[AdNauseam]

Quote:

Originally Posted by Notorious P.I.G. There isn't any resampling. The original sampling is done at just above the Nyquist criterion--thus, data loss from a dynamically sampled signal is inevitable.

Doesn't the Nyquist rate just say that the maximum frequency is half the sample rate? There should be no data loss from what is read off of the CD and what is output by the digital out(excluding compression).

 

[Notorious P.I.G.]

Quote:

Originally Posted by AdNauseam Doesn't the Nyquist rate just say that the maximum frequency is half the sample rate? There should be no data loss from what is read off of the CD and what is output by the digital out(excluding compression).

Not exactly. It say that the minimum sampling frequency Fs must be greater than twice the highest analog frequency.

I meant data loss from the compression, not the read.

 

[erikzen]

Sorry if this thread is dead and this may be slightly off-topic, but does anyone know how the digital out on the D-303 is? Is it worth it to get a DAC for that use?

 

[EyeAmEye]

Well, I recieved the Sony D-EJ885 in the mail today (white model, it looks awesome). Plugged the HD595 into the line out, and the same results as the D-E551. The HD595 gets driven to equal volume on both D-EJ885 and D-E551, which is way more than any other portable with a line out that I own, which is quite a few ranging from older PCDPs from Sony, Panasonic, Aiwa, Kenwood, RCA and even a Fisher.
Again it has me puzzled as to why the HD595, which from the headphone out of all the PCDPs is the least efficient of all the cans I have owned (save the Koss Pro4-AAT, but that is a dreadful tale for another day

), is by far the most easily driven from the lineouts of just the D-E551 and D-EJ885.
To add some to the g-protection/no g-protection debate, from the lineout of both D-E551 (defeatable ESP) and D-EJ885 (non-defeatable g-protection on setting "1"), there is no discernable sound difference, atleast that I can hear. If there is compression of the sound from the line of the D-EJ885, it is very subtle. What effect it may have on the digital signal vs. non-compressed from the D-E551, I will have to experiment by using both as sources to record to minidisc.
BTW, I dug up the manual to the D-E551, and it is the same model as the D-E555, save some accesories. So for those that have a D-E555 and like the lineout, the D-EJ885 has similar performance. Headphone out has to go to the D-E555/551, more power and a cleaner top end, less mud on the low end. Distortion on the D-EJ885 is horrible at max/near max volume, which is only 5mW. The lineout is impressive though.

 

[halcyon]

Corrections:

1) Not all digital output (whether coax or optical) are the same.

2) Differences in digital outputs can be due to: jitter, internal player (lossy) compression and bit-truncation (due to poor design and/or volume control). Maybe there are other reasons, but I'm not aware of them.

3) Jitter means that the bit-wise signal itself is theoretically identical, but with frequency modulation in the signal. This modulation can cause measurable distortion (mostly noise, if uncorrelated with original data) at the DAC, because most cheap DACs reclock to the incoming spdif stream (with it's jitter). This has been detailed in Julian Dunn's "Jitter Theory" paper.

4) Many cd players have internal anti-shock buffers that compress the music, in order to fit more music into the memory. This compression may in theory (if it is lossy) cause loss of data via the digital out. I haven't seen a measurement for this, but it's not impossible theoretically.

5) Some cd players (as an example) employ digital volume control that also messes up the digital ouput bitstream. When this is done, the bit depth is truncated and dynamic range is lost (i.e one doesn't get the identical bitstream off the disc when the volume position is changed. In some situations it is impossible to get the 1:1 bitstream off the disc even with the volume in full position). This has also been measured/proven.

BTW, error correction does not mess up the data after it is gone out from the digital output. Error correction only works when reading from the disc and before the data is pumped out through the digital output. If bit level errors happen after the digital output, they must be concealed as there is no mechnism to correct them.

Now, whether these differences can account for audible differences in the audio, remains to be proven, but I'd be surprised if 4/5 weren't audible.

regards,
halcyon

 

[bangraman]

Coalescent, I don't believe the 'normal' mode on the iMP-550 is unbuffered nor uncompressed. Correct me if I'm wrong though.

 

[thomas]

First off, G-protection does not use compression, though many of the older "classics" with 40 second buffers did use compresion

I have a 5 year old Sony D-ej01, the first with G-protection. It uses a 16 mbit DRAM chip for the antiskip buffer, and gives about 11 seconds between physically stopping the disc from spinning and the music to stop playing. This matches perfectly with the 11.3 seconds of cd audio that could theoretically fit in a 16 mbit chip without compression. Since the G-protection transport can recover from shocks about 10 times faster than older mechanisms, it really doesn't need the 40 sec + buffers of older models, and i've never had the player skip on me.... With newer technology and cheaper RAM prices, i don't see any reason for the new high-end G-protection players like the d-e2000 to use compression...

Of course, any player that uses 40 second buffers must use compression, as 16 mbits has been the practical limit for all CDPs. To fit 40 seconds of audio into a 16 mbit chip, 4:1 compression would be required, and this is certainly lossy (the best lossless codecs manage at best 2:1) and would certainly degrade the sound. But the panasonics i tested shut off their antiskip buffers when an optical cable was attatched. The cheaper Sony's might use a 4:1 compression with a 4mbit DRAM to cut costs- and the datasheet of the g-protection controller states that it is capable of compression/decompression- but such a cheap player wouln't have optical outs anyways...

So the only players that might use lossy compression through their optical outs are the "classic" sony cdp's that supposedly have the best sound quality....


Halcyon brings up an intersting point in that a digital volume control might degrade the sound- and virtually every player made since mine has used some form of digital volume control rather than a potentiometer- But i think the "digital volume controls" are actually digitally controled analog stepped attenuators, placed after the DAC, rather than a DSP that changes the bits of the sample- since most players have both fixed line outs and variable headphone outs. But I'll have to do some more testing on this to be sure....


Error correction can result in a difference in the bits output by the transport, but i really doubt it could account for significant differences in sound, esp. on clean discs. The error correction block, in theory, should only apply the reed-solomon code to mathematically calculate the EXACT bits found in the original. I've seen some audiophiles mention that the error-correction chip "guesses" samples when it isn't sure, but i've never seen any mention of this in technical documentation, and it sounds too complex to implement anyways. And with $20 cd-rom drives that can read discs with virtually 100% accuracy using EAC, i don't see error correction being a significan't factor. If error correction was responsible for differences, then the IHP-120 should have the best sound quality if .wavs ripped by EAC was played on it.

Error correction after the transport is useless, which is why there isn't any. You can send bits down a wire/optical cable with no loss, unless there is a freak burst of cosmic radiation...(wait, now that i think of it, the D-ej01 and other first gen. G-protection players did have a slight anomoly in the s/pdif digital stream when the motor speed changed (causing a large voltage fluctuation thoughout the player), but this only occured once every 9 seconds, which shouldn't be enough to cause differences in sound quality and i'm sure it was solved in newer models)

SO the only thing i can think of that can cause differences in sound is jitter, assuming that the volume control is not in the digital domain. This would also be the only explanation for the poor performance of the IHP if it is indeed bit-perfect. I'm sure most portables have tons of jitter, due to the low cost components.... But jitter ONLY comes into play at the point of digital to analog conversion... So the fact that jitter affects the sound is the fault of the DAC, not the portable! Different DACs have different sensitivites to source jitter, so a poorly designed DAC may show all sorts of sound quality changes with different sources, while a good DAc could sound the same. Could it be that the digital upsampling phones- and upsampling process is known to be much more sensitive to jitter than normal d/a conversion- are revealing the differences in jitter between players that other DACs may not?

 

[Eagle_Driver]

Quote:

Originally Posted by thomas With newer technology and cheaper RAM prices, i don't see any reason for the new high-end G-protection players like the d-e2000 to use compression...

Yep, the compression on the most recent high-end G-protection D-EJ2000 player - and other new G-protection players, as well - can be switched on or off. The G-Pro 1 mode on those players doesn't use compression (in other words, the sound is 'noncompressed'), while the G-Pro 2 mode does. However, the G-Pro 2 mode uses a 2:1 compression alogarithm rather than the crappier-sounding 4:1 alogarithm. (On these newer players, the G-protection laser recovery mechanical circuit stays on all the time, and cannot be turned off at all.)

Many of the older G-protection Sony PCDPs have an On/Off switch for the G-protection circuit. In those players, there is no compressed mode (or put it this way, all buffering is 'noncompressed') - but the laser mechanism would be slower to react to bumps with the G-protection switched off than with the G-protection switched on.

The D-EJ01, D-EJ815 and D-EJ915 have no G-protection switch at all whatsoever. The G-protection is always on when using only the headphone out, or it's always off when you're using the Line/Optical out.

In all of the above cases, since the advent of G-protection, the buffering on the Sony players is always on (and in 'noncompressed' mode in most cases), and cannot be turned off at all whatsoever.

The iRiver SlimX series PCDPs are practically the only PCDPs that can claim up to 45 or 90 seconds of 'noncompressed' anti-skip. That's due to their use of an unusually large memory buffer - 64Mbit on the iMP-350, 128Mbit on the iMP-400 and the iMP-550 (as opposed to the usual 16Mbit on most of today's PCDPs). But those larger buffers contribute significantly to their relatively poor battery life.

 

[halcyon]

<sigh>

Why do people have to guess, when there is perfectly good information available that proves things.

0. Real-time error correction algorithms used on cd players on Red Book Audio discs do not theoretically detect all possible errors nor can they correct all errors. When a player does detect a mismatch in the ECD block that it cannot correct, it will conceal the error. There are various ways to conceal this error as I've explained earlier. This is all part of a normal compact audio disc playback. Proof: all this is explained in Sony's technical book about compact disc standard called "Digital Audio Technology" all down to the nitty gritty details of the block level diagram off the CIRC encoder/decoder and description of the concealment methods.

1. Digital volume controls that truncate bits
Proof of an example player that does this:
http://peufeu.free.fr/audio/extremist_dac/bit_accuracy/

2. Transport jitter is not the fault of a DAC. Proof: read a good book like "Principles of Digital Audio" or just Julian Dunn's tech note on Jitter (Audio Precision, free on the web). This includes simplistic estimates of jitter audibility (better ones from AES papers).

3. Many new Panasonic (and other "high end" players costing as much as expensive Sonys) have up to 40 second playback buffers (and more). My CT-570 produces distorted optical digital output with it's 40 secs compressed anti-shock (not only that, it overdrive's it's own DAC into distortion).

So yes, all three of the above are measured and proven ways to mess up a digital output.

regards,
halcyon

PS Analog stepped attenuators are quite expensive... it's just easier (cheaper, more convenient) to trunacate bits esp. in portables.

 

[bangraman]

OK, I read that... some of it anyway

Having read the specs on the iRiver, it does indeed seem to offer uncompressed buffering... But in practical terms, you take advantage of it if you like dropouts throughout your listening. The 'Normal Mode' had problems in either reading discs or flushing the buffer when the player was on a table (i.e no shocks).


The D-EJ2000 to my knowledge has a 2Mb buffer, which when translated to 40 second anti-shock would seem to indicate 4:1 compression, no? How long was G-protection 1 anyway? 10 seconds? I don't recall it being that short.

 

[Duncan]

Certainly, from a Sony perspective - I would have thought that only 'ESP-Max' players would suffer from compression...

ESP-Max was 40 second anti-shock iirc, which would require an 8mb buffer to be uncompressed... other than that, before the event of G-Pro and ESP-Max, the best anti-shock there was on offer was 20 seconds... or, 4mb.

There was ESP2 added into the equation, but that was merely an algorithm update to the best of my knowledge.

 

[Coalescent]

Quote:

Originally Posted by bangraman OK, I read that... some of it anyway Having read the specs on the iRiver, it does indeed seem to offer uncompressed buffering... But in practical terms, you take advantage of it if you like dropouts throughout your listening. The 'Normal Mode' had problems in either reading discs or flushing the buffer when the player was on a table (i.e no shocks). The D-EJ2000 to my knowledge has a 2Mb buffer, which when translated to 40 second anti-shock would seem to indicate 4:1 compression, no? How long was G-protection 1 anyway? 10 seconds? I don't recall it being that short.

Bangraman,

Have you had problems with dropouts while listening to the iRiver on 'normal' mode? I use mine that way exclusively, and I rarely have such problems... even in my recent car trip across the 'states, during which I listened to my iRiver for many hours while I was a passenger. If I had experienced issues like you describe, I would've stopped using normal mode long ago... I'm just wondering what might account for our different experiences.