Yeah true. The idea is to get the bus to operate without bottlenecks at 2.0 speed.
Sent from my ASUS_Z00AD using Tapatalk
OK
I give up
Ok, let me put my understanding here
- A DAC will not start cooking up information if none is being sent to it. It will simply have no output (this is what used to happen in ye olde times when portable cd players "skipped"
- To cure skipping, a buffer was incorporated using RAM so that a certain amount of data could be kept in reserve when there were issues with reading, which would later be topped up once the head was stable enough to read
- A Rpi playing audio from a usb drive and playing it over a piano dac uses the USB interface, and the DAC uses and I2S interface
- I2S is a synchronous interface, which means that data transmission happens at fixed intervals. if it misses any data, your song will 'skip'
- To prevent this, the OS running on the rpi will maintain a buffer pool of audio data, to ensure that it always has something to send to the DAC.
- This is a software buffer, backed up by your computer architecture memory (L1 Cache, L2 Cache, Primary memory.
- This is still not foolproof as sometimes the CPU gets tied up with some other priority task, or there data may not be available in cache/ram, which may cause the audio to skip
- on the input side, the audio is being read from a USB. pure, uncompressed LPCM audio - 48khz, stereo needs a bitrate of 1.5 Mbps, which is well under even USB 1.1 bandwidth which is 12 Mbps, and insignificant compared to 2.0 which is 480Mbps
- for any USB drive, the bottleneck is usually the Flash memory itself, which maxes out around 60 MBPS usually, and even the fastest 3.0 drives max out at 150 or so. The bus is not really what is making it faster. these are just high performance drives which happen to have usb 3.0.
- So in my opinion, given the actual bandwidth requirements for audio, and what current devices provide, any difference, is purely academic
- Even if you bring up the latency issue, the existence of a buffer makes any latency problems insignificant.
- Finally, mp3 audio does not exist as samples in time like LPCM. it exists in the frequency domain (imagine a shot of a graphical equalizer, except that it has 576 bands, and the encoder decides which are the frequencies), and then they are put into frames.
- Typically when playing an mp3, the media player will simply read the whole thing into its cache, and start decoding and playing it
What he means to say is the Pi is USB 2.0 and while the USB dongle may be rated 2.0, it might actually be 1.0 or deliver less speeds than the Pi's 2.0 specs - in which case it will become the bottleneck. So in that context it's better to use a USB 3.0 pen drive.
@SanjeevM - Why even use a USB pendrive? Unless it's for testing purposes. I'd venture to say most collections would be larger than the capacity of a pen drive. I'm myself using an externally powered HDD.
Please correct me where I am wrong, but when you say 480Mbps, dont you mean 'bits per second', which would translate to 60M bytes per second. I have yet to see a USB 2.0 which actually does 60MBps on file transfer !! (my portable HDD just did 33 MBps xfer to PC).
Will the 2.0 pendrive work if one was playing WAV files (40 MB for a 5 min song) instead of MP3, or worse, a DSF file whose size is 270 MB for 6.3 minutes (Hotel California).
How much of free RAM does a RPi (or the HAT DAC) actually have to use as a buffer is also a question.
I really don't know from where have you got all your misconceptions
I have used RPI to play wav files from a usb 2 pen drive outputting to a dac via usb , without any issues ever
The read and write speed of even usb 1 is enough to play a wav file to play without any hiccups
So a slower pen drive is the least of your worry
Think about it, even a 100 megabit ethernet is enough to stream 4k video
Ok, let me put my understanding here
- on the input side, the audio is being read from a USB. pure, uncompressed LPCM audio - 48khz, stereo needs a bitrate of 1.5 Mbps, which is well under even USB 1.1 bandwidth which is 12 Mbps, and insignificant compared to 2.0 which is 480Mbps
- for any USB drive, the bottleneck is usually the Flash memory itself, which maxes out around 60 MBPS usually, and even the fastest 3.0 drives max out at 150 or so. The bus is not really what is making it faster. these are just high performance drives which happen to have usb 3.0.
the second point should read as Mbps - sorry about thatPlease correct me where I am wrong, but when you say 480Mbps, dont you mean 'bits per second', which would translate to 60M bytes per second. I have yet to see a USB 2.0 which actually does 60MBps on file transfer !! (my portable HDD just did 33 MBps xfer to PC).
theoretically, even 1.1 should work fine. I know of plenty of car head units which play Wav over a USB driveWill the 2.0 pendrive work if one was playing WAV files (40 MB for a 5 min song) instead of MP3, or worse, a DSF file whose size is 270 MB for 6.3 minutes (Hotel California).
Ok, let me put my understanding here
- A DAC will not start cooking up information if none is being sent to it. It will simply have no output (this is what used to happen in ye olde times when portable cd players "skipped"
- To cure skipping, a buffer was incorporated using RAM so that a certain amount of data could be kept in reserve when there were issues with reading, which would later be topped up once the head was stable enough to read
- A Rpi playing audio from a usb drive and playing it over a piano dac uses the USB interface, and the DAC uses and I2S interface
- I2S is a synchronous interface, which means that data transmission happens at fixed intervals. if it misses any data, your song will 'skip'
- To prevent this, the OS running on the rpi will maintain a buffer pool of audio data, to ensure that it always has something to send to the DAC.
- This is a software buffer, backed up by your computer architecture memory (L1 Cache, L2 Cache, Primary memory.
- This is still not foolproof as sometimes the CPU gets tied up with some other priority task, or there data may not be available in cache/ram, which may cause the audio to skip
- on the input side, the audio is being read from a USB. pure, uncompressed LPCM audio - 48khz, stereo needs a bitrate of 1.5 Mbps, which is well under even USB 1.1 bandwidth which is 12 Mbps, and insignificant compared to 2.0 which is 480Mbps
- for any USB drive, the bottleneck is usually the Flash memory itself, which maxes out around 60 MBPS usually, and even the fastest 3.0 drives max out at 150 or so. The bus is not really what is making it faster. these are just high performance drives which happen to have usb 3.0.
- So in my opinion, given the actual bandwidth requirements for audio, and what current devices provide, any difference, is purely academic
- Even if you bring up the latency issue, the existence of a buffer makes any latency problems insignificant.
- Finally, mp3 audio does not exist as samples in time like LPCM. it exists in the frequency domain (imagine a shot of a graphical equalizer, except that it has 576 bands, and the encoder decides which are the frequencies), and then they are put into frames.
- Typically when playing an mp3, the media player will simply read the whole thing into its cache, and start decoding and playing it
I really don't know from where have you got all your misconceptions
I have used RPI to play wav files from a usb 2 pen drive outputting to a dac via usb , without any issues ever
The read and write speed of even usb 1 is enough to play a wav file to play without any hiccups
So a slower pen drive is the least of your worry
Think about it, even a 100 megabit ethernet is enough to stream 4k video
I agree and it's the truth.
But then why do folks have an issue with the shared bus of the Pi, especially self professed audiophiles? Shouldn't the same logic kick in? The shared bus really is a non-issue and I've seen the Pi used as a source even in $10,000 and more setups (check Reddit for proof and pics), but the very same Pi also gets a bad rep on cred-less sites like CA and other so called audiophile review sites that advertise cables as audiophile magic fairy dust.
Let's keep the shortcomings of raspberry pi for audio for another discussion
ok my latest test :
Configuration 1
Equipment 1 : Raspberry Pi 3, Allo Kali, Allo Piano 2.1 Power supply Linear PSU and powered directly to the Kali card only
Software 1 : Volumio : 2.272 released on 9th Sept 2017
Configuration : Output Device 12s is set as Allo Piano 2.1 and Mode is Dual Mono
Configuration 2
Equipment 2 : Allo Sparky, Power supply Linear PSU connected to a UPS Hat and 2.7v battery that then powers the Sparky. SMSL M8a powered by Linear PSU and Ifi Micro USB as input to the dac .
Software 1 : Volumio : 2.255 released on 9th August 2017
Configuration : Sparky USB is used as Audio out and that feeds to the Ify Usb Micro and that in turn is used for input to the S.M.S.L M8a
Common configuration :
Cables IC : DAC ARX01A 2RCA TO 2RCA AUDIO INTERCONNECT CABLE
Amplifier : Marantz PM 6006 Direct Mode
Speakers : Paradigm FS Monitor 7
Played files 2 types :
1. Flac 48KHz 16bit
2. DSD 5.64MHz 1 bit
My inferences :
I heard both systems back to back as well as switch between inputs, and quite frankly there was very little that my 40years old ear could differentiate, both of them sounded open and there was only a subtle hint of the SMSL sounding a tad better to my ears
so was wondering can a RPI, kali reclocker and piano dac sound so close to a dedicated dac
Btw i used the Sparky for the USB output as there apparently less noise on the sparky usb port compared to a raspberry pi. I think i may need to try the SMSL m8a with a windows PC as there are drivers for it maybe that will give a much better difference to the sound.
( i also feel the amp and speakers maybe not the best to give reference quality feedback and maybe the bottle neck here)
Hello Kane,
Could I please request one more testing.
Sparky-Kali-Piano 2.1-Volt(optional)
Vs.
Raspberry Pi-Kali_Piano 2.1-Volt(optional)
I would like to know if the Pi produces better sound with the combo or Sparky.
Also if you have Moode, would it be possible to let me know if it works with Sparky.
Would really appreciate any feedback. I am eagerly waiting for my Vana to be delivered by the 25th.
Hello Debzcool,
I had tried both in earlier iterations of volumio
Pitting both the sparky against the Rpi
i would say the sparky did sound a tad better, probably because all the components were of the same manufacturer. perhaps the only let down is that Volumio etc they push updates out more often for RPI than the sparky as the user base is more. so to use sparky as the base system would mean that not all new releases would support it. If you know how to compile etc then maybe it would make sense i guess.
I have not tried moode lately though i did purchase and try out version 3.6 i guess. let me download that and see.
One advantage the sparky has is that you can use an eMMC card and that is way faster than the micro sd card.
I have one system with the rpi 3 with boss dac and volt and was comparing that to the RPI-kali-piano2.1 which is connected to a marantz pm 6006. the marantz combo sounds fab, also its not a fair comparison as the volt are connected to a pioneer AJ Book shelves, so the speakers are different.
All i can say is that you will not be disappointed with either.
But beware as once this new clearer open sound get to you your thirst for more purity will make you want to spend more to squeeze every ounce of bliss that you can extract -Speaker cables, interconnects, Linear psu, etc
if you are looking only at enjoying your music for a fraction of a cost this does bring you close to a decent system.
I will download moode over the weekend and let you know how it goes .