Linear Powered Rips & flash drives sound better - Alex was right !

[sandyk]

Well, this is more akin to your first analogy but did Alex say that the affect disappears when audio files are transmitted in this way?

Martin Colloms performed a series of 6 separate Blind A/B/A/ 3minute sessions with the files that I had uploaded, and his panel of listeners were still able to correctly identify which version of the track in each instance. Eloise and others refuse to accept those results though, because the tests didn't go as deeply as they wanted. At the very least, Martin Colloms results performed to fit in with magazine production deadlines, should have indicated a much more in depth investigation was needed. Martin Colloms later used the same tracks stored on USB memory with a later model Naim Unitiserve and the differences were reported as even greater. If I was to do this again (unlikely) I would send the files directly on either a USB memory stick or CD-Rs as I did for Barry D. I have found that Internet transmission and saving again causes degradation of the SQ of both versions .

That is also why Cookie Marenco from BlueCiast Records provided her DSD (etc.) files as UNCOMPRESSED .Zips in order to reduce SQ degradation.

[DarrenMc]

Alex... You are talking complete rubbish. There is absolutely no mechanism where information can be stored within files in this way.

 

This discussion is pretty much identical to the one you'd have if you tried to argue a believer in homeopathy out of their delusions.

1. It's impossible in theory.

2. It fails all properly conducted (ie: blind) tests.

3. Ergo, it probably isn't true. Move on, and leave the believers to it.

[mmerrill99]

No it could not.

 

The transmission of a file across a network is not weak or malformed. It is read accurately or not.

Of course - that is how digital works - at the logical level. It allows some leeway in the electrical signalling.

Any malformation of the wave sent gets converted and any malformation after that will bear no relation to the initial malformation if one exists. There is no way for the variation in original signal quality to propagate to the next version (copy) of the original. That such is the case is one of the strengths of digital storage for information including music. Again that is not possible.

Again, you are looking at this at the logical level & I'm trying to examine the underlying physical/electrical level as a possible mechanism.

[alfe]

Story does have a lotta holes, very few electrons.

 

Then keep doping it you may see the light.

[JohnSwenson]

Hi all,

I see I'm being referenced here, I haven't had much time to keep up with these threads, but I thought I should quickly go over my thoughts on this. First off I have NOT done extensive actual research on the subject, this is conjecture on my part from some knowledge on how things work and how it MIGHT affect sound quality.

 

My experience is in the magnetic media side of this (ie HDDs not flash). Modern drives do not magnetize their media to saturation. The field generated by the write head does not completely switch all the domains to one orientation, it changes orientation population to be something between 50% and 75% in a particular orientation. This is well within the "linear range" of the material. The upshot is that fluctuations in the power supply in the drive when it is writing the data DO get recorded in the magnetic material.

 

The tracks in modern drives are REALLY thin, which means the analog signal coming off the read head is very small, it takes some very complex systems to extract the ones and zeros out of this noise. (if you look at the raw signal coming off the head you would not say it is digital data, it just looks like random noise). Variations in that input signal do cause varying current draw in the read electronics. So things that cause power supply noise in the computer when it is writing data can definitely cause noise in the power system when reading said data off the disk.

 

This noise when reading is not going to be exactly the same as when writing, it has some correlation but it is not the same. The part I have discussed up to now I know is true, it is measurable.

 

Whether that noise affects sound quality is conjecture. The tools necessary to trace all this down from beginning to end are quite expensive, no way I can afford them personally. The noise mentioned above has been measured at work using the expensive equipment there, but I cannot use that stuff for "playing around with audio". So at least from me it is going to remain conjecture.

 

Given what I have been able to hear with changing noise in power systems, it does make sense to say it seems possible that this noise generated by the receive electronics could have an affect on the USB system in a computer which could affect the sound quality produced by a DAC.

 

As to whether this noise could affect a copy, it's possible if the reading was happening at the same time as the writing. If the data was read off one disk into memory, the read drive was turned off, then another disk was written with the data from ram, I don't see how the above mechanism could come into play. Not to say there is NO mechanism, just not the one described above.

 

Whether these effects are stored in RAM is another issue. I have not really looked into this in any measurements, so again it is conjecture. It seems to me it depends very much on the type of memory. In static ram the data is stored by mechanisms that ARE saturated so it is essentially impossible to store analog info from signal strength or PS variations at the time of write. DRAM is different, it stores data as electrical charge which CAN store PS variations etc, BUT the charge leaks away quickly so it must be refreshed periodically, the process of refreshing reads the data and re-writes to a charge dependent on the PS voltage at the time of refresh. So after a fairly short time any variations due to PS at write time have been over-written with PS at refresh time. So if copying from HDD to DRAM it seems that any influence from reading the HDD would be gone by a second after the drive shuts down.

 

And I do mean shut down, just not officially reading off it may not be enough to stop influence. If the drive is still powered up the read electronics are frequently still processing the last track read, even if the computer is not doing anything with the data it is supplying. So even if the computer is just reading from the RAM, the noise from the HDD could still be affecting the SQ.

 

One of the other threads right now is talking about vibrations and ripping, I gave a reply on seismic noise and DVD servo loops, the electrical noise produced by seismic noise (vibrations) inside the DVD drive is quite strong, I would think this would be a strong component of any noise being "recorded" on an HDD.

 

John S.

[esldude]

I'm not sure that saying "it's in the bits themselves" is accurate - in fact I believe it confuses matters. A bit is just an agreed upon interpretation of an electrical signal when it's voltage crosses the agreed thresholds. There are just two voltage thresholds for interpretation of 0 or 1. This means that once the signal voltage crosses above the agreed threshold it is interpreted as a 0 & below another agreed threshold it is interpreted as a 1. This means that there are many different electrical waveforms that will be "bit identical" but not "electrically identical". So something can be bit identical but be represented by different waveforms. So the "noise" will be encoded in the electrical waveform but the bits are still correctly interpreted as a digital bit stream Yes that bitstream is the only connection between the optical disc & the DAC. So this bitstream gets written to HDD or SSD. Now, my question is can these two bit-identical but electrically-different bitstreams result in a difference in the signal strengths of bits written to HDD or SSD because there is a different profile to the current draw during the processing of these different bitstreams?

 

I don't know how many ways we can tell you know, what you are imagining might be happening cannot happen.

 

Okay, I will make one last attempt and then follow DarrenMC's excellent advice.

 

Okay lets us say 1's are any voltage between 2.5 and 5 volts. Zeroes are anything between .8 v and zero. In between will not be recognized as either. Now the actual 1's can vary between 2.5 and 5 volts hopefully with most near 5 volts. One drive or machine may have different variance patterns between the acceptable voltages. Due to any number of things. So we read from a hard drive to a memory stick. The hard drive reads this stuff onto data buses other chips doing lots of things will route onto the correct area to go to the USB memory stick. If say the original signal off the hard drive is kind of low with lots of bits between 3 and 3.5 volts will that pattern go on through to the memory stick? No the two will not be related. And even if they were, once you then put that memory stick into another machine it is storing bits at levels that aren't related to the other machine. There will be no connection.

 

Further in the case of memory sticks, they no longer hold one bit per voltage value. Depending on whether it is MLC, SLC or whatever it stores more than one bit per cell. It may have four possible states, and will use 4 voltage levels to store those. So the conversion from a binary voltage level where it is either one or zero will get turned into one with 4 voltage levels representing more than one bit per voltage level. Meaning any pattern from earlier will not transfer directly in any case.

 

I don't know if I am making any sense to you. But what you think might be happening simply cannot happen.

 

And I would ask, if this is not convincing to you, what would convince you? If you can't be convinced then we wasting our time.

[mmerrill99]

Just to try to explain myself a bit further - I'm not saying that the exact same weakness in the waveform is transmitted from one step to the next, what I'm trying to say is that the original weak waveform when being interpreted as a digital signal at the next receive/transmit step in the signal path results in a different current draw profile on the PS than a well formed waveform causes. This different current draw profile can result in voltage fluctuations which causes the transmitted waveform to be somewhat malformed (but not exactly a copy of the original waveform - just a similar level of malformation). This same process can occur from one step to the next.

 

One thing I have noticed is that digital audio is very sensitive to PS quality & not just at the DAC, also at the computer (source) end. This is assumed to be the result of decreased noise being transmitted to the DAC but I'm also considering that it could possibly be the result of a cleaner waveform. I've heard these audible improvements continuing to occur as each PS in the PC is separated & addressed with a higher quality PS.

 

In stock form the PC is perfectly capable of all the digital processing needed. With separated & improved PS something has changed which results in audible improvements in DACs. Now I understand noise can be transmitted in many ways & is difficult to isolate but from my reading of J Swenson & others, I believe that even with all the isolation techniques used, there is still something that gets through from the PC which has an impact on the audio output.

 

Can this be possibly explained by the differences in the actual electrical waveform that represent the bitstream?

 

Oh, I see while I was writing this other posts were made - one from JS - great look forward to reading it

[sandyk]

And I would ask, if this is not convincing to you, what would convince you? If you can't be convinced then we wasting our time

 

That's rich coming from someone who believes that everybody who reports hearing improvements with 24/96 or 192, DSD or double DSD is delusional, and that manufacturers of equipment capable of playing these formats are rogues and charlatans who rip off people. Unfortunately for you, many members are now hearing and appreciating the transparency of, and even further improvements from double DSD !

[Audiophile Neuroscience]

Oh dear, I think people did not understand or have ignored the spirit of my original posts with requests not to turn this into yet another pointless bits is bits (or related) argument. This is not to imply that people are not entitled to their opinion or that valid points have not been made on either side but simply that it has all been argued before. Moreover, without resolution.

My humble suggestion to Chris Connacker or other admins is that when parts of threads have gone off topic or have been hijacked, that they are moved to a new topic. I appreciate that this may not be possible due to time constraints but without such a mechanism I fear that any topic dealing with subjective perceptions will be smothered by the same argument.

You can always start a new thread and link back to the original reference thread. Indeed a whole forum could be created with the relevant topics dealing with subjective perception vs subjective interpretation of data ( and yes I realise that statement is in itself controversial). It could be kicked off with the thread I started, "Where is Audio Truth". So I am not averse to these discussions per se just disappointed that nearly every thread seems to get suffocated.

I am simply asking people to respect that if an OP asks to stay on topic and to quarantine areas of discussion, that this is respected.

That said, a) it is too late for this thread and b) I think both tran #3 and Peter(thesurfingalien) #4 ask legitimate questions (my apologies to others as I have not read too much beyond that). I can say that neither Dennis nor Alex where responsible for orchestrating the listening order of music or equipment. We had available both linear and nonlinear rips of the same track and played back from a HDD or Corsair flash with and without the JLH/LPSU. At different sessions ( with 4 of the same 5 people present) everyone preferred linear ripped material over nonlinear ripped material (all other variables being held equal) with the strongest preference for linear ripped tracks to a corsair flash drive played back via a JLH/LPSU…..as Ripley said, believe it or not!

As for Dennis, present only at the last session, he heard

Linear Rip + Corsair/JLH-LPSU/PC

Linear Rip + HDD/JLH-LPSU/PC

Non Linear Rip + HDD/JLH/LPSU/PC

Linear Rip + HDD/Macbook Pro

His clear preference was for Linear Rip + Corsair/JLH-LPSU/PC. I was accused of sabotage as I inadvertently forgot to use the JLH/LPSU with the mac….but, as we all know pc's are better than macs anyway …just kidding ;-). IIRC his preference in Linear Rip + HDD/JLH-LPSU/PC vs Non Linear Rip + HDD/JLH/LPSU/PC was for the former. No point in asking Dennis as he often did not know what combinations he was listening to.

FWIW the pc and mac have similar specs, OS not tweaked (yet), use JRMC20 using memory playback.

Once again, this is not a scientific test, just an observation. It is not aimed at sparking measurements vs listening debates or whether differences are real or imagined.

David

[mmerrill99]

Thanks JS, as always, an interesting & informative read.

I agree it's all conjecture & what I am suggesting is probably wrong but we probably have to go down many blind alleys before finding a way through. Until such time as these highly complex systems can be analysed at the very low levels needed to gather sufficient data, it will all probably remain conjecture.

Appreciate your willingness to comment on these controversial topics & relate your experience in your areas of expertise.

[mmerrill99]

David,

Apologies if I've been instrumental in dragging this thread down an unwanted avenue.

I've said all I need to now so will desist from further posting

[wgscott]

Oh dear, I think people did not understand or have ignored the spirit of my original posts with requests not to turn this into yet another pointless bits is bits (or related) argument.

 

That is how all troll-posts begin.

[wgscott]

It is not aimed at sparking measurements vs listening debates or whether differences are real or imagined.

 

Hence the "Alex was Right!" proclamation in your title.

[jabbr]

Okay this sounds like 5 dudes got together to drink some scotch and watch a magic show.

 

Do you understand why it isn't convincing?

[jabbr]

Martin Colloms performed a series of 6 separate Blind A/B/A/ 3minute sessions with the files that I had uploaded, and his panel of listeners were still able to correctly identify which version of the track in each instance. Eloise and others refuse to accept those results though, because the tests didn't go as deeply as they wanted. At the very least, Martin Colloms results performed to fit in with magazine production deadlines, should have indicated a much more in depth investigation was needed. Martin Colloms later used the same tracks stored on USB memory with a later model Naim Unitiserve and the differences were reported as even greater. If I was to do this again (unlikely) I would send the files directly on either a USB memory stick or CD-Rs as I did for Barry D. I have found that Internet transmission and saving again causes degradation of the SQ of both versions .

That is also why Cookie Marenco from BlueCiast Records provided her DSD (etc.) files as UNCOMPRESSED .Zips in order to reduce SQ degradation.

 

"Uncompressed .zips" -- just love it

 

I understand. You are demonstrating that all internet communication becomes degraded as a function of time. Entropy. xkcd: Argument

[miguelito]

My experience is in the magnetic media side of this (ie HDDs not flash). Modern drives do not magnetize their media to saturation. The field generated by the write head does not completely switch all the domains to one orientation, it changes orientation population to be something between 50% and 75% in a particular orientation. This is well within the "linear range" of the material. The upshot is that fluctuations in the power supply in the drive when it is writing the data DO get recorded in the magnetic material.

Please correct me if I'm wrong...

 

The writing process utilizes a DAC to effectively make the input logical data into a waveform of 1's and 0's. This is not a literal translation, error correction is added.

 

http://en.wikipedia.org/wiki/Hard_disk_drive#Error_rates_and_handling

 

This waveform is analog and is what is recorded onto the surface. There's sector information, etc etc in the packets as well, all recorded as a series of magnetic variations. There is NO time information, you're just recording data onto a medium. When you read this waveform (which could be very "rounded", not square waves, and you likely want to avoid square waves anyway) that voltage info is then passed through an ADC process off of which the same series of bits that went in come out. Nothing more, nothing less. Like a CD, it probably uses Reed-Solomon error correction to know for a fact that the data is either accurate (after error correction) or known to be corrupt. All in all, the same exact bits come out as went in. And there's no time information of any sort, all of this is asynchronous. So there is no possible way that the data is any different if the bits match.

 

I do agree with you that there's possibly electrical noise filtering to the power bus, and this electrical noise can affect the very last stage of the --audio-- DAC process. So for example if a file is very fragmented such that the HDD heads needs to hop around a lot, the noise into the power bus would be higher. But as long as we are talking logical data, that data is the same regardless. BITS IS FRIGGING BITS!

 

One of the other threads right now is talking about vibrations and ripping, I gave a reply on seismic noise and DVD servo loops, the electrical noise produced by seismic noise (vibrations) inside the DVD drive is quite strong, I would think this would be a strong component of any noise being "recorded" on an HDD.

I don't think this is possible. If you rip a CD and there are no errors detected in the rip, this rip is a perfect replica of the original CD master. And copies of these files will be identical as well. The probability of a read error that is not detected as an error is 2^-19, so impossible in practical terms.

[kumakuma]

That's rich coming from someone who believes that everybody who reports hearing improvements with 24/96 or 192, DSD or double DSD is delusional, and that manufacturers of equipment capable of playing these formats are rogues and charlatans who rip off people. Unfortunately for you, many members are now hearing and appreciating the transparency of, and even further improvements from double DSD !

 

What does this have to do with anything we are talking about here?

[sandyk]

Okay this sounds like 5 dudes got together to drink some scotch and watch a magic show.

 

Do you understand why it isn't convincing?

 

Like the acerbic wgscott, you really are a sarcastic piece of work !

The only drinks consumed were cold glasses of water kindly provided by the host. Both audiophile neuroscience and a friend of his were also present at the previous session at his friend's house, (a different location) and no alcohol was consumed then either.

[miguelito]

"Uncompressed .zips" -- just love it

 

I understand. You are demonstrating that all internet communication becomes degraded as a function of time. Entropy. xkcd: Argument

++1

[alfe]

I don't think this is possible. If you rip a CD and there are no errors detected in the rip, this rip is a perfect replica of the original CD master. And copies of these files will be identical as well. The probability of a read error that is not detected as an error is 2^-19, so impossible in practical terms.

 

Perfect replica of the original CD master I'm still learning.

[sandyk]

What does this have to do with anything we are talking about here?

 

It demonstrates the same lack of credibility that you possess. Further smart ass/ sarcastic comments from you will be ignored by me.

[miguelito]

Perfect replica of the original CD master I'm still learning.

No comment here on the sound quality of those bits. I'm amused by the lack of content of your post.

[kumakuma]

It demonstrates the same lack of credibility that you possess. Further smart ass/ sarcastic comments from you will be ignored by me.

 

Actually it doesn't. It simply demonstrates that you are unable to respond to the technical issues of this discussion and have resorted to ad hominem attacks on various individuals who don't agree with you.

[elcorso]

Okay this sounds like 5 dudes got together to drink some scotch and watch a magic show.

 

Do you understand why it isn't convincing?

 

Who talks to convince anyone?

 

It is a simple tale of experiences of people who bothered to do a test taken and reported.

 

Innocent and well-meaning people like David, who had the courage to report in this impolite forum what he perceived, not have to be attacked by someone who doubts about their obsolete electrostatic speakers (and throwing a bunch of numbers in the air and even cites white papers to impress the newbies), the deaf Professor (who at least has a sense of humor, black, but humor at least), or the alien who has no experience in high quality audio gear. Ah, I missed some IT "experts", who believe their degree or experience in this field proves to review the musical perception of dedicated music lovers, but based only in digital theories.

 

Sheers for the test!

 

Roch

[prot]

I'm out of popcorn, heeelp!

 

Btw, how does one make obsesive-audiophile popcorn? . Is it safe to eat popcorn if the machine uses a lowquality switchin power supply? Will a popcorn piece pop 2x better if I put it alone in the machine? Or is there an optimal no of pieces that you have to cook simultaneously? (based of course on the most relevant params: the color of the machine and ur wife's name).

 

So many "important" Qs, so little time