Just tried DSD/DoP on the Mamboberry LS.
Ok.
I just own a single DSD album from OPUS3 in DSD128. That's not much of a DSD collection. And lets not talk about the type of music on that disk.
DSD128 requires a samplerate of 352.8 for DoP.
I thought I might give it a try. 384 works quite well now.
I installed the DSD plugin on LMS 7.9.0 and restarted squeezelite adding "-D" to rest of the bunch.
That was it!??!? Yep, that's all it was.
As expected, playback runs at samplerate 352.8:
Code:cat /proc/asound/card0/pcm0p/sub0/hw_params access: MMAP_INTERLEAVED format: S24_LE subformat: STD channels: 2 rate: 352800 (352800/1) period_size: 14112 buffer_size: 56448
Great. Now I got DoP on my HAT.
Enjoy.
I think I am missing something here.
As far as I know the 9023 does not support DoP functionality. How were you able to use DoP with this combo?
Edit:
In fact, most of the well known I2S dacs of the market for the RPI should not be able to use the DoP feature that Clivem enabled.
Last edited:
1. Mamboberry lists DxD 352.8 as feature
2. Triode built a DSDDOP plugin for LMS, which I use.
3. I can play my DSD files through LMS - 7.9 . There is sound.
4. Alsa shows 352.8 during playback.
Unless my DSD data get somehow converted to PCM, I assumed DOP to be active.
Let me know if I miss something here.
LMS/squeezelite SW control.
The HifiBerry DAC+ Pro with the PCM51xx offers on-dac volume control.
squeezelite can be configured to use Alsamixer controls for controlling it.
The driver has to make these controls available of course.
Such a feature is not offered by the Sabre DAC.
However. That PCM51xx volume control is just another SW control.
Which to me even sounded worse then the LMS SW control.
In the end I made sure that the on-dac control runs at 100% and left it alone.
I concluded that bypassing the on-DAC DSP functions ( filters/VC/etc) is usually
the better option.
Last edited:
I just turned on debugging and retested everything.
As a matter of fact - as it looks to me know - you seem to be right.
Real DOP streaming (according to the squeezelite debug log) ends up as noise.
I just realized that the DSDplayer plugin on the LMS can either do the
PCM or the DOP conversion.
This would mean DSD gets converted to native PCM in the MamboBerry case.
I hope this is now the correct conclusion.
Tuxx, Thx a lot for letting me/us know.
(Interesting that it took that long, that somebody stepped over this.)
Sorry -- for the confusion folks.
I just turned on debugging and retested everything.
As a matter of fact - as it looks to me know - you seem to be right.
Real DOP streaming (according to the squeezelite debug log) ends up as noise.
I just realized that the DSDplayer plugin on the LMS can either do the
PCM or the DOP conversion.
This would mean DSD gets converted to native PCM in the MamboBerry case.
I hope this is now the correct conclusion.
Tuxx, Thx a lot for letting me/us know.
(Interesting that it took that long, that somebody stepped over this.)
Sorry -- for the confusion folks.
Yes this is correct. AFAIK, there is no HAT DAC yet developed that supports DoP protocol.
-Tim
Not quite so.... I am aware of 2, neither of which have been publicly released. One never will be.
+1 !... Clive, you may be surprised at the support and standing you have here, especially from many such as myself who struggle to follow the esoterica of audio kernels but benefit immensely from their development...your efforts are enormously appreciated..!
Noli nothis permittere te terere !
I sure as hell hope you meant retired for the *morning*
There's more than just allo going on here!
I just turned on debugging and retested everything.
As a matter of fact - as it looks to me know - you seem to be right.
Real DOP streaming (according to the squeezelite debug log) ends up as noise.
I just realized that the DSDplayer plugin on the LMS can either do the
PCM or the DOP conversion.
This would mean DSD gets converted to native PCM in the MamboBerry case.
I hope this is now the correct conclusion.
Tuxx, Thx a lot for letting me/us know.
(Interesting that it took that long, that somebody stepped over this.)
Sorry -- for the confusion folks.
Thanks for your confirmation Soundcheck!
Started modifying my Mamboberry this weekend...
Why? Because I can!
Seriously, I'd previously written that the unit worked well and sounded good in my setup... not reference quality, but very engaging.
Then with the help of Soundcheck, I got my extreme-modified HiFiBerry DAC+ Pro to finally work and sound good... and it was (and still is) very, very good, starting to push into the bottom tier of my references (and I am hopeful putting it on a Kali in slave mode will elevate it even more!). AND then I put my Mamboberry on a Kali... and the SQ was improved in all parameters. But now comparing the MB/Kali to the mod'd HFBD+P (and other sources in that system including vinyl), the Mamboberry still didn't get into the top tier. Close, but it was held back by a tilt towards the high frequencies (some say this is the signature of an ESS DAC) AND it was missing the level of instrument texture and inner detail that I get with my references, along with a level of composure during high-volume and/or complex passages.
I have a TON of respect for the Mamboberry. I think it is the best bang for buck value in the R-Pi DAC HATs that I've tried. The general implementation of the unit is good AND the power supplies are good to great on both the original MB and the LS version. My main caveat on the unit are the output jacks, which seem to be the lowest quality of all I've tried... but these can be fixed.
So all this makes it a very worthwhile unit to look at for additional performance.
For this set of mods of my Mamboberry, I draw my inspiration from two places:
1. A DAC card that was (and may still be) available from DIYAudio poster EUVL. This card was based on the ES9022, the predecessor to the MB & MBLS ES9023. The two DAC chips are VERY similar and many of the things EUVL did to get the best performance from the ES9022 will translate well to the ES99023. You can see some of the details and development of that card in this thread:
http://www.diyaudio.com/forums/digital-line-level/151846-anybody-using-new-ess-vout-dac-es9022.html
Although if you search for 'EUVL' and 'ES9022', you will see many other mentions (a lot by me!).
2. The Mamboberry LS. I am assuming (and Vasilis can confirm, I hope) that some of the design changes between the two were made to improve the performance of the card... at least in what I can see, that is the conclusion I'd draw.
I started with the input signal path, basically the I2S from the R-Pi AND the clock signal. These lines typically have resistors in-line to help manage the line impedances and matching, shape the waveform, and improve the signal integrity.
In the original Mamboberry, all of these resistors are 100R EXCEPT for the LRClock line at 10R. The Mamboberry LS uses 51R resistors in all lines. AND the latest version of EUVLs ES9022 cards uses 330R resistors in the signal lines and 0R in the clock line.
My first thing to try was changing these resistors out. In this case, I decided to start with the LS setup, replacing all 4 resistors with 51Rs.
Interesting change... but before I say anything about how it sounded afterwards, I'd like to ask Vasilis what was his purpose behind changing these resistors AND what differences SHOULD I hear?
Later I intend to try both a 0R on the clock line and the 330R from the EUVL ES9022 DAC card. But for now, these will stay in... and I'll say more hopefully after we hear from Vasilis on this.
Thanks for reading!
Greg in Mississippi
Why? Because I can!
Seriously, I'd previously written that the unit worked well and sounded good in my setup... not reference quality, but very engaging.
Then with the help of Soundcheck, I got my extreme-modified HiFiBerry DAC+ Pro to finally work and sound good... and it was (and still is) very, very good, starting to push into the bottom tier of my references (and I am hopeful putting it on a Kali in slave mode will elevate it even more!). AND then I put my Mamboberry on a Kali... and the SQ was improved in all parameters. But now comparing the MB/Kali to the mod'd HFBD+P (and other sources in that system including vinyl), the Mamboberry still didn't get into the top tier. Close, but it was held back by a tilt towards the high frequencies (some say this is the signature of an ESS DAC) AND it was missing the level of instrument texture and inner detail that I get with my references, along with a level of composure during high-volume and/or complex passages.
I have a TON of respect for the Mamboberry. I think it is the best bang for buck value in the R-Pi DAC HATs that I've tried. The general implementation of the unit is good AND the power supplies are good to great on both the original MB and the LS version. My main caveat on the unit are the output jacks, which seem to be the lowest quality of all I've tried... but these can be fixed.
So all this makes it a very worthwhile unit to look at for additional performance.
For this set of mods of my Mamboberry, I draw my inspiration from two places:
1. A DAC card that was (and may still be) available from DIYAudio poster EUVL. This card was based on the ES9022, the predecessor to the MB & MBLS ES9023. The two DAC chips are VERY similar and many of the things EUVL did to get the best performance from the ES9022 will translate well to the ES99023. You can see some of the details and development of that card in this thread:
http://www.diyaudio.com/forums/digital-line-level/151846-anybody-using-new-ess-vout-dac-es9022.html
Although if you search for 'EUVL' and 'ES9022', you will see many other mentions (a lot by me!).
2. The Mamboberry LS. I am assuming (and Vasilis can confirm, I hope) that some of the design changes between the two were made to improve the performance of the card... at least in what I can see, that is the conclusion I'd draw.
I started with the input signal path, basically the I2S from the R-Pi AND the clock signal. These lines typically have resistors in-line to help manage the line impedances and matching, shape the waveform, and improve the signal integrity.
In the original Mamboberry, all of these resistors are 100R EXCEPT for the LRClock line at 10R. The Mamboberry LS uses 51R resistors in all lines. AND the latest version of EUVLs ES9022 cards uses 330R resistors in the signal lines and 0R in the clock line.
My first thing to try was changing these resistors out. In this case, I decided to start with the LS setup, replacing all 4 resistors with 51Rs.
Interesting change... but before I say anything about how it sounded afterwards, I'd like to ask Vasilis what was his purpose behind changing these resistors AND what differences SHOULD I hear?
Later I intend to try both a 0R on the clock line and the 330R from the EUVL ES9022 DAC card. But for now, these will stay in... and I'll say more hopefully after we hear from Vasilis on this.
Thanks for reading!
Greg in Mississippi
The choice of i2s terminal resistor its a " aftermath choice " *
first off all the basic criterion in a digital design its the right measure , in the case of Mamboberry we see ( and general in sabre DAC,s) as lower is the terminal resistors so better the final sound result ,,
BUT ! if you go too much lover there is the danger of ringing , distortion e.t.c. of signal
* the resistors choices happened after the PCB design .. in theory i try to make the I2S lines proof ,, and practical we made ,, i put 50 ohm in pcb and workink & measure very well !
also some improve of jitter ..
some ideas about the PCB routing its almost in the same theory as my big son..
Ideon Audio Ayazi | Review Technical Details - AVMENTOR.net
about the sound off sabre and texas instruments ,, its a BIIIIG discussion !
personal the only good think of texas its the electrical guitar .. emphasis and bright ,, and thats all .. ( but i repeat ,, FOR ME ) the lower its like rubber ,, the middle like old paper midrange and the upper full of artefacts , and a general a boring banal sound with out soul ,,
in the other side have guys , and get sick with the detail and tactical like army right sound of sabre ..
opinions and taste ...
all its good ... if you like ...
but in case of RPI HAT BOARD , the cost its ridiculous .. take all and keep the right , or modificative the right !
( disclaimer ! im not the seller of mamboberry , i have design that .. as i have design and medical thinks electronics .. but the sound its my first love )
first off all the basic criterion in a digital design its the right measure , in the case of Mamboberry we see ( and general in sabre DAC,s) as lower is the terminal resistors so better the final sound result ,,
BUT ! if you go too much lover there is the danger of ringing , distortion e.t.c. of signal
* the resistors choices happened after the PCB design .. in theory i try to make the I2S lines proof ,, and practical we made ,, i put 50 ohm in pcb and workink & measure very well !
also some improve of jitter ..
some ideas about the PCB routing its almost in the same theory as my big son..
Ideon Audio Ayazi | Review Technical Details - AVMENTOR.net
about the sound off sabre and texas instruments ,, its a BIIIIG discussion !
personal the only good think of texas its the electrical guitar .. emphasis and bright ,, and thats all .. ( but i repeat ,, FOR ME ) the lower its like rubber ,, the middle like old paper midrange and the upper full of artefacts , and a general a boring banal sound with out soul ,,
in the other side have guys , and get sick with the detail and tactical like army right sound of sabre ..
opinions and taste ...
all its good ... if you like ...
but in case of RPI HAT BOARD , the cost its ridiculous .. take all and keep the right , or modificative the right !
( disclaimer ! im not the seller of mamboberry , i have design that .. as i have design and medical thinks electronics .. but the sound its my first love )
Last edited:
Changing impedance matching resistors in data lines without measuring their integrity.. okay...
With the result of better sound? Hmm..
From my experience, 22R worked best on my layout in conjunction of having all lines at the same length. Difference in sound? Probably not, but signals looking excellent.
With the result of better sound? Hmm..
From my experience, 22R worked best on my layout in conjunction of having all lines at the same length. Difference in sound? Probably not, but signals looking excellent.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.