DAC AD1862: Almost THT, I2S input, NOS, R-2R

Miro, All,

Seeking your opinions on reclocking. Most reclock solutions out there will be placed before the i2s input on the DAC board - and then afterwards gets divided left and right via the shift registers. But would it not be better to do the channel splitting first, and then do the reclocking. So the schema would look something like:

bck and lrck in to the flip flop via 2 inputs each
split data taken after the shift registers L and R goes into another 2 lanes
Alignment via MCK
and then BCK, LRCK and Data L and Data R go to each DAC chip.

So the reclock is done last thing just before the DAC chips?

What do you think?
 
with jlsounds i2soverusb you can bypass shift registers and feed directly to DAC chips if this is what you are loking for...


i2soverusb.png
 
Yep, that's true, and I've used that feature in a PCM63 dac I have here, plus I have a spare JLsounds board as well. However, the source I'm thinking of is an SD card player so there is no such options with that.

I am thinking I will first try a reclock before the shift register, and if there is any hint of difference, I will to do it after the split. Interested in any other comments you might have.

Fran
 
@woodturner-fran
... the only thing you can improve with a new clock is LRCK (for AD1862 and AD1865). Only this signal is directly responsible for the resulting sound. (of course, BCK and DATA must be reasonably aligned to avoid bit loss, but in most cases this is fulfilled)
 
Thanks Miro - that is really useful to know. I'll knock up a board tomorrow night hopefully and give it a whirl. This thing sounds so good right now, any improvement is well into bonus territory.

A few friends here are embarking on their own builds..... I think demand for AD1862 chips is going up!
 
The very high quality sound and simplicity of building this AD1862 DAC is very appealing to me. We have seen several options for I/V and output stage experiments with excellent results. For me, I didn’t want to complicate the layout with the addition of external reclocking, that’s just me though. If you can squeeze out a bit more performance going this route, I say go for it! 😉
 
@woodturner-fran
... the only thing you can improve with a new clock is LRCK (for AD1862 and AD1865). Only this signal is directly responsible for the resulting sound. (of course, BCK and DATA must be reasonably aligned to avoid bit loss, but in most cases this is fulfilled)

IME on other projects reclocking all three has distinct SQ benefits - with digital, timing is everything!

For me, I didn’t want to complicate the layout with the addition of external reclocking, that’s just me though.

Compact, even simpler - no shift registers - and reclocked i2s inputs... Just add your choice of output stage and power supplies.

CUKDDHT.jpg


I'm wondering if it might be possible to do a version of the board on the same footprint but with Pedja's IV stage incorporated.
 
Last edited:
A quick bit of research has turned up that there was previously a group buy for PCBs to build Pedja's circuit and gerber files were made available. I think I'll order a small batch from JLC...

I'll happily obtain some for others if there is any interest but I think it would be sensible to restrict that to Europe as with shipping costs it will probably be more cost efficient to order boards direct from JLC for citizens of US/Aus/etc.

I've attached a zip file with the gerber package.

I checked Mouser and it looks like the only part not available from them is the single 2SK170 that is required.
 

Attachments

Compact, even simpler - no shift registers - and reclocked i2s inputs... Just add your choice of output stage and power supplies.



I'm wondering if it might be possible to do a version of the board on the same footprint but with Pedja's IV stage incorporated.

I love your stacked board solution NB, have you powered up yet?
Pedja’s OPA861 I/V output stage is fantastic also 😉
 
does the gerber file of Pedja's OPA861 I/V output stage exit?
I would like to give a try.

Buy the way, if using JLsound USB board to skip shift registers and using 2 Miro's PCB for right and left channel separately with own power supplies, will it be better for lowering cross talk between channels?
(as the pcb is relatively small and cheap, it may only cost several dollars more for one additional PCB, but will have one additional power supply)