimates βˆ0 , βˆ1 and βˆ2 are obtained by solving 110 linear-regression equations of group A, B, C, D and E. Let σˆ lsm  be the unbiased variance obtained by the least-squares method, then 2

the estimate σˆ lsm  is given by 2

σˆ lsm 2 

(y  yˆ ) ij

Se j 1 i  1  (n  3) (n  3)

where, n =  n j and it denotes the total number of specimens. j 1

n  3  n  2  1 . -1 is for the limited number of the sampling and -2 is for the number of degrees of freedom (temperature and humidity). NOTE This clause shows the case for the Eyring method. In case of the Arrhenius method the degree of freedom is 1 (temperature only) and n －3 becomes n－2. 2 The estimated regression-coefficients βˆ0 , βˆ1 and β̂2 and estimated variance of residual errors σˆ lsm  are obtained using regression analysis statistics software tools.

B50 Life, B5 Life and the 95 % lower confidence bound of B5 Life are described as follows. B50 Life = exp ( lnBˆ 50 ) = exp ( βˆ0  βˆ1x10  βˆ2 x20 ), B5 Life = exp ( lnBˆ 5 ) = exp ( βˆ0  βˆ1x10  βˆ 2 x20  1,64 σˆ ), = exp ( βˆ0  βˆ1x10  βˆ 2 x20  1,64 σˆ lsm ), where, x10 , x20 denotes the Controlled storage-condition ( 25 °C and 50 % RH ). By substituting σ̂ lsm for σ̂ , var ( lnBˆ 50 ) and var ( lnBˆ5 ) are obtained as follows (see E.3).

 n  2 σˆ lsm  J  1 var ( lnBˆ 50 )   1 x10 x 20   2 n j x1 j ˆ  σ lsm jJ1  1 n j x2 j  σˆ 2  lsm j 1



1 n j x1 j 2 σˆ lsm j 1 J 1 n j x12j 2 σˆ lsm j 1 J 1 n j x1 j x 2 j 2 σˆ lsm j 1



J  1 n j x2 j  2 σˆ lsm j 1  J  1 n j x1 j x 2 j  2 ˆ σ lsm j 1  J 1  2 n j x2 j  2 ˆ σ lsm j 1 



1 1 x   10   x 20 

 n  2 ˆ σ lsm  J  1 n j x1 j 2  σˆ var ( lnBˆ 5 )   1 x10 x 20  1,64  lsm jJ1  1 nx  σˆ 2 j 1 j 2 j lsm   0 



1 n j x1 j 2 ˆ σ lsm j 1 J 1 n j x12j 2 ˆ σ lsm j 1 J 1 n j x1 j x 2 j 2 σˆ lsm j 1



1 n j x2 j 2 ˆ σ lsm j 1 J 1 n j x1 j x 2 j 2 σˆ lsm j 1 J 1 n j x 22 j 2 σˆ lsm j 1



 0    0   0    2n  2  σˆ lsm

1  1   x   10   x 20      1,64

where, n =  n j and it denotes the total number of specimens. j 1

Using the result of the above equation, the lower confidence bound of log lifetime lnBˆ 5 = (B5 Life)L is given by the following equation. (B5 Life)L  exp (( lnBˆ5 )L )  exp ( lnBˆ5  1,64 var(ln Bˆ5 ) )

A.2

Data analysis steps for lifetime estimation

A.2.1

Judgment of effectiveness of test data and time-to-failure determination

Before the lifetime-estimation calculation, the effectiveness of the test data shall be checked, following the procedure listed below. Step 1: Calculate the linear regression or the polynomial regression of the logarithm of test-data error rate (Errort) = ln(Errort) against incubation time and plot ln(Errort) versus the incubation time and their best-fit line on the linear-scale graph for each test-condition specimen. Step 2: Check the following three conditions: a) The best-fit line increases monotonously. b) All ln(Errort) are almost on the best-fit line. c) The best-fit line has reasonable increase and is not flat nor having a negative slope. If all three conditions are satisfied, then go to Step 3. If the three conditions are not satisfied, then that time-to-failure shall not be determined. There are two cases where the above three conditions are not satisfied. - The first case is that there is a sample that shows unexpected deterioration during the first subinterval time of the accelerated-aging test while other samples satisfy the three conditions. In this case the deterioration mechanism of the abnormal sample can be different from that of other samples. The sample whose error rate cannot be obtained after the first sub-interval time shall

be treated as having the missing time-to-failure. Then go to Step 4 in A.2.2. In this case keep the number of specimens as it is. - The second case is that there is a sample in a group which does not deteriorate within the minimum total incubation-time and its best-fit line does not show reasonable increase while other samples in that group satisfy the three conditions. The case when the time-to-failure of a sample that does not satisfy the three conditions shall be treated as the missing time-to-failure and the procedure shall continue at Step 4 in A.2.2. In this case keep the number of specimens as it is. If there are some ln(Errort)s that show abnormal values, it is recommended to check the reason why those values are abnormal, if possible, and it is also recommended to judge whether to adopt those values or not. Step 3: For each test-condition specimen, determine the time-to-failure where the best-fit line crosses the failure criteria. For a test-condition specimen for which the measured error rate did not reach the failure criteria within the minimum total incubation-time, the time-to-failure may be determined using the extrapolation of the best-fit line of ln(Errort) as a predicted time-to-failure. After Step 3, go to the procedure in A.2.2.

A.2.2

Judgment of complete data

Follow the procedure listed below. Step 4: For each specimen of a stress group, order the time-to-failure values by increasing incubation time. Calculate the median rank of each specimen for each time-to-failure (see B.1 Step 2). If there is a sample that shows unexpected deterioration during the first sub-interval time of the accelerated-aging then its missing time-to-failure shall be given the median rank smaller than that of the shortest time-to-failure when sorting times-to-failure for the determination of the median rank. If there is a sample that does not deteriorate within the minimum total incubation-time then its missing time-to-failure shall be given the median rank larger than that of the longest time-to-failure when sorting times-to-failure for the determination of the median rank. Step 5: Plot the median rank versus the time-to-failure on a lognormal graph, with time-to-failure on the abscissa and median rank on the ordinate, for each specimen of the stress group. Plot the best-fit straight line for each specimen of the stress group. Step 6: Check the following conditions. a) All the times-to-failure corresponding to each median rank are almost on the best-fit straightline of each stress group. b) The best-fit straight lines of all stress groups are reasonably parallel with each other. If both conditions listed above are satisfied, then the data is deemed complete. Proceed to the procedure in A.2.5 (Maximum-likelihood method with least-squares method) or in A.2.6 (acceleration-

factor method). For the precise analysis or the precise interval estimation, go to the procedure in A.2.5. If a time-to-failure is away from the best-fit straight line, then that time-to-failure shall not be used for the lifetime estimation. That time-to-failure is treated as a missing time-to-failure. If at least one condition listed above is not satisfied, then go to the procedure in A.2.3.

A.2.3

Condition for lifetime-estimation effectiveness

Follow the procedure listed below. Step 7: Check the following three conditions and judge the effectiveness of the time-to-failure. a) The lognormal data plots of each stress group are almost on the best-fit straight-line. b) Exclude the missing times-to-failure, then check the specimens of each stress group have effective times-to-failure that span over one-half of a median rank point. c) The best-fit straight lines of all stress groups are reasonably parallel with one another. If these three conditions are satisfied, we can assume that the lifetime distribution is lognormal. In case there are missing times-to-failure the calculation based on the maximum-likelihood method can be possible, but the method is complicated and is not easy to apply. There is a possibility that it is not as precise as the maximum-likelihood method but the other method in which the missing times-tofailure are substituted is shown in A.2.4. In A.1, it was assumed that the lifetime data has a lognormal distribution. If the three conditions are not satisfied, it is proven that the assumption is not effective and a reliable lifetime estimation cannot be obtained.

A.2.4

Life-time estimation when there are missing times-to-failure (Informative)

As shown in Step 7 in A.2.3 there are cases that lifetime distribution is lognormal but missing times-to-failure exist. The method to substitute those missing times-to-failure is shown in this clause. Be aware that this method uses substituted data in the best-fit straight lines and the estimated lifetime may be longer. a) Substitution of times-to-failure For each missing time-to-failure, check the corresponding median rank and substitute the missing time-to-failure value with the value where the best-fit straight line crosses the corresponding median rank on the lognormal graph. b) Maximum-likelihood method with least-squares method application Substitute all the missing times-to-failure and prepare the complete data set. Then follow the steps in A.2.5. c) Acceleration-factor method application Substitute all the missing times-to-failure and prepare the complete data set. Then follow the steps in A.2.6. For the precise analysis or the precise interval estimation, go to the steps in A.2.5.

A.2.5 Lifetime-estimation calculation method (Maximum-likelihood method with leastsquares method) Calculation of the maximum likelihood method with the least-squares method can be done as listed below. a) Calculate the multiple regression coefficients and standard error using the least-squares method across all times-to-failure. This calculation can be performed by multiple regression analysis using statistics software tools. The coefficient of determination is expected to be over 0,8. If the coefficient of determination of the multiple regression analysis is too small then it is recommended to reconsider the accelerated aging-test condition. b) B50 Life, B5 Life and 95 % lower confidence bound of B5 Life at the Controlled storage condition are calculated using the multiple regression-coefficients and standard error obtained by the least-squares method and the equations of maximum-likelihood method (see B.33 and E.4).

A.2.6

Lifetime-estimation calculation method (acceleration-factor method)

Calculation of the conventional acceleration-factor method can be done as follows. a) Calculate regression coefficients using the log-mean failure time. b) Calculate acceleration factors from the difference between the estimated log-mean at each stress condition. c) Calculate the normalized time-to-failure at the ambient condition for each specimen group using the acceleration factors, and plot these data on a lognormal graph. d) Assuming that the normal distribution of the population varies according to the 95% lower confidence bound of the normal distribution, B50 Life, B5 Life and the point estimates of the percentile with variation (= B5V Life) at the Controlled storage-condition are calculated using μˆ and σˆ obtained from the fitting line (see B.33). NOTE The data-analysis steps using the Arrhenius method are almost the same as with the Eyring method. A single regression at the Harsh storage temperature can be used with the Arrhenius method.

Annex B (normative) Disk-life estimation for Controlled storage-condition (Eyring method)

B.1

General

In this annex, the analysis of the complete data case using the results of a least-squares method and a conventional acceleration-factor method for the Rigorous stress-condition testing are shown.

B.2

Data analysis and lifetime estimation using least-squares method

Step 1 Determine the time-to-failure for each specimen at the stress applied following the procedure described below. The data error to be measured is defined in 7.1.3: BD Recordable disks and BD Rewritable disks: DVD-R/RW, +R/+RW disks: DVD-RAM disks: CD-R/RW disks:

Max RSER, Max PI Sum 8, Max BER, Max C1 Ave 10.

Use the initial data-errors measured prior to accelerated aging plus the data errors measured after each specified accelerated-aging incubation sub-interval. For each specimen, a linear regression is performed with the natural logarithm of measured data-errors as the dependent variable and time as the independent variable. The time-to-failure of the specimen is calculated from the slope and intercept of the regression as the time at which the specimen would have a Max RSER of 10-3, Max PI Sum 8 of 280, Max BER of 10-3 or Max C1 Ave 10 of 220. Table B.1 shows calculations leading to an estimated time-to-failure from a hypothetical data set. The data for five stress conditions (Group A, Group B, Group C, Group D and Group E) are offered solely as an example of the mathematical methodology used in this test procedure. Step 2 For each stress condition, the specimens are ordered by increasing log time-to-failure values. The median rank of each specimen is calculated using the estimate (i  0,3) /(n  0,4) , where i is the time-tofailure order and n is the total number of specimens at the stress condition. Table B.2 shows the ordered log time-to-failure and the median rank for the example data.

Table B.1 — Ordered estimated time-to-failure for example data (Rigorous stress-condition) Order number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Group A

Group B

Temp = 85 °C and RH = 80 %

Temp = 85 °C and RH = 70 %

429 451 476 484 493 495 501 512 521 526 534 540 542 548 557 576 579 586 618 645

613 640 649 675 679 696 703 709 719 732 739 743 747 751 766 778 785 804 856 896

Group C Temp = 85 °C

Group D

Group E

and RH = 60 %

Temp = 75 °C and RH = 80 %H

Temp = 65 °C and RH = 80 %

864 913 915 945 951 993 994 998 1 009 1 014 1 027 1 030 1 037 1 049 1 069 1 080 1 098 1 125 1 222 1 249

1 728 1 882 1 907 1 989 2 020 2 076 2 129 2 151 2 180 2 227 2 277 2 318 2 352 2 404 2 443 2 512 2 589 2 590 2 776 2 891

5 455 5 730 5 908 6 114 6 326 6 431 6 544 6 632 6 711 6 779 6 860 6 935 7 038 7 108 7 202 7 285 7 362 7 454 7 562 7 569 7 710 7 827 7 955 8 067 8 250 8 405 8 546 8 700 8 953 9 452

Table B.2 (1 of 2) — Log time-to-failure and median rank for example data Group A Order number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Mean

Temp = 85 °C and RH = 80 %

Time-to-failure Ht (h) 429 451 476 484 493 495 501 512 521 526 534 540 542 548 557 576 579 586 618 645 531

ln(Ht) 6,061 1 6,111 5 6,165 4 6,182 2 6,200 5 6,204 6 6,216 6 6,238 3 6,255 8 6,265 3 6,280 4 6,291 3 6,295 3 6,306 3 6,322 6 6,356 1 6,361 3 6,373 3 6,426 5 6,469 3 6,269 2

Median rank 0,034 0,083 0,131 0,181 0,230 0,279 0,328 0,377 0,426 0,475 0,525 0,574 0,623 0,672 0,721 0,770 0,819 0,869 0,917 0,966

Group B Order number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Mean

Temp = 85 °C and RH = 70 %

Time-to-failure Ht (h) 613 640 649 675 679 696 703 709 719 732 739 743 747 751 766 778 785 804 856 896 734

ln(Ht) 6,418 4 6,461 5 6,475 4 6,514 7 6,520 6 6,545 3 6,555 4 6,563 9 6,577 9 6,595 8 6,605 3 6,610 7 6,616 1 6,621 4 6,641 2 6,656 7 6,665 7 6,689 6 6,752 3 6,797 9 6,594 3

Median rank 0,034 0,083 0,131 0,181 0,230 0,279 0,328 0,377 0,426 0,475 0,525 0,574 0,623 0,672 0,721 0,770 0,819 0,869 0,917 0,966

Table B.2 (2 of 2) — Log time-to-failure and median rank for example data Group C Order number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Mean

Temp = 85 °C and RH = 60 %

Time-to-failure Ht (h) 864 913 915 945 951 993 994 998 1 009 1 014 1 027 1 030 1 037 1 049 1 069 1 080 1 098 1 125 1 222 1 249 1 029

ln(Ht) 6,761 6 6,816 7 6,818 9 6,851 2 6,857 5 6,900 7 6,901 7 6,905 8 6,916 7 6,921 7 6,934 4 6,937 3 6,944 1 6,955 6 6,974 5 6,984 7 7,001 2 7,025 5 7,108 2 7,130 1 6,932 4

Median rank 0,034 0,083 0,131 0,181 0,230 0,279 0,328 0,377 0,426 0,475 0,525 0,574 0,623 0,672 0,721 0,770 0,819 0,869 0,917 0,966

Group D Order number １ 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Mean

Temp = 75 °C and RH = 80 %

Time-to-failure Ht (h) 1 728 1 882 1 907 1 989 2 020 2 076 2 129 2 151 2 180 2 227 2 277 2 318 2 352 2 404 2 443 2 512 2 589 2 590 2 776 2 891 2 272

ln(Ht) 7,454 9 7,540 3 7,553 4 7,595 3 7,610 6 7,638 1 7,663 2 7,673 9 7,687 1 7,708 5 7,730 8 7,748 4 7,763 2 7,785 0 7,800 8 7,828 7 7,859 2 7,859 4 7,928 6 7,969 5 7,719 9

Median rank 0,034 0,083 0,131 0,181 0,230 0,279 0,328 0,377 0,426 0,475 0,525 0,574 0,623 0,672 0,721 0,770 0,819 0,869 0,917 0,966

Group E Order number

Time-to-failure Ht (h)

ln(Ht)

Median rank

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

5 455 5 730 5 908 6 114 6 326 6 431 6 544 6 632 6 711 6 779 6 860 6 935 7 038 7 108 7 202 7 285 7 362 7 454 7 562 7 569 7 710 7 827 7 955 8 067 8 250 8 405 8 546 8 700 8 953 9 452

8,604 3 8,653 5 8,684 1 8,718 3 8,752 5 8,768 9 8,786 4 8,799 7 8,811 5 8,821 6 8,833 5 8,844 3 8,859 1 8,869 0 8,882 2 8,893 6 8,904 1 8,916 5 8,930 9 8,931 9 8,950 3 8,965 3 8,981 6 8,995 5 9,018 0 9,036 6 9,053 2 9,071 1 9,099 7 9,154 0

0,023 0,056 0,089 0,122 0,155 0,188 0,220 0,253 0,286 0,319 0,352 0,385 0,418 0,451 0,484 0,516 0,549 0,582 0,615 0,648 0,681 0,714 0,747 0,780 0,813 0,845 0,878 0,911 0,944 0,977

Mean

7 296

8,886 4

Temp = 65 °C and RH = 80 %

NOTE 1 Some tables in this document show values with many digits. Those digits are retained during the calculation in order to estimate the lifetime without introducing excessive round-off errors. However the resulting estimated lifetime is not intended to be quoted or relied on to the same high level of precision.

Step 3 The data can be plotted in different ways. If lognormal-graph paper is employed, the data is plotted with timeto-failure on the abscissa and median rank on the ordinate. NOTE 2 On most lognormal-graph paper, the actual ordinate scale is the probability of failure, and the median rank is converted to the probability of failure by multiplying by 100.

Figure B.1 shows lognormal plots of specimen groups A, B, C, D and E from Table B.2. The ordinate scale is the probability of failure. Each best-fit straight line is drawn through the plotted data. If the lines are judged to be reasonably parallel, the assumption of equivalent log standard deviation applicable to the individual data sets is verified. An estimate of the log standard deviation can be obtained from the graphical treatment of the failure data. First, for each stress condition, estimate the times corresponding to 15,9 % and 84,1 % failure based on the best-fit straight line through the time-to-failure data. The estimated log standard deviation σ̂ is then calculated as follows.

σˆ  ( lnt0,841  lnt0,159 ) / 2

Verify

that the fitting lines for all stress conditions are reasonably parallel to one another) Figure B.1 — Best-fit lines of specimen groups A, B, C, D and E on lognormal paper

The averaged log standard deviation estimate σ̂ m of the five groups is then calculated as

σˆ m  ( σˆ A  σˆ B  σˆC  σˆ D  σˆ E ) / 5  (0,1036  0,09759  0,09633  0,1407  0,1378 ) / 5  0,1152. Step 4 Table B.3 shows all 110 sample data points belonging to specimen groups A, B, C, D and E for regression analysis. The regression coefficients and error variance are calculated by applying the least-squares method. Table B.4 shows the result of regression analysis performed by the statistics software tool. Estimated variance 2 of residual errors σˆ lsm  , estimated log standard deviation σ̂ lsm and estimated regression coefficients

βˆ0 , βˆ1 and β̂2 are quickly obtained. Other statistics tools can also be used for regression analysis. NOTE

The estimated log standard deviation σ̂ lsm (=0,132 35) at the Controlled storage-condition is fairly large in

comparison with the averaged log standard deviation estimate σ̂ m of the five specimen groups. Variation in the best-fit lines among the five groups and the lognormal distributions of each group are among the anomalies that can affect the estimated log standard deviation.

Table B.3 — 110 sample data for regression analysis

Table B.4 — Regression analysis results Estimated log standard deviation

Estimated regression coefficients β̂0

β̂1

β̂2

σ̂ lsm

-35,381 1

15 789,57

-0,029 74

0,132 35

Step 5

lnBˆ 50 and lnBˆ 5 at the Controlled storage-condition (25 °C / 50 % RH) are obtained using the estimated regression coefficients βˆ , βˆ and β̂ and estimated log standard deviation σ̂ that were obtained in Step 4. 0

Then B50 Life, B5 Life and 95% lower confidence bound of B5 Life at the Controlled storage-condition (25 °C /50 % RH) can be calculated using lnBˆ and lnBˆ (see A.1.3). 50

lnBˆ50  βˆ0  βˆ1x10  βˆ2 x20 = - 35,381 1 + 15 789,57 х 0,003 354 - 0,029 74 х 50

= 16,090 1. B50 Life = exp (16,090 1) = 9 724 120 hours (1 110 years).,

lnBˆ5  βˆ0  βˆ1x10  βˆ 2 x20  1,64σˆ  lnBˆ50  1,64 σˆ lsm = 16,090 1 – 1,64 х 0,132 35 = 15,873 0 B5 Life = exp (15,873 0) = 7 826 297 hours (893 years).

The 95% lower confidence bound of B5 Life is therefore (B5 Life)L  exp((ln Bˆ 5 )L )  exp (ln Bˆ 5  z5 / 100 var(ln Bˆ 5 ) )  exp (ln Bˆ 5  1,64 var(ln Bˆ 5 ) ) = exp (15,873 0 – 1,64 х 0,021129 ) = exp (15,634 6) = 6 166 241 hours (704 years) (see E.4).

B.3 Data analysis and lifetime estimation using conventional acceleration-factor method (Step 4-7) Step 4 Table B.5 shows the average of log time-to-failure for each stress group A, B, C, D, and E (see Table B.2). Table B.5 — Average log failure time for each stress condition Temp

1/T

Group

Average

6,269 2

0,002 792

6,594 3

0,002 792

6,932 4

0,002 792

7.719 9

0,002 872

8,886 4

0,002 957

°C

NOTE The average in Table B.5 shows the average value of the ln(Ht)s of each acceleration condition in the mean row of Table B.2. Instead of such the average value the centre value of the median rank can also be used for log-meani in the following equation.

To determine the coefficients A, ΔH / k and B of the reduced Eyring equation, regression analysis is done using five average values obtained at the temperature values and relative humidity values in Table B.5 as log-mean i.

log-mean i  ln(A )  (

H 1 )  ( )  B  RH i  ε i k Ti

where i = 1 ~ 5.

The estimated values are determined as follows.

ln( Â )

 βˆ 0  35,688 9

Hˆ / k  βˆ 1  1 5 904,21 B̂

 βˆ 2  0,029 968

Step 5 The acceleration factors are calculated from the difference between the estimated log-mean at each stress condition and the estimated log-mean at the Controlled storage-condition (25 °C / 50 % RH). They are listed in Table B.6. Table B.6 — Calculated lifetime and acceleration factors for each stress condition

Stress condition Temp/RH 85 °C/80 % 85 °C/70 % 85 °C/60 % 75 °C/80 % 65 °C/80 % 25 °C/50 %

1/T 0,002 792 0,002 792 0,002 792 0,002 872 0,002 957 0,003 354

Calculated lifetime Ln (Lifetime) 6,320 2 6,619 9 6,919 6 7,595 7 8,946 7 16,155 7

Lifetime (h) 556 750 1 012 1 990 7 682 10 383 119

Acceleration factor 18 685 13 846 10 261 5 218 1 352

Step 6 Using the acceleration factors in Table B.6, calculate normalized time-to-failure at Temp=25 °C and RH=50 % for each specimen group A, B, C, D and E. Table B.7 shows data for a composite lognormal plot before sorting. Table B.8 shows data for a composite lognormal plot sorted in ascending order. Figure B.2 shows a lognormal plot using the composite data of Table B.8. The ordinate scale is the probability of failure. From the fitting line for those data, the log-mean ( μˆ acf  16,15 ) and standard deviation ( σˆ acf  0,132 4) can be obtained. These values are almost the same as the values that were calculated in Table B.7. NOTE

(σˆ acf )2 is the the estimated variance of population.

Table B.7 — Data before sorting for composite lognormal plot Time to Failure

Group

Normalized to Temp = 25 °C and RH = 50 %

Time to Failure

Group

Normalized to Temp = 25 °C and RH = 50 %

429 451 476 484 493 495 501 512 521 526 534 540 542 548 557 576 579 586 618 645 613 640 649 675 679 696 703 709 719 732 739 743 747 751 766 778 785 804 856 896 864 913 915 945 951 993

A A A A A A A A A A A A A A A A A A A A B B B B B B B B B B B B B B B B B B B B C C C C C C

8 015 865 8 426 935 8 894 060 9 043 540 9 211 705 9 249 075 9 361 185 9 566 720 9 734 885 9 828 310 9 977 790 10 089 900 10 127 270 10 239 380 10 407 545 10 762 560 10 818 615 10 949 410 11 547 330 12 051 825 8 487 598 8 861 440 8 986 054 9 346 050 9 401 434 9 636 816 9 733 738 9 816 814 9 955 274 10 135 272 10 232 194 10 287 578 10 342 962 10 398 346 10 606 036 10 772 188 10 869 110 11 132 184 11 852 176 12 406 016 8 865 504 9 368 293 9 388 815 9 696 645 9 758 211 10 189 173

15,896 933 3 15,946 943 7 16,000 894 2 16,017 561 2 16,035 985 5 16,040 034 1 16,052 082 4 16,073 801 0 16,091 226 4 16,100 777 6 16,115 872 2 16,127 045 5 16,130 742 3 16,141 751 6 16,158 041 6 16,191 584 0 16,196 778 8 16,208 796 1 16,261 964 8 16,304 726 7 15,954 116 6 15,997 219 8 16,011 184 4 16,050 464 4 16,056 372 8 16,081 101 3 16,091 108 6 16,099 607 2 16,113 613 0 16,131 532 2 16,141 049 6 16,146 447 7 16,151 816 8 16,157 157 3 16,176 933 8 16,192 478 2 16,201 435 4 16,225 350 9 16,288 022 0 16,333 692 1 15,997 678 3 16,052 841 5 16,055 029 6 16,087 290 5 16,093 619 6 16,136 836 2

1728 1882 1907 1989 2020 2076 2129 2151 2180 2227 2277 2318 2352 2404 2443 2512 2589 2590 2776 2891 5455 5730 5908 6114 6326 6431 6544 6632 6711 6779 6860 6935 7038 7108 7202 7285 7362 7454 7562 7569 7710 7827 7955 8067 8250 8405

D D D D D D D D D D D D D D D D D D D D E E E E E E E E E E E E E E E E E E E E E E E E E E

9 016 704 9 820 276 9 950 726 10 378 602 10 540 360 10 832 568 11 109 122 11 223 918 11 375 240 11 620 486 11 881 386 12 095 324 12 272 736 12 544 072 12 747 574 13 107 616 13 509 402 13 514 620 14 485 168 15 085 238 7 375 160 7 746 960 7 987 616 8 266 128 8 552 752 8 694 712 8 847 488 8 966 464 9 073 272 9 165 208 9 274 720 9 376 120 9 515 376 9 610 016 9 737 104 9 849 320 9 953 424 10 077 808 10 223 824 10 233 288 10 423 920 10 582 104 10 755 160 10 906 584 11 154 000 11 363 560

16,014 589 4 16,099 959 8 16,113 156 1 16,155 256 7 16,170 722 3 16,198 067 7 16,223 277 1 16,233 557 6 16,246 949 6 16,268 280 1 16,290 483 5 16,308 329 5 16,322 890 8 16,344 758 8 16,360 851 5 16,388 704 0 16,418 896 4 16,419 282 6 16,488 635 8 16,529 227 2 15,813 628 2 15,862 811 1 15,893 402 9 15,927 676 8 15,961 763 7 15,978 225 6 15,995 644 1 16,009 002 0 16,020 843 5 16,030 925 1 16,042 803 0 16,053 676 6 16,068 419 6 16,078 316 4 16,091 454 3 16,102 913 0 16,113 427 2 16,125 846 3 16,140 231 2 16,141 156 5 16,159 613 7 16,174 674 8 16,190 896 2 16,204 877 2 16,227 308 7 16,245 922 3

Table B.7 (continued) Time to Failure

Group

Normalized to Temp = 25 °C and RH = 50 %

994 998 1009 1014 1027 1030 1037 1049 1069 1080 1098 1125 1222 1249

C C C C C C C C C C C C C C

10 199 434 10 240 478 10 353 349 10 404 654 10 538 047 10 568 830 10 640 657 10 763 789 10 969 009 11 081 880 11 266 578 11 543 625 12 538 942 12 815 989

16,137 842 8 16,141 858 9 16,152 820 6 16,157 763 8 16,170 502 8 16,173 419 7 16,180 192 8 16,191 698 2 16,210 584 5 16,220 821 9 16,237 351 2 16,261 643 9 16,344 349 7 16,366 204 1

Time to Failure

Group

8546 8700 8953 9452

E E E E

Normalized to Temp = 25 °C and RH = 50 %

11 554 192 11 762 400 12 104 456 12 779 104 Mean Deviation

16,262 558 9 16,280 418 6 16,309 084 2 16,363 321 9 16,150 284 7 0,130 956 0

Table B.8 — Data sorted in ascending order for composite lognormal plot

Group

Normalized to Temp = 25 °C and RH = 50 %

Order

Median rank

Group

Normalized to Temp = 25 °C and RH = 50 %

Order

Median rank

E E E A E A B E E E B C A E B D A E E A A E B A C E C B E A E B C B A E

7 375 160 7 746 960 7 987 616 8 015 865 8 266 128 8 426 935 8 487 598 8 552 752 8 694 712 8 847 488 8 861 440 8 865 504 8 894 060 8 966 464 8 986 054 9 016 704 9 043 540 9 073 272 9 165 208 9 211 705 9 249 075 9 274 720 9 346 050 9 361 185 9 368 293 9 376 120 9 388 815 9 401 434 9 515 376 9 566 720 9 610 016 9 636 816 9 696 645 9 733 738 9 734 885 9 737 104

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36

0,006 3 0,015 4 0,024 5 0,033 5 0,042 6 0,051 6 0,060 7 0,069 7 0,078 8 0,087 9 0,096 9 0,106 0 0,115 0 0,124 1 0,133 2 0,142 2 0,151 3 0,160 3 0,169 4 0,178 4 0,187 5 0,196 6 0,205 6 0,214 7 0,223 7 0,232 8 0,241 8 0,250 9 0,260 0 0,269 0 0,278 1 0,287 1 0,296 2 0,305 3 0,314 3 0,323 4

B C A E C D C E B C E A C B A D B E A C C D B E D C E D C A E D E B D A

10 398 346 10 404 654 10 407 545 10 423 920 10 538 047 10 540 360 10 568 830 10 582 104 10 606 036 10 640 657 10 755 160 10 762 560 10 763 789 10 772 188 10 818 615 10 832 568 10 869 110 10 906 584 10 949 410 10 969 009 11 081 880 11 109 122 11 132 184 11 154 000 11 223 918 11 266 578 11 363 560 11 375 240 11 543 625 11 547 330 11 554 192 11 620 486 11 762 400 11 852 176 11 881 386 12 051 825

61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96

0,549 8 0,558 9 0,567 9 0,577 0 0,586 1 0,595 1 0,604 2 0,613 2 0,622 3 0,631 3 0,640 4 0,649 5 0,658 5 0,667 6 0,676 6 0,685 7 0,694 7 0,703 8 0,712 9 0,721 9 0,731 0 0,740 0 0,749 1 0,758 2 0,767 2 0,776 3 0,785 3 0,794 4 0,803 4 0,812 5 0,821 6 0,830 6 0,839 7 0,848 7 0,857 8 0,866 8

Table B.8 (continued) Group

Normalized to Temp = 25 °C and RH = 50 %

Order

Median rank

C B D A E D E B A E A A B C C E B E A C B B C D

9 758 211 9 816 814 9 820 276 9 828 310 9 849 320 9 950 726 9 953 424 9 955 274 9 977 790 10 077 808 10 089 900 10 127 270 10 135 272 10 189 173 10 199 434 10 223 824 10 232 194 10 233 288 10 239 380 10 240 478 10 287 578 10 342 962 10 353 349 10 378 602

37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

0,332 4 0,341 5 0,350 5 0,359 6 0,368 7 0,377 7 0,386 8 0,395 8 0,404 9 0,413 9 0,423 0 0,432 1 0,441 1 0,450 2 0,459 2 0,468 3 0,477 4 0,486 4 0,495 5 0,504 5 0,513 6 0,522 6 0,531 7 0,540 8

Group

Normalized to Temp = 25 °C and RH = 50 %

Order

Median rank

D E D B C D D E C D D D D D

12 095 324 12 104 456 12 272 736 12 406 016 12 538 942 12 544 072 12 747 574 12 779 104 12 815 989 13 107 616 13 509 402 13 514 620 14 485 168 15 085 238

97 98 99 100 101 102 103 104 105 106 107 108 109 110

0,875 9 0,885 0 0,894 0 0,903 1 0,912 1 0,921 2 0,930 3 0,939 3 0,948 4 0,957 4 0,966 5 0,975 5 0,984 6 0,993 7

Figure B.2 — Plot of composite data on lognormal paper Step 7 B 50 Life, B 5 Life and B 5V Life at the Controlled storage-condition (25 °C/50 % RH) can be calculated as follows. B50 Life = exp ( μ̂acf ) = exp (16,15) = 10 324 187 hours (1 179 years) B5 Life = exp ( μˆ acf - 1.64σˆ acf ) = exp (16,15  1,64 х 0,132 4) = exp(15,933) =8 309 118 hours (949 years) 2 The 95% lower confidence bound of the normal distribution with variation σˆ acf is - 1.64σ̂ acf .

Assuming that the population has a normal distribution, the population shifted by  1,64 σ̂ acf with the mean value μ̂acf - 1,64 σ̂ acf and the standard deviation σ̂ acf is considered to be the 95% lower limit of the normal distribution of the population. On this normal distribution, the point estimates of the 5 percentile is defined as "the point estimates of the 5 percentile with variation." It is calculated as follows: B5V Life = exp(lnBˆ5V )  exp(μˆ acf  1,64σˆ acf  1,64 σˆ acf ) = exp (16,15 – 1,64 х 0,132 4 - 1,64 х 0,132 4) = 6 687 348 hours (763 years). If the precise analysis or the precise interval estimation is required then the calculation based on maximumlikelihood method is recommended (see A.1.3 and E.3).

Annex C (normative) Disk-life estimation for Harsh storage-condition (Arrhenius method)

C.1

Stress conditions and data-analysis steps for Arrhenius method

Here, a test method is shown for the Harsh storage-condition at higher temperature and relative humidity than that of the Controlled storage-condition (25 °C and 50 % RH). This test method follows the scope in this document, which is based on an environment of 30 °C and 80 % RH representing the most-severe condition in which users handle and store optical disks. This test method also uses a different stress-test design that makes the use of the Arrhenius method possible. The same assumptions and data-analysis methods apply for the ambient storage-condition, stress design, and Arrhenius equation. The Controlled storage-condition of 25 °C and 50 % RH is replaced by an expected harsher user environment of 30 °C and 80 % RH. Table C.1 and C.2 summarize the stress design for the Arrhenius method. In case a stress condition would be destructive for the disk to be tested see Annex D. Table C.1 — Rigorous stress-condition for use with Arrhenius method

Test specimen group

Test stress condition (incubation)

Number of specimens

Maximum incubation sub-interval time

Minimum total incubation time

Intermediate relative humidity

Minimum equilibration duration time

Temp °C 85

RH % 80

300

1 500

RH % 30

400

2 000

600

3 000

800

4 000

h 5 7

Table C.2 — Basic stress-condition for use with Arrhenius method

Test specimen group

Test stress condition (incubation)

Number of specimens

Maximum incubation sub-interval time

Minimum total incubation time

Intermediate relative humidity

Minimum equilibration duration time

Temp °C 85

RH % 80

250

1 000

RH % 30

425

1 700

600

2 400

h 5

Regarding the data-analysis steps in Annex B, Step 4 is replaced as follows.

Regression coefficients and the standard error can be calculated using the least-squares method across all log time-to-failure data, which were obtained at the four or three stress conditions. This calculation can be performed by regression-analysis features of statistics software tools.

C.2

Data analysis

Step 1 and Step 2 For each stress condition, the specimens are ordered by increasing time-to-failure values. The median rank of the specimens is calculated using the estimate (i  0,3) /(n  0,4) . Table C.3 shows the result of ordered timeto-failure and median rank for the four stress groups A (85 °C), B (80 °C), C (75 °C) and D (65 °C) with relative humidity kept constant at 80 %. Table C.3 — Ordered time-to-failure and median rank for example data (Rigorous testing) Sample group and stress conditions (RH = 80 %) Sample number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Group A (85 °C) Time-toMedian failure (h) rank 429 0,034 451 0,083 476 0,132 484 0,181 493 0,230 495 0,279 501 0,328 512 0,377 521 0,426 526 0,475 534 0,525 540 0,574 542 0,623 548 0,672 557 0,721 576 0,770 579 0,819 586 0,868 618 0,917 645 0,966

Group B (80 °C) Time-toMedian failure (h) rank 1 015 0,034 1 040 0,083 1 080 0,132 1 203 0,181 1 151 0,23 1 165 0,279 1 193 0,328 1 215 0,377 1 230 0,426 1 239 0,475 1 260 0,525 1 295 0,574 1 310 0,623 1 425 0,672 1 360 0,721 1 388 0,770 1 420 0,819 1 472 0,868 1 540 0,917 1 625 0,966

Group C (75 °C) Time-toMedian failure (h) rank 1 728 0,034 1 882 0,083 1 907 0,132 1 989 0,181 2 020 0,230 2 076 0,279 2 129 0,328 2 151 0,377 2 180 0,426 2 227 0,475 2 277 0,525 2 318 0,574 2 352 0,623 2 404 0,672 2 443 0,721 2 512 0,770 2 589 0,819 2 590 0,868 2 776 0,917 2 891 0,966

Group D (65 °C) Time-toMedian failure (h) rank 5 455 0,023 5 730 0,056 5 908 0,089 6 114 0,122 6 326 0,155 6 431 0,188 6 544 0,220 6 632 0,253 6 711 0,286 6 779 0,319 6 860 0,352 6 935 0,385 7 038 0,418 7 108 0,451 7 202 0,484 7 285 0,516 7 362 0,549 7 454 0,582 7 562 0,615 7 569 0,648 7 710 0,681 7 827 0,714 7 955 0,747 8 067 0,780 8 250 0,813 8 405 0,845 8 546 0,878 8 700 0,911 8 953 0,944 9 452 0,977

Step 3 Figure C.1 shows the lognormal plot of groups A, B, C and D from Table C.3. The ordinate scale is the probability of failure. Best-fit straight lines are drawn through the data plotted for each group. If the lines are judged to be sufficiently parallel, the assumption of equivalent log standard deviations among the individual data sets is verified.

NOTE

Verify that the fitting lines for all stress conditions are reasonably parallel to one another

Figure C.1 — Best-fit lines of groups A, B, C and D on lognormal paper

Step 4 Table C.4 shows a total of 90 sample data values belonging to specimen groups A, B, C and D for regression analysis. The regression coefficients and error variance are calculated by applying the least-squares method to 90 failure data sets that were obtained under the four stress conditions. Table C.5 shows the result of regression analysis using a statistics software tool. The estimated log standard deviation σ̂ and estimated regression coefficients β̂0 and β̂1 are obtained.

Table C.4 — 90 sample data values for regression analysis Number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

ln t 6,061 05 6,111 47 6,165 42 6,182 18 6,200 51 6,204 56 6,216 61 6,238 32 6,255 75 6,265 30 6,280 40 6,291 31 6,295 27 6,306 28 6,322 57 6,356 11 6,361 30 6,373 32 6,426 49 6,469 25 6,922 64 6,946 98 6,984 72 7,092 57 7,048 39 7,060 48 7,084 23 7,102 50 7,114 77 7,122 06 7,138 87 7,166 27 7,177 78 7,261 93 7,215 24 7,235 62 7,258 41 7,294 38 7,339 54 7,393 26

x1 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 792 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832 0,002 832

Group A

Group B

Number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

ln t 7,454 92 7,540 28 7,553 36 7,595 32 7,610 63 7,638 06 7,663 17 7,673 91 7,687 12 7,708 53 7,730 83 7,748 37 7,763 20 7,785 04 7,800 85 7,828 69 7,859 16 7,859 35 7,928 61 7,969 48 8,604 29 8,653 47 8,684 06 8,718 34 8,752 50 8,768 89 8,786 36 8,799 66 8,811 50 8,821 63 8,833 46 8,844 34 8,859 08 8,868 98 8,882 17 8,893 57 8,904 09 8,916 51 8,930 89 8,931 86 8,950 27 8,965 33 8,981 56 8,995 55 9,017 97 9,036 58 9,053 22 9,071 08 9,099 74 9,153 98

x1 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 872 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957 0,002 957

Group C

Group D

Table C.5 ― Results of regression analysis Estimated log standard deviation

Estimated regression coefficients

β̂0

β̂1

σ̂ lsm

-36,321 5

15 304,74

0,161 52

Step 5 Using the estimated regression coefficients β̂0 and β̂1 and the estimated log standard deviation σ̂ in Table C.5, lnBˆ 5 and lnBˆ 50 can be calculated (see A.1.2). The B5 Life, B50 Life and the 95 % lower confidence bound of B5 Life at the Harsh storage-condition (30 °C and 80 % RH) are then obtained using the calculated values of lnBˆ and lnBˆ as follows (see A.1.3), 5

lnBˆ50  βˆ0  βˆ1x10 = - 36,321 5 + 15 304,74 х 0,003 298 7 = 14,164 25,

B50 Life = exp (14,164 25) = 1 417 280 hours (162 years),

lnBˆ5  βˆ0  βˆ1x10  1,64 σˆ lsm = 14,164 25 - 1,64 х 0,161 52 = 13,899 36, B5 Life = exp (13,899 36) = 1 087 462 hours (124 years).

The 95 % lower confidence bound of B5 Life becomes (B5 Life)L  exp(( lnBˆ 5 )L )  exp ( lnBˆ 5  z5 / 100 var(ln Bˆ 5 ) )  exp ( lnBˆ 5  1,64 var(ln Bˆ 5 ) ) = exp (13,899 36 - 1,64 х

0,016 598 ) = exp (13,688 1)

= 880 372 hours (100 years) NOTE

see E.3.

Annex D (normative) Alternative non destructive stress-condition

In case a stress condition is destructive for the disk to be tested, an alternative stress condition shall be applied. The stress conditions of Temp/RH used in this Ecma Standard are 85 °C/80 %, 85 °C/70 %, 85 °C/60 %, 80 °C/80 %, 75 °C/80 %, 70 °C/75 % and 65 °C/80 %. Among these stress conditions, the most severe temperature is 85 °C. In case the stress condition with temperature 85 °C is considered to be destructive for a disk to be tested regardless of the relative humidity, it should be replaced with 80 °C in all the stress conditions with 85 °C. Recommended alternative stressconditions for the Eyring and Arrhenius methods are shown in Table D.1, D.2, D.3 and D.4. Table D.1 — Alternative Rigorous stress-condition for use with Eyring method

Test specimen group

Test stress condition (incubation)

Number of specimens

Maximum incubation sub-interval time

Minimum total incubation time

Intermediate relative humidity

Minimum equilibration duration time

Temp °C 80

RH % 80

300

1 500

RH % 31

400

2 000

600

3 000

600

3 000

800

4 000

h 7

Table D.2 — Alternative Basic stress-condition for use with Eyring method

Test specimen group

Test stress condition (incubation)

Number of specimens

Maximum incubation sub-interval time

Minimum total incubation time

Intermediate relative humidity

Minimum equilibration duration time

RH %

Temp °C

RH %

250

1 000

250

1 000

500

2 000

625

2 500

Table D.3 — Alternative Rigorous stress-condition for use with Arrhenius method

Test specimen group

Test stress condition (incubation)

Number of specimens

Maximum incubation sub-interval time

Minimum total incubation time

Intermediate relative humidity

Minimum equilibration duration time

Temp °C 80

RH % 80

300

1 500

RH % 31

400

2 000

600

3 000

800

4 000

h 7 8

Table D.4 — Alternative Basic stress-condition for use with Arrhenius method

Test specimen group

Test stress condition (incubation)

Number of specimens

Maximum incubation sub-interval time

Minimum total incubation time

Intermediate relative humidity

Minimum equilibration duration time

Temp °C 80

RH % 80

250

1 000

RH % 31

425

1 700

600

2 400

h 7

The amount of water absorbed by the inkjet printing label due to the stress conditions far exceeds what would be experienced in normal use and operating conditions. A disk with a label layer for inkjet printing can show a large tilt due to the excessive moisture in the label layer. Thus it is recommended to use disks without the inkjet printing label layer for the test specimen. In case to use disks with the inkjet printing label layer that shows a large tilt by the excessive moisture for the accelerated-aging test, it is recommended to set the minimum equilibration duration time long enough, as an alternative stress-condition, so that the excessive humidity goes out from that inkjet printing label layer and the large tilt disappears. To leave the disks in the test environment for a while until the large tilt disappears may also be applied.

Annex E (informative) Interval Estimation for B5 Life using Maximum Likelihood

E.1

Lower confidence bound

Lifetime-estimation analysis (point estimation and simple interval estimation) for B5 Life and B50 Life are described in Annex A. In this Annex, a more precise analysis method for interval estimation is introduced. One may consider only the lower bound of the confidence interval to estimate lifetime. NOTE

E.2

The equations shown in this annex are for the case of complete data.

Maximum-likelihood method

To ensure that log lifetime ( y = lnt ) follows the normal distribution described in A.1.2, the likelihood function of parameters β and σ can be defined by the following equation. J

L ( β,σ ) 

 

f y ij | x i

j 1 i 1

   j 1 i 1

 1  y  x・ β  2    ij   exp   2 σ 2π σ       1

where J denotes the number of specimen groups, n j denotes the number of specimens in the specimen group j and σ is the standard deviation of the population. The log likelihood function is then J





1 lnL( β , σ )   ln 2π σ n j  ( y ij  ( β0  β1x1j  β2 x2 j )) 2 2 2 σ j 1 j  1 i 1 The maximum-likelihood estimators β and σ can be obtained by maximizing the second member of the equation. The estimates βˆ0 , βˆ1 and βˆ2 are coefficients in the multiple regression equation, and the estimate σ̂ is the standard deviation. The point estimation of lnB̂p can be obtained using the estimates βˆ0 , βˆ1, βˆ2 and σˆ as

lnBˆ p  βˆ0  βˆ1x1  βˆ2 x2  zp / 100 σˆ . Then the point estimates of 5 percentile and 50 percentile of the lifetime distribution are

lnBˆ5  βˆ0  βˆ1x10  βˆ2 x20  1,64 σˆ and

lnBˆ50  βˆ0  βˆ1x10  βˆ2 x20

where x10 , x 20 denotes the Controlled storage-condition ( 25 °C and 50 % RH ). For interval estimation of the population for lnB̂p of an optical disk, one may consider only the lower bound. Therefore, the (100  α ) % lower confidence bound of the log lifetime lnB̂p is given as

lnBˆ   lnBˆ  z p L

α / 100

var(ln Bˆ p ) ,

The equation lnBˆp  βˆ0  βˆ1x10  βˆ 2 x20  zp / 100σˆ can be modified as follows.

lnBˆ p  x p θ̂



x p  1, x10 , x20 , zp / 100

where





ˆ  βˆ , βˆ , βˆ , σˆ  . θ 0 1 2





Then var lnB̂p can be given as







var lnBˆ p  x p  var θ̂  x p



where var θ̂ is given by the inverse matrix of the Fisher information matrix as

  covβˆ , βˆ  covβˆ , βˆ  covβˆ , σˆ  var βˆ  covβˆ , βˆ  covβˆ , σˆ  var βˆ  covβˆ , σˆ  

var βˆ 0  ˆ var θ     



var σˆ  

and cov ( βˆa , βˆb ) denotes the covariance between βˆa and βˆb . As the variances of lnBˆ 5 and lnBˆ50 are represented by var (ln Βˆ 5 ) and var (ln Βˆ 50 ) respectively, the 95 % lower confidence bounds of lnBˆ 5 and lnBˆ 50 are given as follows,

lnBˆ   lnBˆ  1,64 var(lnBˆ ) 5 L

lnBˆ   lnBˆ  1,64 var(lnBˆ ) 50 L

where var ( lnΒˆ 5 ) and var ( lnΒˆ 50 ) can be calculated by the following equations.

var ( βˆ 0 ) cov ( βˆ 0 , βˆ1 ) cov ( βˆ 0 , βˆ 2 ) cov ( βˆ 0 , σˆ )  1      var ( βˆ1 ) cov ( βˆ1 , βˆ 2 ) cov ( βˆ1 , σˆ )   x10  var ( lnBˆ 5 )   1 x10 x 20  1,64   var ( βˆ 2 ) cov( βˆ 2 , σˆ )   x 20      var ( σˆ )   1,64 

var ( β0 ) cov ( βˆ0 , βˆ1) cov ( βˆ0 , βˆ2 )  1     var ( lnBˆ 50 )  1 x10 x20   var( βˆ1) cov ( βˆ1 , βˆ 2 )   x10   var ( βˆ 2 )   x20    Then the 95 % lower confidence bound of B5 Life is obtained as follows. (B5 Life)L  exp((ln Bˆ5 )L )  exp (ln Bˆ5  1,64 var(ln Bˆ5 ) )

E.3

Calculation method of Fisher information matrix and variance

By using the function ln L = ln L(β,σ ), the Fisher information matrix I in E.2 can be expressed as follows.

  2 lnL  β 2  2 0   lnL  β β I  E  2 0 1   lnL  β0 β 2   2 lnL   β0 σ

 2 lnL β0 β1  2 lnL  2 β12  2 lnL β1β 2  2 lnL β1σ

 2 lnL β0 β 2  2 lnL β1β 2  2 lnL β 22  2 lnL β 2 σ

 2 lnL  β0 σ    2 lnL  β1σ    2 lnL  β 2 σ   2 lnL   σ 2 

where, E(xi) is the expectation of xi. Components in the matrix I are as follows;

  2 lnL  n  E  2 2   β0  σ

J   2 lnL  1  E  n j x1 j  2  β0 β1  σ j 1

J   2 lnL  1  E  n j x2 j  2  β0 β2  σ j 1

  2 lnL  1 J  E  2 n j x12j 2   β σ j 1 1  

J   2 lnL  1  E  n j x1 j x 2 j  2  β1β2  σ j 1

  2 lnL   E 0  β1σ 

J   2 lnL  1  E  2 n j x 22 j 2   β σ j 1 2  

  2 lnL   E 0  β2 σ 



  2 lnL   E  0  β0 σ 

  2 lnL  2n  E  2 2   σ  σ J

where, n =  n j and it denotes the total number of specimens. j 1

Then I becomes,

 n  2 σ  J  1 n j x1 j σ2 j 1 I J  1 nx  σ 2 j 1 j 2 j   0 



1 n j x2 j σ 2 j 1 J 1 n j x1 j x 2 j 2 σ j 1 J 1 n j x 22 j σ 2 j 1



 0    0   0    2n  σ 2 



1 n j x1 j σ 2 j 1 J 1 n j x12j 2 σ j 1 J 1 n j x1 j x 2 j σ 2 j 1



By using the inverse matrix I-1, var ( lnBˆ 50 ) and var ( lnBˆ5 ) can be expressed as follows.

 n  σ2   1 J var ( lnBˆ 50 )   1 x10 x 20   2 n j x1 j  σ jJ1  1 n j x2 j σ 2 j 1 



J  1 n j x2 j  2 σ j 1  J  1 n j x1 j x 2 j  2 σ j 1  J 1  2 n j x2 j  2 σ j 1 



1 n j x1 j σ 2 j 1 J 1 n j x12j σ 2 j 1 J 1 n j x1 j x 2 j 2 σ j 1



 n  2 σ  J  1 n j x1 j  2 ˆ var ( lnB5 )   1 x10 x 20  1,64  σ jJ1  1 nx  σ 2 j 1 j 2 j   0 



1



1 n j x1 j σ 2 j 1 J 1 n j x12j σ 2 j 1 1 J n j x1 j x 2 j σ 2 j 1

1 n j x2 j σ 2 j 1 J 1 n j x1 j x 2 j σ 2 j 1 1 J n j x 22 j σ 2 j 1



1 x   10   x 20 

 0    0   0    2n  σ 2 

1  1   x   10   x 20      1,64

For the Arrhenius method temperature is the only variable and humidity is fixed, then the equation for estimation of the variances for the Arrhenius method are as follows:.

 n  2 σ var ( lnBˆ 50 )   1 x10   J  1 n j x1 j σ 2 j 1 



J  1 n j x1 j  2 σ j 1  J 1 2  n j x1 j  σ 2 j 1 

 n  σ2  J  1 var ( lnBˆ 5 )   1 x10  1,64  2 n j x1 j σ j 1   0  



 

 J

1 1    x10 

1 n j x1 j σ 2 j 1 J 1 n j x12j σ 2 j 1

 0

 0   0  2n  σ 2 

1  1   x   10   1,64

E.4

Example of variance calculation

Variances for the data in B.2 is calculated using Table E.1 to E.4 as follows: Table E.1 — Group data

Table E.2 — x1j and x2j

Table E.3 — Fisher-information matrix I

Table E.4 — Inverse matrix I-1

NOTE

The inverse matrix can be obtained by using a spreadsheet such as the EXCEL function "MINVERSE".

Then var ( lnBˆ 50 ) and

var ( ln Bˆ 5 ) are calculated as,

var ( lnBˆ 50 ) = [1, 0,003 354 016, 50] ·

 0,303 502 4   117,650 6  4,315 019E - 4

 117,650 6

4,315 019E - 4   1    39 119,189  0,254 791 9  0,003 354 016   0,254 791 9 3,958 656E - 6   50

= 0,020 915

var ( lnBˆ5 ) =[1, 0,003 354 016, 50, -1,64] ·

 117,650 6 4,315 019E - 4 0 1  0,303 502 4     117,650 6    39 119 , 189  0 , 254 791 9 0 0 , 003 354 016    4,315 019E - 4  0,254 791 9 3,958 656E - 6   0 50    0 0 0 7,962 056 E - 5   1,64   = 0,021 129. NOTE

The matrix multiplication can be calculated using a spreadsheet such as the EXCEL function "MMULT".

Annex F (informative) RSER measurement of BD disks

The Max RSER value of 10-3 was adopted as suitable for evaluating the time-to-failure in accelerated stress testing of BD disks. The ECC used for BD is powerful enough and has better error-correction capability than that of DVD at RSER = 10-3.[8][9] RSER excludes burst errors of length ≥ 40 bytes. But it is still affected by bursts shorter than 40 bytes. When measuring disks, manual handling of disks is inevitable. In order to avoid introducing short bursts, it is important to take care not to leave fingerprints on the surface of disks, especially before recording initial data. If the RSER should increase unexpectedly (especially near the outer edge of disk), it is recommended to wipe off any fingerprints and re-measure the RSER.

Bibliography

[1]

ISO/IEC 10149, Information technology — Data interchange on read-only 120 mm optical data disks (CD-ROM)

[2]

ISO/IEC 16448, Information technology — 120 mm DVD — Read-only disk

[3]

ISO/IEC 16449, Information technology — 80 mm DVD — Read-only disk

[4]

ISO 18927, Imaging materials — Recordable compact disc systems — Method for estimating the life expectancy based on the effects of temperature and relative humidity

[5]

Experimental statistics, US National Bureau of Standards Handbook 91, 1963

[6]

Applied Regression Analysis, Draper and Smith, Wiley Edition 2

[7]

Statistical Methods for Reliability Data, Meeker, Escobar, 1998, John Wiley & Sons Inc.

[8]

V. Bagdonavičius and M. Nikulin, Accelerated Life Models: Modeling and statistical analysis, Chapman & Hall/CRC, 2002

[9]

J. F. Lawless, Statistical Models and Methods for Lifetime Data, 2nd Ed., Wiley, 2003

[10]

W. Yamamoto, C. Kumazaki, and K. Suzuki, On Estimation of Archival Lifetime Distribution of Writable Optical Discs, Proceeding of the 21st Symposium on Phase Change Optical Information Storage PCOS 2009, pp.72-75, 2009

[11]

White Paper Blu-ray Disc™ Format BD-ROM 9th Ed., August, 2015. Available from http://www.bluraydisc.com/Assets/Downloadablefile/White%20Paper%20BDROM_Part1_9th_20150806_clean.pdfWhite Paper Blu-ray Disc™ Format BD-ROM 8th December,

[12]

White Paper Blu-ray Disc™ Format General 4th Ed., August, 2015. Available from http://www.bluraydisc.com/Assets/Downloadablefile/White_Paper_General_4th_20150817_clean.pdf

ECMA-396 — Test method for the estimation of lifetime of optical disks for long-term data storage (June 2017)

Related documents