Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: KProbe2 and reality (Read 6678 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

KProbe2 and reality

Several month ago, when my mail order Mitsui supply was running low, I bought a pack of Sony blanks at a local store. I've been using them mainly on things when I'm less concerned about how long they will last.

Recently I got a Lite-On CD-RW drive so I can run KProbe2 to monitor disk quality. I tested a number of disks from back to the time I started using CD-R and I frequently run a test after writing a new disk.

I wrote about 70 minutes of music onto one of the few remaining Sonys. It tested at 4200 C1 errors, about ten times what I've seen before. I pulled out some disks written a few months ago. The next one tested even worse, the third one tallied 42,000 C1 errors, almost 13 average,  and half a dozen C2 errors. Its length was 50 minutes and the graph seemed to show a very even distribution of (many) errors across the entire surface.

I ran a couple Mitsui from the same several months ago time. They tested the normal few hundred total C1 errors, 0.1 to 0.25 Average. Apparently the drive isn't blown. One more Sony, from that same time period: its managed the normal low numbers.

Returning to the first 70 minute disk, I extracted the tracks with EAC in burst mode, test and copy, not wanting to spin the disk for hours in secure mode. Everything came off without complaint or delay and all checksums matched. I selected a long track to do a mix-Paste Inverted in CoolEdit against the source file. It was a perfect match, a complete cancel.

I verified the 42,000 errors disk with KProbe2 again, then took that disk to the living room. I put it into the player that fails miserably on level 2, as soon as the test get above the standard manufacturing errors (for those who read the CD-Check thread). I sat down and listened through so I would be less likely to miss anything. Of course I don't know what anyone else might have heard, but I detected nothing amiss, certainly no clicks or pops from mis-reads. Music on that player always sounds good.

I understand that C1 errors are correctable. In various places the opinion is expressed that audio quality suffers from these errors even if they are corrected. I'm reasonably sure that opinion is not terribly meaningful, but I would not have been surprised if there was some limit on tolerance. If there is, it appears to be way beyond what normal discussions about testing might suggest.

KProbe2 and reality

Reply #1 –
Beyond a certain point of error correction (C2 I would imagine), a player/ripper can opt to do -interpolation-, which gracefully degrades the audio quality. CIRC was designed to give that ability so that CDs would not have annoying pops and stuff when playing.

KProbe2 and reality

Reply #2 –
the scannigsmeans alot.

but you have to consider tat you are only testeting the media

- NOW
- in this driver
- at this speed

different drives (read & write), media & speeds will ave different errors numbers
Sven Bent - Denmark

KProbe2 and reality

Reply #3 –
Sure, different readers, different media. However, this reader does fine on most media, as evidenced by most CDs put into it. This reader even does fine on some disks out of that one pack. But, there should be bit of a difference between 327 errors and 42,000 errors -- if the test really means anything. If the error count doesn't differentiate anything meaningful, what does it tell one that is worth knowing?

KProbe2 and reality

Reply #4 –
Slightly OT with regards to your original post but to answer the question as to if it really means anything, it may potentially give a vague idea of the lifespan of the disk.  The higher the error count, the sooner the disk will become unreadable by the majority of drives, etc.

KProbe2 and reality

Reply #5 –
I know that longevity is a tenant of the ‘official' propaganda. I guess I am in a position to investigate that since I have a number of disks from the same package, written on the same writer, at the same speed, in the same time period. I can check them again in a year, two years, etc. and see if the numbers seems to mean anything more over time. Running that disk again I got
C1 Max                  : 53
C1 Average              : 15.31
C1 Total                : 41839
C2 Max                  :
C2 Average              : 0.00
C2 Total                : 2
Another CD written three days after that one tested
C1 Max                  : 7
C1 Average              : 0.20
C1 Total                : 481
C2 Max                  : 0
C2 Average              : 0.00
C2 Total                : 0
I also tested several others dated both a little before and a little after. Numbers vary but all are close to what I've come to expect as "normal" rather than being like the "bad" one above or the 4000+ error one that initiated this consideration.

I extracted from that high error disk with EAC in burst mode. Speed varies across the disk but gets up to about 42X. Checksums matched (Test & Copy). The source for that disk exists in Monkey's Audio form so I have a reference for comparison. I chose the longest track to do a Mix-Paste Inverted test. Resut: another bit perfect match.

Thus this disk's error rate -- two orders of magnitude greater than most others -- does not effect playback, even on a CD player that tests as marginal in error correction ability (according to CD-CHECK). It does not interfere with perfectly DAE at about the highest speed available. So far it doesn't seem to mean anything in a practical sense.

KProbe2 and reality

Reply #6 –
I've had variable results with Sonys as well. Discs from one 50pcs spindle I bought burned as good as TY. Encouraged I bought another, and it was good as well. Then tried a 100pcs spindle, and had C1s out of wazoo, with occasional C2s, with speeds over 8X. That's on two Plextors - a Premium and a PX-712A. You don't mention what speed you burned at; it may be worthwhile to try a few and see if you get better results.

40,000 C1s overall is not bad, though not great either, on PlexTools at least. I have an impression that KProbe reports consistently lower values. There have been comparison tests between the two and professional scanners, with the verdict being that while no results matched, PlexTools produced figures/patterns closer to those by the pro h/w.

I usually ignore the total number. As long as there are no C2s, and the C1s don't go over 30-50, I'm happy.

Quote from: AndyH-ha on
I understand that C1 errors are correctable. In various places the opinion is expressed that audio quality suffers from these errors even if they are corrected.

Naturally. The hardware gets upset at all the errors it has to correct, which is liable to add harsh coloration to the sound; persistent abuse of the device's error correction facilities will inevitably lead to loss in detail. Fortunately, the audiophile industry offers very reasonably priced solutions to these and numerous other  problems.

KProbe2 and reality

Reply #7 –
Is that machinadynamica site for real?
If you told me it was a joke, I'd believe it...

KProbe2 and reality

Reply #8 –
Quote from: Firon on
Is that machinadynamica site for real? :blink:
If you told me it was a joke, I'd believe it...

Read his answer carefully and you may spot the sarcasm

BTW: no, the site is no joke.... although it should be.
I am arrogant and I can afford it because I deliver.

KProbe2 and reality

Reply #9 –
[Referring to machinadynamica]
Welll... d'oh... obviously placebo effect taken to the extreme.

Well, huh, if they want to spend their money buying something that makes their enjoyment better... go ahead.

Unfortunately -- and I am 120% sure of that -- I would consider such nonsense to be a waste of my hard-earned monies...

Well, let's answer with the standard HA reply: ABX please!

KProbe2 and reality

Reply #10 –
Lyx: I did spot the sarcasm in his post, I just wasn't sure if that site was really real because it seems ludicrous that anyone would believe in the crap there. I mean, come on, pebbles in a bottle to improve audio quality?

KProbe2 and reality

Reply #11 –
What about a magic solution that you paint on the LP center label, and on your light fixture switch plates, to improve audio quality? I've seen that for sale too. At least pebbles have some mass -- and you know about E=mc**2.

KProbe2 and reality

Reply #12 –
Back to the main topic

I recently found FujiFilm blanks on sale. I figured they might be good enough for projects where longevity isn't a concern. Of course one wants them to work properly for the duration, even if that duration isn't going to be too long. These were plainly and prominently labeled "archival quality."

I wrote a project onto three of them. None of the three tested very good in KProbe2, but the prize was this one, containing one continuous audio track of a little more than 72 minutes.
C1 Max                  : 438
C1 Average              : 57.41
C1 Total                : 172,299
C2 Max                  : 141
C2 Average              : 1.98
C2 Total                : 5,942

I remember reading something about the KProbe2's graphs being automatically scaled. Perhaps that accounts for their unusual appearance (unique in my limited experience). Towards the left hand beginning side there is just a thin colored layer on the upper graph and possibly a few small bumps on the lower graph.

Around 20 minutes in the colored part starts to thicken. One can definitely see there is something there on the upper graph but it is a continuous wash of color, not the collection of vertical lines I've seen from other CD scans.

Beyond that the graph color starts rising in earnest, although not exactly steadily. The really heavy error displays don't start until about 50 minutes in, with the lower graph always trailing the upper (there are less than 6000 C2 errors, after all).

Nero's CDSpeed Disk Quality test gave similar, though not identical, results. It's ScanDisk test produced a completely green display.

I listened to the very beginning of this CD, then the last part, via Winamp from the CD-Rom drive. Then I played it on a HiFi CD player. Who can account for taste, but I am unable to hear anything off color or at all suspicious.

Finally I let EAC loose on it. I didn't want to get tied up for hours as EAC fought with errors so first I did a test and copy in burst mode. That started out around 20X and gradually increased (as the error density increases, according to the test programs).The maximum speed reached was at the end of the disk, about 35X. I've seen speeds almost at the drive's claimed 52X on some CDs. Anyway, the CRC match was perfect.

Encouraged, I switched to secure mode. That started out at 3X and gradually increased to 14.2X at the end. The speed never backed off anywhere across the disk and was again higher further in where the supposed error density is higher. I don't recall exactly what EAC's red Error Correction grid is about. I've seen parts of it "light up" on some CDs but that never happened on this error filled one.

And you still tell me the quality tests mean something?

KProbe2 and reality

Reply #13 –
Quote from: AndyH-ha on
And you still tell me the quality tests mean something?
With a Plex and PlexTools, a resounding yes. I never owned a Lite-On, but if KProbe is right, yours is quite the monster of a drive. =)

To complete the picture please list:
-drive model
-recording speed
-scanning speed
-scan itself - a picture is worth a thousand words
-EAC's cache flush setting

It would be interesting to see what CDFreaks' Lite-On forum regulars would have to say about this. There have been some Lite-Ons that produced meaningless Kprobe scans, yours could be one of them.

KProbe2 and reality

Reply #14 –
Quote
With a Plex and PlexTools, a resounding yes.
Perhaps this is a true statement, but what do you mean by it? That when PlexTools produces "good" numbers, the disk is indeed good and when PlexTools produces "bad" numbers the disk doesn't work very well? That is what this thread is about, after all.

KProbe2 measures my Mitsui disks as "good" while some other brands frequently measure "bad" to "really terrible." However, in terms of any actual use, the best and the worse seem to be the same, giving evidence that the KProbe2 numbers don't mean much. Do you have some other evidence that the PlexTools measurements are valid and useful?

The Lite-On is model SOHR-5239V. I believe I verified that it is supposed to work properly with KProbe2, but I suppose I could be mistaken. However, one must come up with some reasonable explanation as to why some brands of CD-R measure so differently from others if one is going to condemn the drive based on its performance with KProbe2.

I've been writing CD-Rs at from 8X to 40X in order to compare results. So far, 24X seems to do the best -- with the greatest frequency -- for the Mitsui, but the FujiFilm with the awesome numbers was also written at 24X. The FujiFilm that measured best was written at 16X but so far I've only used about a half dozen blanks of that package. None have been very impressive in a good way.

The scanning speed in KProbe2 is set at Max. Of course, the actual speed varies across the disk. I don't pay a lot of attention, but I'm pretty sure I've seen speeds in the 40s and perhaps even at the drive's rated 52X. Same for EAC, it is whatever the program does. I know I have seen at least 49X, but each CD is somewhat different.

What would the graphs tell you that the numbers don't? I looked at them for awhile but found them pretty useless. Of course, if the total error number is very small, one can see if the errors are clustered in a relatively short section or are more evenly distributed. Also, the graph on that disk in my May 11th post was quite unusual with very small Y-axis values on the left-hand side that increased markedly on the right-hand side.

Still, what could one do with the graph? I know I spent several sentences of my 5/11 post describing what I saw in this case, but upon reflection I don't know that it would not have been just as good left out.

I don't think there is an EAC "cache flush setting" per se so I would guess you are asking if "Drive caches audio data" is checked under Secure mode. It is.

I did not attempt any tests on that FujiFilm disk beyond what I reported but some other experiences speak well of the Lite-On drive's DEA ability (a major feature of my report). A few days ago I extracted Alex deGrassi's Southern Exposure to the hard drive using Burst mode (at high speed) with test and copy. This is exquisite acoustical guitar, with nothing else to distract one. I sat and listened to it all from the hard drive. It sounded very good.

One might speculate on my musical inclinations, taste, or ability to distinguish one note from another, but for the past several years my most frequent music related undertaking has been listening to LP transfers to detect (and repair) clicks and pops. I feel reasonably confident about my ability to do that, but I know of no way to make an objective measure.

Anyway, while there might have been subtle errors with the DeGrassi extraction that I did not detect, clicks and pops are the most likely result of any very gross problems. There were none. This, of course was not the FujiFilm disk which measured so badly in KProbe2 while measuring essential perfection in EAC, but it is a verification of sorts that EAC is doing what it says it is doing. As I wrote then, the FujiFilm disk also played with no problem I could hear in both the Lite-On drive and a Sony CD player.

KProbe2 and reality

Reply #15 –
Quote from: AndyH-ha on
Perhaps this is a true statement, but what do you mean by it?
With PTP scans, if I get a CU it's a guarantee I won't be able to rip the corresponding track without errors. When I see hundreds of C2s in the scan, I know better than try and rip that CD with the NEC 3520. If I see a number of C2, it's an indication that Burst mode is likely to fail.

Quote from: AndyH-ha on
Do you have some other evidence that the PlexTools measurements are valid and useful?
They are only useful to someone who considers CD audio standards: max 220/s BLER, 1C2/hr. To me, a disc that scans with one single C2 is a coaster. I have little trust in a disc's readability five years down the line if it burns with C2s freshoff the spindle.

Quote from: AndyH-ha on
if one is going to condemn the drive based on its performance with KProbe2.
I think the drive is fine. It's KProbe that looks suspect.
What's the ATIP on those Fujis? If they are Made in Japan, they are Taiyo Yuden; if in Taiwan, could be CMC or something else (RiTEK or Prodisc?).

Quote from: AndyH-ha on
What would the graphs tell you that the numbers don't? I looked at them for awhile but found them pretty useless. Of course, if the total error number is very small, one can see if the errors are clustered in a relatively short section or are more evenly distributed.
That and also huge pin-thin spikes that drive the max number up, but can be safely discounted. I'd rather see just the graph than the numbers alone, but it's not terribly important.

Quote from: AndyH-ha on
I don't think there is an EAC "cache flush setting" per se so I would guess you are asking if "Drive caches audio data" is checked under Secure mode. It is.
Yes, that checkbox activates EAC's cache flushing routine. The "Drive caches audio data" label is nonsense.

Quote from: AndyH-ha on
A few days ago I extracted Alex deGrassi's Southern Exposure to the hard drive using Burst mode (at high speed) with test and copy.<...>  I sat and listened to it all from the hard drive. It sounded very good. <...>
Anyway, while there might have been subtle errors with the DeGrassi extraction that I did not detect, clicks and pops are the most likely result of any very gross problems. There were none.
That's very good, but you only get clicks when big chunks of data are completely gone and interpolating neighboring samples produces silence. When data has been damaged just enough for neighboring samples to survive, chances are  you won't hear a thing in that spot.

Do an MD5 hash on the original WAV, and then on the WAV extracted from the wonder disc. Run both through DAE Quality test's Analyse and C2Extract - there is objective measurement for you.

You should definitely ping CDFreaks users who have the same drive for similar problems. There could be a  simple explanation to all this.