hi everybody,

does anybody have in-depth information about the reliability of cd-r and cd-rw discs concerning longevity of backup data? is there a general advice you may give, or is the quality of the discs solely dependant on its manufacturer?

thanks a log
MS

The FAQ gives links about user reports on CDR longevity, that seems to be around 2-4 years in average. CD-RW are said to last shorter once recorded : about 1 year.
But it can vary much. In fact, the extreme I saw reported were 2 weeks and 8 years for CDRs.

For reliable data backup, there are long discussions that you can search for in this forum. People rather use external hard drives to regularly backup their own hard drive. When one fails, it is detected when you backup, and the data can be retrieved from the other one.

.... I hope only, that DVD-medias have a longer lifetime than CD, ´cause I already backuped on them around 430 GB of data (= 100 DVD+R´s). The error-recovery shall be better at DVD.

From this you can conclude that CD-R has a longer life.

CD-R/RW may be risky & I'm not sure the error rate for CD's for a the suggested life but I have e few very cheap (& CMC) CD's from 3+ years which still work fine, I think there are many elements at play which can influence CD-R/RW;
-Exposure to sunlight
-Storage Temperature/humidity
-Burner/burn speed
-Chemical type/manufacturer
& Of course physical damage.

So I'm not sure user reports can be too trusted.

FYI:
I recently found that the data on one of my CD-RWs had 'died' after 2 or 3 months. (it was a CD-RW that came with my burner)

A quick summary

1
Same CDRs stored at my father's home were all in perfect state while all mine had began to fail, so I posted here and told that the temperature and light exposure were important (the main differences between my father's home and mine). Now, several monthes after mine, all my father's ones are failing too (mostly Mitsui golden dye).

2
One CD burned in 2000, scanned with EAC at full speed in april or may 2002, showed 100% quality, no C2 errors.
I wanted to listen to it in august 2002, but the sound was crackling. EAC would fail at correcting errors in my two drives, the Teac being much better at reading old CDRs than the Memorex. I bought the original (the copy was burned from a library original).
This shows that checking regularly for quality, or keeping a good drive and testing on a bad one is useless, unless you scan all your collection once a month at least.

3
Identical gold and silver CDRs from the same manufacturer and burned the same month on the same burner at the same speed and stored together aged differently. The gold one shows no errors while the silver one has uncorrectable tracks. In fact my 4 gold CDRs are still readable without errors while the about 25 silver ones more than two years old I tested have uncorrectable parts.

4
But... among the dead CDR reports collected at Afterdawn, there are two gold CDRs clearly described as having decayed without physical damage, while no Tayo Yuden have ever been reported to die (there was a false alert about it, but it turned out to be the drive that was broken when the CDR was tested).
HiSpace Carbon (MPO) that I burned at 24x, showing some C2 errors in Nero CDSpeed when freshly burned (one or two per CD) are failing after one year.

5
Memorex black CDRs (Unknown manufacturer code in CDRidentifier) showing 10s of C2 errors after burning one year ago are reported dead to me by my mother (can't listen to them) who still keep Verbatim and Mitsui from 4 years ago. And my father just had a black Memorex from one year ago skipping in the Yamaha CD Player of the hifi, while most of the Mitsuis from 2 or 3 years ago, though showing uncorrectable errors, are still playable in the hifi (only 2 of about 10 are already unplayable, and they don't skip, they just click).

Conclusion

I think because of 1 that good storage conditions can only delay the dead of CDRs, but it should be compared with CDRs stored in a closed box.
I think because of 3 that gold CDRs last longer than silver CDRs, but not much longer, because of 4.
I think because of 4 and 5, and because Tayo Yuden CDRs show a low error rate (can't find again the C1 scans I'm thinking of, but these ones are good too), that the initial error rate plays a crucial role.
I think because of 2 that CDRs in perfect state after 4 or 5 years are not necessary much better than some failing after 3 years. They still may die within 6 monthes.

Is it helpful at all to create clone cd images of your older cd-r backups before transfering back to your harddrive or re-burning them?

I use this method and set the software error correction to at least 10 passes. I set the read speed to 8X only.

My first back-ups are coming up on 5 years old and I get no errors. They are a mixture of verbatim an maxell golds. This was before I switched to Kpdack Gold.

All back ups are burned at 4 times speed.

I've not understood the question.
Are they data or audio ?
Do you want to copy them and keep an image, store them on hard drive and keep an image, store them on hard drive as CloneCD images, or copy them with CloneCD ?

Oh, I forgot Artemis3's one :

http://www.hydrogenaudio.org/forums/index....2301#entry80918

Well... I seem to myself as Alice in Wonderland I started to burn Audio CDs on CDRs in 1996 and NONE of my CDRs has died so far. All CDRs are kept in their original plastic boxes, no direct sunlight, the approx. temperature around 24 C. All disks were burned using various Yamaha burners at max. 4x. And almost all CDRs are golden or silver or silvergold Kodaks...
Just my 2 cents.

Pio2001 & Budgie,

It might be interesting to know how you have labeled your discs and whether you think that has had any effect on longevity -- do you use Sharpie pens or other pens, with oil- or alcohol- or water-based inks? Or do you use adhesive labels, and if so what kind? Or do you not label the discs at all, and rely on labeling the packaging?

mrosscook:

I avoid labelling CDRs with audio... don't use anything... just put down unique ID of each CD onto the booklet enclosed to CD together with the author & title... that's all. And have two separate printouts with the list of all CDs.

No labeling of any kind. All my CDRs are left bare, save for my fat fingerprints on the label side, extending two milimeters from the outer edge, and near the center when I press the CD in its jewel case. You can see the effect on the uncoated silver of this CDR : scanned bare - with tracing paper). The protective layer of these CDRs don't cover all the metallic layer. The inner ring is bare (detected with an ohm-meter).

The CDRs actually turn unreadable from both edges, the outer first :


Loudness in logarithmic scaling of the difference between the extraction from the CDR and the reference file, against playback time from 0 to 79 min

but the read errors extend far beyond my fingerprints, and the same behaviour, outer edge with higher error rates, can also be seen with brand new CDR (it occured to me with old Fnac silver CDR).

Too bad! Pio, I was hoping that something in the way that you labeled CD-R's could explain the sudden degradation of the discs at your apartment and at your father's place; but since neither you nor budgie use ANY labeling, the difference in your experience can't be attributed to inks or solvents or adhesives.

The very sharp boundaries between high and low error rates on Pio's graph make it look as if the degradation is caused by some corrosive agent, like oxygen or sulfur dioxide, diffusing inward from the outer edge and outward from the inner edge, and oxidizing the metal. If the metal layer extended all the way to the rim of the disc, that could provide an avenue for rapid radial diffusion of oxygen; but I don't think that the metal layer does reach all the way to the edges of the disc. Looking at the CD-Rs that I have handy, there seems to be at least 0.5 mm or so of fused plastic at the outer rim, with no metal included. It's hard to see why diffusion through this thick solid edge should be so much greater than diffusion through the thin top surface of plastic. Very odd.

The sharp boundaries also don't allow for an explanation based on decay of the dye layer; that is supposedly caused by heat and/or light changing the refractive index of the dye, and I can't see any reason for that to depend so strongly on the radius.

If there is anyone lurking here who has experience in CD manufacturing, please feel free to pitch in and explain Pio's graph. I'm stumped.

The sharp edges of the graph depends strongly on the audio interpolation behaviour, because it is the loudness of the clics, that is plotted here. Graphs plotting error rate in logarithmic scale for the same CDR show a constant rise, that is an exponetial increase in linear scaling.

Much more graphs and analysis : Error detection, error correction and error concealment reading audio CDs

I guess I will finally have to swallow the bitter pain and buy two additional Multi-GB-Monster-Harddisks (thinking of 5400 RPM 120GByte Samsung's) to store my Audio Copies alongside with the cuesheets in a lossless Format ...

Dying CDR's are slowly becoming my worst nightmare with more than 1000 Audio CD's (around 500 copies) ...

I wonder if the CD-R disc failures reported here by Pio2001 and others could be related to the "CD rot" problem that afflicted a number of releases of commercial pressed CD's a while ago; see, e.g., here.

In that case, a poor formulation of the lacquer sealing the top surface of the CDs allowed residual acid and sulfur from the booklet paper to penetrate below the surface and oxidize the aluminum reflective layer. The oxidation started at the outer rim of the CD and progressed inwards.

What do you think? Does the picture in the link above resemble any degradation that you've seen?

I experienced some (silver) discs turning yellow on the upper side over time, beginning at the outer diameter. You have to remember that cdr reflectors are mostly made of silver or gold ... not aluminium so I don't know if this can be called "CD Rot" in the sense of it's definition, although it surely is some kind of "rot", but not caused by acid or sulfur (my burned CD's that turned yellow were stored with AND without paper).

Some people claim that insuffiecient sealing of the outer rim will allow oxygen to diffuse into the area between polycarbonate layer and reflector to oxydize the reflector itself and make the CD degrade and become unreadable as a final result.

UV radiation can destroy the dye formation (and thus, the pit and land structures) ... enough destruction on the dye formation and the CD will be unreadable as well. Phtalocyanine (the "clear" dye) is believed to be the dye with the best UV resistance possible ... Azo or Cyanine (the deep or light blue/green discs) will suffer much more from UV radiation ...

Some cheap media show a pit geometry totally out of specification (can be made visible via atomic force microscopy) ... the writing quality article (absolutely recommended) from www.cdrinfo.com explains this quite good ... this is a problem of the production process itself, some manufacturers sell the batches from production startup where production conditions mostly haven't stabilized (the prodcution chain will come to balance only after some hours of "warm-up") - other manufacturers (like Taiyo Yuden, I guess) will throw these batches away and only use/sell the batches that were produced on a perfecly running and balanced production chain ...

Burner and Write Speed is another parameter that may influence CDR lifespan ... writing strategies (laser power, burst length) are stored within the burner's firmware and MUST match the appropriate media (which will be recognized by reading the ATIP) ... tests show that many burners do not adjust themselves as good as they could which explains different results with the same media type on different burners ... write speed is important as well ... too fast is never good, but burning too slow can lead to thermal problems (think about longer laser pulses over a longer time when writing at 2x or 4x where e.g. heat exchange from the disc to the surrounding air is lower due to lower rotation speeds)

In theory, a pure gold reflector, in combination with phatlocyanine dye and good sealing (like e.g. Mitsui Gold Ultra or Hi-Space Gold), written at 8-16x CLV, would be the best choice for long-term data storage on CDR ... but these Mitsui or Kodak Gold discs are reported as slowly dying, too ...

just my 2 cents ...

Right ! I've just realized that the first dead CDR I found was actually more yellow than more recent ones.



Top, a CDR already unplayable in a hifi Cd Player in february 2002, and unrippable in secure mode several monthes before.

Bottom, a CDR with no C2 errors in august 2002.

Here's a CDR that developped a yellow color early, without any effect on its readability :



It's a noname CDR. It has no special protection on top, while the Mitsui have the "diamond coating" that "protects them against any external attack". The laquer can be seen (in reality, not in the picture)extending on the transparent part, thus its not the laquer that is yellow.

Note the way the yellow color is progressing : exept the very outer edge that is unharmed, it attacks the beginning and ending of the CD.

I'm currently running CDSpeed on them. It progresses at 0.6x on the first CD, so don't expect the result before long

EDIT : I don't think UV plays an important role. They should be very low inside homes (blocked by windows glass), and black CDRs, with the anti-UV tint, decay as fast as clear CDRs.

For the yellow Mitsui, I stopped CDSpeed when it reached 176400 errors per seconds, that is, not a single user byte decodable anymore.

Here are the screenshots, in the same order than the above pictures :







The errors in the Ignition CDR comes from the joint between the audio and data sessions, Nero CDSpeed always displays the lead out/lead in of an enhanced CD as erroneous.

I would like to point out that I have many Mitsui CD-R's (look like pio's 3, recording, storage conditions unknown) which have unrecoverable errors, I would not recommend the brand.

The colorations are not visible on the optical side

For me, Ricoh and HiSpace are not better. Have you got other brands as old as your Mitsuis, but that have no errors ?

I have many however they where not burnt by me, all I know is that they are 4+ years (most probability some are even older 10 years+), by various burners I would say at 4x but once again I can't be sure so no scientific conclusion can be made other than CD-R's can last 10+ years all I know is that most of the Mitsui's are unreadable or have huge errors.