Hi,
I am looking for some actual "error concealment" measurements for audio CD players.
I know that this is highly dependent on the the CD used, whether it is badly scratched or not. But still, I would like to get a feel that how often has this "error concealment" been employed in a normal audio CD player, say with a real world CD disc, clean with little or no scratch by the standard of human eyes.
I have searched the net and could only find this:
http://perso.numericable.fr/~laguill2/dae/interpolation/interpolation.htm
which was mainly focus on the PC CDROM drive.
Many thanks,
P
I am looking for some actual "error concealment" measurements for audio CD players.
I know that this is highly dependent on the the CD used, whether it is badly scratched or not. But still, I would like to get a feel that how often has this "error concealment" been employed in a normal audio CD player, say with a real world CD disc, clean with little or no scratch by the standard of human eyes.
I have searched the net and could only find this:
http://perso.numericable.fr/~laguill2/dae/interpolation/interpolation.htm
which was mainly focus on the PC CDROM drive.
Many thanks,
P
E22 and CD transport sound quality
(As I have been proved wrong more than once, please feel free to point out if I have been wrong again).
I have also found the following E22 (i.e. error concealment) measurements on the net:
http://www.hifi-tuning.com/furutech_rd2.htm
It's done by FURUTECH to market their DISC DEMAGNETIZER. I have known this company for long time, so personally I would take Furutech's measurements with a gain of salt.
Anyhow, just for an example, it seems that both the E22 and E32 errors are relatively uncommon. (Difficult to say for sure without knowing the condition of the CD disc which Furutech actually used).
I think it is very true that the more E22 errors that you have the worst sound quality you will get, but just how important is this E22 will affect the sound of CD transport in normal life is still a bit uncertain.
Why asking the question? It is because all E11 E21 E31 E12 can be corrected (in theory at least), and E32 cannot be corrected nor concealed and hence one will likely to hear a break in the music passage and knew that something is wrong; so the big mystery 😕 is the E22 errors, which are concealed by the CD transport but in the process also degraded the sound quality fo the CD.
Sure, the more errors E11 E21 E31 E12 E22 E32 the CD transport is facing with, the harder the CD transport circuit has to work and hence will affect the power supply and hence jitter on a global scale etc etc .... but just for arguement sake, let's leave this "jitter" on the side for the moment.
So just to rephase my question and confine it within the context of E22:
Q. Is there any measurements which shows the average E22 one should expect in an average joe system?
From that, we may able to see a bit more (although by no way conclusive) on how E22 is actually affecting the sound quality of the CD transports, and hence how much "work/effort" should be done to minimize the E22.
Take care.
P
(As I have been proved wrong more than once, please feel free to point out if I have been wrong again).
I have also found the following E22 (i.e. error concealment) measurements on the net:
http://www.hifi-tuning.com/furutech_rd2.htm
It's done by FURUTECH to market their DISC DEMAGNETIZER. I have known this company for long time, so personally I would take Furutech's measurements with a gain of salt.
Anyhow, just for an example, it seems that both the E22 and E32 errors are relatively uncommon. (Difficult to say for sure without knowing the condition of the CD disc which Furutech actually used).
I think it is very true that the more E22 errors that you have the worst sound quality you will get, but just how important is this E22 will affect the sound of CD transport in normal life is still a bit uncertain.
Why asking the question? It is because all E11 E21 E31 E12 can be corrected (in theory at least), and E32 cannot be corrected nor concealed and hence one will likely to hear a break in the music passage and knew that something is wrong; so the big mystery 😕 is the E22 errors, which are concealed by the CD transport but in the process also degraded the sound quality fo the CD.
Sure, the more errors E11 E21 E31 E12 E22 E32 the CD transport is facing with, the harder the CD transport circuit has to work and hence will affect the power supply and hence jitter on a global scale etc etc .... but just for arguement sake, let's leave this "jitter" on the side for the moment.
So just to rephase my question and confine it within the context of E22:
Q. Is there any measurements which shows the average E22 one should expect in an average joe system?
From that, we may able to see a bit more (although by no way conclusive) on how E22 is actually affecting the sound quality of the CD transports, and hence how much "work/effort" should be done to minimize the E22.
Take care.
P
One more reference
Hi,
I have found another E22 (plus others) measurements on the net:
http://www.digido.com/portal/pmodule_id=11/pmdmode=fullscreen/pageadder_page_id=75
This was done a while ago on CDRs, but more importantly, "all test cuts are typically 45 minutes long, and verification in the StageTech EC2 Media Tester are run at REAL time, the same speed we listen to music!"
So this may give us a better idea of the amount of E22 errors we are facing. Again, sadly these tests were based on CDRs and computer CDROM (I presume that StageTech EC2 Media Tester uses computer CDROM and not Philips CDM 1/4/9/12).
So still looking for more measurement done on average joe audio CD player ....
P
Hi,
I have found another E22 (plus others) measurements on the net:
http://www.digido.com/portal/pmodule_id=11/pmdmode=fullscreen/pageadder_page_id=75
This was done a while ago on CDRs, but more importantly, "all test cuts are typically 45 minutes long, and verification in the StageTech EC2 Media Tester are run at REAL time, the same speed we listen to music!"
So this may give us a better idea of the amount of E22 errors we are facing. Again, sadly these tests were based on CDRs and computer CDROM (I presume that StageTech EC2 Media Tester uses computer CDROM and not Philips CDM 1/4/9/12).
So still looking for more measurement done on average joe audio CD player ....
P
Hello patwen
Confusing isn't it🙂
You ask the right questions, however this knowledge about CD behaviour is not commonly known.
As I worked for the Philips Compact Disc System Evaluation Group I did many measurements for CD licencees (record companies, player manufacturers).
If you have an old Philips CD204 it is easy to get the error flags out. For newer player the signals might be not so easy available or the interpretation is more difficult.
So having a system able to read errorflags and playing a clean disc you will find out that BLER is almost the same as E11. This means you will find in general no uncorrectable errors on a CD....
If you take a special disc with errors (Fingerprints, black dots, wedge damage) you will see how surprisingly strong the error correction is. You will hardly see uncorrectables...
For factory use there are discs which will cause uncorrectables but they are not free available as they are an indication for product quality (which is company confidental)
You probably have a disc which do generates audible problems. It is very easy to jump on conclusions if you use that disc for "measurements". A correct judgement of measurements with a disc like that can take several days of work and a lot of experience in CD-CD player interaction.
So having a clean disc and a proper working CD player you will not have any uncorrectable error leading to audible differences...😀
Audible differences are not in this department😀
Ward
Confusing isn't it🙂
You ask the right questions, however this knowledge about CD behaviour is not commonly known.
As I worked for the Philips Compact Disc System Evaluation Group I did many measurements for CD licencees (record companies, player manufacturers).
If you have an old Philips CD204 it is easy to get the error flags out. For newer player the signals might be not so easy available or the interpretation is more difficult.
So having a system able to read errorflags and playing a clean disc you will find out that BLER is almost the same as E11. This means you will find in general no uncorrectable errors on a CD....
If you take a special disc with errors (Fingerprints, black dots, wedge damage) you will see how surprisingly strong the error correction is. You will hardly see uncorrectables...
For factory use there are discs which will cause uncorrectables but they are not free available as they are an indication for product quality (which is company confidental)
You probably have a disc which do generates audible problems. It is very easy to jump on conclusions if you use that disc for "measurements". A correct judgement of measurements with a disc like that can take several days of work and a lot of experience in CD-CD player interaction.
So having a clean disc and a proper working CD player you will not have any uncorrectable error leading to audible differences...😀
Audible differences are not in this department😀
Ward
I don’t understand these E numbers; maybe they refer to later Data CDROM drives. In the older days – and with plain old audio CD there are two types of Error flags available to the outside world.
C1 Data reading Error - 100% corrected by the Error correction system. This does not result in audio degradation as the data is truly corrected. An average disk may contain say 10000 – 20000 of these errors
C2 Data reading Error - this error cannot be corrected by the error correction system (Too much data loss), therefore the CD decoder chipset has been forced to “Conceal” the Error by interpolation or in the worst cases Muting the output. An average reasonable condition CD will normally contain less then 10 per disk if any.
Philips & Sony servo decoder chipsets normally have these outputs available on the decoder pins. You can add a fast counter (such as an option on the original Cambridge Audio CD1), or add a monostable pulse stretcher (due to pulses indicating a "Frame error" the pulse widths are inherently very narrow) and LED to observe these errors.
I once designed a product with the LED’s, but it was left off the production unit as it was felt the customer would be distracted (concerned) by the constant flashing of the C1 LED.
C1 Data reading Error - 100% corrected by the Error correction system. This does not result in audio degradation as the data is truly corrected. An average disk may contain say 10000 – 20000 of these errors
C2 Data reading Error - this error cannot be corrected by the error correction system (Too much data loss), therefore the CD decoder chipset has been forced to “Conceal” the Error by interpolation or in the worst cases Muting the output. An average reasonable condition CD will normally contain less then 10 per disk if any.
Philips & Sony servo decoder chipsets normally have these outputs available on the decoder pins. You can add a fast counter (such as an option on the original Cambridge Audio CD1), or add a monostable pulse stretcher (due to pulses indicating a "Frame error" the pulse widths are inherently very narrow) and LED to observe these errors.
I once designed a product with the LED’s, but it was left off the production unit as it was felt the customer would be distracted (concerned) by the constant flashing of the C1 LED.
This is a TEST CD the A-BEX TCD-725A. You should just be able to make out (sorry for the poor Photo) the increasing “Scratch”, simulated by Screen-printing on the disk, and above the “Scratch” the simulated “finger print”.
Despite the simulated scratch having a 1mm interruption at its worst point, the finger print test is tougher as this results in many reading error in quick succession – leaving the Error correction system without enough “Correct” data to correct for later errors.
CD Data is recorded on the disk in a Non Sequential pattern; so that a Scratch does not drop one large continues chuck of data, but rather a much smaller part of preceding and future data, which can be corrected for. This is why the finger print test presents such a challenge – many sequential errors.
The Test CD cost about $200 each – and usefully warns: -
“Caution! Please don’t wipe off the Black dots and fingerprints part with water and chemical solution because the fingerprints are printed on the compact disc”
Now there’s sound advice for your $200!
Track 11 (1mm interruption) and Track 15 (fingerprint 0.075mm) are Karaoke renditions of “Danny Boy” & “Heartbreak Hotel” so for more then 15 years of my life I’ve been working in factories in Asia – normally without air conditioning hearing these SAME tracks blearing out from 50 Test station in a row!
To make matters worst, there’s even an “Elvis” who goes around the bars in HK serenading you and your “Date” (for a small fee). He’s been going about his happy little business for at least 10 years, singing his dear little heart out, and is just about getting a little better – well at least for a small Chinese guy. If one day something REALLY BAD happens to this guy – you all now know whom to blame!
Despite the simulated scratch having a 1mm interruption at its worst point, the finger print test is tougher as this results in many reading error in quick succession – leaving the Error correction system without enough “Correct” data to correct for later errors.
CD Data is recorded on the disk in a Non Sequential pattern; so that a Scratch does not drop one large continues chuck of data, but rather a much smaller part of preceding and future data, which can be corrected for. This is why the finger print test presents such a challenge – many sequential errors.
The Test CD cost about $200 each – and usefully warns: -
“Caution! Please don’t wipe off the Black dots and fingerprints part with water and chemical solution because the fingerprints are printed on the compact disc”
Now there’s sound advice for your $200!
Track 11 (1mm interruption) and Track 15 (fingerprint 0.075mm) are Karaoke renditions of “Danny Boy” & “Heartbreak Hotel” so for more then 15 years of my life I’ve been working in factories in Asia – normally without air conditioning hearing these SAME tracks blearing out from 50 Test station in a row!
To make matters worst, there’s even an “Elvis” who goes around the bars in HK serenading you and your “Date” (for a small fee). He’s been going about his happy little business for at least 10 years, singing his dear little heart out, and is just about getting a little better – well at least for a small Chinese guy. If one day something REALLY BAD happens to this guy – you all now know whom to blame!
Attachments
Hello JohnW
It seems that my older days are even older😉
The E flags are used typical in first generation CD players. Philips could not make enough chipsets at that time so some first generation Philips players use Sony chipsets (the SoPhi boards). These boards are easy to interface for measurements. The C flags you refer to are the ones I mentioned which are less easy to use for measurements.
In a first generation chipset it comes handy to know "what happens inside" For later generations the availability of this info related to cost could not be justified.
For CD measurements this Cflag information is less valuable. As far as I remember you could bring these type of chipsets in a testmode where more info became available. But as we had the first generation full info available nobody spend that much time on it.
Ward
It seems that my older days are even older😉
The E flags are used typical in first generation CD players. Philips could not make enough chipsets at that time so some first generation Philips players use Sony chipsets (the SoPhi boards). These boards are easy to interface for measurements. The C flags you refer to are the ones I mentioned which are less easy to use for measurements.
In a first generation chipset it comes handy to know "what happens inside" For later generations the availability of this info related to cost could not be justified.
For CD measurements this Cflag information is less valuable. As far as I remember you could bring these type of chipsets in a testmode where more info became available. But as we had the first generation full info available nobody spend that much time on it.
Ward
Ward,
Yes – don’t scare me, I still remember designing turntables – which in my humble opinion still sound better than CD…
Granted, there maybe more information available from the “E codes”, but is there an issue with monitoring the C1 & C2, as I believed (still do) that they give a reliable indication of corrected (C1) and un-correctable (C2) errors – which at the end of the day, is what we are concerned about?
John
Yes – don’t scare me, I still remember designing turntables – which in my humble opinion still sound better than CD…
Granted, there maybe more information available from the “E codes”, but is there an issue with monitoring the C1 & C2, as I believed (still do) that they give a reliable indication of corrected (C1) and un-correctable (C2) errors – which at the end of the day, is what we are concerned about?
John
I know exactly what you are feeling JohnW😀
I have the same for Philips testdisc 5A
Track9 "An die Musik- Schubert" 900um interuption in informationlayer
Track17 "At the jazzband ball- Dutch Swing College Band" 800um blackdot
Track18 "Ricochet-Teresa Brewer" fingerprints
Track19 "To pazari- Yannis Markopoulos" fingerprints
If I hear it
Ward
I have the same for Philips testdisc 5A
Track9 "An die Musik- Schubert" 900um interuption in informationlayer
Track17 "At the jazzband ball- Dutch Swing College Band" 800um blackdot
Track18 "Ricochet-Teresa Brewer" fingerprints
Track19 "To pazari- Yannis Markopoulos" fingerprints
If I hear it
Ward
John
Regarding the Cflags I always had that feeling that:
C1 indicates that there is something but as it is correctable I do not need to be informed.
C2 indicates normally nothing.
So I prefer the Eflags as they give more information about disc quality.
In the end of the day-with a good player and a clean disc- there is nothing to be concerned about regarding uncorrectable errors.
Ward
Regarding the Cflags I always had that feeling that:
C1 indicates that there is something but as it is correctable I do not need to be informed.
C2 indicates normally nothing.
So I prefer the Eflags as they give more information about disc quality.
In the end of the day-with a good player and a clean disc- there is nothing to be concerned about regarding uncorrectable errors.
Ward
Ward,
Yeh – but I bet you don’t have to listen to 1000 workers happly practicing for their evenings merry making in “ChingEnglish” to the demented (dementing) little tunes!
John
Yeh – but I bet you don’t have to listen to 1000 workers happly practicing for their evenings merry making in “ChingEnglish” to the demented (dementing) little tunes!
John
Ward,
"C2 indicates normally nothing" Oh dear! - why? - Or do you mean theirs just not many of them?
John
"C2 indicates normally nothing" Oh dear! - why? - Or do you mean theirs just not many of them?
John
E11 E12 E13 E21 E22 E23 and C1 C2
I think the relation between them is something like this:
"C1 is defined as the sum of E11+E21+E31 per second within the inspection range.
The block error rate (BLER) equals with the sum of E11 + E21 + E31 per second averaged over ten seconds."
"C2 is defined either as the total of
* E32 per second within the inspection range for some manufacturers
* E12+E22 per second within the inspection range for other manufacturers"
E11, E21, E31 and E12 are not problematic, as they all can be perfectly corrected (in theory at least).
But E22 and E32 are not correctable and hence with affect the sound quality (interpolation,data-hold and muting used for amending E22 errors are NOT actually correcting the errors, instead they just hide the E22 errors, if you know what I mean).
Since C2 is loosely defined, and hence cause a bit of confusion.
I think the relation between them is something like this:
"C1 is defined as the sum of E11+E21+E31 per second within the inspection range.
The block error rate (BLER) equals with the sum of E11 + E21 + E31 per second averaged over ten seconds."
"C2 is defined either as the total of
* E32 per second within the inspection range for some manufacturers
* E12+E22 per second within the inspection range for other manufacturers"
E11, E21, E31 and E12 are not problematic, as they all can be perfectly corrected (in theory at least).
But E22 and E32 are not correctable and hence with affect the sound quality (interpolation,data-hold and muting used for amending E22 errors are NOT actually correcting the errors, instead they just hide the E22 errors, if you know what I mean).
Since C2 is loosely defined, and hence cause a bit of confusion.
John,
Yes indeed there are not many of them (on a normal disc zero).
patwen,
The relation is indeed roughly the same. As the Eflags are realy straight forward they are easy to handle/ interprete.
If you think about error detection and error correction in relation to strategies to tackle common errors caused by fi. scratches and fingerprints there are many more methods. The results however are not so easy to understand anymore.
Ward
Yes indeed there are not many of them (on a normal disc zero).
patwen,
The relation is indeed roughly the same. As the Eflags are realy straight forward they are easy to handle/ interprete.
If you think about error detection and error correction in relation to strategies to tackle common errors caused by fi. scratches and fingerprints there are many more methods. The results however are not so easy to understand anymore.
Ward
- Status
- Not open for further replies.