NAS Super-Low Noise Linear Power Supply

[Macca]

What he says^.

I have a dedicated mains feed to my modest system and could swear it makes a 5-10% (whatever that means - audible, basically) difference. My NAS sits on the general house supply, so the only way I would get an improved performance through running it off a linear PSU (as I do my Squeezebox and iRDAC which are on the dedicated mains) would be by somehow cleaning up the 0s and 1s being read off the drives and transmitted along the Cat6 cabling to the Squeezebox.

I struggle with how I might improve a stream of 0s and 1s, even if it's from a typically dirty house mains to a clean dedicated hifi one. Do the 0s and 1s change shape? Do extra ones sneak in? Do some of them get tired and give up on the journey? I genuinely "get" the benefits of linear PSUs over switched, I just can't apply that to a NAS as I can't see what it would change at the 0s and 1s level.

I'm speculating obviously but whilst the 0s and 1s might stay the same the amount of noise transmitted with them might be altered. That won't make any difference if you are sending a Word document but if you are sending a signal that will ultimately be translated into sound waves and heard then I think it conceivable that it might make a difference. The 0s and 1s are just variations in the voltage of an electrical transmission in the same way as the signal from amp to speakers is. If the latter can transmit noise why not the former?

[Marco]

Impressive, Stuart. Looks like you've used the same case as me

I'm sure you'll achieve major sonic benefits if you build one for your RPi and SSD, and if you do, please chronicle the build on here, for the perusal of the membership

Marco.

[Paul Hynes]

Hi Marco,

Originally Posted by Paul Hynes
I’m more or less OK again now, thanks. Unfortunately it would appear that I am no longer invincible and I will have to pay more attention to my health in future. I am working on getting more help with the business, which in the longer term will ease my workload, but it is a slow process getting things done up here.

Good to hear, Paul, Yes, as we all get older, we're definitely not as 'invincible' as we once thought we were! If there's anything I can do to help with the latter, let me know and I'd be glad to.

I am not sure there is anything that you could do to help with getting things done up here, but I am all ears if you have any ideas. This area is designated a priority area for development by Scottish development agencies, to increase local employment opportunities, in a bid to arrest the depopulation that has been going on for decades. I was planning on retiring last August, but a chance meeting last autumn, with a retired development officer in the local supermarket, provided encouragement to make an effort to help, by setting up a replacement manufacturing facility here for the now closed Paul Hynes Audio Ltd. This would provide much needed industry diversification on the island and provide both the younger and older generations with local employment opportunities. Historically they have had to leave the island for College and University education, but after this education they do not return to the island due to lack of local employment opportunity. The net result of this is a declining population of ever increasing age and there are only around 1800 people living on the island at present. It would be sad to see South Uist go the same way as St Kilda, where the island population became unsustainable and everyone left, with the last person literally switching the light off.

There is a plan in place based on local development funding access and manpower support with a proposed site for the facility at the new Lochboisdale Marina development. The time consuming part is the acquisition of the very comprehensive data required for preparing the drawings for the building and the data required for the business plan and funding applications. I am also still building power supplies for audio on a custom basis at present, so I will be rather busy this over the next couple of years, although this will change once the project set up is accomplished.

1. Out of interest there has been much discussion about digital interface items, audio servers, NAS storage, and power supplies over on Computer Audiophile in the rather long thread :-

https://www.computeraudiophile.com/f...nd-microrendu/

Well worth the read with much pioneering work on improving digital audio systems.

Indeed, and a rather novel use for an SR7! Somewhat 'extravagant' in that application, no?

Not really extravagant as I have been providing SR7 and SR7MR (Multirail) power supplies for powering audio and video servers for years. They are used for both the server and the peripheral digital signal storage and processing items with considerable improvement in sound and video quality. The improvements are very similar to those you found with the SL1200 and Raspberry Pi power supplies. Essentially they remove a major cause of “digititis” and provide a robust musically engaging and dynamic three-dimensional presentation to music program. They also clean up power supply related video signal processing hash improving the screen image rendition.

We're all fine, thanks. Looking forward to the summer, and also the completion of our new (bespoke) kitchen!

I sense there will be some exceptional culinary work later this year at Marco Towers. Enjoy the moment.

Regards
Paul

[WAD62]

I'm speculating obviously but whilst the 0s and 1s might stay the same the amount of noise transmitted with them might be altered. That won't make any difference if you are sending a Word document but if you are sending a signal that will ultimately be translated into sound waves and heard then I think it conceivable that it might make a difference. The 0s and 1s are just variations in the voltage of an electrical transmission in the same way as the signal from amp to speakers is. If the latter can transmit noise why not the former?

My guess is that Marco's NAS is streaming PCM to the Pi, so we're already outside the IT domain IMHO, hence the negative effects of a low quality NAS PSU...if he could do the transcoding at the Pi end, and therefore stream the 'File', I doubt the PSU on the NAS would make much of an impact...

[Marco]

Sorry, Will, I'm trying to wrap my nut around what you're getting at... For starters, what's PCM and transcoding?

And can you describe, in the clearest laymen terms possible, what distinction you're trying to make between 'transcoding' at the Pi and NAS ends respectively, and why you consider one method potentially more sonically advantageous (less liable to be influenced by noise, and hence the benefits of a linear PSU) than the other?

Marco.

[TheFlash]

I'm speculating obviously but whilst the 0s and 1s might stay the same the amount of noise transmitted with them might be altered. That won't make any difference if you are sending a Word document but if you are sending a signal that will ultimately be translated into sound waves and heard then I think it conceivable that it might make a difference. The 0s and 1s are just variations in the voltage of an electrical transmission in the same way as the signal from amp to speakers is. If the latter can transmit noise why not the former?

I'm way out of my depth here technically, but do understand that a "digital" cable is actually an analogue cable down which digital signals are sent. Between a NAS and a Squeezebox or DAC, that signal will start as a series of 0s and 1s and end as a series of 0s and 1s. Any "noise" must therefore manifest itself as corrupt digital data - a 0 where a 1 should be, or t'other way round, or maybe some extra 0s or 1s sneaking in, or some 0s and 1s falling by the wayside, or some sort of jumbled sequence. Is that correct?

[WAD62]

Sorry, Will, I'm trying to wrap my nut around what you're getting at... For starters, what's PCM and transcoding?

And can you describe, in the clearest laymen terms possible, what distinction you're trying to make between 'transcoding' at the Pi and NAS ends respectively, and why you consider one method potentially more sonically advantageous (less liable to be influenced by noise, and hence the benefits of a linear PSU) than the other?

Marco.

Hi Marco, I'll try...

Whilst the data being transmitted is in 'File' format, eg the compressed FLAC file, or the WAV file etc. we are still in the realms of IT, I.E. a reliable File Transfer(TCP/IP), so there's all of the standard FTP protocols in place, so the file received by the player is bit for bit the same as the file sent.

Ultimately the 'File' will need to be converted to a data stream that can be consumed by a DAC, in this case PCM (pulse code modulation, DSD is a newer format), to my mind this is where we start moving out of the IT world and into the analogue world, where interference can occur, and bits are no longer bits etc. The process of converting a 'File' to a PCM stream is transcoding.

With LMS (logitech media server) running on my NAS, I can choose whether to transcode to PCM at the Server running on the NAS, or at the player/Pi end, I choose to transcode at the Pi end, firstly for network efficiency (I'm only transmitting a small FLAC file not a 2 * 1411 Khz PCM stream), and secondly I want to decouple my player from the NAS, all it does is send the unpolluted FLAC file.

In the earlier days of Squeezebox, when the players were a much lower spec than a Pi, some people preferred to Transcode to PCM at the server in order to take the load off the puny player (the CPU on the early SBs is a fraction of the Pi's power)...and felt they gained an improvement, hence this option exists in LMS.

I can't find anything about the streaming protocol used in Moode, but it would appear that Roon streams in PCM by default, something to do with synchronisation issues...

However I would like to be wrong in this instance, the idea of improving the SQ of 3k FLAC cd's with a relatively inexpensive NAS/PSU is very appealing...but I can't see how with my streaming architecture



EDIT: I've had a quick look at the moode installation instructions, and it would suggest that it's a stand alone player, that can refer to the NAS, so I'm even more confused!!!

[TheFlash]

Hi Marco, I'll try...

Whilst the data being transmitted is in 'File' format, eg the compressed FLAC file, or the WAV file etc. we are still in the realms of IT, I.E. a reliable File Transfer(TCP/IP), so there's all of the standard FTP protocols in place, so the file received by the player is bit for bit the same as the file sent.

Ultimately the 'File' will need to be converted to a data stream that can be consumed by a DAC, in this case PCM (pulse code modulation, DSD is a newer format), to my mind this is where we start moving out of the IT world and into the analogue world, where interference can occur, and bits are no longer bits etc. The process of converting a 'File' to a PCM stream is transcoding.

With LMS (logitech media server) running on my NAS, I can choose whether to transcode to PCM at the Server running on the NAS, or at the player/Pi end, I choose to transcode at the Pi end, firstly for network efficiency (I'm only transmitting a small FLAC file not a 2 * 1411 Khz PCM stream), and secondly I want to decouple my player from the NAS, all it does is send the unpolluted FLAC file.

In the earlier days of Squeezebox, when the players were a much lower spec than a Pi, some people preferred to Transcode to PCM at the server in order to take the load off the puny player (the CPU on the early SBs is a fraction of the Pi's power)...and felt they gained an improvement, hence this option exists in LMS.

I can't find anything about the streaming protocol used in Moode, but it would appear that Roon streams in PCM by default, something to do with synchronisation issues...

However I would like to be wrong in this instance, the idea of improving the SQ of 3k FLAC cd's with a relatively inexpensive NAS/PSU is very appealing...but I can't see how with my streaming architecture

Great post, thanks!

[WESTLOWER]

With LMS (logitech media server) running on my NAS, I can choose whether to transcode to PCM at the Server running on the NAS, or at the player/Pi end,

Will....Please, What are the settings for the PCM conversion on LMS to Transcode at the Pi end? I would like to switch this to the Pi end if it is currently
Transcoding at the NAS...(i'm not near the equipment at the moment!)

[Smegger68]

A few minor errors in understanding here.

Whilst the data being transmitted is in 'File' format, eg the compressed FLAC file, or the WAV file etc. we are still in the realms of IT, I.E. a reliable File Transfer, so there's all of the standard FTP protocols in place, so the file received by the player is bit for bit the same as the file sent.

Data between the NAS and the Pi isn't in any form other than TCP/IP packets. This is the common internet transport protocol responsible for the webpage in front of you and millions of other reliable information transfers done every second. As you say, this ensures that the data on the NAS arrives at the Pi exactly the same. FTP is a Transfer protocol, not a Transport protocol.

Ultimately the 'File' will need to be converted to a data stream that can be consumed by a DAC, in this case PCM (pulse code modulation, DSD is a newer format), to my mind this is where we start moving out of the IT world and into the analogue world, where interference can occur, and bits are no longer bits etc. The process of converting a 'File' to a PCM stream is transcoding.

The process of converting one encoded data set to another is transcoding. PCM will be derived from the file by the codec built into the Pi and fed into the DAC. This is demodulation. Files are simply containers for various types of data - A/V file containers can contain data in several diffeerent types which is decided by the codec used at encoding time. This is important to know because when transcoding, the final file type will depend upon the new data set. For example, if you convert a FLAC @ 24/192 to 16/44.1, you can either choose to make the new file a FLAC or a WAV. File type is much less important than the actual encoding used within the file container. Video files are MUCH more complicated again

With LMS (logitech media server) running on my NAS, I can choose whether to transcode to PCM at the Server running on the NAS, or at the player/Pi end, I choose to transcode at the Pi end, firstly for network efficiency (I'm only transmitting a small FLAC file not a 2 * 1411 Khz PCM stream), and secondly I want to decouple my player from the NAS, all it does is send the unpolluted FLAC file.

This is essentially correct, the file could be transcoded at the NAS end into a WAV file (which is a file container for PCM) and sent to the Pi over the network for demodulation, or as you have it set up which is indeed more efficient.


It's an interesting thread, but it should be pointed out that improving power supplies to a NAS can only improve the sound by reducing mains noise that affects the analogue components of an audio system. The NAS operates in the digital domain which is agnostic to noise of this sort unless it's so bad that the system will simply just not work anymore. There is no gradual downgrade of a digital network signal - it works right up until the point at which error correction cannot cope and then it stops functioning. It should be remembered that the inside of your PC (or any digital device) is rife with RF noise and it doesn't stop you from logging on the AoS. Error correction within TCP/IP works all the time, it's effects are not felt in the time domain because TCP/IP is a buffered protocol. If anyone tells you that they are getting jitter from their NAS, you are safe to laugh at them. Jitter only occurs in streaming protocols like S/PDIF which are clock locked.