Soekris' DAC implementations

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
My understanding about SqueezeBox Duet is than all tunes which are above 48 Khz is downsampled thenn compressed (or is the word compressed equal to downsampled in the threads I read ?) before to be sent to the Duet receiver (The server is embeded in an other device and makes this job, in my case a Synology NAS)
 
My understanding about SqueezeBox Duet is than all tunes which are above 48 Khz is downsampled thenn compressed (or is the word compressed equal to downsampled in the threads I read ?) before to be sent to the Duet receiver (The server is embeded in an other device and makes this job, in my case a Synology NAS)

Compression and downsampling are two entirely different things. Your server downsamples but does not compress.

Have you ever done a double-blind ABX test to see if you can hear a difference between 48k and 96k?
 
Compression and downsampling are two entirely different things.
:D I know that ! I report what I read without understanding why the word "compressing" was writed ! IIRC, they talk about the decompressing task at the reception in the devices (do they mix up with bufferering term?). Odd. This downsampling woulg give bad sounding results (due to the treatment tasks ?).

Many report a better quality with the SPIDF output of the SB Touch in relation to the Duet. It would give subjectiv better results with 24/96 Khz on the Touch without showed scope measurement of the spidf outputs of the both devices. Threads ae old, they didn't comment result with spidf output at 192 Khz on the Duet in relation to the still native 44 Khz of the Duet spidf !

Btw, the questions are: is there limitations due to the SB server soft and/or with the USB output devices embeded in some NAS like the Synologys ?
 
Last edited:
:D I know that ! I report what I read without understanding why the word "compressing" was writed ! IIRC, they talk about the decompressing task at the reception in the devices (do they mix up with bufferering term?). Odd. This downsampling woulg give bad sounding results (due to the treatment tasks ?).

All I can say is "don't believe everything you read, especially not on audiophile forums" :)

Talking about compression there is data compression (lossy and lossless), and dynamic compression, and some people actually manage to confuse all 3! FLAC is a losslessly compressed format (just a smart way to package the information tighter without throwing anything away), and decoding the FLAC in the DAC/streamer can be viewed as one form of "decompression" (but one with no audible effect).

Downsampling to 48k with a competent algorithm (and the SoX default algorithm used by the Logitech server is one of the best) is inaudible.

Many report a better quality with the SPIDF output of the SB Touch in relation to the Duet. It would give subjectiv better results with 24/96 Khz on the Touch without showed scope measurement of the spidf outputs of the both devices. Threads ae old, they didn't comment result with spidf output at 192 Khz on the Duet in relation to the still native 44 Khz of the Duet spidf!
I am sure you can find people on the Internet who swear music sounds better if you wear a wool hat. I suggest you look at the measurements done by Archimago.

Btw, the questions are: is there limitations due to the SB server soft and/or with the USB output devices embeded in some NAS like the Synologys ?
Some NAS boxes have limited CPU capacity and bog down when having to process silly sample rates.

A 384k sample rate implies 8 times as much data needs to be transmitted and processed compared to 48k. That causes much more CPU load, and that can possibly cause more noise and might even result in dropouts.
 
Some NAS boxes have limited CPU capacity and bog down when having to process silly sample rates.

A 384k sample rate implies 8 times as much data needs to be transmitted and processed compared to 48k. That causes much more CPU load, and that can possibly cause more noise and might even result in dropouts.

I have to dig here. Indeed my NAS have (like many home appliances) a poor processor and small RAM while not knowing how the Duet receiver is managing the packet at the arrival... a small buffer at least I surmise, but I really don't know !

Let's return (if you want) to the main subject. References have been updated in first page to help.
Thanks guys for the given references and links.

Anyone talked about power traffo, PS and EMC ?distance minimum between the traffo and the "huge" resistors of the R2R DAC ?

Do I ask too much questions, last outrages to the drosophiles ? I assume the building document Soren will provide will answer to many questions !
 
The Linux kernal used by the NAS always do SRC to 48Khz before outputting digital audio. HD audio playback on NAS is not bottlenecked by CPU resources.

Hi,

Thank you for your input.

So now bios downscaling due to some limitations of the hards in my USB Nas output !

Are you meaning whatever the outputport (RJ45 or USB of those appliances NAS) those small embeded kernel is always limited to 48 Khz sample rate by a sample rate compression (=SRC ?).

Here is a problem of understanding of me : Sampling Rate Compression is not downsampling ?

Does the final device (SB Touch e.g. as the Duet is limited to 44 Khz) which are receiving the 48 Khz SRC can "reconstruct" the original sampling rate sent compressed by those mini-Kernels ?

I.e : 96 khz in the material file then sent at 44 Khz by the NAS (SRC) then re "extrapoled" at 96 Khz by the device (SB touch? hummm, bad example as the SBserver is embeded in the Touch iirc! But in any case if a NAS before the SB touch : the linux kernel will SRC to 44 Khz before sending to the Touch ?!)

Or do I need to garbadge the NAS then put a full Linux/windows distrib in a mini pc to stream more than 44 Khz (to 384 Khz) to an asynch USB to I2S device like the Wave I/O, Amareno, etc ? Or finding a little device which have a big rom enough to embeding a better Linux distrib ? I'm lost here ! Oh no not a MAC :D !

Hummm, was never involved by such problem in my working environment (no multimedia files and devices involved), so sorry if some confusions in my questions !

Thank you for your interresting inputs
 
Last edited:
Are you meaning whatever the outputport (RJ45 or USB of those appliances NAS) those small embeded kernel is always limited to 48 Khz sample rate by a sample rate compression (=SRC ?).

Here is a problem of understanding of me : Sampling Rate Compression is not downsampling ?

SRC=Sampling Rate Conversion

RJ45 on NAS is not affected by the 48KHz limitation as it only transports data, not digital audio, and most likely the audio data is played by some external player.

To overcome the issue, we can:-
1) Ask the NAS vendor to do something on the Linux ALSA driver to remove the 48KHz limitation.
2) Use an external network streamer / networked FPGA audio decoder / miniPC.

Anyway, from my experience the 48KHz limitation does not affect to sound quality much.
 
The Linux kernal used by the NAS always do SRC to 48Khz before outputting digital audio. HD audio playback on NAS is not bottlenecked by CPU resources.

Really? They actually crippled the device drivers? Why would they have done that?

What version of kernel and ALSA does the synology use?

Actually I can't find any indication of a 48k limit - some versions seem to have a 96/24 limit.
 
Last edited:
So now bios downscaling due to some limitations of the hards in my USB Nas output !

Not bios, but device driver, and I do wonder about if even that is true.

Are you meaning whatever the outputport (RJ45 or USB of those appliances NAS) those small embeded kernel is always limited to 48 Khz sample rate by a sample rate compression (=SRC ?).
If there is a limitation, it is in the sound system, so only affects sound devices.

Here is a problem of understanding of me : Sampling Rate Compression is not downsampling?
It is sample rate conversion, so can be either up or down.


Does the final device (SB Touch e.g. as the Duet is limited to 44 Khz) which are receiving the 48 Khz SRC can "reconstruct" the original sampling rate sent compressed by those mini-Kernels ?

I.e : 96 khz in the material file then sent at 44 Khz by the NAS (SRC) then re "extrapoled" at 96 Khz by the device (SB touch?
When something is downsampled to a 48 kHz sample rate, anything above a 24 kHz frequency has to be thrown away (not that there is much there usually, and not that it matters to your ear), so there is no way to get it back. Any upsampling will just generate empty data bytes.

hummm, bad example as the SBserver is embeded in the Touch iirc!
Yes, there is a server embedded in the Touch, but I do not recommend using it, as it is very limited.

But in any case if a NAS before the SB touch : the linux kernel will SRC to 44 Khz before sending to the Touch ?!)
No, as the data stream to the touch isn't audio to the NAS, it is just a data stream.

Or do I need to garbadge the NAS then put a full Linux/windows distrib in a mini pc to stream more than 44 Khz (to 384 Khz) to an asynch USB to I2S device like the Wave I/O, Amareno, etc ? Or finding a little device which have a big rom enough to embeding a better Linux distrib ?
You could, but first I would check if you can actually hear a difference between 48k and 96k. What is the point of all that hassle if it doesn't make any difference? How many 384k recordings have you come across?

I suggest you ask in the 3rd party hardware subforum of the squeezebox community forum where they have a lot of experience with running NAS boxes as music servers.
http://forums.slimdevices.com/forumdisplay.php?18-3rd-Party-Hardware
 
Re reading Slimdevices threads I saw spidf is giving better subjective results than Toslinsk (spidf advised by some users' testimonials) with the SqueezeBoxs. In the same time I have Duet which stream only 24/48... yes 48 Khz while the Touch I haven't can be uptdated to stream 192 Khz ! All the upper materials being compressed before to be sent from the server to the SB receiver :eek: !

Why bother with the inferior S/PDIF interface when we finally got a dac with I2S input?
It turns out you’ve even got a NAS server, so I suggest you ditch the SB Duet and buy a cheap second hand SB Touch (can be had for about €100). Then you can even go up to 192Khz hires files. Or get an even cheaper Raspberry PI and install the Squeezelite freeware.
Combine it with an Amanero board and you’re all set. There’s an Amanero group buy here on diyaudio.
http://www.diyaudio.com/forums/vendors-bazaar/216474-usb-i2s-384khz-dsd-converter.html

IMO I2S is the right path although I myself might implement a S/PDIF connection as well, just as sort of a backup and for convenience purposes.
 
@Eldam: In fact, Synology NAS are able to output bitperfect through usb from 44.1 to 384khz since DSM 5.0 with DS audio / Audiostation app
French resource link: Test "Bitperfect" d'un NAS Synology via l'USB et le DLNA. : Musique dématérialisée - Page 22

If you don't like this app, there are alternative mpd packages that can do the same thing.

That should close the OT on NAS :D

Since dual dac raw balanced output is not possible yet, I was thinking about using Soekris unbuffered output to an OPC Wire BAL-BAL headphone amp with negative inputs grounded so that the OPA1632 does the single ended to differential conversion. (Figure 11 of OPA1632 datasheet page 7)
It should be higher end than the included LME49710 + LME49724 for driving headphones requiring large current.

My concern is about the impedance output of the raw dac outputs versus input impedance of the Wire BAL-BAL.

Default input resistor of the Wire is 1 kohm, and raw dac output is 1.25 kohm.

Based on this post - I was wondering if 10kohm input resistor such as used in the calculation would be enough ?
 
Last edited:
So now bios downscaling due to some limitations of the hards in my USB Nas output !

Are you meaning whatever the outputport (RJ45 or USB of those appliances NAS) those small embeded kernel is always limited to 48 Khz sample rate by a sample rate compression (=SRC ?).

I can confirm that Synology NAS' can output HD audio through their USB ports to Amanero boards. I have a DS211j running DSM 5.1 outputting audio to a Buffalo III through an Amanero:

attachment.php


IMG_8310_resize1.jpg


In theory it works up to 384kHz, even though I couldn't go that high with my NAS since it is a bit old and underpowered (the files did play but with dropouts).
 

Attachments

  • dsm 5.1 audio 3.png
    dsm 5.1 audio 3.png
    496 KB · Views: 753
Member
Joined 2006
Paid Member
Synology NAS units will run Minimserver also.

I am completing a Win8.1 based server for my house. Minimserver on a headless PC in the network closet. This unit will allow me to keep my whole library with some hires files for use in the music room, and keep a separate directory with only 14.1/44 files for the Sonos units elsewhere in the house. Backup duties as well.
 
Wow, many answers.

thank you fellows for all your answers about streaming! Very interesting, I've been looking for a RJ45 to I2S (with a good I2S output layout because RF : GSGSGS... ) to swap my Duet but still no existing in DIY World ? I mean instead of the world of the ROM based apliances RaspBerry likes which not seem to have a so good I2S output and when they have one ?

Humm, based on the answers above ,I believe i putt a hand in RaspB. world... not too much expensive... while re-put the hand in Linux is not my cup of tea ! We do a lot of effort for our Hifis :)

Funny, many of us are focusing on 384 upsampling... while finaly some of us seems ok with 44 or 96 Khz should be Toslink or spidf or USB NAS limited !

Where do I put my Rega Planar III ! :) . I mean for the moment we can surely upsampling by a soft our red book libraries and Stream it at 44 Khz waiting for more High Res recordings !

PS: @GuruLubrik (what a name ! t'es du coin du mandarom plus bas dans les ALpes de Haute-Provence? ouarf !) : I find it works very well since the DSM 5 ! I've the last one 5.1 and the last squeezebox server revisions : works fine. like SB Duet for the remote controll : both the genuine one like the Android one which is even faster !

So that's fine if If my DSJ 210+ can Stream at 384, I would go for a Wave/IO which or an Amareno as the Soekris'dac has an isolator (I sleep behind the front door heating pizzas while waiting fo UPS and wife is fulling its holyday case:D)...
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.