I have played successfully 384/32 with Amanero384 and AD1862.
Unrelated, I’ve played 384/32 with RPi3 and RPi4 and FifoPi v2. I since rolled back to 192 KHz as there is an almost imperceptible sound similar to EMI interference. It is very quiet and I can only hear it only when the music plays (nothing during the pause - hence it’s not an electrical issue) and with quiet passages. And easier with the electrostatic headphones...
I can try over the holidays with the RPi and AD1862, but I see no reason for any surprises.
Unrelated, I’ve played 384/32 with RPi3 and RPi4 and FifoPi v2. I since rolled back to 192 KHz as there is an almost imperceptible sound similar to EMI interference. It is very quiet and I can only hear it only when the music plays (nothing during the pause - hence it’s not an electrical issue) and with quiet passages. And easier with the electrostatic headphones...
I can try over the holidays with the RPi and AD1862, but I see no reason for any surprises.
Miro,
what Is maximum allowable length of I2S wires using shielded and non shielded wires
what Is maximum allowable length of I2S wires using shielded and non shielded wires
@samoloko
it is not possible to say the exact dimensions in general
non shielded wires as short as possible (below 15cm)
shielded can be longer (but not more than 2m)
use an spdif for longer distances (spdif-i2s) ... it can be very long (even more than 10 meters)
it is not possible to say the exact dimensions in general
non shielded wires as short as possible (below 15cm)
shielded can be longer (but not more than 2m)
use an spdif for longer distances (spdif-i2s) ... it can be very long (even more than 10 meters)
Hi RbW,
The I2S protocol was not made for long distance. Remember this is the Philips Sony protocol to interlink the cd laser lense/ic with the dac ic board inside the cd players ! It must be short, shielded wires or at least use flat wires with galvanic separation : signal-gnd-signal-gnd-signal-gnd.
imho beyond 8", this is not serious anymore. Frankly it should stay bellow 4" interconnect and even shorter as you may consider the pcbs traces... We talked about that in the thread if you return back in the thread in the photographs comets.
Think also to get a power supplies traffo > 10 cm from the ICs. Torroids and C-Core are less radiating than classic cheap traffos.
The Rpis boards are not made for audio really, i.e. their I2S output comes from a ppl and single crystal. Hence the need of an asynchronous reclockig board if you don't use one already between the Rpi and the AD1862 board as we talked in the thread. Look at the Luchoh member post just above for illustration.
For instance a cheap Allo Kali will improve lot the sound and fix imho your problem if you follow the advice of short connexion, shielded or at least with a gnd separation between each i2s signal like a flat cable : gnd-signal-gnd-signal-gnd-signal-gnd ; but shielded cable are better. All those asynchronous reclocker board use galvanic isolation and that is mandatory as the streamer are noisy whatever the source : the laptop, the NAS, the Internet Provider box, the tower PC, the Rpi boards...
The I2S protocol was not made for long distance. Remember this is the Philips Sony protocol to interlink the cd laser lense/ic with the dac ic board inside the cd players ! It must be short, shielded wires or at least use flat wires with galvanic separation : signal-gnd-signal-gnd-signal-gnd.
imho beyond 8", this is not serious anymore. Frankly it should stay bellow 4" interconnect and even shorter as you may consider the pcbs traces... We talked about that in the thread if you return back in the thread in the photographs comets.
Think also to get a power supplies traffo > 10 cm from the ICs. Torroids and C-Core are less radiating than classic cheap traffos.
The Rpis boards are not made for audio really, i.e. their I2S output comes from a ppl and single crystal. Hence the need of an asynchronous reclockig board if you don't use one already between the Rpi and the AD1862 board as we talked in the thread. Look at the Luchoh member post just above for illustration.
For instance a cheap Allo Kali will improve lot the sound and fix imho your problem if you follow the advice of short connexion, shielded or at least with a gnd separation between each i2s signal like a flat cable : gnd-signal-gnd-signal-gnd-signal-gnd ; but shielded cable are better. All those asynchronous reclocker board use galvanic isolation and that is mandatory as the streamer are noisy whatever the source : the laptop, the NAS, the Internet Provider box, the tower PC, the Rpi boards...
Last edited:
Thanks for everyone's input! Wires are 6 inches long, see attached picture. I am using fifopi with accusilicon clocks as input. It is all fine with 24/96 except 24/192, it heavily distorts, why would that be "hardware" related?
Maybe ask the thread for the besr I2S input for your case in the Moode thread. But first, make better the connexion/layout (shielded & shorter)
@Robin De Wolf
simple-bclk-64fs
... well, it can be tricky 😀
... look for someone who finded settings for DDDAC to run on 24b/192kHz, apply the same settings for this AD1862 and it should work ...
... or just simply change the settings - select DDDAC
simple-bclk-64fs
... well, it can be tricky 😀
... look for someone who finded settings for DDDAC to run on 24b/192kHz, apply the same settings for this AD1862 and it should work ...
... or just simply change the settings - select DDDAC
Attachments
Last edited:
Btw, how is the aculisons in the FifoPi vs NDK or Laptechs please ? Did you hear a difference between them ?
There is room enough in the casing of the picture, nothing to change but putting the ad1862 side to side to the Fifo Pi with 3 ultra short shielded 50 ohms cables with gnd tied like Miro's shematic above for the boards gnd conections ! Try 1" length... challenging 🙂
There is room enough in the casing of the picture, nothing to change but putting the ad1862 side to side to the Fifo Pi with 3 ultra short shielded 50 ohms cables with gnd tied like Miro's shematic above for the boards gnd conections ! Try 1" length... challenging 🙂
Last edited:
Hmm.. So I tried the dddac selection in moode with fifopi.
Same result. Then I tried hifiberry digpro hat with moode and various selections. Also same result. Finally I installed volumio with hifiberry driver selected, again same result. The hifiberry has shorter i2s wires btw, so indeed as suggested, I have moved the rpi to the dac.
I'm a bit at loss here, might need to sleep over it..
I don't play a lot of 24/192 tracks, but got triggered by you guys that it should be able to work properly 😉
Ps. I only tested the stock clocks vs the accusilicons. I did that along with other upgrades so can't comment on the improvements of the clocks alone. I will do that when some more extended listening time is on hand.
Same result. Then I tried hifiberry digpro hat with moode and various selections. Also same result. Finally I installed volumio with hifiberry driver selected, again same result. The hifiberry has shorter i2s wires btw, so indeed as suggested, I have moved the rpi to the dac.
I'm a bit at loss here, might need to sleep over it..
I don't play a lot of 24/192 tracks, but got triggered by you guys that it should be able to work properly 😉
Ps. I only tested the stock clocks vs the accusilicons. I did that along with other upgrades so can't comment on the improvements of the clocks alone. I will do that when some more extended listening time is on hand.
Last edited:
Thanks Miro for an excellent design.
Can we use 220uF/25V electrolytic caps instead of 100uF for C7,8,11,12,17,18,21,22 locations ? As i have some Pana FC already with me.
C15,16,25,26 will be maintained as per your BoM.
Thanks in advance !
Can we use 220uF/25V electrolytic caps instead of 100uF for C7,8,11,12,17,18,21,22 locations ? As i have some Pana FC already with me.
C15,16,25,26 will be maintained as per your BoM.
Thanks in advance !
I concur and have the same issue of distortion in the music when I played a track of 24/192 and I use a DIYINHK XMOS dac via I2S short 10cm jumper wires to the AD1862. I tried on Volumio as well as RopieXXL Roon transport on Pi3. Plays fine till 96khz very clean and no noise or distortion.
No issue with 24bit 176kHz with a DIYinHK interface, at least no obvious distortion, though I thought 176kHz OS sounded worse than 88.2kHz but prefer NOS the most.
But maybe 192kHz had problems, I have vague memory of noise with a 192kHz test file.
But maybe 192kHz had problems, I have vague memory of noise with a 192kHz test file.
Is the OPA1656 compatible with this board as I see some of them using it? But when I check, it says 2 channel but the Miro's dac board is compatible only with 1 channel opamp per side. Here is the one I am planning to get OPA1656ID Texas Instruments | Mouser.
Thanks
Thanks
> ... the DAC logic does support a 64-bit frame I2S ... it does not support shorter bit frames 32 (2x16)
So what would you do with 32-but frame ?
Patrick
So what would you do with 32-but frame ?
Patrick
Try to change these resistors:
R1=100R
R2=1k
R3=100R
maybe there really is this limitation ...
I'll do some testing in the near future 🙂
@manniraj
@laserscrape
thanks guys for the confirmation, we'll see what can be done about it
@EUVL
configure the I2S source to the 64-fs (which is default for most cases)
... the only I2S I didn't find how to remake was CP2615 from silabs (it uses 48-fs) - for this case, the input logic for the DAC can be changed
@laserscrape
thanks guys for the confirmation, we'll see what can be done about it
@EUVL
configure the I2S source to the 64-fs (which is default for most cases)
... the only I2S I didn't find how to remake was CP2615 from silabs (it uses 48-fs) - for this case, the input logic for the DAC can be changed
you have to use some adapter from dual to single ... @Vunce has some experience ...
see the post #459 (it is untested 😀)
I can only talk about my experience with two 4499 based dacs:
The Yanasoft kit I had assembled with a modified Amanero/ external oscillator board, with full galvanic decoupling all directions. The external oscillator module is based on NDK SDA XO.
The other one is a Topping D90, with Accusilicon clocks.
Both have their merits, but I like better the Japan aircraft carrier design 🙂 🙂 more full, & vivid lively sound.
In order to use a dual opamp with Miro’s Dac board, the second opamp channel needs to be dealt with properly. I had an adapter board designed following TI’s recomendation. I’m currently using OPA1656 opamps.
