cmiu

quick question regarding tidal windows application: I've noticed that if the device sampling rate is set in windows audio device preferences on other value different from 44.1/16 the output to usb DAC is upsampled to the specified sampling rate. This raise the following question: is windows Tidal player it perfect? I've tried to find o definitive statement on Tidal site but despite my efforts I was unable to find a clear statement about windows player implementation, all the information found is referring to lossless source files. Thank you

hi there, glad to see another project around mpd. here are few things from my experience with mpd: 1. you should opt between a high power horse cpu or solution based on arm (rpi; bbb etc...). use high (x64) when you plan to do stream processing like upsampling/drc or go for lower (arm - BBB or something with quad core) if you plan for bit perfect and trust your dac (your case). pros for low power arch - everything in cheaper eg: power source ; case ; small physical footprint. 2. ditch everything that is unnecessary from your computer. eg: hdd - with enough ram you can make a ram disk (the system I'm running consumes aprox 90MB), you can perfectly boot from PXE from your nas or you can use voyage mpd 3. tweak irq smp affinity and monitor /proc/interrupts 4. force cpu to a stable frequency with cpufreq-utils, ussualy the lowest. frequency adjustments are the source of a lot of electrical noise inside the computer. 5. RT kernel is somelike obsolete with kernels > 4, and you are not recording music so low latency is not a must 6. if you can avoid usb data transmission entirely eg: http://bbb.ieero.com/ (BBB+reclock+bufallo II) in my opinion, for an entry level diy-er, is the best aproach. for linux based audio solutions you can find valuable info here: Linux Audio the way to go!? - diyAudio PS: IMO Jude&all are spot on about upsampling in computer vs letting dac chip do whatever upsampling likes; listening now mpd -> sox-r upsampling from 16/44.1K to 24/176.4K -> brutefir (processing at 64/176.4k bytes for a little DRC to tame my room) and an old i3-3225 running at 1.60GHz hava a load around avg 10% per core and a lot in consumed by irq/23-ehci_hcd ...

Thanks, I'm asking because I'm using different ARM boards. For example for BBB I used volumio as a starting point, updated it to jessie, compiled latest kernel and latest git mpd. I'm just trying to reinvent the wheel. I've build some power souces as well.

Could you pleas detail a little bit more: - did you use vanilla version of volumio? if not - kernel version - mpd version - tweaks could you please detail a little bit more IBM's kernel tweaks? or any other tweaks? Thank you

@atreyu linux is best fitted for a headless server with a separate controler thru ios/android app or web page the solution depends on your linux skills: A. you can use ssh console and follow a simple how to - go for basic debian testing and custom install of mpd or logitech squeeze box B. you need mouse - try to find a dedicated distribution for your HW my system x64 - hdplex case with core i3. only cpu, ram, memory / fanless, headless, no hdd attached. debian jessie, boots from network. only alsa and mpd installed. Previously I used voyage mpd, booted from usb stick. and when my little one needs it I boot openelec for movies and youtube ARM - BBB - best in class for audio - cheap, low consumption (cheaper power supply), and able to play all the common hires from NAS without problems. Looking forward for I2S reclocker from twisted pear. wandboard quad - a lot of cpu horse power (able to upsample till 192kHz with SOX), a little bit power hungry and no premade mpd based audio distribution only squezee box. cubietruck - bougt it for I2S output, but is not worth without a reclocker board cubox - the nicest visual presentation as it comes as a very small box it is suported by volumio if I remember corectly. For ARMs I prefer only mpd and alsa because they have limited cpu processing power and I prefer to spare it. If I understand correctly the latest raspberrypi version was supposed to solve eth/usb irq share and the consequent USB DAC poor performance. Botom line BBB + twisted pear reclocker + a I2S DAC may be a winner in long term, in short term I love my BBB - USBtoI2S converter + I2S DAC and x64 - USB DAC combo

Yup, I just got it from the picture, I was curious because is the first USB-I2S that has the FPGA in the galvanic isolated side along with the clocks, usually USB receiver and FPGA were on the "dirty" side . But maybe this approach works better due to lack of isolator between FPGA and clocks. a. I like to tweak things so from my POV I like to have access as much as I can. In my case I control ESS directly with BBB and a python script. b. taking in consideration point a in my case I prefer to upsample and use my custom FIR oin the computer side and let the DAC do what it do best convert digital signal to analog signal in best conditions:) . as you probably noticed in the r2r thread from diyaudio it is easy to create and upload you custom filter in a FPGA so I see no point to use a "closed" aproach. as a conclusion: it sounds nice (based on your experience), it has a new aproach from my point of view it deserves a test. on the other hand the lack of (software) flexibility can drive advanced diyers away. Thanks

1. dead quiet (fanless, no moving parts). 2. I don't see the reason to include HDD as most of the users are using NASes so it is relatively simple to PXE boot the computer and run everything from memory 3. maybe it will good to include (make) a couple links to a basic network and nas configuration creation. I'm not a fan of closed boxes when you can make a better NAS with fewer money. 4. as for software (in case of PXE boot) switching between daphile, voyage for example, or any user configured image is one reboot away windows can be booted as well from PXE but because MS made a mess with UEFI is a little bit harder. If you can "split" the configuration like this: Level I bare computer (case, mobo, proc, memory) Level II HDD, sata cables Level III USB / SPDIF HiFi adapters Level IV Power source Leve V EMI treatment for case etc .. etc ...

Those two were meant as examples. What was your previous USB to I2S converter? From technical point only Spartan chip should stay in usb part rather than clocks part, IMO. any special usb cable? did you tried any arm based device apart from sonore server? Because we are talking about a diy project (board) maybe a little more versatility to control DAC settings will be fun. For example some 16/44.1 tracks need deemph, IIR bandwidht or FIR rolloff . One ideea should be to let Spartan control ESS chip via I2C. My project is built (for the moment) with BBB + Joro's USB to I2S + Bufallo III. http://comprami.ro/upload/streamerstein_beta_01.jpg the picture is from a previous version with wandboard quad. Thanks for your answer and sorry for short answers but as you probably guesses English is not my primary language.

Hi @barrows, three questions: 1. did you had the chance to compared it to other galvanic isolated USB - I2S interfaces? eg: XMOS DSD 384 kHz / 32bit USB - diyAudio or Isolated XMOS DSD DXD 384kHz high-quality USB to I2S/DSD PCB with ultralow noise regulator - DIYINHK 2. could you please describe the connection between sonore server and USB-I2S interface? 3. what registry setting do you use for ESS dac chip? default or do you control it with I2C?

You don't have to recompile mpd. a) you can use pipe for binary sox b) you can replace mpd resampler with libsoxr www.audiofaidate.org • View topic - MPD (Music Player Daemon) I prefer to split the dsd iso files because I can add the tags and keep the mpd library "clean".

You can split the ISO's with the app used for ripping the SACDs. Based on your preferences you can split the ISO in dsf or dff files. You can tag the files with Picard and SACD tag plugin. You can play the files with MPD or squeeze server. Keep the ISO's for backup and for the multi channel tracks.

With all due respect but your lack of skill and deep understanding of hw and software (for example you failed to configure the cisco router vlan for files server and renderer, change the switch MTU etc ..., your test software setup on the renderer machine is based only on OSX) makes your tests almost null from a scientific point of view. You fail to make a correlation in a standard environment and extrapolate your findings to different systems and replicate the differences in other setups. From HW point of view ethernet interface is different from USB your analogy does not stand. Not to mention the differences between the data layers .... Trivia: did you know that ethernet hw is galvanic isolated? and for synced clocks and real time ethernet data transmission the delta of time the reconstructed stream is zero in both cases. John Swenson has a nice theory but it fails when it comes to galvanic the newer isolated USB with reclock interfaces, or maybe I didn't find any post regarding this type. And horror a lot of audiophile reviewers found that the diferences between cables tend to got to zero with this USB topology (ofc the results cannot be zero because the economics involved)

Just to be sure I took the NAS aproach - freenas. I'm using six hdds configured as zfs raid-Z and scrub my data periodically (30 days). and smartctl weekly checks for every hdd in order to predict hdds failures. It is not bullet proof but add another layer of checks for reliability of the hardware. and most important I use off site back-up as well. google search "bit roting zfs" ZFS - Wikipedia, the free encyclopedia - Data integrity http://www.freenas.org/ pretty easy to install and configure and the documentation/community support is very good. You can even create one click virtual machines for upnp server or various services.

IMO Not so easy to explain as ethernet is galvanic isolated and have a data layer added for data integrity. More probable the audio eth chain is not bugged with eth brodcasts and similar packets. And I don't presume that the switch is located anywhere near the audio rack. If I understood correctly TB-eth controller is used for the connection to the rest of his network.

Hi Superdad, did you try to make a separate 2 port vlan in the switch for the computer and the storage? Later Edit: and could you please copy paste an ifconfig from console?