Problems
Important process and procurement decisions are frequently based on product quality as defined by test data, but what is the quality of that data. Tester A fails the disc but tester B passes it. Why? Which result is correct? is a frequent complaint indicating that correlation practices are either absent or flawed. This is a critical problem because accurate test results are a vital part of high quality and low cost. Bad data results in flawed decisions, often while engineers and managers are unaware of the limitations of their information.
Although the industry accepts that frequent media testing is necessary, users neglect regular evaluation of test equipment, a critical element of media testing. All testers are wrongly assumed to be accurate, resulting in unpleasant surprises when results from various test systems disagree.
Causes
Different hardware, software, and test procedures all generate variances. Although users need not become experts in test equipment design and construction, they should be knowledgeable about the principles so that they can intelligently select test equipment and then implement the necessary correlation procedures.
Optical paths in test equipment usually consist of a complex assembly of lasers, gratings, mirrors, beam-splitters, lenses, polarizers, waveplates, detectors, and servo actuators. Each element has manufacturing tolerances, and their alignment introduces additional variations. Consequently, two identical testers often generate different results. Although the focused spot is ideally circular, this is rarely achieved in practice, and distortions can vary from unit to unit. Operating or ambient temperature changes also introduce additional complications. More variations are introduced by design differences. For example, holographic and discrete element optics can give different results, as can polarizing and non-polarizing beamsplitters. Optics designed for writers may have push-pull radial tracking methods and higher numerical apertures that differ from specified test methods.
Drive components used for radial tracking, axial focus, spindle speed, data recovery, and error correction in test drives may not conform to public and private standards (ISO, DVD Forum, Philips, Sony, etc.) Modified radial, axial, and data recovery servos differ in their response to defects. Systems may use electro-mechanical methods to correct for tilt, or electro-optical techniques that compensate for tilt or spherical aberration. High speed testing (above 1X) creates additional challenges when test drive servos react differently to defects.
Calibration is another source of variation. Too frequent calibration can introduce variations. Calibration discs provided by the equipment supplier may not agree with those from master laboratories (Philips, Pioneer, Sony etc.) A different risk arises from single point calibration. Figure 1 demonstrates that inaccuracies can occur away from a calibration point in nonlinear test systems, causing false results near specification limits.
System nonlinearities can only be detected by evaluating special test discs having a wide range of quality characteristics. Figure 2 shows that very useful test discs have parameters that approach or even exceed specification limits.
Test software differences introduce variations. Parametric measurements are usually averaged over a pre-established distance or time, and results from software that uses one second averaging may differ from those using ten seconds, one revolution averaging, or that average over a one mm radial band. Error test results are also affected by the measurement interval; DVD inner parity failures (PIF) differ when one system reports errors from one ECC block and another sums over eight consecutive ECC blocks. Nomenclature can also cause confusion when equipment manufacturers fail to use the same label for a test, or use terms that differ from public or private standards.
Uncontrolled hardware and software changes, measurement techniques, and test limits can be sources of poor correlation. Test results are also affected by handling methods that contaminate or damage test samples and distort test results.
Tester Types
Electrical testers use an accurately focused spot to evaluate pit geometry, defects, and track quality. As many as forty different quality indicators can result from simultaneous measurements at various drive test points, creating a major challenge for comprehensive correlation tests.
Optical testers use unfocussed, collimated laser beams that do not resolve individual tracks. Off-line testers initially evaluated only birefringence, but now may include transmission, reflectivity, radial and tangential angular deviation, and some axial tests, and also may measure various layer thicknesses. In-line optical testers conduct 100% screening for physical defects such as black spots, bonding gaps, and cracks, and some evaluate reflectivity. These important functions require correlation of optical test systems.
Mechanical tests include unbalance and various dimensional measurements. The latter should be correlated with national or regional standards organizations. Unbalance testers normally evaluate dynamic unbalance, while media standards specify static unbalance limits, resulting in a correlation issue.
Correlation Procedures
Three types of correlation tests are necessary. The first type, accuracy, qualifies calibration discs and methods by evaluating reference discs from master laboratories (Philips, Pioneer, Sony, etc.), while non-linearity problems are identified using reference discs that have a wide range of values for each quality indicator. Reproducibility is the second type that detects test system instabilities or drifting with temperature or time.
The third type is playability that ensures the proper identification of media flaws. The complexity of test systems, especially electrical testers, can result in adverse interactions between different test paths in the presence of defects. For example, misleading high error rates and jitter shown in Figure 3 are not related to pit geometry but instead are caused by a combination of high eccentricity and disc unbalance. Surface defects, such as long tangential scratches, result in high PIF and POF errors shown in Figure 4 that may not be reported by other testers that have different drive servos.
Playability tests are critically important, but are often overlooked either because of unawareness or because evaluation is time consuming. Confidence in test results can be achieved only when discs having a wide variety of known pit geometry, physical, optical, and mechanical flaws are evaluated to assure that test systems identify only the actual flaw and do not generate false results.
Objectives
In the ideal state, all test methods would conform to public and private standards (ISO, DVD Forum, Phillips, Sony, etc.), test results would agree with those of master laboratories (Philips, Pioneer, Sony, etc.), linear test systems would maintain accuracy over the full measurement range, data would be precisely repeatable, and test results would identify only the actual defect. Although this perfect state cannot be realized because of equipment limitations, reasonable conformance to accuracy, repeatability, and playability requirements can be achieved, enabling managers to avoid false alarms by guard-banding test limits. These requirements should be established pre-purchase using either values supplied by the test equipment manufacturer or generated by the purchasing organization and accepted by the supplier. Only test equipment that conforms to these requirements should be used for media testing, providing the using organization with reliable data that accurately reflects product quality, not tester limitations.
Solutions
Focus on test equipment correlation. Do not be distracted by equipment design and manufacturing details. Realize that many accuracy, reproducibility, and playability tests are required in order to avoid correlation problems. Do not be misled by results from just one or a few test discs. Calibration discs from an equipment manufacturer are useful but do not prove accuracy, linearity, and playability Comprehensive evaluation using many different test discs is necessary. Although time consuming, it is more effective than constantly fighting problems created by false data.
Initial acceptance testing to pre-established specifications is an obvious, but little used, method that avoids future problems. Ongoing correlation tests are necessary because test systems can and do degrade with use, and flawed systems must be identified and then repaired or replaced before their incorrect results adversely affect product quality. Acceptance tests must again be conducted when repaired or new components are received.
Consistent measurement techniques are necessary if correlation is to be maintained. Test procedures should be defined by controlled documents and verified by periodic audits. Test limits must be centrally controlled, and not vary between testers. Maintain logs that document calibration and correlation discs, hardware, software, and correlation results. Carefully handle test samples and correlation discs to avoid damage or contamination that could affect test results. Do not use or attempt to clean or repair damaged correlation discs. Instead replace them and correct the handling methods that caused the damage.
Although quality departments or test equipment suppliers may be capable of performing some correlation tests, use of an independent laboratory provides the advantages of independently certified test discs, time and cost savings, and confidence in the results. Media Sciences has more than twenty years of experience in correlating optical and magnetic media testers, both for internal use and as a service to other facilities. Certified test discs, manuals, training, and in-house correlation testing provided by Media Sciences. Media Sciences result in a high level of confidence in test systems that are an essential part of high process and product quality.