Again a comparison with 1541A nonos.
Upper graph SAA7220, lower graph Kwak clock.
SAA7220 is about 0,5 dB better ( I double checked ) for both distortion and noise.
Difference in distortion can be seen best on K3 which is close left of cursor.
Noise is shown in the readout on the top of the picture.
-68,3 for SAA7220.
-67,7 for Kwak clock.
K1 is equal (0,0dB).
Upper graph SAA7220, lower graph Kwak clock.
SAA7220 is about 0,5 dB better ( I double checked ) for both distortion and noise.
Difference in distortion can be seen best on K3 which is close left of cursor.
Noise is shown in the readout on the top of the picture.
-68,3 for SAA7220.
-67,7 for Kwak clock.
K1 is equal (0,0dB).
Attachments
Guido Tent said:
Guido,
I honestly can not believe this.
Who said that I measure jitter ?
You are right, I measure something else:
The DAC @ -60dB signal level.
This is distortion caused by: TDA1541 + jitter + op amps + signal itself.
So if distortion looks exactly the same with to different clocks there are two possible conclusions:
1. Both clocks have same jitter. Very unlikely.
2. Jitter of the magnitude of SAA7220 is of no importance.
Konnichiwa,
Actually, there is a third possibility you omit to mention.
Your measurement technique, due to the use of a single tone is not sensitive (enough) to jitter.
Why not use a signal that makes sense, like -6db below full scale 11.025KHz tone and a 229Hz tone just toggeling the LSB. Cooledit should easily generate this kind of signal.
Then you need a spectrum analyser to "zoom in" onto the spectrum around the central (11.025KHz) peak. A broad peak suggests a lot of low frequency uncorrelated noise, distinct sidebands suggest the presence of jitter components with Fsignal - Fsideband / Fsignal + Fsideband.
Of course, as your aim is to pretend that jitter does not exist such a measurement is useless to you, hence you choose to use one that is comparably insenstive to jitter. Well done.
Sayonara
Bernhard said:
Actually, there is a third possibility you omit to mention.
Your measurement technique, due to the use of a single tone is not sensitive (enough) to jitter.
Why not use a signal that makes sense, like -6db below full scale 11.025KHz tone and a 229Hz tone just toggeling the LSB. Cooledit should easily generate this kind of signal.
Then you need a spectrum analyser to "zoom in" onto the spectrum around the central (11.025KHz) peak. A broad peak suggests a lot of low frequency uncorrelated noise, distinct sidebands suggest the presence of jitter components with Fsignal - Fsideband / Fsignal + Fsideband.
Of course, as your aim is to pretend that jitter does not exist such a measurement is useless to you, hence you choose to use one that is comparably insenstive to jitter. Well done.
Sayonara
Bernhard said:
It is fact.
But believe what you like, I don't want to destroy your illusions.
Deeper bass, bigger soundstage, smoother highs...
Could be the result would be different for the best DAC chip in the world, the TDA1543
I suggest you try a known decent clock first before again drawing premature conclusions. I don't have any illusions Bernhard after having built several tens of low jitter clocks into cdplayers of all different kinds. It is not a question of belief. When even untrained ( non technical ! ) music lovers are very enthousiastic I know enough. I once did a test with 2 CD634 cdplayers. One with a well implemented clock and one original. The owner ( of both ) could not believe that I only replaced the clock. We were clever enough to compare them both in original state and we couldn't detect any difference between the two.
I take your remark concerning TDA1543 for granted as I know you never heard the chip non os. BTW I don't have TDA1543 DAC's in use so please find another argument to make me ridiculous. Your subtle remarks prove you are the one that needs them to win a discussion.
There is nothing to gain for me by stating a low jitter clocks make a great difference in positive sense. If you don't hear any difference there some possibilities left:
a: Your setup has flaws.
b: Your low jitter clock is non optimal built or non optimal
implemented.
Concerning measurements and not sonical conclusions there might be a third option
c: Your measurements are not the right ones to measure jitter.
When a *lot* of people immediately hear that a better clock is mounted you can ask yourself that if you don't hear ( let alone measure ) any difference there must be something not OK. My guess is that you believe that low jitter clocks are nonsense and any measurement to "prove" that fixed belief is enough to convince you of your thoughts.
jean-paul said:
This is real boring.
Always the same arguments like "not well implemented".
Isn't it strange that I hear differences between different chips. And measure as well ?
If hundreds of people did see an UFO, that does not necessaryly mean there was one.
Why does Elso not want to post his results ???
Tell you something:
My dog shakes his tail when listening to a silvercrown
Silvercrowns must be good
Why everybody doubts my measurements, but nobody does it better ?
Konnichiwa,
Actually, a guy of my aquaintance by the name of Paul Miller build a test set around the methode outlined. The test software (analyser) and the test signal is used by a number of UK Magazines for measurements, John Atkinson of Stereophile also uses it.
I personally am actually of the opinion that all this "clock" stuff is more hype than substance and use a different methode of measuring clock jitter (one that tells me much less than Paul Millers approach, but one I can do using simple analogue hardware).
However, just because a "clock module" often does not lead to lower but higher jitter it would be foolish to discount the effect of jitter on the sound. For fun, feed an oscillator a flat white noise (AF range) polluted supply and then repeat the test using a signal that is made of AF pink noise with the same peak level. Surprisingly the two different "noises" at levels causing only a few 100pS additional jitter sound distinctly different.
Of course, most of the jitter in CD-Players and DAC's has sources other than the clock and this must be addressed if we wish to hear what CD Digital Audio is truely capable of. Adding a clock module is often merely a change, not a genuine reduction in jitter. Of course, it may sound good with more jitter, it often does, nice low rate, pink noise type jitter that broadens the signal peak has a tendency to make for a warm, pleasant sound, more warm and pleasant than the original recording.
Ann all of this, just like dither at levels that should be at or below the level of perception, yet make audible contribution nevertheless. Go figger.
Anyway, you need to decide what you wish to achieve with your measurements and then proceed to implement them suitable to that purpose. If you wish to investigate the possible differences than you are required to use a technique that will actually show them reliably.
If you wish to proove that certain, comparably irelevant (with respect to audibility) technical parameters are not or only marginally affected by jitter differences (BTW, I suspect that your add in clock has a much poorer jitter levels than the clock originally present, a great danger with add-in clocks - this may be related to layout, implementation, circuit design or the way you feed the signal to the servo IC) you have chosen the right measurement, but you have choosen a measurement whose results show no congruence with audibility.
Sayonara
Bernhard said:
Actually, a guy of my aquaintance by the name of Paul Miller build a test set around the methode outlined. The test software (analyser) and the test signal is used by a number of UK Magazines for measurements, John Atkinson of Stereophile also uses it.
I personally am actually of the opinion that all this "clock" stuff is more hype than substance and use a different methode of measuring clock jitter (one that tells me much less than Paul Millers approach, but one I can do using simple analogue hardware).
However, just because a "clock module" often does not lead to lower but higher jitter it would be foolish to discount the effect of jitter on the sound. For fun, feed an oscillator a flat white noise (AF range) polluted supply and then repeat the test using a signal that is made of AF pink noise with the same peak level. Surprisingly the two different "noises" at levels causing only a few 100pS additional jitter sound distinctly different.
Of course, most of the jitter in CD-Players and DAC's has sources other than the clock and this must be addressed if we wish to hear what CD Digital Audio is truely capable of. Adding a clock module is often merely a change, not a genuine reduction in jitter. Of course, it may sound good with more jitter, it often does, nice low rate, pink noise type jitter that broadens the signal peak has a tendency to make for a warm, pleasant sound, more warm and pleasant than the original recording.
Ann all of this, just like dither at levels that should be at or below the level of perception, yet make audible contribution nevertheless. Go figger.
Anyway, you need to decide what you wish to achieve with your measurements and then proceed to implement them suitable to that purpose. If you wish to investigate the possible differences than you are required to use a technique that will actually show them reliably.
If you wish to proove that certain, comparably irelevant (with respect to audibility) technical parameters are not or only marginally affected by jitter differences (BTW, I suspect that your add in clock has a much poorer jitter levels than the clock originally present, a great danger with add-in clocks - this may be related to layout, implementation, circuit design or the way you feed the signal to the servo IC) you have chosen the right measurement, but you have choosen a measurement whose results show no congruence with audibility.
Sayonara
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.