SSD for NAS

So, you really believe the difference in isolation between my original Mac Mini setup and yours explains this? Sorry, you can hypothesise all you like but I cannot accept this. The Melco is significantly different to a Mac Mini and if you heard no difference between the two then that points to something else.

My explanation is no less likely than yours! However, clearly we're not going to agree, and as I said it's unlikely any conclusive evidence will emerge for one explanation over the other, so we'll just have to respect each other's different views. The important thing is having whatever works best for us in our own systems.

Yes, indeed, but I think I need to throw something else into the ring here, which I neglected to mention. Until recently there was no USB driver available for the QBD DAC, so when I first home-trialled the Melco it was via my iFi supply into a Wave Luckit IO USB to SPDIF board, optical out, using a Paul Hynes power supply. To draw the comparison as accurately as possible (I didn't really want to pay out for another bit of kit!) I was swapping the USB cable across from the Mac Mini to the Melco, the DAC being locked to the Wave board. When using the Mac Mini (because of the custom nature of the QBD's USB input this also required a driver) the USB input (which, by the way, is galvanically isolated) sounded better than via the Wave board. But even using USB direct from the Mini vs. the Melco via the Wave board, the Melco still sounded considerably better. Soooo...The most recent software update to the Melco included the driver for the QBD76HDSD (the gentleman who heads up Chord in the UK worked for Chord and holds the QBD in high regard). The result is a further increase in sound quality and one less box.

What to make of it all? I think I did sufficient fiddling about with cabling and kit (including other things - the Mini had an optical out, so tried this. Tried without the iFi) to establish that the Melco gave a surprisingly wonderful improvement in sound whatever I did.

Reading through this thread, one of the key differences between HDD and SSD that I have used professionally is power cycling reliability. If HDD are left powered up and spun up they are incredibly reliable, however if they are regularly power cycled then the reliability reduces and the SSD  handles this better. SSD also consumes less power when operating and so is ideal for battery powered equipment.

But if used on a music store NAS that is left running I really think there is little to choose between them and I would be inclined to mirror the largest devices you can get.

From an engineering perspective, if the noise floor modulation from different disk technologies had an observable/audible effect on the system, that I would be a little concerned about the system architecture and noise decoupling. In my experience such finely tuned system optimisation has the risk of producing differeing audible output on system or firmware software changes. There is increasing understanding that our brains are impacted/confused by noise floor modulation in audio replay systems whilst undertaking its auditory processing.

On streaming technology, I am undertaking some tests, as I have discovered, certainly with Naim streamers, that network I/O timing seems to have quite an audible effect almost certainly caused by noise floor modulation from the streaming board. I will be analysing the streams now that I think I have worked out what to analyse and measure. I believe this accounts from differeing SQ from media servers such as from Melco, different NAS, RPi etc when streaming.... quite exciting stuff (for me). If and when I get meaningful results I'll share here.

Simon-in-Suffolk posted:

that network I/O timing seems to have quite an audible effect almost certainly caused by noise floor modulation from the streaming board. I will be analysing the streams now that I think I have worked out what to analyse and measure.

Nice work - I'll be very interested in your findings Simon

TonyM, I get the impression that you take my comments to be an affront on your system or your hearing, which it certainly isn't. I don't doubt for one moment that your system sounds wonderful. For the sake of clarity, I have no reason whatsoever to doubt that the QBD76 is great, that its latest firmware has improved it, and that it sounds better than Hugo, and that at least within the synergy of your system it sounds better than Dave, and I have no reason to doubt that in your system Melco sounds better than the Mac Mini setup you had been using.

But it does seem to me that you have missed the point of my observations, which were suggesting an explanation for the Melco/MM difference, and why not everyone has the same experience. Your response this morning merely reinforces my impression that you have somehow missed the thrust of what I've been saying.

My point is simply that the sound quality of some, maybe many, DACs can be affected by electrical noise, most notably RF, superimposed on the input signal, and that the effectiveness of different approaches to removing the noise in the Mac Mini's (or any other computer's) output can vary, and that if whatever approach is used is less than fully effective there could be an effect on SQ. I then suggested that a difference in effectiveness of the measures to remove RF noise employed on your MM setup and those employed internally within the Melco N1A could account for why you heard a clear difference (likewise others who have found the same). The description of the measures employed in your MM setup via usb does not negate this because there is no clarity as to the effectiveness of RF filtering on the usb signal, meanwhile feeding the Melco through the same iFi device is irrelevant if the Melco's output is already clean. This is not suggesting that you didn't go to some lengths to try to clean up the MM's output, following what at the time may have been promoted as state of the art solutions to the problem (and of course solutions improve with time, so some options available now may not have been, or at least may not have been so widely known when you set up yours).

The fact that I did not hear an obvious difference between Melco and MM usb output without the Gustard is easily explained if Dave (That was the only DAC it was played through) has such effective RF filtering that the noise in the MM's output is reduced to the same level as Melco's (or indeed both reduced to an even lower level). The only lack of certainty in all this is the difference in noise (RF) filtering in Dave's input compared to the QBD, and I base my hypothesis on the development path by Chord, and Rob Watts' comments on RF and his evident learning along the way - and I'm not sure that firmware updates can fix a lack of input RF filtering, though others with more specialist knowledge may be able to confirm.

I hope this has clarified things a bit!

james n posted:

Simon-in-Suffolk posted:

that network I/O timing seems to have quite an audible effect almost certainly caused by noise floor modulation from the streaming board. I will be analysing the streams now that I think I have worked out what to analyse and measure.

Nice work - I'll be very interested in your findings Simon

I will be utterly astonished if noise floor modulation has anything to do with I/o streaming. Simon, as you know, this is a very theoretical debate that has arisen elsewhere. The current perceived applicability of noise floor modulation is believed to be within the DAC and has to have some feature attributable to the ambient surroundings eg mother board heat, RFI, etc. I still cannot see or understand how that noise floor can be translated by the internal processing of the DAC.

However, if you say that post processing the analogue signal is corrupted by noise floor modulation then that is a different story. If noise floor modulation had any discernible impact on the I/o of data other than destroying it completely then it is highly unlikely we could watch a digitised and or streamed film. There is a difference, obviously, between I/0 interference causing annoying interruptions but when the data is properly transmitted and received only the packaged data can and  should be processed. Injecting noisome flavouring IMHO can only be done from the DAC onwards. (All disgusting allusions purely intended.)

I certainly don't take your comments as an affront Innocent Bystander, merely trying to understand why you didn't hear the considerable improvement in sound quality I experienced. You do seem intent on pointing to RF shielding as the reason, which, given the lengths I went through to ensure this wasn't a factor, really doesn't make much sense, particularly when I was using optical digital direct from my Mac Mini as part of the comparisons I made. You make several assumptions to justify your viewpoint regarding the QBD76's USB input, and the iFi supply. If DAVE has a better USB implementation than my QBD, then the cause of DAVE sounding less good must lie elsewhere. You've not mentioned the Wave IO board which I used in my original evaluations, or that I wasn't in fact, using the QBD76's USB input until very recently.

Should you wish to journey to Suffolk, you'd be very welcome to visit me & bring your Mac Mini setup over with you. I would certainly be interested to hear it in my system.

james n posted:

Simon-in-Suffolk posted:

that network I/O timing seems to have quite an audible effect almost certainly caused by noise floor modulation from the streaming board. I will be analysing the streams now that I think I have worked out what to analyse and measure.

Nice work - I'll be very interested in your findings Simon

+1

When considered alongside RFI, I believe these to be the most significant uncontrolled variables that affect how a given streamer will perform in practice in any given individual network.

The fact that RFI influences analogue electronics is well established, and approaches to ameliorate its effects are known.

However, despite there being a very good theoretical rationale indicating that there may be an effect of network timings, to the best of my knowledge Simon may be the first to pursue an authoritative investigation into this.

Tony,

Please excuse my inconsistent reading ability, but I missed the details of your approach to ensure RFI wasn't a problem.

AFAIK RFI is always a problem, it's a question of how influential it might have been when making the comparison between a Mac Mini & a Melco.

Thanks for the invitation to Suffolk - if I lived nearer I'd be very intrrested in taking you up on it, but I'm on an island quite a distance away, so unlikely to be an opportunity at least in the near future.

I'm not familiar with the Wave Luckit, but the same principle applies - if the Melco has better filtering in its own output than MM+Wave, then the RF presented to the DAC would be less with Melco whether or not connected through the Wave. Re optical output from the Mac, that is known to be poorer SQ than the usb output, the optical relying on the Mac's sound card, which is bypassed completely when using the usb output (which can also be a dedicated usb bus).

andarkian posted:.

However, if you say that post processing the analogue signal is corrupted by noise floor modulation then that is a different story. 

That's essentially my hypothesis - but also including the digital DSP/DAC clock which is hyper sensitive to noise modulation and  well understood.

I don't have any advanced spectrum analysis equipment at home currently - and so my evaluations will be be based on empirical listening.

Huge - my findings I hope will be interesting - but I am not sure how authoritative they will be   and of course it might lead me down a dead end. But I have certainly discovered some very observable differences that warrant investigation because at one level they could be considered identical. I have also, by reading some engineering development papers published by TI developed a greater awareness of digital transports including Ethernet and USB and ground plane modulation causing noise that has given me some of the ideas to explore. TI of course make the DAC chips that Naim use in their higher end devices. I have been experimenting over the weekend on detecting and measuring TCP flow timing variations. There is more to do,  and I will be exploring the variation and patterns of this on the empirical listening evaluations of streaming from different media servers... so we shall see how we go and have to fit around the day job...

Innocent Bystander posted:

Thanks for the invitation to Suffolk - if I lived nearer I'd be very intrrested in taking you up on it, but I'm on an island quite a distance away, so unlikely to be an opportunity at least in the near future.

I'm not familiar with the Wave Luckit, but the same principle applies - if the Melco has better filtering in its own output than MM+Wave, then the RF presented to the DAC would be less with Melco whether or not connected through the Wave. Re optical output from the Mac, that is known to be poorer SQ than the usb output, the optical relying on the Mac's sound card, which is bypassed completely when using the usb output (which can also be a dedicated usb bus).

Oh well, maybe some other time. Yes, understand the sound quality issues around the Mini's optical output (although mine was a later model with apparently improved specifications). I was only using it for comparisons. You see, I don't doubt the Melco has better filtering than the Mac Mini, but I'd assert the Wave board (link here) gives excellent isolation, leaving aside that built into the iFi device. It really doesn't explain why you heard little or no difference whilst I heard a substantial one, no matter what I did.

Anyway, enough. I would suggest though, if you're considering a new DAC, you try the QBD76HDSD. It's still current, but seems to have been forgotten since the advent of DAVE.

Couldn't find an online version of the review so here is a picture out of my magazine copy of Paul Millers conclusion from the N1A review (HFN Aug 15) which mentions measurable differences via USB. 

Richard not sure if this contravenes forum posting rules so please delete if it does.