But it is still used in optical communications. And the Cd is in that domain momentarily. The spacial filter for the CD basically limited the field of view of the lens in the receive direction. We were able to measure and show a marked improvement of the optical signal. This was still in the optical domain so clock errors nor corrections were included. It made a pretty significant change to the CD player's sound, the cheaper ones had more improvement
i need to find that old stereophile article
i need to find that old stereophile article
John:
That is a mistake I see too often here and its misleading. Distortion is not a cause or specific property of an amp. Its a symptom of its linearity. You are right in that the IM can be caused by the same nonlinearity that the higher harmonics are.
Optical filter
If there was any change to the sound it would have to be due to a reduction in uncorrectable errors or E32s. This can be observed easily in many players by finding the pin on the error-correction chip which flags E32s, and driving a pulse-stretcher/LED driver. If on that single player, between two bit-identical discs there are no E32s there can be no difference in sound.
It's late, so GN, and thanks for opening old wounds! All joking aside, optical discs were a fascinating way to spend a portion of an engineering career.
Howie
Howard Hoyt
CE - WXYC-FM 89.3
UNC Chapel Hill, NC
www.wxyc.org
1st on the internet
If there was any change to the sound it would have to be due to a reduction in uncorrectable errors or E32s. This can be observed easily in many players by finding the pin on the error-correction chip which flags E32s, and driving a pulse-stretcher/LED driver. If on that single player, between two bit-identical discs there are no E32s there can be no difference in sound.
It's late, so GN, and thanks for opening old wounds! All joking aside, optical discs were a fascinating way to spend a portion of an engineering career.
Howie
Howard Hoyt
CE - WXYC-FM 89.3
UNC Chapel Hill, NC
www.wxyc.org
1st on the internet
High feedback vs low feedback, op amps vs discrete vs transformers.
In the last three months I have tried about 10 different analogue output stages on my Sabre DAC ( running in voltage mode ). They included 2 highly regarded transformers, 3 highly regarded op amps and several versions of a very simple discrete line level amp.
The discrete circuits were designed in spice. All versions were judged by ear ( not just my ears )
Active versions were all powered with the same shunt regulated supply.
There are still some variations I want to try but thus far the circuit that was subjectively judged clearly superior is one of the discrete line level amps - the main positive characteristic was the natural presentation of extreme fine level detail.
This particular cct was designed to be completely stable without any Lag ( miller ) compensation. It has heavily degenerated 1st & 2nd stages, has about 14db of feedback, uses 7 solid state devices ( 5 of which are jfets ) and good quality resistors throughout.
After I implemented it I was somewhat bewildered that such a simple circuit could sound so much better than the op amps that had such amazing specs. After that I was wondering why is it op amps have such extraordinary high levels of gain. One of the op amps was using over 100db more feedback than this discrete cct.
Charles Hansen has stated that using any loop feedback to any part of a audio circuit ( even in the PSU ) brings a degradation the sound - not sure if he still holds to that view.
I don't have such a strong feeling against FB but based on my recent experience I have concluded that IF you have goal of resolving very fine low level detail then feedback should be employed only to just below the level where lag compensation would be necessary to maintain very good stability.
I know that to ardent objectivists this anecdotal report will mean next to nothing - sorry about that.
In the last three months I have tried about 10 different analogue output stages on my Sabre DAC ( running in voltage mode ). They included 2 highly regarded transformers, 3 highly regarded op amps and several versions of a very simple discrete line level amp.
The discrete circuits were designed in spice. All versions were judged by ear ( not just my ears )
Active versions were all powered with the same shunt regulated supply.
There are still some variations I want to try but thus far the circuit that was subjectively judged clearly superior is one of the discrete line level amps - the main positive characteristic was the natural presentation of extreme fine level detail.
This particular cct was designed to be completely stable without any Lag ( miller ) compensation. It has heavily degenerated 1st & 2nd stages, has about 14db of feedback, uses 7 solid state devices ( 5 of which are jfets ) and good quality resistors throughout.
After I implemented it I was somewhat bewildered that such a simple circuit could sound so much better than the op amps that had such amazing specs. After that I was wondering why is it op amps have such extraordinary high levels of gain. One of the op amps was using over 100db more feedback than this discrete cct.
Charles Hansen has stated that using any loop feedback to any part of a audio circuit ( even in the PSU ) brings a degradation the sound - not sure if he still holds to that view.
I don't have such a strong feeling against FB but based on my recent experience I have concluded that IF you have goal of resolving very fine low level detail then feedback should be employed only to just below the level where lag compensation would be necessary to maintain very good stability.
I know that to ardent objectivists this anecdotal report will mean next to nothing - sorry about that.
Last edited:
This is how I and many others determined that this is a basically non-existing problem. It was particularly easy to do in the old Magnavox/Philips players (I used a 582 at that time), and if I recall correctly, John Atkinson published a how-to squib. I think they also ran an article surveying CD tweaks, found no significant differences, then danced for a while about how mysterious the reported sonic differences were.
I'll be kind and not bring up the Armor-All fiasco.
I beg to differ, because it seems to be a bit misleading.
Uncorrectable errors are a quite unlikely cause, as the number during any length of replay is usually to low to have a certain (reproducible) effect soundwise.
So, if there is any audible effect, it could still be caused by something like LIM or power supply related or .....or....
Instead of monitoring every other circuitry inside the player it might be easier to monitor the analog audio output.
There are those SHM, super hard material?, CD's and I've read about glass CD's once so there should be some literature there, but it most likely won't discuss the transferal of 1's and 0's back and forth, if we copy a CD twenty times there should be software which can compare the result to the original or check for jitter performance.
As far as 1's and 0's are concerned then jitter seems to be the topic when looking at say PCM27xx versus CM6631, I.e. USB transfer and we can allege that the latter sounds more accurate. Likewise the same applies to coax and optical decoders. The sound quality in optical is mostly decided by DIR9001 or the Cirrus Logic alternative, not the optical cable unless it can't transfer the data correctly for some reason but it should transfer the data without any EMI, RFI or power supply effects, unlike USB which may be suspect to that and interfere with the PSRR or EMIRR of the external device it's sending to.
Just a few viewpoints.
This would be extremely easy to test for and get actual data from the electronics. Where is it?!?!?
You have to make sure something's real before you generate theories for it ... the exact opposite of science.
Why don't you do some work to actually back up the very large amount of fluff you post? That would actually be worth something ...
Hypothetically glass CD's could have a higher error rate and thus sound different.
I'd suspect not.
The paradox is we can send data via USB back and forth perfectly so why should audio not follow the same rules? I'd suspect an external hard drive "doesn't care" about the error or interference but a few audio components may pick up the power supply noise, EMI or jitter and then under-perform a little.
With that said CM6631 is most likely more immune or accurate than DIR9001 in the comments I've skimmed over so USB may still perform slightly better than optical as long as the interference isn't severe, due to the receival chips, not the medium of transfer itself.
I've heard the difference between optical and coax in a basic blind test, just hitting the select input switch quickly xx times and then identifying, but I suspect it's the nature of the receiver chip and unlikely the cable itself.
This is often asserted. I strongly suspect that it is false.canyoncruz said:
The idea seems to be that the error correction circuitry (or software) has to work harder when there are errors to correct - as though the circuit checks for errors, and if it finds any then has to work to correct them. My understanding is that the decoding logic does not work like this. It is presented with a buffer-full of raw data, which may or may not include some bit errors, and it outputs another buffer-full of processed data which contains no errors. The first buffer is longer than the second buffer, as the first buffer contains redundancy. The processing time and effort is exactly the same regardless of how many errors are present. Essentially, the circuit just maps a bit pattern from the input to a different bit pattern at the output. Redundancy and eror correction simply means that several different input patterns all map to the same output pattern. Anthropomorphic approaches to understanding CD players can be misleading.
One exception to the above: if error correction is unable to correct the errors then interpolation happens. This is rare, but may be audible when it happens. Apart from this I say the amount of error correction cannot affect the sound.
We are not yet at a point where classical distortion measurements tell very much useful about how equipment sounds, how well a system performs subjectively. Having listened to equipment over the decades sound all the way from superb, to atrocious, with a big fat zero of useful information telling me what's happening, why this is so - mainly because of a high level of disinterest of the people in the game for doing intelligent research into such ... I have zero interest in doing the work that plenty of others should have already invested some serious time in looking into.
Something's real if I'm aware of it - if a car rolls over when it goes around a corner, it doesn't impress me that people spend decades shrieking that you have to supply "data", to prove that cars go tipsy turvy on occasion. I can hear audio systems going rolly polly all the time, and it bores me intensely hearing their aural incompetence - I'm only interested in fixing the mess.
If you were vaguely paying attention you would have noted that I am actually looking at a few things worthwhile - rather than rigorously inspecting navel fluff at a microscopic level for the 5,000th time ...
In what kind of error correction could interpolation happen? Any photo or video analogy for this?
Howard explained this very carefully a few posts back.
Well, yes, in the sense that if you painted over the front of your headlights with black paint, it could affect your ability to drive at night.
All of this can easily be googled or found through a search on this site. Please work hard to improve your basic understanding of how these things work. Lernen!
When there are too many errors for error correction to work. Google will help you find when this occurs. Because of the data interleaving (you can Google that too) you need a lot of disc damage or player malfunction for interpolation to happen.Kastor L said:
Note that there is not a gradual decline in sound quality; there are either enough valid bits for a 100% correct decode to take place, or not enough valid bits and no decode can occur so interpolation happens instead. No half-way house of a roughly correct decode.
- Status
- Not open for further replies.