Time Base Analysis
Is a fundamental measurement in the many areas of digital transmission and storage systems. It characterises the stability of digital signals and dictates a decoding system's ability to reconstruct the original waveform.
Jitter first appeared as a required measurement for optical disc in the Orange book. Subsequently it was amended to the Red Book. In the case of CD the measurement was data to data jitter, a many-valued affair relating to pit and land lengths as seen by a drive compared to their correct - or intended - lengths. Each of the discrete pit or land lengths was examined in turn yielding over thirty measurements. A critical component of the preceding statements is the fact that the measurement is dependent on the qualifier "as seen by a drive." This has been a recurring issue with the measurement of jitter, which instead of going away, has recently returned to the DVD arena with a vengeance.
The Red Book stipulated that the measurements specified must be performed on a single beam drive. This relates to the method of tracking and was the early standard for consumer players. Early tester equipment was based on the Philips CDM1 or CDM4 mechanisms, both single beam. However, by the time the jitter amendment to the Red Book was ready, most drives were triple beam. Research revealed that there was low correlation between single and triple beam measurements of jitter. The standard for jitter measurement became another Philips drive, the CDM12, which was triple beam; a solid drive that gave good results reflecting what a triple beam drive in the field would see. This becomes an important point. The pragmatic approach of adopting hardware that measures what drives in the field will measure crops up again later.
Data to data jitter had a drive dependency due to the method of measuring pit and land lengths in time. A drive contributed variations from its electronics and optics. This was one of the main reasons that drive selection was important to CD test equipment manufacturers: they had to try to constrain the range of variation by selecting the drives they used.
When the time came to define jitter as a measurement for DVD, two areas were addressed. Firstly, the thirty-plus measurements were replaced with a single aggregate parameter. Secondly, the measurement is with respect to a clock signal generated from the same data channel as the HF. This data to data jitter cares only about variations between a clock period generated from the HF and the actual data transitions. Couple these facts with an almost universal acceptance of the Pulstec SDP-1000 drive and you have reduced variation of measurements considerably. A lot of diagnostic information has also been hidden in an aggregate measurement, but for QC purposes you have one place to hang your hat.
This certainly simplified jitter conceptually but there was a physical aspect of DVD waiting in the wings ready to muddy the waters. Since DVDs have only 0.6mm of substrate as opposed to 1.2mm in CD, the light from the pick-up head travels half the distance through the polycarbonate. This makes DVDs more sensitive to warp in the disc. Tilt in the disc introduces coma or a change in the spot shape depending on the direction and magnitude of the tilt.
There were many discussions regarding the measurement of jitter. Should radial or tangential or both tilts be removed? Should neither be removed as that is what a drive would see, presuming it did not have any tilt compensation? In the end it was agreed that the radial tilt should be removed. The tangential tilt was not removed because physically this was much harder to do. The removal of tangential tilt would require dynamic head movement as the disc was rotating. Radial tilt or dishing at least remains constant as the disc rotates. So we have an approach where the jitter is measured in a way to characterise the quality of the pit structure excluding the radial tilt, but tangential tilt is accepted. This is the way the specification stands and is reflected in jitter measurements on test systems which characterise minimum jitter as the tilt optimised jitter, and another jitter which is not tilt optimised. The names given to these two jitters vary but one is usually called Optimised or Minimum Jitter, and the other System Jitter or just Jitter.
Things have remained stable on the jitter front until this year. In its efforts to reduce variation of measurements across the optical media industry, the ODMA began a campaign to generate calibration discs several years ago. Obviously this was quite a task for an organisation independent of the Forum. It had the universal support of the test equipment manufacturers who all wanted a standard they could work to. Eventually, as the ODMA came close to actually making the discs Philips stepped up and offered to do it. A huge help to the industry and not without some pain to themselves. As this is published, the first of the discs will be out in the field. At present the DVD-5 is finalised and later this year it will be joined by a DVD-9. This comes as some relief to the industry but does it signify the putting to bed of variation? Possibly not.
Philips has produced a smart, succinct little booklet to accompany these discs. In these booklets it describes how the measurements should be performed. This again is excellent help in attaining consistency across the industry. However on the page titled Operational Signals a single sentence could be a source of variation that is immune to normal attempts at calibration. The following sentence sits under the title of Jitter [2.7.1.a]: "The parameters radial tilt, focus, tracking and slicing offsets are optimised for each measurement." The effect of this sentence is to remove many physical aspects of the disc from the measurement of jitter. That is the measurement relates more to mastered jitter. In some ways the arguments for measuring in this manner are strong. When a consumer drive looks at a disc its job is to get the best out of that disc regardless of what the disc is like. Hence optimising focus, tracking or slicing is all part of the strategy. As far as I know radial tilt is still not optimised in all but the very high-end consumer drives; this technology has been around for some years on Laser Disc players but has not made it onto low cost drives.
The point is academic anyway as this kind of optimisation is done on HF amplitude which does not necessarily give the same optimum as the jitter. Indeed optimum in this context becomes a shifting point. When you have so many parameters to tweak to minimise a dependent value an iterative approach is the simplest method. It is well understood however that such simplex descending value methods find a minimum, not necessarily the minimum. Global optimisation is a huge field with techniques that you can fry an egg on. It is no simple thing. By considering what is happening from a physical point of view the results are also questionable. The radial tilt optimisation changes the spot shape to compensate for coma. The focus optimisation changes the focus point which, depending on whether you are looking at an angle or not may give the same value over quite a range. These two parameters alone interact with each other in a very complex way. The practical result of the optimisations is to increase the time of measurement for optimised jitter considerably.
A further issue worthy of mention is that for discs with optimum substrate thickness, good reflectivity and flatness, the optimised results equal the non-optimised. For discs that are warped, of low reflectivity, variable or wrong substrate thickness or bad space layer thickness, the optimisation improves the jitter. This destroys the QC aspect of the system where assistance is needed in getting to the bottom of such issues.
It is not clear whether this jitter, optimised in the described way, is to take over from the specification jitter. The name for the new jitter is Standard Jitter. So when faced with a jitter value you now need to ask is it:
Jitter or System Jitter
Optimum or Minimised Jitter