Hi,
i try to reuse my old Sony CDP 710 that i liked very much as a network player. The cd mechanism is not useable anymore, so i came up with the idea to use it as a network based player with the help of a raspberry pi. The i2s lines of the Pi are reclocked, so jitter is a bit lower than on pure pi.
Now my question is :
Can i just connect the I2S lines form my pi directly to the TDA1541 ? Pin 1,2 and 3 (LE/WS, BCK, DATA) should do the trick if i understand it correctly.
Or is it more complex than this ?
If i connect the pi directly to the TDA1541 i would go NOS. Do i need to alter something else, besides the 3 input pins ?
I would be very happy for any help, because music over all the PCM51X2 DACs is not really fun compared to my old Sony CD player.
Second thing is : The TDA datasheet shows, that it can go way higher that 44.1khz. If i upsample in the pi via software, would this be some kind of OS emulation ? Dont know if i understand it right, but if i upsample in the pi via software to 176,4khz for example this would be 4xOS or not ?
i try to reuse my old Sony CDP 710 that i liked very much as a network player. The cd mechanism is not useable anymore, so i came up with the idea to use it as a network based player with the help of a raspberry pi. The i2s lines of the Pi are reclocked, so jitter is a bit lower than on pure pi.
Now my question is :
Can i just connect the I2S lines form my pi directly to the TDA1541 ? Pin 1,2 and 3 (LE/WS, BCK, DATA) should do the trick if i understand it correctly.
Or is it more complex than this ?
If i connect the pi directly to the TDA1541 i would go NOS. Do i need to alter something else, besides the 3 input pins ?
I would be very happy for any help, because music over all the PCM51X2 DACs is not really fun compared to my old Sony CD player.
Second thing is : The TDA datasheet shows, that it can go way higher that 44.1khz. If i upsample in the pi via software, would this be some kind of OS emulation ? Dont know if i understand it right, but if i upsample in the pi via software to 176,4khz for example this would be 4xOS or not ?
Last edited:
But the "desperately low quality" will be far more the adequate for the super huge vast majority.
Last edited:
Feeding such a no-frills NOS DAC with poor I2S signals taken straight from RPI is not a good idea.
From your statements I read you're not interested in pc-speaker quality sound.
Inmate DimDim did some measurements.
All of us who have worked with these 1543/1541 DACs know how sensitive these devices are.
And it's not just jitter. It's also noise. It's power. ....
The currently IMO best option to attach such a Dac to a PI is iancanada's FifoPi. (Allo is not offering such a device) The FifoPi is an isolator and reclocker and the clean side can be powered separately. You'll get the major
issues under control with this device.
FifoPi also provides high quality UFL sockets for I2S interfacing to the dac.
Using some cheap jumpercables from GPIO is not the best of all options
to run I2S.
*****
Resampling.
Yes that's an option. It won't get you around the poor RPI I2S though. It rather makes things worse.
Let's assume you'd go for the FifoPi.
The upsampling will have a positive effect on a NOS dac. But in this case it's actually no longer a NOS dac you'll talk about. Every DSP introduces it's own losses.
Not any resampler is lossless. You need to find the best resampler and use that one at highest quality settings.
Squeezelite and LMS would both allow to do so. These would even support synchronous (x44,1 + x48) resampling.
However.
Then you'll run into the 16bit issue. If you do resampling you need some headroom. You'll loose on the level to avoid clipping. And then there's
digital volume control. On a 16 bit DAC running digital volume control and
resampling simply doesn't make any sense. If you have an analog volume control it'll do.
Good luck with your choices. Don't waste to much time and resources.
been there done that.
I recently tried the FifoPi and the Audiophonics I-TDA1387 TCXO.
Wow. What a sound. You really get the feeling that there's something seriously wrong with these delta-sigma DACs.
Enjoy.
From your statements I read you're not interested in pc-speaker quality sound.
Inmate DimDim did some measurements.
All of us who have worked with these 1543/1541 DACs know how sensitive these devices are.
And it's not just jitter. It's also noise. It's power. ....
The currently IMO best option to attach such a Dac to a PI is iancanada's FifoPi. (Allo is not offering such a device) The FifoPi is an isolator and reclocker and the clean side can be powered separately. You'll get the major
issues under control with this device.
FifoPi also provides high quality UFL sockets for I2S interfacing to the dac.
Using some cheap jumpercables from GPIO is not the best of all options
to run I2S.
*****
Resampling.
Yes that's an option. It won't get you around the poor RPI I2S though. It rather makes things worse.
Let's assume you'd go for the FifoPi.
The upsampling will have a positive effect on a NOS dac. But in this case it's actually no longer a NOS dac you'll talk about. Every DSP introduces it's own losses.
Not any resampler is lossless. You need to find the best resampler and use that one at highest quality settings.
Squeezelite and LMS would both allow to do so. These would even support synchronous (x44,1 + x48) resampling.
However.
Then you'll run into the 16bit issue. If you do resampling you need some headroom. You'll loose on the level to avoid clipping. And then there's
digital volume control. On a 16 bit DAC running digital volume control and
resampling simply doesn't make any sense. If you have an analog volume control it'll do.
Good luck with your choices. Don't waste to much time and resources.
been there done that.
I recently tried the FifoPi and the Audiophonics I-TDA1387 TCXO.
Wow. What a sound. You really get the feeling that there's something seriously wrong with these delta-sigma DACs.
Enjoy.
I see the problem. I've got a IL715E isolator and some SRC4192 SRCs flying around, i will go and mill an isolator and an relock PCB tomorrow to check how much this will improve the sound. The power rails from the Sony CDP 710 are very nice, good regualtors very short tracks from regs to ICs and high quality caps from the beginning, so this should be covered.
I will power the isolation/relock PCB from LiFePo4 first to ensure highest possible quality of power with lowest effort.
Yes i know, Ians products are very nice, but at the moment i tried with the kali reclocker. It was a very good improvement, but still not the quality i was used to hear from this CD Player. But as you stated, it can be the missing isolation, that interferes with the DAC.
You are right, cables make a greater difference than i thought. i changed the cables from simple copper cables to high quality shielded cables and i have to say, that this little mod made a difference you can hear. Never thought this short cables can pick up so much hf noise...
Ok, seems very plausible to me. I will skip the software implementation of upsampling and will try a SRC4192 (as a sideeffect i can detach the reclocker then, because the SRC4192 does reclocking itself). I used this one with an old PCM63K DAC and it did its job fantasticly, perhaps i should give it a try with the TDA1541 too. Never thought about it, but as i saw the link you posted, now i know that the TDA is even more problematic in terms of clean power and hf noise than the PCM63.
If i do not like the upsampled sound, i can still configure it to just do reclocking and no upsampling (16 bit, 44,1 kHz on IN and OUT, so no processing occurs, just reclocking).
I've been using a PCM1794A dual Mono DAC with USB->I2S Wave IO for a long time and i have to say that delta-sigma suffers very much from active I/V. Go passive and you will experience a whole new world. Lundahl tranformators did the trick for me, but some weeks ago a friend showed me, that there is another way for comparable quality with lower price tag. He used battery powered opamps for IV and it was simply beautiful what came out of the speakers. Very very nice. IV on Lifepo4 simply rocks. LiFePo4 has such a low internal resistance, that you can even go capless if your traces are short enough. This was astonishsing quality for opamps. He used LM6172 on batteries.
I will check the isolation/relocking route and will write some lines about it. Thanks for your input, sadly i have NO experience with the TDA1541, but im not shy of experimenting and/or creating PCBs by myself...
Im still thinking about DEM timing on the TDA1541. Anyone some experince with it ? Is it worth the effort ?
Last edited:
The LiFePO4 approach is
1. a bit dated
2. annoying to handle
You'll find similar performing solutions using supercaps as buffers.
These will/can IMO outperform the batteries.
Let see if the upcoming Allo supplies (Shanti/Nirvana) will shuffle the market.
Then you shouldn't miss the rather exhausting ecdesigns thread about it.
I'm pretty sure you'll learn ""everything"" about this chip.
Good luck.
1. a bit dated
2. annoying to handle
You'll find similar performing solutions using supercaps as buffers.
These will/can IMO outperform the batteries.
Let see if the upcoming Allo supplies (Shanti/Nirvana) will shuffle the market.
Then you shouldn't miss the rather exhausting ecdesigns thread about it.
I'm pretty sure you'll learn ""everything"" about this chip.
Good luck.
I know, LiFePo4 can be very strange because of its charging nightmare. But its very good for testing purposes. And like i said, it simply rocks together with IV opamps.
Of course you can build very high end shunt PSUs and supercap powered, but not for the price of a LiFePo4 and some charging curcuit.
My nect approach will be supercap based PSU, but my experience with them is rather limited.
Of course you can build very high end shunt PSUs and supercap powered, but not for the price of a LiFePo4 and some charging curcuit.
My nect approach will be supercap based PSU, but my experience with them is rather limited.
Yep. That's what use them for. It acts as a kind of reference and doesn't come with groundloop issues.
Don't forgett the Kali outputs x4 upsampling if I'm correct.
Dem time reclocking is not worthing it imho, many things are more important.
Of course I2S lines asks individual ground return path and the best is shielded cable doing the task like the uf-l ones that keep the 50 ohms impedance between the boards. Those I2S cables should not be longer than 4", the shorter the better. In your situation I'm not sure the fifo will a great enough improvment over the Kali because of the layout problem between the sony and the streamer/reclocker combo
But the TDA1541 needs a very good layout and pcb I'm not sure the old cd players deserve all the care you targett.
Easier to start from a good proof design as Audial DIY board for instance (imho the best so far on the diy markett). front end can be sota in board usb towards simultaneous mode (improved I2S) or I2S/Sim mode through uf-l plugs. Then indeed a fifo can surely improves the sound vs a Kali (fifo is not upsampled so you must do it from the Pi or a NAS server if you have a segmented solution.
A good thing should be to ask Abraxalito to putt 3 uf-l inputt plug on his TDA1387 diy board then feed it from the Fifo. If you have a doubt, read the long TDA1541 thread Souncheck is talking about and plan for 3 years of experiments, tries & errors... the dac chip is not all, better a good TDA1387 board design than an aproximative TDA1541 one!
As usual YMMV.
Dem time reclocking is not worthing it imho, many things are more important.
Of course I2S lines asks individual ground return path and the best is shielded cable doing the task like the uf-l ones that keep the 50 ohms impedance between the boards. Those I2S cables should not be longer than 4", the shorter the better. In your situation I'm not sure the fifo will a great enough improvment over the Kali because of the layout problem between the sony and the streamer/reclocker combo
But the TDA1541 needs a very good layout and pcb I'm not sure the old cd players deserve all the care you targett.
Easier to start from a good proof design as Audial DIY board for instance (imho the best so far on the diy markett). front end can be sota in board usb towards simultaneous mode (improved I2S) or I2S/Sim mode through uf-l plugs. Then indeed a fifo can surely improves the sound vs a Kali (fifo is not upsampled so you must do it from the Pi or a NAS server if you have a segmented solution.
A good thing should be to ask Abraxalito to putt 3 uf-l inputt plug on his TDA1387 diy board then feed it from the Fifo. If you have a doubt, read the long TDA1541 thread Souncheck is talking about and plan for 3 years of experiments, tries & errors... the dac chip is not all, better a good TDA1387 board design than an aproximative TDA1541 one!
As usual YMMV.
Don't forgett the Kali outputs x4 upsampling if I'm correct.
Dem time reclocking is not worthing it imho, many things are more important.
Of course I2S lines asks individual ground return path and the best is shielded cable doing the task like the uf-l ones that keep the 50 ohms impedance between the boards. Those I2S cables should not be longer than 4", the shorter the better. In your situation I'm not sure the fifo will a great enough improvment over the Kali because of the layout problem between the sony and the streamer/reclocker combo
But the TDA1541 needs a very good layout and pcb I'm not sure the old cd players deserve all the care you targett.
Easier to start from a good proof design as Audial DIY board for instance (imho the best so far on the diy markett). front end can be sota in board usb towards simultaneous mode (improved I2S) or I2S/Sim mode through uf-l plugs. Then indeed a fifo can surely improves the sound vs a Kali (fifo is not upsampled so you must do it from the Pi or a NAS server if you have a segmented solution.
A good thing should be to ask Abraxalito to putt 3 uf-l inputt plug on his TDA1387 diy board then feed it from the Fifo. If you have a doubt, read the long TDA1541 thread Souncheck is talking about and plan for 3 years of experiments, tries & errors... the dac chip is not all, better a good TDA1387 board design than an aproximative TDA1541 one!
As usual YMMV.
More and more i'm getting what you mean. All my tries until now show a slight imrovement, but not what the CD Player istself was able to do.
I think it will be a design problem to marry the pi with the old CD player.
The IL715E though brought a huge difference, i think it will be because of the galvanic barrier. Last try for me will be the ecm i2s to sync mode after an isolator. The SRC4193 approach was not what i hoped for, with and without isolation.
If Kali does x4 upsample, this would explain some of my pains. Will check it on monday when i can use it on an osci.
Where did you get that info?
This is what Allo outlines as output FS for Kali:
According to this you can't be right. Please check your source.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.