Precision, accuracy, and reliability of disc quality (PI/PO/jitter) tests

vbimport

#1

This thread was prompted by some discussions about disc quality (PI/PO/jitter) tests in the thread “CD-DVD Speed Disc Quality (PI/PIF) Test”, and I hope that some better understanding of the significances and limitations in such tests may be gained through more dialogues. Contradictory opinions, occasionally even by the same person at about the same time, about such tests have been expressed in many places. Obviously much remains to be done. Of course each person is free to believe whatever he/she chooses to, but we are supposed to deal with science rather than religion at this website.<O:p</O:p

To begin with, I shall provide some basic information. <O:p</O:p

“Accurate” is a word frequently misused and abused. The following material is a must read for a proper understanding of the use of the words “accuracy” and “precision” and the distinction between them, if you don’t know yet:

http://phoenix.phys.clemson.edu/tutorials/ap/

<O:p</O:p<O:p</O:p
Definitions of some terms by ECMA 337:<O:p</O:p

PI error (PIE): “A byte error occurs when one or more bits in a byte have a wrong value, as compared to their original recorded value. A row of an ECC Block as defined in 13.3 that has at least 1 byte in error constitutes a PI error, in any 8 consecutive ECC Blocks the total number of PI errors before correction shall not exceed 280.”
<O:p</O:p
PI-uncorrectable, also referred to as PI failure (PIF): “If a row of an ECC Block as defined in 13.3 contains more than 5 erroneous bytes, the row is said to be “PI-uncorrectable”.”
<O:p</O:p
Jitter: “Jitter is the standard deviation σ of the time variations of the binary read signal. This binary read signal is created by a slicer, after feeding the HF signal from the HF read channel through an equalizer and LPF (see Annex D). The jitter of the leading and trailing edges is measured relative to the PLL clock and normalized by the Channel bit clock period.”
<O:p</O:p
For graphical presentations essential to a proper understanding of DVD jitter and its measurement by time interval analyzer (TIA):
<O:p</O:p
http://www.yokogawa.com/tm/ta/ta320/tm-ta320_09.htm<O:p</O:p


#2

None of these terms are appropriate for error scanning. Scans are what they are, and it’s all a matter of interpretation. In a nutshell, there’s no such thing as an “inaccurate” scan, assuming that the software is doing what it’s supposed to and sample rates are adequate.


#3

This may seem like splitting hairs to some people, but I both somewhat agree and disagree with the above statement:

First of all there are inaccurate scans from drives that show impossible values, such as NEC drives showing more than 1664 PIE per 8 ECC blocks, or more than 1664/208 PIF per 8/1 ECC block. There is no way such a result can be considered accurate because it is theoretically impossible!

Second of all I think there is a closely related concept which I will call a “useful scan”. Also the “precision” (using muchin’s linked article about accuracy and precision) or repeatability of scans differ between drives.

I define a “useful scan” as a scan that helps to discriminate between media that can easily be read in a number of drives and media that cannot easily be read in a number of drives. If the scans from a drive cannot be used for that purpose, then I deem it to be “not useful”. Usefulness doesn’t necessarily depend on “accurate” PIE, PIF, Jitter or other values, but it has to be somewhat precise or repeatable.

An example of a “not useful” scan would be e.g. scanning CDs (not DVDs) on a LiteOn SHW-1635S drive. Such scans can show extremely low C1 errors and a limited number of C2 errors, and the disc is still unreadable in almost all drives!

An example of a useful scan that is precise but not accurate is a Read Transfer scan, provided that it is performed on a drive that is not defective, too picky, or which reads everything. Read Transfer scans from my NEC 3500 have proven useful: It is repeatable (i.e. precise) and it is a good predictor for whether the disc can be read back on other drives. A Read Transfer test on a LiteOn drive might not be as useful, because LiteOn drives will often read discs without any drop in speed, where the same discs cannot be read in many other drives - hence it isn’t as useful.

Scans at 5x and 8x on my NEC 4551 drive (not necessarily all drives of this family) are not always useful, because they are not precise, and the drive or scanning software will sometimes randomly report extreme peaks of PIE and PIF errors. Scanning on the same drive at 12x seems more precise in practice with the limited number of tests I have done, and scans at 1x are even better (but very slow).

The point on which I agree with rdgrimes and many other people is, that the PIE and PIF errors are not on the disc per se, but are generated during the reading process. So it is wrong to think of scanning as a true measurement of a precise and accurate value on the disc itself.


#4

First of all there are inaccurate scans from drives that show impossible values, such as NEC drives showing more than 1664 PIE per 8 ECC blocks, or more than 1664/208 PIF per 8/1 ECC block. There is no way such a result can be considered accurate because it is theoretically impossible!

True, but not really relevant. There’s always exceptions, which only serves to further demonstrate my point.

An example of a “not useful” scan would be e.g. scanning CDs (no DVDs) on a LiteOn SHW-1635S drive. Such scans can show extremely low C1 errors and a limited number of C2 errors, and the disc is still unreadable in almost all drives!

Also true, but the disc IS perfectly readable in that drive.

Error scans do not measure ANY physical characteristic on a disc. One can make interpretations and assumptions about such physical qualities based on scans, but that’s all one can do.

A lot of the flap about which drives are “better” scanners is due to underlying misconceptions about the differences in what the different drives are reporting, and how. Erik Depp has done a wonderful job of trying to accomodate those differences in drives, but the nature of those differences MUST be taken into account when interpreting scans.

I go by my own experience, not what I read on forums. I’ve yet to see any reason to question the scan results I see with LiteOn drives. I use a CD burner for CD’s, and a DVD burner for DVD’s, and not once has a scan led me astray.
Of course, I don’t use LiteOn for burning either. :wink: (except for CD’s) A well-burned disc ALWAYS scans well in any of the Litey’s I’ve used.

Where the problems always arise with scan interpretations is with the discs that are not burned so well. Marginal quality discs will always confound scan interpretation. They will scan well once, and not the next time, or scan well in one drive and not another. Which is partly why I do not accept discs with anything but excellent burn quality. It’s either excellent, or it’s in the trash.


#5

An example of a “not useful” scan would be e.g. scanning CDs (not DVDs) on a LiteOn SHW-1635S drive. Such scans can show extremely low C1 errors and a limited number of C2 errors, and the disc is still unreadable in almost all drives!

No, actually it isnt! A ScanDisc shows unreadable sectors! :a

Here’s “proof” of what I talk about. This is why I say that CD Quality Scanning in a LiteOn SHW-1635S is not useful:


#6

It’s gonna be the most enthralling thread on this board for me :smiley: :smiley:

Thanks to the three of you for the great input so far. :clap: I’ll actively participate as soon as I have more time for very carefully formed posts.


#7

I agree fully here - a good burn on good media causes no problem. I would also add that scnning older disks ( 2 to 3 years old) produces some interesting results. The differences between one driv and another and one scanning program and another can become quite marked


#8

In years past many of us have often wondered if the introduction of error scanning was a mistake. There’s far more “wrong” applications of this stuff than “right”. I’ll wager that 90% of the folks doing it have no idea what they’re measuring or what it really means. For those 90%, a plain old transfer rate or scandisc test is probably more appropriate.

There’s really only one legitimate use for scans, and that is to compare disc A to disc B, both of which are burned in the same drive with the only difference being the burn speed or the media type. Any other use of a scan is suspect, and probably mis-guided. With the above stated purpose, it really doesn’t matter what the scanning drive is reporting, or how, only that it does so consistantly. In other words, the actual error levels or counts are NOT the main point. That disc B is “better than” disc A is the main point.

Use of error counts and averages for some sort of “benchmark” of relative burn quality is totally unsupported in the DVD specs. It can be used within the above stated example, but again only to compare disc A to disc B. And often even then it will be misleading. The exception might be where you are scanning only a set small range of blocks and comparing total counts. But this gets too technical, and far beyond the range of what the average end user would be doing.


#9

This is the second time in the past two days, that I’ve seen a comment like this… and It strikes me as odd, and I have to question what this is actualy suposed to mean, with regard to scanning as it’s being discussed here.

The other time a simmilar comment was made, was here… with regard to BenQ drives somehow showing theoreticlly impoissible numbers of errors.

http://club.cdfreaks.com/showpost.php?p=1245331&postcount=1207

Now, I know knothing about the NEC drives spoken of, but do own a BenQ 1640 drive that I’ve come to ‘trust’ to some degree… so talk of this kind of thing is interesting to me. If my BenQ is reporting the impossible, I’d like to know about it… NEC… well that’s a lesser concern of mine

Just exactly what is theoretcially impossible? is that theoreticly impossible at 16X, 8X, 4X, 1X, or all of the above?


#10

@rdgrimes :bow: :clap: Thanks for this input! I’ll keep the link handy! :bigsmile:

I find another use for scanning, maybe you’ll disagree. I use it to check for media “natural” degradation. I keep the original scan, then after some time (2 months, or 6, or 8…: the less reputable the media, the sooner I do it) re-scan the disc (with the same drive of course) and look for any significant increase in errors. Of course I dismiss small differences, and even small solid PIF peaks, that are most of the time due to tiny scratches and not to the dye degradation.

This way I started suspecting an issue with some G05 discs I had (this was before the G05 issue was widely spread on this board), and thanks to this method I could reburn them in time (these discs are now unreadable). I also discovered the unstability of many cheap MIDs (GSC, Infodisc, Multimedia…).

I even performed some “homemade” accelerated aging tests, mainly for fun as my method was rather sloppy (LOL), but scanning the discs put to the torture tests did teach me a lot. :bigsmile:

For CDRs, I also like to scan (with a Benq drive) to find the sweet spot for writing speed, based on jitter reporting.


#11

This was with the 1620, I think they’ve fixed it with the 1640 as I haven’t been able to reproduce this issue with it. But I started with f/w BSLB, so maybe early releases were still affected. :confused:

As for NEC drives, it gets worse with the latest family (4550/3550): PIEs reported can be in the stratosphere (600+) though all drives read the disc without a glitch. :rolleyes:


#12

Hopefully someone will give you the numbers in a short while - but basically the point is that for a given disk or part of a disk there are a maximum number of errors that can be reported - assuming that everythig was wrong. when a scanner reports more errors than are possible it should set off warning bells.

As to worry - I wouln’t. To my mind one of the best( if not the best) burners currently available is the BenQ1640. If you burn good media ( Taiyo Yuden unbranded 8X +R for me) at the best speed and with good software ( DVDdecrypter to burn ISO) you could probably forget scanning. I have not had a bad burn for 6 months to a year - I began using Tiayo Yuden about a year ago.

If, however, you like messin around then why not buy a Liteon1693S ? I would never use it to burn a DVD in anger but think it is a great scanner


#13

I think the real point is that if a burner reports greater than the theoretical max number of errors, yes that’s “suspect”, but the disc is still crap. So the drive was, in fact, correct in calling it a crappy disc. Once again, the actual numbers are not the main point.


#14

If you were familiar with scientific investigations, you would not have said so. And you have contradicted yourself, for you also cited ECMA for the definition of PIE in a thread started by you.

Scans are what they are, and it’s all a matter of interpretation. In a nutshell, there’s no such thing as an “inaccurate” scan, assuming that the software is doing what it’s supposed to and sample rates are adequate.
It seems that you are not aware that DVDInfoPro is not doing what it’s supposed to at least in some cases, information is in:

DVDInfoPro PIE value concerns”.<O:p</O:p

DrageMester has given an example to show that LiteOn 1635 is inaccurate for scanning CD. I have some related data: Even when a LiteOn CD (including combo) drive could not extract the tracks in audio CD correctly, it still reported very low C1 and no C2 errors. I shall talk about other cases later.<O:p</O:p


#15

Sorry - have read this thread and can’t see your point. using version 4.52 and comparing with DVD speed and Kprobe I have so far been unable to find any evidence to suggest any PIE concerns. This may have been a problem in the past I wouldn’t know

To my mind the attached scans are boringly similar ?





#16

If you were familiar with scientific investigations, you would not have said so. And you have contradicted yourself, for you also cited ECMA for the definition of PIE in a thread started by you.

I’m pretty sure that I don’t need you to tell me what I have said or not, or what I should say or not say. I also don’t see your point.
Perhaps if you would devote your time to reading ALL of a thread instead of just quoting what you think will support your own “cause”, you contribute more to this discussion.

It seems that you are not aware that DVDInfoPro is not doing what it’s supposed to

I’ve been aware of that for some years.

The point I’m trying to put froward is that it doesn’t matter what is considered “Accurate” by anyone. The purpose of scanning is not to acheive “Accuracy” in error reporting.


#17

Now I hope I can inject my opinion. In regards to the word ‘precision’, I would not use the word in regards to disc quality testing that is done on your home PC. The closest we would get to precise results would be in a controlled environment, with top of the line testing equipment, which is being done by the NIST.
The words accuracy and reliability would be more relevant to CD Speed. I have no experience with Kprobe so I can’t comment on that. But I have had conflicting results from CD Speed on the exact same disc.

Therefore my question;
What is the main purpose of CD Speed , if for example: I have a DVD copy, one of the first movies I copied, It plays absolutely flawless on the screen even after all this time.
Then I test it on the CD Speed and it comes out at a very low PIE count, so low in fact that I run it again just to see, but the second result almost doubles in the error count?.
Now I know that the disc is flawless and that I recorded on top media and kept it very good care, So why did I test it for in the first place?

The answer is because of all this frenzy about how the disc quality testing will show that our copies are deteriorating quickly. How do you Know, when the test results vary so much from software to software and drive to drive and test to test? Isn’t the bottom line that you record on good quality media, take great care in handling your copies, and that it plays like it did with the original?

So in my opinion, while these tests are useful to a degree, it is not the absolute measure of your DVDs. :wink:

Also,
Francksoy, I am offering a public apology to you because I seemed to have rubbed you the wrong way. That was never my intention, I apologize.


#18

Without the actual scans, it’s not possible to answer this question. Variations in scans are perfrectly normal. And, at lower error levels, variations of 2x are not unexpected. Once again, you are not measuring something on the disc.


#19

Thanks for your input, I didn’t save those test results, but maybe that accounts for the variation, And yes I understand the test results are what the drive is reading and reporting.


#20

What a scan is displaying is the activity in the 2 levels of error correction, or whatever that given drive is ABLE to report in that regard. It’s not reporting anything that’s being read, and in the strictest sense nothing is being read from the disc at all. It’s really just “going through the motions” of reading. Hope that makes sense.

In transfer rate and scandisc tests, the data is actually being read from the disc. It seems that the 2 shall never meet, drives cannot do both at once. If you think about it, that makes sense. Can’t have 2 different realtime datastreams heading down the cable at once.