SSD for NAS

Fundamentaly my understanding is that SSDs are more reliable than HDDs, particularly bearing in mind the minimal re-writing in normal music store usage. And as for warning of failure, in my experience SMART status certainly does not always warn of HDD failure - and even when it may I'm  unclear as to whether in a music NAS situation any warning would necessarily be evident to the user when typically the user only  normally accesses via the music library software, only using the NAS control software very rarely. 

As for noise, whether to minimise potential network interference or simply to minimise wiring requirements if all it is used for is music I believe it is not uncommon for people to have the NAS in the listening room - and in the case of music store such as Mac Mini, which has the advantages of not streaming the music files across the network when playing, and direct short cabling to (and effectively part of) the hifi, it is highly likely to be in the listening room. 

Regarding RAID, it should not be regarded as a backup, and given the choice of a second disk in the NAS as a mirrored RAID copy, or the same second disk elsewhere as a separate backup, it should be the latter.

Of course everyone has different experiences of these things, but I think the general view is more likely to be that SSDs are better than HDDs for this application, though cost is a factor at 3-4 times - but I accept may presently be very different in a server used for multiplevpurposes other than music, especially in any intensive usage applications such as commercial setups, where music storage might be a minor function.

I may not have followed all related threads by now, but my primary driver would still remain SQ. Full, possibly off-site, back-ups will always be required. Hence my request for a US2 SSD 2TB under 4k£ to respond to Melco, Aurender and others.

Chag - 

In a properly designed system there's no difference is SQ between a HDD and an SSD.

With my ND5 XS (fed from a Synology NAS via an unmanaged Netgear switch) after taking care of RFI in the network, I now can't tell the difference between UPnP streaming from the NAS and playing WAVE files locally from a USB stick (including 3 different ones!); ergo the network, NAS and WD Red HDD are irrelevant in SQ terms.

The NAS, PC, Minimserver, Melco, Unitiserve whatever, are just that: servers. They deliver a file to your DAC  which then translates it into an analogue signal. As long as the components within the file can be translated without corruption on reception then it is the hardware shaping the received data that affects the listening experience. Obviously RFI comingled with the processed data will impact the music's output quality., but as the packaged data has all the relevant information I still cannot see how it can be 'coloured' between server and DAC. 

Melco is owned by Buffalo and my most catastrophic loss of data including music, documentation and pictures was a Buffalo NAS with Buffalo drives where the controller failed and totalled the lot. This is equally applicable to Raided drives, although I see no point in Raid 0 because if one drive goes you lose the lot. Multiple controllers with Raid 5 is the only satisfactory solution.

Andarkian, +1.

I also believe that the data in the digital domain are unaffected, thus the detrimental effects transmitted from the network to the audio chain are due to RFI impacting the analogue circuitry.  I also believe this can be handled in a much cheaper manner than the extreme effort required to create an inherently low noise specialist computer such as the Melco, Vortexbox or US.

Furthermore in the case of the Melco, they've messed about with the programming in the SSDs, to reduce to amount of internal processing they do.  There's a good reason SSDs do all that data processing internally - it's to preserve data integrity!  So with a Melco you'll need another NAS on the network as an online backup to the Melco - hence introducing all the RFI again, so you'll still have to filter it out to get the SQ back (or ignore this and risk your data!).

What is needed in terms of backup depends on how much you add or change your music files. I don't add much/often, and what I do is download or rip onto a computer, then copy to the music store (in my case a Mac Mini) and to a duplicate store on a NAS (not otherwise anything to do with my music play. I then back up the MM about every 6 months or so to an offline disc stored remotely, after which I clear out the folder on the computer containing anything additional that had been added. For anyone adding to or changing their music store on a daily basis a more rigorous approach to backup would seem appropriate.

Huge,

I wholeheartedly concur with what you say and must correct one thing that I said. I said that it was the hardware that shaped the music on reception which is not strictly correct. The DAC software, as we all know now, has a huge impact, as well as the quality of the subsequent hardware components. As to Melco processing within the SSDs, apart from the inherent risk of non standard hardware, there is the problem of future support, upgrade and replacement. Seems to me that you would be taking your standard format CDs and loading them on to a non standard medium. Smart money seems to be spent and start at the DAC level but that is only my opinion. Whatever it is, if it enhances your listening experience go very much with your own opinion.

IB, that SSDs only fail due to user write wear on the cells isn't actually true.  The SSDs internal processing is slowly and repeatedly refreshing the NAND flash cells (typically a full refresh every few months) and so they have a finite lifespan as well as a finite number of user writes.  This process itself will give a typical MTBF in the low tens of years.  This limitation is balanced against the Uncorrectable Error Bit Ratio (UEBR) to maximise the expected life.  It should also be noted that 'bathtub failures' can occur sooner:  Reducing the internal processing increases UBER and raises the floor of the bathtub due to accumulated UBER errors.

Thanks for SSD clarification Huge.

So the Melco SSD approach seeks to 'fix' the internal re-write, which they claim has beneficial effect on sound quality (though it is indeed hard to understand why it should have any bearing on SQ). That Melco's approach is non-standard shouldn't be of concern per se as the disks dont get moved into, other machines, and the N1Z can take standard HDDs. If they fail they fail, and still require backups.

I haven't heard a Melco N1Z SSD so can't comment on it. I have very briefly (about 10 mins)  heard their base model, N1A, compared to my Mac Mini/Audirvana when I was auditioning the Chord Dave (through Bryston 4Bsst3 and PMC Fact12). There was no immediately obvious difference. That is unlike my previous comparison of ND5XS front end networked to Mac Mini with Serviio UPnP server as music store, into Hugo, when Mac Mini/Audirvana/Gustard into Hugo sounded better. Of course, several factors in play there, so the precise reason(s) for the difference can only be a matter of conjecture.

Trying the Melco N1A in place of the Mac Mini/Audirvana/Amarra Symphony iRC / iFi supply in my system revealed a considerable uplift in sound quality, quite shocking really, which I wasn't expecting. I stupidly thought I'd better borrow the N1Z, which was disturbingly even better. How much a part the SSDs play in this I couldn't say; the N!Z differs in several areas from the base model, such as power supply and isolation.

Hi Tony; I don't remember reading if you have tested US in or not. If you have, would you be able to comment on how N1A and N1Z respectively compare with US HDD and US SSD, hopefully both on continuous PSU? 

Chag -

tonym posted:

Trying the Melco N1A in place of the Mac Mini/Audirvana/Amarra Symphony iRC / iFi supply in my system revealed a considerable uplift in sound quality, quite shocking really, which I wasn't expecting. I stupidly thought I'd better borrow the N1Z, which was disturbingly even better. How much a part the SSDs play in this I couldn't say; the N!Z differs in several areas from the base model, such as power supply and isolation.

Given my experiences with Mac Mini/Audirvana into Hugo via electrical SPDIF, optical SPDIF, usb direct, and usb via Gustard U12 isolator/converter, and my recent (admittedly very brief) comparison of MM/Audiv vs Melco N1A into Dave, i wonder if the clearly very distinct improvement you noted with the base Melco may have been due to better noise (especially rf) filtering by the Melco, which would be relevant if your QBD76 might have lesser input isolation by whatever connection you had been using with MM, than does Dave, or Hugo via Gustard.

Not a factor - the QBD76 uses a FPGA Async. USB input. I've tried DAVE on a couple of occasions & ultimately it didn't sound as good as my DAC. I've been using the iFi isolated USB supply for a number of years, and it certainly made an improvement with the Mac Mini.

Hugo also has async usb input, but that's to stop jitter. Hugo doesn't have galvanic isolation, nor effective rf filtering, and according to Rob Watts, The Chord DACs' designer, rf interference is a prime cause of degraded sound quality. I understand Hugo TT and Dave have such isolation/filtering, but I don't know about the QBD76?

Improving the Mac Mini's powersupply can indeed reduce noise on the output, so give noticeable improvement in SQ, but the power supply is not the only source, and the remainder of the computer also adds noise, so in fact something that adequately isolates/filters out the noise before reaching the DAC stages is likely to be more effective than tweaking the power supply, hence my thought that if the QBD76 doesn't filter as well as Dave, or Gustard before Hugo, then that may account for Melco sounding better if it has isolation/filtering built in.

As for your (and others) preference of QBD over Dave, that may be nothing to do with the interference from the source (was Melco in use at the time?) , and of course could be down to synergy with the remainder of the system. I haven't had the opportunity to compare, so have no view.

ChrisSU posted:

Graham Russell posted:

SSDs tend not to fail gracefully like spinning disks mostly do.

I'm really struggling with the concept of a graceful disc failure! The expletives that inevitably fill the room don't help, but in my world, any disc failure is pretty awkward. 

Just extend the little finger . . . instead of the middle finger ;-)  Much more graceful that way!!

Innocent Bystander posted:

Hugo also has async usb input, but that's to stop jitter. Hugo doesn't have galvanic isolation, nor effective rf filtering, and according to Rob Watts, The Chord DACs' designer, rf interference is a prime cause of degraded sound quality. I understand Hugo TT and Dave have such isolation/filtering, but I don't know about the QBD76?

Improving the Mac Mini's powersupply can indeed reduce noise on the output, so give noticeable improvement in SQ, but the power supply is not the only source, and the remainder of the computer also adds noise, so in fact something that adequately isolates/filters out the noise before reaching the DAC stages is likely to be more effective than tweaking the power supply, hence my thought that if the QBD76 doesn't filter as well as Dave, or Gustard before Hugo, then that may account for Melco sounding better if it has isolation/filtering built in.

As for your (and others) preference of QBD over Dave, that may be nothing to do with the interference from the source (was Melco in use at the time?) , and of course could be down to synergy with the remainder of the system. I haven't had the opportunity to compare, so have no view.

You need to look at the iFi supply. It most certainly does filter out noise, indeed that's it's sole raison d'être. It's not for powering the Mini, I was using a Paul Hynes power supply for that. QBD76 has a very fine USB input, considerably better than the Hugo's.

It's no good looking for these sort of explanations, they don't figure; it's down to system synergy, and mine is particularly fine at revealing how good or not any component might be.

If you didn't notice a significant increase of sound quality with the Melco I'd look at your own system for problems.

Unless I've missed something it seems the iFi usb power supply is intended for devices powered via the usb, which is not relevant to Mac Mini to HUgo, QSB76 or Dave as they all have their own power supplies, and It is not evident that it has any RF filtering on the data side of the USB link, therefore any RF on the Mac Mini USB output - which is inevitable - will affect the DAC unless either the DAC adequately filters, or some other RF filtering is employed. The impression I have from reading some notes by Rob Watts is that it was only after developing Hugo that he became fully aware of the impact of RF, so that that later products than the QBD and Hugo had more attention to that aspect. I obviously don't know that RF interefernce IS the reason why you found the Melco N1A to be so much better than your implementation of Mac Mini into QBD76, when my comparison - yes, in a different system, but still a revealing one - gave no immediately evident difference (not full enough time to assess if any minor difference) between the N1A abpnd a well isolated MM/Audirvana implementation. Yes there can be other explanations, but RF noise on the Mac Mini output does seem to me to remain a possible explanation.

As for Dave vs QBD, I agree that could very well be down to overall system synergy - or simply that it is better in respect of the musical appreciation of whoever compared, and as I said, I haven't had the opportunity to compare so for all I know I may agree. 

Sorry to persist; I use the iFi Gemini dual cable which separates signal from power supply. The QBD76HDSD is still a better DAC than the Hugo, based on feedback from someone I really trust. When I last trialled DAVE in my system I had several forum members round, (J.N., Foot Tapper, Darke Bear) who came to the same conclusion as myself, that my DAC sounded better than DAVE. Others on here have experienced a more positive experience than yourself with the Melco. There you go, it's down to system differences.

Re QBD vs Dave, I haven't suggested otherwise! all I've said is I haven't heard, the point I have npbeen trying to make being a possible explanation as to why some people regard melco as better than Mac Mini source, while others report no difference, which I suggest is down to the individual setup and the signal presented to the DAC.

You may believe that the difference in experience with Melco N1A is down to system differences, but there is no proof either way, so either you or I could be right. I would observe that separation of signal from power supply from signal in usb does not itself remove RF noise superimposed on the signal, which is the issue. However unless someone can do the necessary direct assessments/measurements there is no clearcut answer - it's not even a matter of agreeing to differ, simply insufficient information to determine.

Noise from the USB connection using my Mac Mini, vs. Melco, into my DAC, given my setup, is extremely unlikely to be an explanation for the difference I heard. None of this stuff is for certain, but differences in system have a profound effect over what we get in terms of sound quality. 

And there we must leave it! (Unless anyone with definitive info on RF undervthe different setups joins in)

Before my chance opportunity to hear the N1A last week I was talking to Melco about possible home trial, of N1A and N1Zssd - I may proceed with that later in the summer, out of interest

TonyM,

Why do you believe that the Mac Mini and the Melco have the same noise (RFI) distribution?
(If the noise distributions are different, than that's a perfectly reasonable explanation for perceived differences in the sound of the system when these two are swapped.)

Undoubtedly, like any other such setup, the influence of RFI will be different. But given how the various sources I've used in my system, and the umpteen DACs I've tried, It's difficult to accept the dramatic improvement in sound quality I experienced with the Melco can be largely down to how well the Melco handles USB noise. When I used the Mac Mini, the iFi supply gave a significant improvement. I tend to look to the most likely explanation for the difference with the Melco in my system compared with Innocent Bystander's experience. 

And I in turn believe that the rf isolation of TonyM's Mac Mini setup may have been less effective than mine, making the two different and so comparison with Melco also different and so this the most likely explaanation - although how significant would depend on the susceptibility of the QBD76, which is an uncertain factor. Interestingly the other respondees to a question I asked about the Melco on another thread, who asserted the Melco was better, though with different DACs from TonyM's, also had what appeared to be lesser isolation of the Mac Mini, which is consistent with my hypothesis.

Bart posted:

ChrisSU posted:

Graham Russell posted:

SSDs tend not to fail gracefully like spinning disks mostly do.

I'm really struggling with the concept of a graceful disc failure! The expletives that inevitably fill the room don't help, but in my world, any disc failure is pretty awkward. 

Just extend the little finger . . . instead of the middle finger ;-)  Much more graceful that way!!

No chance, I might spill my pint