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

Damn, those smd tinny things... A lot of 0805 cases, few 1206... I enjoyed to solder 0605 cases on 0805 pcb foot print...
Beginning with the ad811... then the 0pa861, then op961, then ADA4899.
 

Attachments

  • 20250405_022228.jpg
    20250405_022228.jpg
    388.5 KB · Views: 99
for I, it willl be the same at the end as there is layer more after: a FGPA (miro's) to handle the PCM for the two DAC chips.

Whatever the xtals, each fgpa after add jitter, ground-bounce, decoupling behavior of active devices... and should ask near recloking. The less the better or the last FGPA must be clocked /slaved by the reference clock which is anyway before as embeded on the "clone like".

With shorter solution like Yoke or JLSounds, reclocking at the feet of the dac chips is surely not mandatory or difference certainly hard to hear. The few added jitter comes from the last active fgpa/cpld I assume (?). Ons stage logicallly seems better to me.

I have something like Tesla in my main DAC with IanCANDA I2S to PCM on a third board : but it is slaved by the MCLK and in another clock domain than the first USB to I2S first stage (here a Wave i/o). And all is impedance controlled with uf-l cables from the beginning to the dac chips. Dping combo with two boards that have diffference uf-l arrengment is never good (bad sheiling, impedance problems, more bouncing; not isolated return path and crosstalk most of the time).

But at the end, if you ask me, I really believe only on the final listening test beyond theory !
 
2 to 10 pico second?
Anyway, the problem is not your design is bad , that is not was I said btw but that you have to add a stage more in Tesla scenario.

I'm not sure it is better in that scenario than the 74hcxxx on the same board than the dac chips. Layout matters.
I dunno if someone has the machines to measure that.

Ah, I should be able before the end of the week to test my v5 4 layers board with Grunf's BOM but the feedback cap which is a simple TDK C0G soft terminations).
 

Attachments

  • 20250409_035641.jpg
    20250409_035641.jpg
    360.1 KB · Views: 54
  • Like
Reactions: miro1360
@diyiggy Both are good, either HCT registers (with untouched LRCK) or CPLD

@altor EPM240T100C5N is still available everywhere in a fair price (new or aliexpress) and performs great - this is my choice nr1 😉 ... even the cheapest 5M80ZM64 with 64 macrocells can be enough for this DAC conversion and is very tiny (yes, the supply voltage is only 1.8 for this and you need voltage convertors) 😉