Why high-speed scanning is good / Why high-speed scanning is bad

There have been some discussion as to the value of using high-speed scanning for determining the "quality" of DVD media and burns.

Some say that using high-speed scanning is essential in order to discover problems that are not shown at low or medium scanning speeds.

Some others say that high-speed scanning is misleading because it tends to show problems with the scanning drive instead of problems with the media.

There's a good reason why people don't agree on this subject, and that's because both viewpoints are right...and both viewpoints are wrong.

I will show my best examples supporting both sides of this debate, and then the thread will be open for discussions.

For this discussion, I define "high-speed" scanning of DVD media as scanning at a scanning speed of 12x and above.

High-speed scanning is good because it can show some real problems with the media that are not necessarily shown at low to medium scanning speed.

The disc shown here is one of several that has a media flaw in the middle of the disc that causes mediocre results.

The scanning drives that most people trust, such as LiteOn, Plextor and BenQ don't show any problem in low to medium speed scans. Apparantly these drives are good enough readers to compensate for this media flaw at these speeds.

Some of the less trusted scanning drives, e.g. NEC, Optiarc and Pioneer do show problems at medium scanning speeds.

Some drives slow down when trying to read the disc at maximum speed, while others don't.

By picking the "wrong" scanning speed and/or scanning drive and the "wrong" drive for Transfer Rate Tests, the flaw in this media might not be discovered at all.

* BenQ DW1655 scans and TRT

* NEC ND-4551A scans and TRT

* Pioneer DVR-111L scan and TRT

...continued in next post...

...continued from last post...

* Optiarc AD-7173A scans and TRT (this drive was a very unreliable scanner, now it's dead)

* Plextor PX-712A scans and two TRT tests

...continued in next post...

...continued from previous post...

* Plextor PX-760A scans and TRT

* LiteOn SHW-1635S scans and TRT

...continued in next post...

...continued from previous post...

* LiteOn SHM-165P6S scans and TRT

...continued in next post...

...continued from previous post...

* LiteOn LH-20A1P scans and TRT

Please note that the KL05 firmware version is not so good for reading/scanning at high speed, and this has improved a lot in later firmwares for this drive.

This also serves to illustrate that firmware version in the scanning drive can play a part in how a disc looks when scanned. If the 12x scan in this drive had been the only scan, then the horrible PIE/PIF at the beginning of the disc would have made us believe this to be a really bad burn or disc!

High-speed scanning tends to magnify the impact of the scanning drive on the results, and this makes comparing scans in different units of the same drive model even less useful than when using low to medium scanning speeds.

So if you scan some media in your own drive at high scanning speeds, and compare the results to scans of similar media scanned in the same drive model by someone else, the differences can be much larger than if using low to medium scanning speed in the same drives.

That's a major reason why reviews at CDFreaks don't use high-speed scanning, because reviews are made by different reviewers with different drives (and sometimes different models), so comparing results between drives reviewed would be even more difficult if high-speed scans had been used instead of 4x scans.

But there are more reasons why high-speed scanning can be a bad thing...

Now let's see an example of why relying only on high-speed scanning can be bad.

The following disc has been burned in a drive that for some reason doesn't produce optimal results compared to many other drives when burning the same media (CMC MAG E01).

If 16x speed scanning in LiteOn drives had been used, with no low or medium speed scans or scans in other drives, the results would have looked very good.

For some reason the 16x scans in the LiteOn drives completely hide the less than stellar performance of this media burned in this drive, while any lower scanning speed and scans in other drives show this to be a relatively poor burn. Since CMC MAG E01 media tends to be easily readable, the disc plays flawlessly in my most picky DVD player and it reads without problems in DVD drives despite poor PIE/PIF scans.

* BenQ DW1655 scans and TRT
* Plextor PX-760A scans
* Optiarc AD-7170A TRT

...continued in next post...

---

Last edited by DrageMester; 23-10-2007 at 12:45. Reason: typo

...continued from previous post...

* LiteOn SHM-165P6S scans
* LiteOn LH-20A1P scans

Notice how all the scans basically agree that this is a poor result, except the 16x scans which show this disc to be good (20A1P) or excellent (165P6S)!

I have demonstrated with some examples why high-speed scanning can be good and why high-speed scanning can be bad.

Please comment and discuss on these findings, and please add your own examples of why high-speed scaning of DVD media is good or bad.

Holy crap the high speed scans on that bad E01 showed some ridiculously good results! I have also experienced this with E01 media. I've had spikes of 300 PIE and large PIF simply disappear into thin air when 16x scanning. Funny thing is though, that some other drives, just like yours, showed the problem @ all speeds. Only the Liteons seemed to miss it @ 16x scanning.

I enjoyed your examples, Drage, very clear and thorough.

I might another post with some scans later in the same format as you did.

Excellent once again Dragemester

I use 16X scanning in the 20A1P as part of my testing and I find it useful.
I would like to add an observation about the sometimes great results at 16X speed. (Unfortunately I cannot post my test results as my PC at home has been creamed by an electrical fluctuation and it will take me some time to replace the ruined components.)
I have some discs on which I ran repeated 16X DQ test with the 20A1P KL0N firmware. I was very careful to prevent disc temperature being an issue. The results show variance in PIE totals of as much as 10 times. That is, on one test the DQ result show 20,000+ PIE. In a repeat test, the same disc can show more than 220,000 PIE. The shapes of the graphs are very similar.
The variance is repeatable.
I not sure, but a possibility is the dropped samples. If the dropped samples at 16X are changing considerably from test to test, this could account for the large variance.
Another observation with the 20A1P, but this is only my own intuition, the drive seems more likely to drop samples with high PIE than samples with low PIE. So even comparing PIE average as shown in CDS can be misleading.

The number of samples tested vary very little in my experience (< 5%) when using the same scanning speed and the same system setup, so I don't think that's the explanation.

Heat is a big factor (which you have already taken into account) and others might be focusing, tracking, vibrations, how the disc is clamped etc.

I also sometimes see big variations, but that's the exception and not the rule.

The same is true for all LiteOn drives; more samples are dropped at higher speeds.

It's possible to normalize averages as explained here.

Dragemester
I respect your views highly but I believe you misunderstood what I was trying to say.
I agree that the number of samples tested vary very little from test to test, but what I meant was which samples are dropped are apparently changing from test to test and this seems to affect(significantly) the number of PIE reported at 16X speed. It's curious that the PIF totals vary very little from test to test in comparison. I tried to check this by flashing back to KL0G(as it drops more samples) and comparing changes in PIE totals by the two firmwares in relation to the number of samples dropped. I hope that makes some sense.
In practice, I now usually run at least two 16X DQ tests with the 20A1P on each disc because the variance is occurring frequently. Other brand drives may not have this attribute at 16X, particulary if the number of dropped samples is much less than the LiteON.

Dragemester,
Thanks again, even if I had read that thread previously.
It is possible you misunderstood me again(my English must be decaying, living in Thailand and speaking Thai most of the time.)
I think the post in the link you provided above is assuming that the average PIE in the dropped samples is the same or close to the average PIE in the reported samples. My experience is that this very often not the case. And that is what I meant in crying about the reported averages.

Yes, the normalization calculation assumes that dropped samples follow the same error distribution as the collected samples. This is a fair assumption if there's no evidence to the contrary.

My guess is that the variations you're seeing is the result of normal variation between the collected samples and don't have anything to do with which specific samples are being dropped. That's the simplest theory explaining your observation.

Other drives that hardly drop any samples at all (e.g. BenQ, Plextor) also show variation between scans.

Exceptionally stepping in to applause your research and input on the subject, Dragemester.

I concur with about all of your points cause as you know I came to very similar findings.

Bottom line for me is, using a single scanning speed or worse, a single scanning drive, to perform PIE/PIF/POF/jitter tests, leads to meaningless data and flawed conclusions in a fair number of instances. I always test a burning combination (blank+burner+firmware+burning speed) in several drives at several speeds before considering this combination as better or worse than another combination. Relying on a 1-scan-1-drive method, never again for me. It's good only to compare burns when one has determined the best combination first.

Stepping out now.

Francksoy,
I agree completely.

Dragemester,
It's a shame I can't post my results in this area at this time, because the results appear to indicate that on some discs, that assumption doesn't hold.
It's not an easy thing to check. But perhaps my approach was incorrect?
I was comparing the difference in PIE totals as contributed by the difference in the number of samples dropped by KL0G vs. KL0N firmware. It seems about the only way to check that assumption.
Perhaps it is an unimportant detail to worry about?

I think I might remember something about KL05 and perhaps KL0G dropping more samples than later firmwares, but I'm not certain and I don't think I have saved any scans showing it. Even if one firmware is dropping more samples than another firmware, it's still a fair assumption that random samples are being dropped instead of specific samples. Some sort of evidence that the dropped samples are not random would be needed (Occam's Razor principle).

Thanks, DrageMester, for the balanced opinions. I'm glad to see that these have been received without raised tempers.

For myself, I have found no problems with using a single speed scan on a single drive. I have never actually found that transfer tests have told me anything about the long term playability of the disc. After doing multiple tests over the years I found that consistency was more important than additional variables, especially when it is so difficult to determine exactly what we are measuring when we look at the scans.

My original burns from the good old days still play and scan the same as before, but then, I only play discs at 1X and I have narrowed down to Verbatim and the good TY so some other problematic media may need more testing to determine its quality or lack thereof. With coupons and sales, I just see no need to go digging for "bargain" media that might take more time to burn and scan than to watch.

After all, none of what we do is really up to lab specs so in the end it all comes down to what works or what we enjoy doing.

chas0039,
I second that to a point.
Dragemester,
I agree that "randomness" is sort of maintained with any of the firmware versions. I agree in the sense that I don't think it's possible for the firmware to know in advance which samples have high PIE or have any other decision mechanism which deliberately drops samples with high PIE. It's too far-fetched.


Ah, there is the rub. Evidence, when apparently no one in this forum can explain how samples are dropped by any drive.
The discs I have burned which show the highly variable PIE results at 16X scanning speed were all burned by the LiteOn at 16X. None of my discs burned at 8X show this high(factor 10) PIE totals variation when scanned at 16X. The discs are all Verbatim (MCC004 MIT and T03). I consider all of them from good(not great) spindles.
Despite all I have said above, I still believe it is possible for the drive/firmware to have a tendency to drop samples with high PIE on discs burned at 16X, and as a result skew the avg. PIE on those discs.

Thanks DrageMester for laying this out so nicely.

A while back I started a discussion thread about a small media database project of ours, but it only lead into a few high speed scanning evangelists preaching how 4x CLV scans were all so wrong and faulty.

I concur with the opinion that CLV scanning is the way to go. Plextor preferred over LiteOn's and others, if possible. The idea is to analyze the quality of the recordable surface and for that we need the CLV. Of course the scanning only represents the only the single combination of the disc/recorder/scanner.

Inclusion of TRT tests or high speed scans can add variety to the results, but as said high scanning speed has an impact on the results, and it may tell more about the drive than the media. At very high speed the vibrations become an issue and even the way that the drive is bolted to the PC case has an impact -- I once had an 18x installed to a PC with one of these "no-tools-required" snap-one drive slots, and the drive had constant problems at 16+ X speeds in reading and recording, as the hardware installation was not rigid enough.

DVD recorders are damn could readers in general. Sometimes it sees that they will read a slice of pizza just fine, and as such they often give smooth TRT curves even though the media would be less than perfect. But it is always good to verify that the disc is readable and there are no slowdowns, as if such occur then your data is at immediate risk. Using a wider variety of drives of course helps to bring certainty -- I used to have a very old Samsung DVD-ROM, which was neat for testing as it was extremely media picky.

The MediaBase, with it's obvious limitations, tries to tackle the issue by giving the chance of collecting more data and variations of different combinations of drives/media/scanners, which would help in seeing the "big picture" of the quality that the media is delivering.

I am in the process of conducting a "Road Test" of the new Pioneer DVR-115D. My "Test Kit" includes a BenQ DW1640, Plextor PX-760A, a LiteON LH-18A1P@DH20A4P, and of course the test drive - the Pioneer DVR-115D. Here are my findings so far in regards to scanning abilities of each drive.

1. The BenQ DW1640 is very jitter and beta sensitive and therefore makes a great parity scanner. However it consistantly reports jitter levels that are +/- 1% higher than the LiteON. Note that the BenQ scans PIF @ 8ECC.

2. The Plextor PX-760A is also very jitter and beta sensitive and again, makes for a great parity scanner. If a disc has high or wildly fluctuating jitter or beta, quite often the PI error rate resulting from a PlexTools Q-scan is considerably higher than any of the other two drives I'm using.

3. The LiteON, even at 8X CAV, is nowhere near as sensitive as the other two drives, especially with regards to PI errors. While scanning at 1ECC, the location of PI failures across a disc largely agrees with scans performed by the other drives. I have found that there is little difference between scanning at 4X CLV and 8X CAV.

4. The LiteON is far too lenient for CD scanning, even though the unit's CD error correction appears to be only average. The BenQ and Plextor are much better for this task.

5. The Plextor is very unforgiving when it comes to CD scanning. C2 and CU errors will almost definitely show up when scanning problematic discs.

Regards,
TerminalVeloCD

This thread shows why high-speed scanning can be bad but also why it can be good, so you're not really concurring with my opinion by saying that CLV scanning is the way to go.

Low-to-medium speed CLV scanning makes it easier to compare scans between drives, but it also tends to paint a prettier picture than can be seen if scanning at higher speeds.

That's why low/medium speed scanning and high speed scanning both have their place in a quality testing toolbox.

I prefer LiteOn scanning over Plextor scanning for everday DVD testing, because Plextor scanning is too slow IMO. In my experience Plextor DVD drives are more picky when scanning, so I'd prefer a good scan in a Plextor drive over a good scan in a LiteOn drive, but I rarely perform such scans due to the required scanning time (1½ hours for a full PIE/PIF/POF/Jitter/Beta scan at 2x CLV).

I find that for some discs my BenQ DW1655 shows higher jitter than my LiteOn drives and for some discs it's the other way around. There can also be a difference in reported jitter level for the same disc scanned in the same drive at the same speed, and also a general difference in jitter levels reported by two drives of the same model.

My Plextor drives are in general more critical when reporting PIE/PIF than my other scanning drives (except for the completely flaky NEC/Optiarc drives which I don't trust for scanning DVD media).

I find that 4x and 8x scans are mostly, but not always, similar. By also using high-speed scanning on my LiteOn(s), I can get a more sensitive scanning result from the LiteOn.

LiteOn DVD burnsers (at least recent series) only report C1=E31 instead of reporting C1=BLER=E11+E21+E31 so they are not well suited for CD scanning.

My experience is the same, although for some discs my BenQ DW1655 will be even more picky when scanning (Advanced Disc Quality).