I have been doing some digging on the sine input to the Digital Linear Reference Level SD3.75 SD Player.
There is a quad buffer Nexperia 74AHC125 on the board before where i take off the I2s so i expect this is doing the squaring and reclocking the I2s, i do not know enough to say this categorically and i suspect there are those who know better and can correct me very quickly but maybe this is the key to my success.
I may have been missing a trick with the original Driscoll clocks running them with a squarer on the clock board as the sine output may have been going through 2 squaring stages.
Anyway music is still great today
There is a quad buffer Nexperia 74AHC125 on the board before where i take off the I2s so i expect this is doing the squaring and reclocking the I2s, i do not know enough to say this categorically and i suspect there are those who know better and can correct me very quickly but maybe this is the key to my success.
I may have been missing a trick with the original Driscoll clocks running them with a squarer on the clock board as the sine output may have been going through 2 squaring stages.
Anyway music is still great today
Hi! I have received all the boards! They look absolutely gorgeous!.....Thank you Andrea! 
....
I need to use the x48 CMOS input (PPA OCXO USB) of my FSDO board. According to manual, I need to remove C8 and reinstall in C26 (same cap), would it be possible to know any part-number for C8??, just in case...
....
I need to use the x48 CMOS input (PPA OCXO USB) of my FSDO board. According to manual, I need to remove C8 and reinstall in C26 (same cap), would it be possible to know any part-number for C8??, just in case...
Have you ever tried a lower frequency clock for SD Player, e.g 11MHz or 5MHz? Very unlikely to work but still I've been curious...
I tried the 5Mhz just now and it plays but sssslllloooowwww
I added the first doubler and the SD would not lock/play, then tried it with both doublers and still the same. Made sure they were the correct doublers by the way and had 16v in again but no luck. Clock Out to Doubler In and doubler Out to SD In, etc.
I will come back to this in a week or so and try and figure it out then when I get time as I just want to break in the 22Mhz clock for now and enjoy the music, leaving it running 24/7 and it is getting better.
D
Deleted member 537459
I have a McFifo.
One channel is supplied with a DRIXO, the other still has an NDK.
they both live happy together, and the fifo works as expected. so i must assume that there's some form of switching/multiplexing on board.
But Ian is the man here
Done a little reverse engineering on the McDual.
At the output of each XO there's a fanout 1:4 buffer (NB3L553) with output enable
signal.
The outputs from both Fanouts are in parallel to each other and goes to the MCK1..2..3..4 outputs.
The McFifo(McDualXO) selects the clock by BOTH :
- enabling / disabling the OE pin on the desired XO.
- enabling / disabling the OE pin of the fanout buffer.
So, your are covered both if you have a XO with High-Z output capability or one without any form of shutdown (like the STS).
(FYI, the signal that goes back to the FIFO for clocking the "clean" side and for the reclock stage (74g374) originate both from the MCK4 output, so if you connect something here you are effectively in parallel with the fifo and the reclock)
Hope that helps...
Last edited:
D
Deleted member 537459
In the last 2 years I use roon software. And I see how is important the quality of all network. Cable, switch ethernet, roon core and roon end point.
Now, I already change and try some oscillator for switch ethernet and I find ocxo raltron 5000 very good.
Now I buy TWTMC-PXO_AIO to 25 MHz from andrea, and the result is really impressive, the low frequency guys I's incredible the improvement. is also better every where dynamic, musicality, bigger sound with more focus more three-dimensionality and so on! Next day i try 5.6448 drixo on my dac.
P.s. for me rpi4 sound better than rpi3 and older because rpi have direct ethernet than rpi3 and older because them have ethernet trough usb. And I try change xo 25mhz from rpi4 with great improvement.
OK I understand the concept now but not to the exact details on where you have applied the new clocks to improve an ethernet switch and the Pi4.
I use Roon too as many of us do....
D
Deleted member 537459
at the moment I have only replaced the clock at the ethernet switch. my final project is to make a 25mhz master clock for each electronics that communicate on the network, replacing the original oscillator with the masterclock. in this way all the objects that communicate with the network will have an excellent clock and in sync with each other.
But then I wonder how the network clock affects the DAC timing? 25MHz network clock is never in sync with the audio sample rate.
So what I mean is that the network clock stability should not end-up as being audible when the DAC clock is the final 'master' in the system.
D
Deleted member 537459
the digital audio file is sent from the roon core over the network to the dac, passing through a switch or router. this greatly affects the final result. even if the network clock is not that of the audio. however the file is transferred 10011000101010 if noise or latency is inserted leads to a deterioration of the audio file. you can have the best dac in the world but if you listen to an mp3 you will not have a great sound, as the source is deteriorated. even the ethernet cable in the theoretical line should not impact the sound, instead seeing is believing. if you do not hear any differences, I think it is useless to continue spending money on upgrades, the problem is your ears.
My opinion
My opinion
I read about the impact of the network on the sound from many people. I have no reason to question that their observation is real vs imagined. To me the issue lies with the FIFO/reclock/isolator setup. The signal we listen to is manufactured by the FIFO. The latency can be measured in seconds as the Rpi software (MPD) buffers the data and then FIFO buffers the data and creates the bit stream we ultimately hear. The jitter should be determined by the FIFO, and it only goes downhill from there. The isolator is supposed to eliminate the upstream noise. It should not matter if the data source was a HDD, a memory card, an internet stream or a CD.
Let us hope for the day when smart guys like Ian and Andrea can build a FIFO/reclock/isolator that fully functions as intended. Until then it makes sense to optimize the upstream network if we hear it in the sound, even if we should not need it.
Let us hope for the day when smart guys like Ian and Andrea can build a FIFO/reclock/isolator that fully functions as intended. Until then it makes sense to optimize the upstream network if we hear it in the sound, even if we should not need it.
D
Deleted member 537459
I am sure you are hearing an improvement.
From a technical stand point, that makes no sense.
If we wanna trust what we hear (that's good), I think is useless trying to rationalize the experience.
File format and transport has very little in common.
While is true that an mp3 will never sound like a lossless format, is not the fact that a better "clocked network datastream" on the PHY layer leads to a more accurate reproduction.
There's MANY places in a network stack where error check and recovery take places.
Just think about the global internet network. Datas are exchanged at speeds that a flac will never reach... congestion, arbitration, datapaths switch and they have no global clock to sync on but yet, we are still in the XXI century.
For what matters, a fifo is still the better option. and maybe, at the end of the game, far cheaper
(but... again, if that makes you feel better and makes you enjoy more the listening experience please upgrade everything you can to move along the road of happiness, just don't try to rationalize the process).
my2c.
Last edited: