I think measurements are worth doing in order to find out what are the important parts that are meaningful to (ex)change or upgrade.
It is also interesting to see what are the limits of the measurements you make. For instance, in RightMark Audio Analyzer the measured overall noise performance can vary +/- 0.4 dB by only adjusting very slightly the attenuator at the sound card input while remaining in the "green" area for recording (input level +/- 1dB). I was not able to measure any differences when replacing the output opamps of a ROTEL RCD-971 player and of a PHILIPS CD 723 player. The sound was also the same on my amp+speakers, so I am not very motivated in tweaking opamps and even less motivated to exchange passive elements. The major difference comes from the DAC - this can be easily seen in the plots of the different players.
And now I am experimenting with jitter, which however needs other measurement methods.
It is also interesting to see what are the limits of the measurements you make. For instance, in RightMark Audio Analyzer the measured overall noise performance can vary +/- 0.4 dB by only adjusting very slightly the attenuator at the sound card input while remaining in the "green" area for recording (input level +/- 1dB). I was not able to measure any differences when replacing the output opamps of a ROTEL RCD-971 player and of a PHILIPS CD 723 player. The sound was also the same on my amp+speakers, so I am not very motivated in tweaking opamps and even less motivated to exchange passive elements. The major difference comes from the DAC - this can be easily seen in the plots of the different players.
And now I am experimenting with jitter, which however needs other measurement methods.
sinigersky said:
Whilst I respect your scientifically guided approach you may hit a problem here: there appears to be a phenomenon of audibility of modifications combining.
What this boils down to is that you won't hear subtle tweaks unless the system can resolve the change. This includes amplification, the speaker system & room and even the cabling to some small extent. And it includes all the other parts of your CD player working to a whole.
Op-amps should be audibly different.
Simon
Hi
I have fitted the 15VA for dac and 7VA for decoder along with loads of SPower regs for the driver ics + 2x 40V 10000uF Aerovox caps for each clock supply (same as Audiocom use on there new clock psu). The txs will be changed next week for the clock psu supply.
Here are some pics of the changes.
Pic1
Pic2
Pic3
Pic4
Pic5
Pic6
Brent
I have fitted the 15VA for dac and 7VA for decoder along with loads of SPower regs for the driver ics + 2x 40V 10000uF Aerovox caps for each clock supply (same as Audiocom use on there new clock psu). The txs will be changed next week for the clock psu supply.
Here are some pics of the changes.
Pic1
Pic2
Pic3
Pic4
Pic5
Pic6
Brent
Interesting. I too use RMAA with an M-Audio Firewire Audiophile external box. I haven't ever used it on a CD player and was wondering how you were doing this, given that a CD player doesn't have analog inputs? Are you recording a test CD using RMAA and measuring the playback of it?sinigersky said:
I am currently doing a couple of discrete I/V stages for my Rotel 855 and was using RMAA on these... mainly because I could
. I had started to think about being able to measure the entire CD player once I installed these, and to see what differences between the discrete vs. opamps stages, etc. I could see.This is a most interesting thread, although I have to confess I haven't read ALL of it.
I question the measurements above, because the sound card is not a good one. The DNR for a good soundcard (like my E-MU 1212) is around 117 dB. This USB card is only 1-2 dB above what the CD player is measuring. Please note I am not questioning your mod or the listening results, just the measurement test.
Just to back up my point a little, someone did measurements on my ezDAC a while back with an E-MU 0404 USB, and you can clearly see the difference between the 16-bit and 24-bit tests (with the 24-bit tests 12 dB higher).
http://haowulin.googlepages.com/Comparison.htm
Ray might have some opinions on measuring, since he's been doing quite a bit lately (from what I hear).
Just to back up my point a little, someone did measurements on my ezDAC a while back with an E-MU 0404 USB, and you can clearly see the difference between the 16-bit and 24-bit tests (with the 24-bit tests 12 dB higher).
http://haowulin.googlepages.com/Comparison.htm
Ray might have some opinions on measuring, since he's been doing quite a bit lately (from what I hear).
Jitter measurements
I can answer the first two. The wave on the screen is the slope of the stretched clock signal of a Flea with Tentlabs XO. The measurement is RMS Jitter which requires the color-grade view switched on. Every color stands for a certain range of values, and the scope uses this to gather statistics.
Here's another nice one, this one uses a histogram to display the jitter distribution.
If i'm correct, jitter is the deviation in time from the actual correct slope, which would theoretically be rock solid in the center of the color trace. So the peak-to peak deviation is 163ps, and actual peak-peak jitter would be about half that value (if i'm right?), because the deviation can be positive as well as negative. How this converts to RMS jitter I don't know (yet). The standard deviation 'std dev' or also called sigma, is a statistical value and it describes the spread of the values around the mean value.
I don't exactly know (yet) how to interpret this data, still working on that. Maybe Guido can help us out?
Regards,
Ray
I can answer the first two
. The wave on the screen is the slope of the stretched clock signal of a Flea with Tentlabs XO. The measurement is RMS Jitter which requires the color-grade view switched on. Every color stands for a certain range of values, and the scope uses this to gather statistics.Here's another nice one, this one uses a histogram to display the jitter distribution.
If i'm correct, jitter is the deviation in time from the actual correct slope, which would theoretically be rock solid in the center of the color trace. So the peak-to peak deviation is 163ps, and actual peak-peak jitter would be about half that value (if i'm right?), because the deviation can be positive as well as negative. How this converts to RMS jitter I don't know (yet). The standard deviation 'std dev' or also called sigma, is a statistical value and it describes the spread of the values around the mean value.
I don't exactly know (yet) how to interpret this data, still working on that. Maybe Guido can help us out?
Regards,
Ray
Attachments
I think the important number here may be sigma (std dev). Here's a figure I found:
If this is true, 19 ps seems like a pretty good number, right?
An externally hosted image should be here but it was not working when we last tested it.
If this is true, 19 ps seems like a pretty good number, right?
Pars said:
Interesting. I too use RMAA with an M-Audio Firewire Audiophile external box. I haven't ever used it on a CD player and was wondering how you were doing this, given that a CD player doesn't have analog inputs? Are you recording a test CD using RMAA and measuring the playback of it?
I am currently doing a couple of discrete I/V stages for my Rotel 855 and was using RMAA on these... mainly because I could
This is a most interesting thread, although I have to confess I haven't read ALL of it.
I have recorded a test CD. It's very easy:
1) record CD with two tracks generated by RMAA
2) play first track on repeat to adjust recording level
3) start test by recording the second track
I find it very, very interesting to see what comes out. In this way you can test the effect of all changes you make. Even if the absolute measured levels are not quite correct, you still get some trends.
ezkcdude said:
Well, this is a much better sound card than my previous Creative Audigy 2 NX USB. I can also confirm that there are some differences when recording at differenet sample rates and resolutions. But this is still suitable for comparision and trend measurement. For instance, I was able to clearly see the dynamic range of my old PHILIPS CD 723 changing form 94 dB to 89 dB when adjusting the volume via the digital volume control from 100 % to 50 %. Another good idea is to repeat the tests in order to see if there are some stochastic influences. Before posting those results, I repeated all tests 2-3 times.
Photos of the installed Tentlab XO
Here are some pictures of the newly installed clock.
http://www.sinigersky.com/images/Marantz-CD67-Clock-1.jpg
http://www.sinigersky.com/images/Marantz-CD67-Clock-2.jpg
I have treated the clock as a high frequency device (16.9344 MHz is in the short wave range!) so I use a screen and a coaxial cable.
Here are some pictures of the newly installed clock.
http://www.sinigersky.com/images/Marantz-CD67-Clock-1.jpg
http://www.sinigersky.com/images/Marantz-CD67-Clock-2.jpg
I have treated the clock as a high frequency device (16.9344 MHz is in the short wave range!) so I use a screen and a coaxial cable.