|
Post by MartinT on Jun 30, 2020 13:06:07 GMT
No John, the signal doesn't split and then rejoin. The signal takes a straight path. The clock path is something completely different.
I may have to show you.
|
|
|
Post by MikeMusic on Jun 30, 2020 13:13:42 GMT
The big thing stopping me going streaming is the amount of music I can't get. Plus the huge subscription every month £14.99 per month is piddly for all the music I get. Compared with how much per month I spent on CDs... Catalogue is personal, of course. It's become very infrequent that I cannot play what I want these days. Spotify fills in maybe once every two weeks. Qobuz is that good now. I was thinking of £9.99 ! Rarely spend more than £2 on CDs and still discovering and back filling the collection from the past. Different decision if I didn't have any CDs
|
|
|
Post by MartinT on Jun 30, 2020 13:17:14 GMT
That's for Qobuz HiFi, which is CD quality.
I'm on Qobuz Studio.
|
|
|
Post by John on Jun 30, 2020 13:25:17 GMT
So the clock signal adds as a reference point for the audio signal once it gets to the Mutec. Splitting the Clock signal from the audio signal gives the Tinkerboard one less processing task giving fewer jitter errors when going to the Mutec. So this is helping with the timing and therefore, the sound.
|
|
|
Post by MartinT on Jun 30, 2020 13:43:16 GMT
So the clock signal adds as a reference point for the audio signal once it gets to the Mutec. Splitting the Clock signal from the audio signal gives the Tinkerboard one less processing task giving fewer jitter errors when going to the Mutec. So this is helping with the timing and therefore, the sound. Strictly speaking, the Mutec strips the data from the 'dirty' stream, and clocks it out using the external reference clock to constitute a new 'clean' stream. The Tinker Board doesn't see the reclocking activity either from the incoming data (from the EtherREGEN) or the outgoing data (to the Mutec), it always has the same processing task, using its own crystal to clock the data out to the Mutec. Either side of it, the data is cleaned.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 30, 2020 13:54:36 GMT
Time for Martin to run a separate thread for the principles of data re alignment via the various re-clocking schemes he is currently implementing that way the folks could understand why he decided to go this way .
|
|
|
Post by MartinT on Jun 30, 2020 14:48:43 GMT
Not that I'm the expert!
|
|
|
Post by MartinT on Jun 30, 2020 14:53:52 GMT
In the above diagram, arrows always denote direction. The thick arrows denote signal path. Grey arrows are power. Red arrows are clock. Grey boxes are power. Green box is internet. Blue boxes are core data processing. Orange boxes are analogue.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 30, 2020 15:24:55 GMT
Maybe an explanation of how PCM (digital audio data) is made up from three data lines, Word Bit, Word Clock and the audio data itself and how the word clock is extracted from the incoming signal and used the help with the alignment of bit clock and audio data timing (jitter).
The more accurate the word clock data the potential for jitter reduction in the audio data transmission lines (TIE:- Time interval error).
Jitter isn't some mystical place that eargasams reside, simple put Jitter is the theoretical (mathematically constructed) time the signal should arrive at its destination against the time it actually arrives at. This can range form many 100's nanoseconds to ridiculously low femtoseconds.
However the real test is what actually happens actually at the point of inception of the clock signal rather than the quoted figurers, the REAL world after the cables and connectors, circuitry have taken their toll on the signal fidelity. I've got 50 FS jitter nan nan lol
In reality its nearer 50ps and the measurement point!
|
|
|
Post by MartinT on Jun 30, 2020 17:53:53 GMT
I'll come back to things I've learned in a moment, but one thing that everyone can try is looking at how effective (or not) your supports and footers are, as well as granite mats and weights on top. Install the Hamm Seismograph app on your smart phone (this is an Android app but there must be something similar for Apple). Play music at your normal volume and place the phone anywhere in the room. Look at the amplitude of the vibrations. Now carefully place it on your equipment stand. Is it better? Now rest it on your DAC. Better still? If not, you have work to do. It's an education getting the readings as low as possible and you WILL hear the difference if you do it across your source components. play.google.com/store/apps/details?id=de.inovation.hammgraph&hl=en_US
|
|
|
Post by MikeMusic on Jun 30, 2020 18:12:18 GMT
That's the sort of thing I got my phone for !
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 30, 2020 19:27:04 GMT
Its a good sort to understanding vibration control for sure.
In my tool box is 3 axis accelerometer which is a just finer more accurate version of your phone app, but then I do look at internal microphany.
However it gives you a good idea, something for you to try, take measurement with the system off, the just switched on, then with music playing at say 68-72dB, then Martin's normal volume 92dB and perform a correlation table or XL spread sheet.
See if you can find anything, or whether you hit a brick wall like a couple of the chaps on AA did
|
|
|
Post by MartinT on Jun 30, 2020 19:48:53 GMT
92dB, LOL!
Modern phones have pretty sensitive three-axis sensors. There's always feed-through even on several layers of isolation. It's worth stamping a foot on the floor to see how much gets through and comparing with, say, a coffee table.
|
|
|
Post by MartinT on Jun 30, 2020 20:38:33 GMT
However the real test is what actually happens actually at the point of inception of the clock signal rather than the quoted figurers, the REAL world after the cables and connectors, circuitry have taken their toll on the signal fidelity. I've got 50 FS jitter nan nan lol In reality its nearer 50ps and the measurement point! Actually, this is perfect. I was going to write something similar. Take your DAC - the core of your system, right? What influences it? - 1. mains-borne noise (coming from general mains and local equipment pushing noise back into it) - 2. signal-borne noise (coming from your digital chain in the form of jitter and phase noise) - 3. EMI & RFI (coming from all around, but especially switched-mode power supplies) - 4. vibrations (coming from the music and the environment) What makes it perform better? - 1. regenerators, active filters, passive filters, cables that suppress high frequency noise - 2. reclockers, impedance matching cables, filter devices, grounding boxes - 3. shielded equipment and shielded cables, keeping SMPS away - 4. footers, isolation shelves, weights A few things to think about there.
|
|
|
Post by Clive on Jun 30, 2020 20:54:08 GMT
When I spoke with Norbert Lindemann he was very critical of clock specs being made at irrelevant frequencies. It sounds like some people throw together circuits on theory rather than a deeper understanding of what is really happening.
|
|
|
Post by John on Jun 30, 2020 20:56:30 GMT
Thanks, Martin, that is a good summary.
|
|
|
Post by MikeMusic on Jul 1, 2020 16:03:13 GMT
However the real test is what actually happens actually at the point of inception of the clock signal rather than the quoted figurers, the REAL world after the cables and connectors, circuitry have taken their toll on the signal fidelity. I've got 50 FS jitter nan nan lol In reality its nearer 50ps and the measurement point! Actually, this is perfect. I was going to write something similar. Take your DAC - the core of your system, right? What influences it? - 1. mains-borne noise (coming from general mains and local equipment pushing noise back into it) - 2. signal-borne noise (coming from your digital chain in the form of jitter and phase noise) - 3. EMI & RFI (coming from all around, but especially switched-mode power supplies) - 4. vibrations (coming from the music and the environment) What makes it perform better? - 1. regenerators, active filters, passive filters, cables that suppress high frequency noise - 2. reclockers, impedance matching cables, filter devices, grounding boxes - 3. shielded equipment and shielded cables, keeping SMPS away - 4. footers, isolation shelves, weights A few things to think about there. Excellent Could have a thread per item above, probably more
|
|
|
Post by MartinT on Jul 1, 2020 16:15:06 GMT
Keep going...
What has worked best for you, Mike?
|
|
|
Post by MikeMusic on Jul 1, 2020 18:17:06 GMT
Difficult to compare the big steps up over the years but probably the biggest and totally unexpected jump was the Coherent Grounding Box. Maybe 2nd were the Coherent jumpers on the TAD E1s.
|
|
|
Post by MartinT on Aug 21, 2020 8:45:20 GMT
Further recent lessons:
- Don't skimp on your router's PSU if you're streaming music from the internet. Running my TP-Link MR-6400 router from first a Chinese 12V LPSU and then a Paul Hynes 12V PR3iEHD gave two distinct jumps in sound quality from the original SMPS. Phenomenal, when you think about it. Noise along your ethernet cable is significant and worth paying attention to.
- Splitting my power regenerators into PS Audio P3 for source components and P10 for the power amp was well worth it, as was brought home to me with my recent P3 failure.
- Clock cables really do make a difference. Upgrading from Canare LV-61S to LV-77S made a very audible and significant improvement in sound quality.
|
|