Hello,
I am new to EAC, and somewhat confused at this point. As most guides suggest, I am using "Test and Copy" mode when ripping a CD. Sometimes, in secure mode, I get 100% track quality, with mismatching "Test CRC" and "Copy CRC" values, like in the following example:

Track 2
Filename C:\music\Santana\Supernatural\eac_wave_crc\02. Love of My Life.wav

Peak level 99.3 %
Track quality 100.0 %
Test CRC 0D9C02A0
Copy CRC 28C519ED
Copy OK

Track 3
Filename C:\music\Santana\Supernatural\eac_wave_crc\03. Put Your Lights On.wav

Peak level 99.8 %
Track quality 100.0 %
Test CRC 1F5EEC45
Copy CRC 4485B98E
Copy OK


In fast mode or burst mode I sometimes get (on the same tracks) "Test CRC" and "Copy CRC" values to match each other - does this mean the fast/burst mode was a more accurate rip (than one in secure mode), or it just did not noticed any problems because of less rigid error checking?

P.S. I don't think there are any problems with my hardware, with new CDs I get matching CRC values.

Any help and advice is appreciated.

Maybe your drive doesn't support C2, or does cache audio. This could explain why errors are not detected and the quality is wrongly reported as 100%.
What are the secure mode settings in EAC for your drive ?

Here are my settings (from the same LOG file):

EAC extraction logfile from 23. March 2006, 23:47 for CD
Santana / Supernatural

Used drive : SAMSUNG CD-R/RW SW-252S Adapter: 1 ID: 0
Read mode : Secure with C2, accurate stream, disable cache
Combined read/write offset correction : 0
Overread into Lead-In and Lead-Out : No

Used output format : Internal WAV Routines
44.100 Hz; 16 Bit; Stereo

Other options :
Fill up missing offset samples with silence : Yes
Delete leading and trailing silent blocks : No
Native Win32 interface for Win NT & 2000

P.S. Actually, I noticed that in "Extraction Method" tab the "Drive cashes audio data" box is checked. This is the default value. Not sure why LOG shows "disable cache". Shall I uncheck the "Drive cashes audio data" box?

disable c2 error correction and also make sure that you don't use anything in the background while you rip a cd.

no, at least not before confirming that your drive doesn't cache audio. but even if it doesn't it doesn't matter if you leave it checked.

OK, yes my drive supports "Cache audio". This is why this box was set - to tell the program to override the cashe with actual values. Still confused about the CRC mismatch.

Yes, my drive supports the cache feature. I confirmed it by pressing "Detect Read Features" button. "Caching" - yes, "Accurate Stream" - yes, "C2 Error Info" - yes.

Your example is not so uncommon, it can happen sometimes.
This is the reason to use test & copy and consider a rip only as perfect, if the crcs match (and no read/sync errors of course).
Reason behind: if you used no c2 for error detection, eac reads twice (in each run), but the probability is quite high, that both readings gave same result, ie. 100% ok, but a next reading/run might result to something different.

I see what you say. But it reads twice when I "Test and Copy" in all modes. In my example the "secure mode" resulted in differnt "Test CRC" and "Copy CRC" (it means test was different from copying), but it shows 100% quality because it was able to "reliably" read all the data. At the same time, "fast mode" resulted in identical "Test CRC" and "Copy CRC" values, but how much can I trust this result?
Maybe, it simply did not check for errors so intensively, just skimmed the track - and have not noticed any mismatch. In other words, should I (in this case) prefer matching "CRC" values WAVE (from fast mode) over 100% track quality with mismatching CRS (from secure mode), or not? I have 2 different WAVE files. Is the one created in "Fast mode" more accurate?

The easier solution is to uncheck C2 and see if you get consistent results.

If you still get bad CRC with "no errors occured", then you will have rip the CD until you get two wavefiles from the same track with different CRC.
Then you will have to compare them. If you are lucky, the compare wav option of EAC will just tell you that they are offsetted from each other (and you will have to uncheck "accurate stream"), otherwise you will have to load them is a wave editor and substract them in order to see what went wrong.

I just read this discussion:
http://www.hydrogenaudio.org/forums/index....=ST&f=20&t=9905

about CRCs, C2 and reliability. In all honesty, it just made things more foggy for me. As I understand, CRC is just a checksum generated from a file read or copy process. There's no stored CRC value in the data on the CD iself to check against. So, EAC does check between two different reads - if they match, CRCs match, but both reads still can be wrong (in the same way). Could anyone explain me what exactly C2 is - and why it's desirable to have it ON or OFF?

As a more general question, I'd like to clarify for myself a better way of ripping my CDs. Some people suggest to start with burst mode, "test and copy". If CRCs match, and there was no errors (read/sync errors) consider it a good rip. If CRCs don't match, use secure mode, it's allegedly a more reliable rip.

As my case shows, CRCs can match in burst or fast mode, but differ in secure mode. So, the above method (to start in "bust mode") isn't that reliable, after all.

Should I do this the opposite way? Start with "Secure mode", if CRCs match - good, if CRCs don't match - go to fast and them burst modes, and if they produce identical CRCs (even entirely different from those of "Secure mode") take it as a better rip? What's your advice?

P.S. I found some limited info on C2 here:
http://www.feurio.net/English/faq/faq_vocable_c2error.shtml

As I understand, C2 is a hardware-built capability of a CD-ROM drive to "correct" errors on a CD. But these "corrections" are not always right - hence I get different CRC values in test and copy. But you mentioned that turning C2 off will make EAC to read everything twice. Does not it do it even with C2 on? If not, what are the two different values if compares in my case?

Here is another page suggesting to turn off "C2 error" in "Secure mode":
http://www.teqnilogik.com/tutorials/eac.htm#AboutTestAndCopy

Could anyone explain what sense does it make to have a "C2 error" correction at all, if not in secure mode?

This C2 reporting is not about the ability of the drive to make corrections but its ability to report the errors. Many drives have no way to communication with the outside world about the error correction process.

@m3dsa

Please read some EAC guides...

Liekloo's Essential Ripping guide :
http://users.fulladsl.be/spb2267/

SatCP's EAC Tutorial + Quickstart Guide :
http://users.pandora.be/satcp/tutorials.htm

Okay, guys! I have decided to use AccurateRip, at least it gives me some confidence about how accurate my rips are, based on statistics. So far, I am happy to see many of my results are extracted correctly, according to AccurateRip.

Still not sure what's best way to do - start with "secure mode", or start with "burst mode". Gotta learn it by trial & error. Thanks for your help!

I still suggest starting with Test & Copy in Burst Mode, then if you don't get matching CRCs, try Secure Mode. I would just disable C2 altogether. Some drives are better at it than others. I'm not sure how your drive compares.

m3dsa, Pio2001 gave you sound advice, why not use it?

You're getting lost thanks to EAC's poorly chosen terminology. That Drive is capable of retrieving C2 error information checkbox should have been titled something like "Rely on drive's C2 reporting".

In Secure mode, EAC will re-read a CD sector until (up to a certain limit) two matching reads are obtained. With the option in question enabled it will read once, unless the drive reports that an uncorrectable error occured during the read. If the drive's error detection/reporting aren't reliable, EAC will not re-read where it should and proceed to next sectors. This way you can have consistent CRC mismatches alongside 100% extraction quality reports.

With Test & Copy, the above routine is done twice: once to the hard drive (the Copy part), and once into the memory (Test). Multiply all the re-reads, necissitated by detected errors, by two.

Your Samsung obviously doesn't report errors reliably, so that checkbox should be cleared. Since you are doing Test & Copy, there is no need to stay in Secure mode - because matching CRCs in Burst mode mean a rip more accurate than that done in Secure mode with mismatching CRCs.

Fast mode is the same as Burst mode, but with an extra routine to ensure sector sync when ripping on a drive that does not support Accurate Stream. Such drives are not common these day, so this mode is just slower with no advantage. You can follow Pio's advice if you suspect that your drive doesn't support Accurate Stream; however, as the drive works with AccurateRip, it is clear that it does.

>As my case shows, CRCs can match in burst or fast mode, but differ in secure mode. So, the
>above method (to start in "bust mode") isn't that reliable, after all.

I cannot begin to understand how you arrived to that conclusion. What bigger proof of reliability than matching CRCs do you want? Unlike Secure mode, that can be easily misconfigured, Burst mode is virtually fool-proof: either the CRCs match (success) or they don't (failure). If they don't, use Secure mode with only the Accurate Stream and Drive caches audio data boxes checked and the third one clear.

I don't know, if you read the http://www.High-Quality.ch.vu guides.

They recommend to use EAC secure mode, test & copy (in no c2 modes, though c2 with test copy and matching crcs should be also ok, though the probability can go down to get 2 same crcs if you use c2).
If the crcs match and no read/sync errors, you can feel safe to have got a perfect extraction, at least no clicks.
(Extracting CD-Audio is always about probabilities, there are higher and lower ones.)
The quality rate shown by eac log is irrelevant in this respect, if it is 100% or 95%, who cares,
this percentage told by EAC is a figure to get a rough impression, in which condition the surface of the CD is. A lot of scratches -> eac has to reread more often -> percentage goes downward.
If this extraction can be considered safe, is only told by the matching (or dismatching) crcs in test & copy mode.

So, what can you do, if there are crc mismatches ?

==>> extract those tracks again, best by 2nd copy run, maybe with another, a better drive. If you got 2 same crcs somewhere, feel lucky and consider this extraction as safe.

The comparison by accurate rip db gives additional safety, of course, though it doesn't give you the music content. As you have to live with your (scratched) CD and your rips, you can only try to extract it yourself and apply the methods, EAC gives you, which is 2 same crcs to feel safe for a good rip.



addon about test & copy in secure mode and test & copy in burst mode:

If you get matching crcs in burst mode, you have had good luck in combination with a CD in good condition.
Because a scratched CD will give misreads, so low probability to get same crcs in burst, as burst has no error detection and does not perform rereads in case of failures.
Secure modes do perform error detections during extractions and do perform rereads to evaluate the most probable (=correct) read results by statistics.
So, in case the CD has any scartch, even only 1 light one, better to use secure mode with test & copy to get more safety.

That's the problem, this "most probable = correct" assumption is just a wild guess.
AccurateRip does add some confidence because it checks for consistency between
different sources, but consistency from a single source is not a proof of correctness.

I have been burnt before -- one of my drives reports OK when some samples are not okay. So, now, I rip each disc 3 times with different drives. I consider the rip OK when 3 rips are identifcal (after correcting for the sample offset, of course).