RPi dacs do not use transports that I am aware of. They use files. All test music is in wav file format.
Ian, Yes, hardware SRC immediately before the dac can effectively reduce jitter. To fully take advantage of it, it is necessary to access dac chip control registers to reduce DPLL filter bandwidth as much as possible.
The more expensive AK4137 board has a digital serial audio receiver chip on it to convert SPDIF, TOSLIK, and AES to PCM, I2S or DSD. The low cost board does not.
The more expensive AK4137 board has a digital serial audio receiver chip on it to convert SPDIF, TOSLIK, and AES to PCM, I2S or DSD. The low cost board does not.
Cheers for the clarification. The editing of registers sounds a complex task, but I'm guessing for you to have gone to such lengths, it must be a worthwhile modification. Are there other benefits to this upsampling prior to input to the DAC or is it just for jitter reduction?
Is this the cheap board you refer to?
AK4137 I2S/DSD Sample Rate Conversion Board Support DOP input digital lcd | eBay
Is this the cheap board you refer to?
AK4137 I2S/DSD Sample Rate Conversion Board Support DOP input digital lcd | eBay
sorry, I was wondering why you are not using the same transport when comparing?
Indeed it does, my friend who runs a recording studio is always complaining he can hear jitter in playback of digital recordings.
Was my answer okay a few posts back? Not sure if you mean to ask again or if I am missing something?
may be I am misusing the term "transport", but here I mean the way how the digital stream is fed into the dac. for Katana board it would be RPi-I2S-Katana. my question was about the combination RPi-I2S-modES9038q2m board?
I'm not sure if this is what you mean, but for Katana wav files were put on a USB memory stick and it was plugged into RPi. The Volumio app reads the file and sends it to the Katana linux driver.
For my modded dac, a wav file on a computer hard drive it is sent by USB to an XMOS or Amanero board, then by I2S to an AK4137 board, then via DSD to the dac.
For DAC-3, a wav file on the hard disk is sent to a sound card with a SPDIF output, which then goes to the SPDIF input on the DAC-3.
IMHO, these things probably don't matter, but if they do make any difference they probably reasonably represent how the devices would most likely be used by someone who purchased them. Maybe the DAC-3 would be connected to the computer by USB instead of using SPDIF, but no practical difference I am aware of there.
thanks Mark! however for me (and perhaps for someone else) the relevant question would be should I go with Katana as RPi dac, or "simple" moded Chinese ES9038q2m board is good enough.
connecting RPi I2S to your board would address this question, I guess.
connecting RPi I2S to your board would address this question, I guess.
Not that simple. 'My dac' is everything I have described. It is a modded Chinese dac, it uses AK4137 to convert the input to DSD, and it uses an Arduino to program the dac registers over I2C bus.
If you wanted to plug it into an RPi, you would have to plug the input of the AK4137 to the output of the RPi I2S/PCM bus, and perhaps convert the RPi SPI control bus to I2C, unless it already has I2C, or maybe just use an Arduino to set the Q2M dac registers.
There is a little more I didn't say, but since you are asking questions like this maybe I should say a little more now. It may be as cdsgames said that katana sounds better with other than the apodizing filter for some listening conditions. My modded dac also sounds best with a filter that sounds a little less than the closest fit to DAC-3. If we use Katana slow minimum phase filter and AK4137 short-sharp filter then I like my dac better that way. On the other hand if both Q2M dac filters are chosen to sound as close to DAC-3 as possible then maybe Katana sounds slightly better, but not so much difference as with the first set of filter choices I just described.
Anyway, to me the big questions for someone might be if an RPi dac will meet their needs. If you want to use the dac when you watch a movie on DVD, or in other situations, maybe RPi is not ideal for that. Don't know how it might work with wireless streaming of some kind.
The other big question for someone would be if they really want to do all the work I did including the digital stuff and including using an AK4137 board. If the answer is yes to all that then which will give the best sound quality may come down to which filters you end up preferring.
On the other hand, for $250, an RPi, and some power supplies, a whole lot of work could be skipped, if you view it as work that is. If it seems like a fun project then by all means I would vote for having fun.
If you don't like either of the above options, a used DAC-1 might be a good choice. They still sound very good, maybe roughly around where these two Q2M dacs are at, although I haven't heard one for awhile so my memory is not exact on that. But, there is a reason old DAC-1's tend to maintain pretty good resale value.
Last edited:
Mark4, ezitis was totaly right what he meant.
We want to reveal trully performance of Katana and other 2 dac by using as close same source as possible.
So my idea will be...
use Rpi ...i2s...Katana
use Rpi ...i2s...es9038q2m chineese...if not than rpi...usb...xmos (amanero sound different) ..reclocker..dac
use rpi...usb out...usb xmos in Dac3
If you have in that way sorry to interrupt, because i don"t know your setup in details.
It should be always same source with same software with same settings. I have 3 sd cards with moode 3.8, 4,1 and 4.1 audio optimized from one diyaudio member that tested and optimized it. All play different with same settings.
If i only change buffer size in moode 4.1 from 4096 to x2 or x4 i get different sound if i don mentioned kernels , sox, volume off, hdmi off...and all other settings for optimal sound. We are even not at power supply...i tried at leat 10 different supplies...combinations... rpi can sound harsh and flat but also relaxed analogue full and big 3d stage. That can only be done with unsoldering switching reg from rpi and use low noise linear regulators...all connected to accu or supercaps like Ian try to implement in designes, not using wifi dongle, use metal enclosure for Pi, put far away all wifi routers if rpi is not in enclosure....
Biggest problem with "Boards" is that you can not try it before you buy it or send it back if you are not satisfied, so therefore we try to search for as much possible infos on all sides to deceide easier if product is good or not good for custromers.
What about same price range comparisson Katana vs soekris dac? That will be also one fine comparisson.
After all for less than 300€ for Katana it is ok dac and after all it is not big risk if you are not satisfied with the sound. People loose 10000€ on hi end gear after few years.
Sorry for little messing the topic.
We want to reveal trully performance of Katana and other 2 dac by using as close same source as possible.
So my idea will be...
use Rpi ...i2s...Katana
use Rpi ...i2s...es9038q2m chineese...if not than rpi...usb...xmos (amanero sound different) ..reclocker..dac
use rpi...usb out...usb xmos in Dac3
If you have in that way sorry to interrupt, because i don"t know your setup in details.
It should be always same source with same software with same settings. I have 3 sd cards with moode 3.8, 4,1 and 4.1 audio optimized from one diyaudio member that tested and optimized it. All play different with same settings.
If i only change buffer size in moode 4.1 from 4096 to x2 or x4 i get different sound if i don mentioned kernels , sox, volume off, hdmi off...and all other settings for optimal sound. We are even not at power supply...i tried at leat 10 different supplies...combinations... rpi can sound harsh and flat but also relaxed analogue full and big 3d stage. That can only be done with unsoldering switching reg from rpi and use low noise linear regulators...all connected to accu or supercaps like Ian try to implement in designes, not using wifi dongle, use metal enclosure for Pi, put far away all wifi routers if rpi is not in enclosure....
Biggest problem with "Boards" is that you can not try it before you buy it or send it back if you are not satisfied, so therefore we try to search for as much possible infos on all sides to deceide easier if product is good or not good for custromers.
What about same price range comparisson Katana vs soekris dac? That will be also one fine comparisson.
After all for less than 300€ for Katana it is ok dac and after all it is not big risk if you are not satisfied with the sound. People loose 10000€ on hi end gear after few years.
Sorry for little messing the topic.
Last edited:
Haha! Don't think it will be practical for me to go all that far. Think I will say a RPi dac is a black box that has a USB memory stick connector on the RPi for input and two line level RCA jacks coming out, and I don't know how to break down or take apart whatever is in that black box. I will just have to test or evaluate the sound of the box a whole.
Similarly for my dac, it has a USB cable sticking out for input I can plug into a computer and two line level RCA jacks coming out. Whatever is inside that black box is also a mystery. I will also have to judge how I like its sound as I find it. Viewed this way, maybe the two dacs are not completely comparable. That's okay, I can still describe what I hear coming out and say what I like and don't like about them.
Someone else may have to decide if they have the interest, and time and energy to do something else with them some other time.
Similarly for my dac, it has a USB cable sticking out for input I can plug into a computer and two line level RCA jacks coming out. Whatever is inside that black box is also a mystery. I will also have to judge how I like its sound as I find it. Viewed this way, maybe the two dacs are not completely comparable. That's okay, I can still describe what I hear coming out and say what I like and don't like about them.
Someone else may have to decide if they have the interest, and time and energy to do something else with them some other time.
I agree that SW of Pi has an impact on SQ. Also USB input / output is not said to be the best choice regarding SQ from Raspi due to internal bus capacity (never tried). Probably with a picoreplayer fed by LAN the SQ would be better. Anyway this setup from Mark could be the one many useres would take the Katana for.
Mark, what sample rate do you use for feeding the Raspi / Katana?
I have the same es9038q2m board as Mark, and did similar HW mods on it. I did not modify the digital part by SRC Input or I2C hack. Nevertheless I can hear differences feeding the DAC from my G-Sonos (a modified Sonos device with extremely low jitter on SPDIF output) via SPDIF and feeding the DAC from Raspi -> HifiberryDAC+pro -> I2S -> es9038q2m. The G-sonos sounds slightly better - maybe due to a cleaner PSU and lower Jitter.
Maybe the new Isolator board from Allo would decrease the neg. impact of bad Raspi onboard PSU as androa76 describes on Katana SQ ..
Mark, what sample rate do you use for feeding the Raspi / Katana?
I have the same es9038q2m board as Mark, and did similar HW mods on it. I did not modify the digital part by SRC Input or I2C hack. Nevertheless I can hear differences feeding the DAC from my G-Sonos (a modified Sonos device with extremely low jitter on SPDIF output) via SPDIF and feeding the DAC from Raspi -> HifiberryDAC+pro -> I2S -> es9038q2m. The G-sonos sounds slightly better - maybe due to a cleaner PSU and lower Jitter.
Maybe the new Isolator board from Allo would decrease the neg. impact of bad Raspi onboard PSU as androa76 describes on Katana SQ ..
I am using the native sample rate of the source files, mostly CD rips from especially well-produced CDs, so mostly 16/44. Since that is how most music is recorded and mostly what people have to listen to, it is important to see how well it gets reproduced, IMHO. It turns out that CD audio can be surprising detailed, musical, and easy to listen to, but it is hard to make it sound that way. Few dacs do it very well so in a way it can be kind of a hard test.
Thank you, sir. Likewise, I am sure! If I may say so, I think you started out with a particular concept and purpose in mind, something that was not the case for me. I just kept trying to use good, common sense electronics practices, kept trying things to see what could help improve the sound a little more each time, and ended up where I am now. I think I said a long time ago somewhere in this thread that we didn't even know what system architecture would turn out to be best. Apparently, we still don't know what might be best if we could just get past what appears like it might be filter limitations. Obviously Katana with it's master mode I2S and RPi architecture is pretty different compared to a slave mode upsampling to DSD architecture. Interesting they turned out so close, or at least they seem close given we can never seem to find that elusive perfect resampling filter sound.
eziitis, No puzzle there. I feed the AK4137 16/44 input files. It upsamples and converts to DSD. I have it upsample to remove jitter since my dac runs in slave mode I need to give attention to jitter suppression. The DSD mode is so I can avoid the Q2M interpolation filters being combined with AK4137 filters. Too many uncoordinated filters can add more distortion. I can't get rid of the AK4137 filters, but by using DSD mode I can avoid some of the unneeded Q2M filters. The DSD still goes through the final stage of Q2M IIR filters though, which is appropriate. Also, with DSD mode, there are more notches I can turn down DPLL bandwidth in Q2M which also helps with jitter reduction and sound quality improvement. The point is, my dac is a black box that you don't have to ever know uses DSD. All you have to know is it is using a 16/44 input file and the sound quality coming out is good. I just let you in on the secret of what is inside the black box is all, but you can pretend like you don't know, since you don't really need to know to enjoy your CD.
Once you go past a certain point of inexpensive, just go with something proven. When you start adding up all the little experimental pieces and then the additions to the spare parts parts box, a good commercial offering is the way to go. Now if you are intrigued and don't mind the anguish then by all means experiment for fun. I was really hoping the katana would have done a lot better but it looks like, only larger companies can push the envelope now.
My experiment was the value of VAT tax for a DAC3 which is about $300 USD. I started out in December with a $40 board and worked my way up....to buying oscillators worth nearly that much.
Anything more than approx $300 is not worth it for me. As it is, even if my modded 9028pro chinese board is not the best, it will do darn well for watching videos in stereo 2.1 thru a HTPC which will always be my setup. I was hoping that the Katana would be very close to the DAC3 and thus could sit on top of the preamp in my listening setup and serve all my files from my network without needing more equipment. So it looks like I will still have to serve files via a PC then maybe a DAC3 or 4 in the future. Time to save the pennies again. It's been a rewarding learning experience though.
Last edited: