Linnofil wrote:Music Lover wrote:Any forum reader that can speculate about the reason why some RAM's are better.
Do they have anything in common?
And the opposite, have the bad ones anything in common?
Would be great avoiding buying/testing hundreds of RAM's...
Yes, that would be very good to know! I wonder if it's the memory chips or the pcb design? Is there a controller in RAM sticks? It doesn't look like it. How many manufacturers of memory chips can there be? Does metal covers (design/cooling) act as screening? Maybe they can be grounded?
I have been following most of what has been written about the LSNAS, SSD, switches, cables,... and although I have only taken minor steps towards optimizing my own digital infrastructure I have noted a couple of things. To me it seems that the closer to the DS the more impact the units have, e.g. the type of switch and how it is connected, powered, connected to ground, positioned ... have more impact on the music than the NAS or what disk is used in the NAS. I do not know if this is generally true, just my reflection. It seems though that it is not the standard source-first relationship. Ripping however seems rather important which would contradict what I am suggesting, but it could well be that the important thing here is that the right information is read and the right compression applied.
In a sense the digital data is just that (data) until it is converted to an analogue waveform, but (and that is a BIG BUT :-) the analogue waveform representing the digital data is very much important at that stage and will, to some extent, determine the resulting quality of the waveform genereated by the DAC. Of course the isolation properties of the DAC will also have profound impact. The isolation is however limited (I do not know how much though - only that it is very difficult to reach more than about 60+ dB in a single box - very much frequency dependent). I think though, that most people will agree that no DAC circuit offer total isolation (or imunity if You will) between the analogue representation of the digital data and the analogue output of the DAC. At least the non ideal signal properties of the digital data signal will affect the earth currents around the DAC circuit hence also affecting the analogue side of the circuit.
The fundamental property of a digital system is that as long as the data can be interpreted as either 0 or 1 the data message can be fully recreated (with 100 % accuracy). This applies as long as the data is interpreted as data. When there are a conversion process involved, where the data has to be interpreted before conversion and in real time, the time dependent representation of that data beomes relevant. As it affects the interpretation and thus also the conversion process. The interpretation and isolation properties of each beffering stage between the creation of the data and the final conversion will, to some degree, affect the final result. It is however possible to both improve (mend) and degrade (or destroy) the data. For example it would be possible (if very time consuming) to print out the data and read it back to be stored before feeding it to the DS. This process (as long as the data is correctly interpreted) would be ideal and provide perfect reproduction as long as the storage of the re-written data is ideal. No reference would be made to the earlier representation of the data. As long as there are no real time requirements in any part of the chain the system is close to totally immune to the analogue representation of the data as long as there are no unrecoverable bit errors introduced in the chain. What you and others have dicovered is that when there are real time requirements on the data and on the quality of the analogue representation of the digital signal, it also becomes important to minimize the imperfections in the digital circuits. The quality of the analogue representation of the digital data can however not exceed the circuit topology and implementation of the final circuit closest connected to the DS. It could however be cleaned/restored along the way improving on the original analogue representation of the data (as in the previous example).
I do not think we (or others) have tried (compared) to connect several high quality switches in series in order to try to restore the analogue represenation of the data. We should try that! Anoher thing to try is to do a series of read and write the data from/to a high quality storage device. That could also be beneficial. I have heard of testimonials indicating that this kind of procedures could make for some improvements.
Considerig the quality of the RAM, as the data, after having been read from the disk, is stored in RAM, the analogue representation of the data is analogue waveform read from the storage device, but it will also be determined by the circuit topology, power supply, grounding conditions. Depending on the isolation properties of the RAM toloplogy and implementation in the system the analogue representation of the data as being transmitted from the storage device (e.g. from the HD, SSD over the PC data bus) will affect the data representation in the RAM circuit hence also the analogue waveform after having read the data and transmited it to the ethernet circuit/line-driver. I find it very plausible that the quality of the RAM have at least as much, possibly more, imact on the quality of the data delivered from the music server.
Yadda, yadda...
It seems I had too much time to spend on writing BS rather than on value adding tests... Sorry! Keep up the good work and the positive attitude!