Optical drive

[HenSch]

Yeah, great thread indeed and great information already gathered upon the CD-drive technology.

 

For me it is also not understandable at all, why bit-identical rips do sound differently. And hell yes they do, as I personally found out in my system. However as usual, YMMV.

 

My ripping station consists of a Plextor Premium drive which is powered by a HD-Plex LPS. Both 12V and 5V independently. The drive is connected to my PC via an IDE<>SATA converter which is powered with clean 5V as well. Great that the newest generation of HD-Plex LPS has an adjustable 5-9V output. Therefore it can generate double 5V plus the 12V out of one box. Very neat!

 

I'm so much looking forward to some further technical investigation on this topic.

 

Best wishes,

Henning

[sandyk]

Yeah, great thread indeed and great information already gathered upon the CD-drive technology.

 

For me it is also not understandable at all, why bit-identical rips do sound differently. And hell yes they do, as I personally found out in my system. However as usual, YMMV.

 

My ripping station consists of a Plextor Premium drive which is powered by a HD-Plex LPS. Both 12V and 5V independently. The drive is connected to my PC via an IDE<>SATA converter which is powered with clean 5V as well. Great that the newest generation of HD-Plex LPS has an adjustable 5-9V output. Therefore it can generate double 5V plus the 12V out of one box. Very neat!

 

I'm so much looking forward to some further technical investigation on this topic.

 

Best wishes,

Henning

 

Henning

Have you also tried ripping the files directly to an internal SSD via a +5V low noise voltage regulator powered from the LPS ?

Kind Regards

Alex

 

P.S.

See also

http://www.computeraudiophile.com/f10-music-servers/batter-linear-power-supply-powered-solid-state-drive-what-cable-adapter-use-25636/ Post 10.

[mansr]

For me it is also not understandable at all, why bit-identical rips do sound differently. And hell yes they do, as I personally found out in my system.

 

How did you find out? Did you take any measurements?

[alfe]

Let's go back to P19 of the ECMA document,the tolerance for nominal time is +/- 1s witch represent 7350 frames.

Now if we add to maximises focus an offset is applied:

http://www.ieeecss.org/CSM/library/2008/june08/11-June08ApplicationsOfControl.pdf

(Tom from this document you will understand what I meant for phase shift)

And that usually the drive manufacturers will use the most significant bit of the first left channel of the time frame of the first frame of the sector.

You will get the famous drive offset:)

 

*edbk: I'm not teasing it's just about timing, I'm not retired yet.

[sandyk]

If that were possible, computers as we know them could not exist. In fact, the universe as we know it couldn't exist under such circumstances.

 

mansr

As you are a new member here, you may not be aware that respected E.E. and Technical Journalist Martin Colloms from HiFi Critic Magazine ,has previously performed a series of 6 positive separate Blind A/B/A 3 minute tests with comparison .wav files that I supplied to him. If you are genuinely interested , you will find further information at the attached links, as well as learning something from Alfe who is well qualified in the design of BluRay writers and other related areas. If you have come here to tell us all how stupid we all are, I would suggest that you leave this thread to those who are interested and don't try to disrupt it.

 

Alex

 

Audio Networking: A potpourri of computer network audio findings, including updates on the Naim UnitiServe, gigabit switches, CAT 5/6 cable and ripping issues. Review By Martin Colloms

 

 

Computer Audio deep thinking, not scoffing, required: different sounding rips - General HIFI Discussion - HIFICRITIC FORUM - HIFICRITIC FORUM : hi fi audio systems forum

Initial thread.19th August 2011

 

 

Re differences in audio file quality, HIFICRITIC investigates HDDand SSD sound quality - Page 2 - General HIFI Discussion - HIFICRITIC FORUM - HIFICRITIC FORUM : hi fi audio systems forum

See : Martin Colloms Post 30: Friday, 18 November 2011 10:36:42 AM

 

 

Kethel ripping results second session - General HIFI Discussion - HIFICRITIC FORUM - HIFICRITIC FORUM : hi fi audio systems forum 19/12/2012

http://www.hificritic.com/forum/default.aspx?g=posts&t=897 Originally

 

 

See replies 1 and 10

 

 

Re differences in audio file quality, HIFICRITIC investigates HDDand SSD sound quality - Page 4 - General HIFI Discussion - HIFICRITIC FORUM - HIFICRITIC FORUM : hi fi audio systems forum

See : Martin Colloms Post 68 Monday, 30 January 2012 3:13:48 PM

 

 

Problem when hocus-pocus works? - Page 5 - General HIFI Discussion - HIFICRITIC FORUM - HIFICRITIC FORUM : hi fi audio systems forum

Most recent. See replies 86 to 91

[mansr]

mansr

As you are a new member here, you may not be aware that respected E.E. and Technical Journalist Martin Colloms from HiFi Critic Magazine ,has previously performed a series of 6 positive separate Blind A/B/A 3 minute tests with comparison .wav files that I supplied to him. If you are genuinely interested , you will find further information at the attached links, as well as learning something from Alfe who is well qualified in the design of BluRay writers and other related areas. If you have come here to tell us all how stupid we all are, I would suggest that you leave this thread to those who are interested and don't try to disrupt it.

 

Let's say there is indeed an audible difference between different rips despite the files being bit for bit identical. This means there must be some other difference, beyond the data content, influencing playback. Computers and operating systems being what they are, this is a possibility I'm willing to entertain. It could be related to fragmentation on disk or some other intricacy of the management structures in the OS or hardware. Where things get mysterious is how these differences are able to propagate together with the file data when they are copied to another storage medium or even sent across the internet. A file transfer program (be it browser, email, FTP, or something else) would not, and should not, attempt to discern and convey such hidden properties of the files. It seems, to me, mighty strange that these differences, whatever their nature, are able to undetected pass themselves along through a variety of channels not designed to accommodate them. Nevertheless, people report hearing differences. As an engineer, it goes against my nature to simply leave it at that without any explanation. The first step in obtaining an explanation is to determine precisely what we are trying to explain. That means measuring the signal somewhere along the playback chain and identifying something that correlates with the perceived sonic differences. In all of the discussions about this phenomenon I've read, nobody has yet presented anything along those lines. There is even outright hostility towards the notion of applying a little science to find out what's going on. Until that changes, I'm going to continue believing that it's all in peoples' imagination. Even the best and the brightest are susceptible to tricks of the mind, which is why we must at least try to measure things using objective methods.

[sandyk]

mansr

It may surprise you to hear that when these comparison .wav files with identical check sums are burned to a quality CD-R , that audible differences still remain when played through a good CD/DVD/BR player , yet when ripped back to a HDD, the checksums are still identical.

 

Alex

[HenSch]

Henning

Have you also tried ripping the files directly to an internal SSD via a +5V low noise voltage regulator powered from the LPS ?

Kind Regards

Alex

 

P.S.

See also

http://www.computeraudiophile.com/f10-music-servers/batter-linear-power-supply-powered-solid-state-drive-what-cable-adapter-use-25636/ Post 10.

 

Indeed I haven't yet.

 

But it really could make sense, because today I tried better SATA cables for both the Plextor drive and the SSD and it bettered the SQ of the rips a tiny, but noticable bit.

 

Btw, my wife is a good helper in evaluating these differences, since she doesn't know much about all those technical tweaks, but she has a good musical hearing and understanding as well.

 

So I will give the LPS powered SSD a go, thanks SandyK for this impulse.

 

Henning

[edbk]

From experience I recommend a good choke filtered powersupply for the ssd sounds better then battery!

[mansr]

mansr

It may surprise you to hear that when these comparison .wav files with identical check sums are burned to a quality CD-R , that audible differences still remain when played through a good CD/DVD/BR player , yet when ripped back to a HDD, the checksums are still identical.

 

Alex

 

Enterprise storage systems commonly employ deduplication, a space-saving technique whereby blocks of identical content are stored only once. I wonder how these files would behave after a trip through such a system.

[YashN]

From experience I recommend a good choke filtered powersupply for the ssd sounds better then battery!

 

Seen your PS units a few times in the forum, are you planning to sell these or share the schematics or something like that?

 

Cheers.

[sandyk]

Seen your PS units a few times in the forum, are you planning to sell these or share the schematics or something like that?

 

Cheers.

 

Yashn and edbk

If you are using internal SSDs, it doesn't make sense to use an additional Linear PSU , with choke input or not, as it will introduce a degree of mains earth loop problems via the Neutral line, even if you don't use a 3 pin plug.(Transformer Primary to Secondary winding capacitance)

Great results can be obtained simply by further regulating the existing +12V supply (whether SMPS or Linear) using a low noise voltage regulator for a further improved +5V supply over the existing internal +5V supply . Use of a Super Regulator will give a further small improvement.

 

Regards

Alex

[alfe]

As I see the ongoing comment about internet here is the link again for the ones that don't read them

ftp://ftp.cis.upenn.edu/pub/mbgreen/papers/ton98.pdf

P.8 regaining a uniform distribution: compression.

[One and a half]

mansr

As you are a new member here, you may not be aware that respected E.E. and Technical Journalist Martin Colloms from HiFi Critic Magazine ,has previously performed a series of 6 positive separate Blind A/B/A 3 minute tests with comparison .wav files that I supplied to him. If you are genuinely interested , you will find further information at the attached links, as well as learning something from Alfe who is well qualified in the design of BluRay writers and other related areas. If you have come here to tell us all how stupid we all are, I would suggest that you leave this thread to those who are interested and don't try to disrupt it.

 

By re-representing the statements above, you have disrupted the discussion of ripping, once again, to travel down the rabbit 'ole. Had to come sooner or later i guess. Oh well another thread not to read, pity, it started out very good until post 55. Would have been better as a PM, sandyk.

[sandyk]

By re-representing the statements above, you have disrupted the discussion of ripping, once again, to travel down the rabbit 'ole. Had to come sooner or later i guess. Oh well another thread not to read, pity, it started out very good until post 55. Would have been better as a PM, sandyk.

 

So you expected this quote from reply 37 to Kumakuma to go unchallenged ?

 

If that were possible, computers as we know them could not exist. In fact, the universe as we know it couldn't exist under such circumstances.

 

And , YES, I did take the discussion to PM with that poster to try and get the thread away from yet another meltdown.

[edbk]

Seen your PS units a few times in the forum, are you planning to sell these or share the schematics or something like that?

 

Cheers.

 

I dont have a problem with either, its hardly rocket science. Attached a picture of a beta version Im currently using with mostly off the shelf parts. I have ordered several custom made high current chokes with a much higher (500-5000 times) inductance then off the shelf hammonds to take this concept to the next level. Motherboard / ssd etc are all powered from the same supply but do have their own filter stages, there's just one "ground". I also have several custom isolation transformers to try to create "floating grounds" for different pheripherals. Dont know yet if thats gonna do any good, only so much hours in a week.

 

Sorry for the small Off topic Alfe.

 

Btw notice the leds still burning with the power unplugged, there's 300.000uF of supply storage on the main board :-)

 

[alfe]

Error correction Doc #24:

For CD audio probability of error is on the average of 10^-5, this mean 1 time 10⁵ bits.

Knowing the bit rate of CD audio this falls to 10 times per second.

Luckily we have error correction and with a certain range erroneously read 0’s and 1’s can be corrected.

I said with a certain range because there is a limit to its effectiveness, matching this limit to error distribution is the length and frequency of erroneous 0’s and 1’s.

Digital audio is 0’s and 1’s and if we want to add error correction a small number of extra 0’s and 1’s is necessary.

In other words error correction=extra data and these data are called redundant bits.

The portion that these redundant bits occupy in the total of recorded bits, is called redundancy.

So if m (bits) show 1 unit of information and k units represent the digitalised audio, n unit will be the complete data including the redundant bits with of course n>k

Than for CIRC error correction using 2 steps C1&C2

C1: m=8 n=32 k=28

C2: m=8 n=28 k=24 d=5

The smallest value of the distance among all the possible data is called the minimum distance of the code and expressed by the letter d.

The unit is 8 bits and there are 4 units of data for correction.

The 8 bits unit is called a symbol and recorded data is expressed of n symbols.

Lets take two sets of data X and Y:

X=x_0, x₁, ..., x_n-1

Y=y₀, y₁, ...,y_n-1

Where x_i and y_i (i=0...n-1) shows the symbols.

If a code has a minimum distance of d, for a given X data we can change d symbol in X to get data Y.

So it’s clear that minimum distance is an extremely important concept of error correction.

 

Without going trough a long demonstration we can jump to a burst of errors.

In the case of 2 erroneous data t₁ and t₂

d=t₂+ (2t₁+1) t₂≥ 0

Then evidently a word with 3 or more errors can be considered to be the same as the next data word so it can’ be judged erroneous.

[alfe]

Jumped to fast

 

d=t₂+ (2t₁+1) t₂≥ 0

d=5 for CD audio

 

Witch mean: Correction of 0 symbols t₁=0, t₂=4 no correction occurs even if there are 4 erroneous symbols.

Correction of 1 symbol t₁=1, t₂=2 one symbol is corrected 2 errors can be recognized

Correction of 2 symbols t₁=2, t₂=0 we can no longer define t₂ hence why 3 errors or more are considered as the same as the next data word.

[edbk]

So do I understand you correctly that corrupted data can actually be read and not be detected as being wrong at all? And the to be expected default error rate is 10 times a second? Hmm

[alfe]

And the to be expected default error rate is 10 times a second? Hmm

Before error correction after the average go to 10^-12.

^{With a proper design as stated in the Doc #24}

[edbk]

After error correction as in unrecoverable errors?

 

Still 10 a second, although obviously undesirable, could still be acceptable for audio. I assume this doesnt go for data storage? Data is stored differently then audio?

 

Edit: ofcourse it is but data can be recovered 100% right? Just audio cant?

[mansr]

After error correction as in unrecoverable errors?

 

Still 10 a second, although obviously undesirable, could still be acceptable for audio.

 

No, 1e-12 is roughly one error every 8 days.

 

I assume this doesnt go for data storage? Data is stored differently then audio?

 

Edit: ofcourse it is but data can be recovered 100% right? Just audio cant?

 

Data tracks have an additional level of error correction that reduces the error rate further. It is impossible to get an error rate of zero with any medium. Hard drives typically have an error rate around 1e-15.

[alfe]

Burst errors can be the result of fingerprint, scratches or servo synchronisation.

Now if we go back to the 1^st statement: all form of error checking involves adding something to the digital pattern, guess witch distribution will be used for the CRC checksum.

I think with all these information’s it will be easy to link the dot and find the evident conclusion.

Mine is different bits should sound different

[alfe]

No, 1e-12 is roughly one error every 8 days.

 

 

 

Data tracks have an additional level of error correction that reduces the error rate further. It is impossible to get an error rate of zero with any medium. Hard drives typically have an error rate around 1e-15.

 

Sorry my mistake

CD data 10^-12

CD audio 10^-10

Burst error rate CD audio 10^-4

[alfe]

For finger print and scratches the distance d of the code is correct so CD players will go for error concealment (interpolation) and the drive will reread for ripping.

In the case of overlapping burst errors, d will be affected so different bits same checksum.