Melco Audiophile NAS

[simoncn]

I'm currently listening to MinimServer running on this device.

As we all know, it's impossible for the NAS hardware to make any difference to how the music sounds because the audio is transmitted as a bit-perfect stream of data across a bit-perfect Ethernet connection. This has been true for every NAS or computer that I've used to run MinimServer (12 at the last count).

This NAS is a bit different, though. It has a dedicated opto-isolated Ethernet port for direct connection to my player (a Linn Klimax DS/1) together with other high-quality componentry such as a low-jitter clock and a better-quality power supply.

Edit: The Ethernet port is not opto-isolated. See this post.

When I connect my player to the network using the dedicated player link of the Melco NAS, there is a very noticeable change in the clarity and transparency of the music. I hear this difference with MinimServer running on the Melco NAS and I hear the same difference with MinimServer running on some other NAS elsewhere in my network, as long as my player is connected to the network via the Melco NAS. This confirms that the benefit is caused by something in the Ethernet connection feeding the player rather than by the hardware running the UPnP server.

I can't explain this scientifically and I know that most people reading this will think I'm imagining things. This would have been my reaction as well until very recently.

[rhoering]

Simon,

Welcome to my world.



Ron

[DavidHB]

@simoncn

I'm a bit disappointed that you have had only one response so far to your interesting post. I suspect that on, say, the Linn Forum, your report would by now have sparked all kinds of argument and counter-argument, with people taking up quasi-theological positions on the various technical issues.

I am intrigued as to how you came by this piece of kit and why you chose to test it. Information on the web is remarkably sparse. The post to which you linked is a (rather informative) product announcement, but I can't find anyone who is actually selling the item as yet.

That apart, your post is of particular interest both because it comes from you and because of the way in which you have been able to narrow down the range of possible causes of what you have observed. By now, you must know as much as anyone about the specifics of running music server software on NAS hardware. Moreover, in this forum and elsewhere, you have always taken an open-minded but sceptical approach on the question of whether network components can affect sound quality. So your observations in this instance were, as you implied in your post, counter-intuitive, and therefore worth taking seriously.

Was the "very noticeable change in the clarity and transparency of the music" you observed in any way comparable to the change in SQ you experience as consequence of transcoding your files to WAV24? The reason I ask is that increased clarity and transparency is what I get from transcoding. If improving the network configuration can add more of the same, that is potentially rather good news.

If the change is the result of the configuration of the network components rather than the working of the NAS as such, I certainly agree with you that we are unlikely to be seeing any change in the actual digital content delivered to the player. The only other variable I am aware of that can affect SQ is the timing of that delivery, in other words the dreaded jitter.
One can at least conceive of ways in which the removal of electrical noise from the network connection (which is presumably the purpose of opto-isolating the port) can reduce timing inaccuracy. But I do not have the knowledge to judge whether this could in any way explain your observations.

There is also the fact that you are using a Linn player. AFAIK, Linn ties the DAC in its players to its own clock when playing network streams, and so does not use the clock data provided by the source. (This, I had always understood, was the reason why FLAC files ripped from a CD sound better on my player than the same CD played on my CD player used only as a transport, and so delivering the CD data to the same DAC as the network stream.) If opto-isolating the network connection as you have described improves the performance even of the Linn DAC/clock combination, that is a remarkable result. Would the result be the same if the opto-isolation were at the player end, one wonders.

I look forward to any further information you can glean on this interesting subject. In the meantime, I shall remain as sceptical as ever on the issue of whether network cables can affect SQ ...

David

[simoncn]

(12-12-2014 16:14)DavidHB Wrote:  @simoncn

I'm a bit disappointed that you have had only one response so far to your interesting post. I suspect that on, say, the Linn Forum, your report would by now have sparked all kinds of argument and counter-argument, with people taking up quasi-theological positions on the various technical issues.

Thanks for this thoughtful and constructive response. I was concerned that my post might elicit the kind of responses typical of the Linn forum, and I'm thankful this hasn't happened (yet).

Quote:I am intrigued as to how you came by this piece of kit and why you chose to test it. Information on the web is remarkably sparse. The post to which you linked is a (rather informative) product announcement, but I can't find anyone who is actually selling the item as yet.

It has been loaned to me by Melco. If you want to buy one, a web search tells me that it is available from Analogue Seduction.

Quote:That apart, your post is of particular interest both because it comes from you and because of the way in which you have been able to narrow down the range of possible causes of what you have observed. By now, you must know as much as anyone about the specifics of running music server software on NAS hardware. Moreover, in this forum and elsewhere, you have always taken an open-minded but sceptical approach on the question of whether network components can affect sound quality. So your observations in this instance were, as you implied in your post, counter-intuitive, and therefore worth taking seriously.

My approach to this is scientific, by which I mean formulating a theory, trying an experiment to validate the theory, observing the results and then trying to understand any observed differences from what the theory would predict. My first 12 experiments with different server hardware didn't show any significant difference, as expected from my theory. I was very surprised by the result from experiment number 13 and I am now trying to understand what is causing this.

Quote:Was the "very noticeable change in the clarity and transparency of the music" you observed in any way comparable to the change in SQ you experience as consequence of transcoding your files to WAV24? The reason I ask is that increased clarity and transparency is what I get from transcoding. If improving the network configuration can add more of the same, that is potentially rather good news.

It is very similar but even more noticeable. For example, the first piece in Monterverdi's Vespers (a favourite test track of mine) has powerful choral singing with some trumpets playing. On poor equipment, it isn't easy to hear the trumpets in the "wall of sound". On my pre-Melco setup, the trumpets are clearly audible and blend in with the choral singing. With the Melco, the trumpets are as clear as if they were playing as solo instruments.

Quote:If the change is the result of the configuration of the network components rather than the working of the NAS as such, I certainly agree with you that we are unlikely to be seeing any change in the actual digital content delivered to the player. The only other variable I am aware of that can affect SQ is the timing of that delivery, in other words the dreaded jitter.
One can at least conceive of ways in which the removal of electrical noise from the network connection (which is presumably the purpose of opto-isolating the port) can reduce timing inaccuracy. But I do not have the knowledge to judge whether this could in any way explain your observations.

I am told by Melco that the improvements are caused by reduced jitter in sending the network data, opto-isolation to prevent transmission of electrical noise and a better power supply to prevent common mode interference. My understanding of these technical details is quite limited and I won't attempt to expand on what I have been told.

Quote:There is also the fact that you are using a Linn player. AFAIK, Linn ties the DAC in its players to its own clock when playing network streams, and so does not use the clock data provided by the source. (This, I had always understood, was the reason why FLAC files ripped from a CD sound better on my player than the same CD played on my CD player used only as a transport, and so delivering the CD data to the same DAC as the network stream.) If opto-isolating the network connection as you have described improves the performance even of the Linn DAC/clock combination, that is a remarkable result. Would the result be the same if the opto-isolation were at the player end, one wonders.

Like you, I had thought that the Linn player would be more or less immune from interference by anything at the other end of the Ethernet cable and it was a great surprise that I could hear such a big difference.

Quote:I look forward to any further information you can glean on this interesting subject. In the meantime, I shall remain as sceptical as ever on the issue of whether network cables can affect SQ ...

David

Since you happen to mention cables, I will report one more observation. The Melco player came with a Melco cable whch was claimed to improve sound quality. For the Melco to Linn connection, I have compared the Melco cable with a normal Cat 5 cable and there is a small but noticeable improvment when using the Melco cable. This improvement is much less than the improvement from connecting the Linn via the Melco NAS.

[DavidHB]

(12-12-2014 18:46)simoncn Wrote:  I was concerned that my post might elicit the kind of responses typical of the Linn forum, and I'm thankful this hasn't happened (yet).

Perhaps we're all too nice Seriously, it would be good to know what other folk think.

(12-12-2014 18:46)simoncn Wrote:  [The Melco NAS] has been loaned to me by Melco. If you want to buy one, a web search tells me that it is available from Analogue Seduction.

The N1A HDD version is £1,600, which is a lot for a (2 x 2TB) NAS, but perhaps a not totally surprising price. The N1Z SSD version (with 2 x 512GB "audio quality" SSDs) is an eye-watering £6,200. The specifications for the two models are somewhat differently laid out, but, apart from the drives, I cannot detect any significant difference between them from the text. Which version are you testing? If it is the HDD version, is it suitably free of mechanical noise so that it can reside in the listening room?

Melco is a brand of Buffalo Technology, so the effort that has gone into improving the network behaviour of these devices should come as no surprise. What is surprising, to me as to you, is that it should have such an effect on SQ.

(12-12-2014 18:46)simoncn Wrote:  My approach to this is scientific, by which I mean formulating a theory, trying an experiment to validate the theory, observing the results and then trying to understand any observed differences from what the theory would predict. My first 12 experiments with different server hardware didn't show any significant difference, as expected from my theory. I was very surprised by the result from experiment number 13 and I am now trying to understand what is causing this.

I shall be very interested to know your conclusions.

(12-12-2014 18:46)simoncn Wrote:  

DavidHB Wrote:Was the "very noticeable change in the clarity and transparency of the music" you observed in any way comparable to the change in SQ you experience as consequence of transcoding your files to WAV24?

It is very similar but even more noticeable. For example, the first piece in Monterverdi's Vespers (a favourite test track of mine) has powerful choral singing with some trumpets playing. On poor equipment, it isn't easy to hear the trumpets in the "wall of sound". On my pre-Melco setup, the trumpets are clearly audible and blend in with the choral singing. With the Melco, the trumpets are as clear as if they were playing as solo instruments.

Is this by any chance the second Eliot Gardiner recording (made in St Mark's, Venice)? If so, I understand exactly what you mean, and this is something I would love to experience.

(12-12-2014 18:46)simoncn Wrote:  I am told by Melco that the improvements are caused by reduced jitter in sending the network data, opto-isolation to prevent transmission of electrical noise and a better power supply to prevent common mode interference. My understanding of these technical details is quite limited and I won't attempt to expand on what I have been told.

My knowledge is also limited. I can see how reducing jitter improves SQ, but do not pretend to understand the relationship between improved signal/noise performance and SQ in a digital system.

(12-12-2014 18:46)simoncn Wrote:  Like you, I had thought that the Linn player would be more or less immune from interference by anything at the other end of the Ethernet cable and it was a great surprise that I could hear such a big difference.

I'd dearly love to know what the Linn engineers think of this ...

(12-12-2014 18:46)simoncn Wrote:  Since you happen to mention cables, I will report one more observation. The Melco player came with a Melco cable whch was claimed to improve sound quality. For the Melco to Linn connection, I have compared the Melco cable with a normal Cat 5 cable and there is a small but noticeable improvement when using the Melco cable. This improvement is much less than the improvement from connecting the Linn via the Melco NAS.

This is the first report I trust of network cables having any effect on SQ. It seems not only that higher quality cables do not provide the greatest measure of improvement, but also that they are best used as part of a coherent programme to improve network signal quality. That at least starts to make sense.

David

[simoncn]

(13-12-2014 18:17)DavidHB Wrote:  Perhaps we're all too nice Seriously, it would be good to know what other folk think.

I hope they will be as open-minided as you have been.

Quote:Which version are you testing? If it is the HDD version, is it suitably free of mechanical noise so that it can reside in the listening room?

It's the N1Z (SSD version) which is fanless and completely silent.

Quote:Melco is a brand of Buffalo Technology, so the effort that has gone into improving the network behaviour of these devices should come as no surprise. What is surprising, to me as to you, is that it should have such an effect on SQ.

It's actually the other way round. Melco is the parent company and produced some serious hi-fi products in the 1970s/80s before they created the Buffalo brand and headed in a different direction.

Quote:Is this by any chance the second Eliot Gardiner recording (made in St Mark's, Venice)? If so, I understand exactly what you mean, and this is something I would love to experience.

Yes, that's the one. I have Gardiner's first Vespers recording on LP and the second is clearly superior. I was fortunate enough to be at the Proms for Gardiner's live performance on the penultimate night in 2010, so I know what it really sounds like.

[beckphotonik]

I like the ideas behind this. The streamer being isolated from the rest of the network by this NAS and for the NAS to have some form of opto isolation to the rest of the network certainly appeals. I can then see the point of a special ethernet cable from NAS to streamer. Shame I didn't read this last week as I have just taken delivery of a new Qnap NAS but such is life!

[DavidHB]

(13-12-2014 19:05)simoncn Wrote:  

(13-12-2014 18:17)DavidHB Wrote:  Which version are you testing? If it is the HDD version, is it suitably free of mechanical noise so that it can reside in the listening room?

It's the N1Z (SSD version) which is fanless and completely silent.

... and costs £6,200. Ouch. How much of the benefit does one get from the £1,600 N1A, I wonder?

(13-12-2014 19:05)simoncn Wrote:  Melco is the parent company and produced some serious hi-fi products in the 1970s/80s before they created the Buffalo brand and headed in a different direction.

Thanks for this correction, and the information. Up to now, I have only encountered the Buffalo brand.

Quote:I have Gardiner's first Vespers recording on LP and the second is clearly superior. I was fortunate enough to be at the Proms for Gardiner's live performance on the penultimate night in 2010, so I know what it really sounds like.

Well, the Albert Hall (in any of its acoustic incarnations) is not like St Mark's - where Gardiner was also free to play around with the spacing of the performance - but I know what you mean. There's an interesting account in Gardiner's recent book ("Music in the Castle of Heaven" - it's really about J.S. Bach) of his pioneering performance of the Vespers in Cambridge when he was still an undergraduate there.

David

[DavidHB]

(13-12-2014 19:17)beckphotonik Wrote:  Shame I didn't read this last week as I have just taken delivery of a new Qnap NAS but such is life!

You can console yourself with the thought that you didn't have to pay £6,200 or equivalent (or even £1,600 - see above) for your QNAP.

David

[beckphotonik]

(13-12-2014 19:41)DavidHB Wrote:  

(13-12-2014 19:17)beckphotonik Wrote:  Shame I didn't read this last week as I have just taken delivery of a new Qnap NAS but such is life!

You can console yourself with the thought that you didn't have to pay £6,200 or equivalent (or even £1,600 - see above) for your QNAP.

David

I was considering an SSD loaded Qnap that would have set me back £1100! In the end I chickened out and bought WD red HDDs instead. Fortunately they are quiet enough to live next to the streamer. The Qnap has 2 ethernet ports so I could put the streamer on a separate sub net but I would have to dedicate a tablet control point to that sub net and buy a wifi transceiver for it to communicate with. Even so it would not be as elegant as the Melco solution.