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

Jump to Latest
diyiggy

diyiggy

Disabled Account
Joined 2019
manniraj said:
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
Click to expand...


We talk about that in the thread, same with opa861 and all not pins compatible aops like the non single ones : it needs some diy and care with decoupling. We also talked about discrete and EUVL even gave us a discrete scheme :cool::).
 
diyiggy

diyiggy

Disabled Account
Joined 2019
miro1360 said:
@har297
add 0.22R or 0.33R after LM317/337 ... it will increase the stability for those regulators :)
DC -> 4700uF -> 10 Ohm -> 4700uF||100nF -> LM317/337 -> 0.33R -> 10uF||100nF -> DAC Board
Click to expand...


Idem for the 78xx/79xx regs : they needs it at their output. I prefer to use an hold lytics between 47 to 220 uF there that has lost its characteristic, i.e. that now shows very bad esr instead of the resistor !


I liked the old Nichicon Vs or Vx caps for the job I get from old CD players boards.


so with that regs what works fine is : low esr at the input side near the legs (a standard good modern lytic, Simic 2, Panasonic, Nichicon, what suits your tastes) and the bad old cap at the reg output. Then near the load the low capacitance decoupling in the nF and few uF range, // if you want.


For the DC rectifier that can be far from the reg sometimes : This gave me rxcelent results if we take the Miro1360 example CRC : 4700 uF-1 to 10 ohms-4700//1200 or 800 uF.... imho the 100 nf should be more on the legs of the talked regs at their input (Vin-Gnd)...


Anyway , trust your ears, experiments are fun (as sims with power supplies)
 
M

miro1360

Member
Joined 2010
Ok guys, I already tested AD1862 with the same glue-logic on jlsounds (32b/384kHz) and diyinhk xmos (24b/192kHz). easyDAC ... it has the same schematic for AD1862 and it’s data input.
I used 10cm data wires. R1=R2=R3=0R (it should work with other values too)

It worked perfect on diyinhk xmos:
16b: 44.1, 48, 88.2, 96, 176.4, 192 kHz
24b: 44.1, 48, 88.2, 96, 176.4, 192 kHz
(my diyinhk xmos can’t go higher, it’s an older version)

It worked perfect on older jlsounds xmos (with heatsink :D) :
16b: 44.1, 48, 88.2, 96, 176.4, 192 kHz
24b: 44.1, 48, 88.2, 96, 176.4, 192 kHz
32b: 44.1, 48, 88.2, 96, 176.4, 192 kHz

It failed (distorted) on jlsounds xmos:
16b: 352.8, 384 kHz
24b: 352.8, 384 kHz
32b: 352.8, 384 kHz
... it would fail on any xmos from the frequency 352.8kHz and higher
... so the limit is 192kHz


I will test this PCB in the future because I haven't soldered it yet :D
 
M

miro1360

Member
Joined 2010
har297 said:
@miro1360
...
Do you feel 317/337 is better or 78xx and 79xx regs are better in this design.
...
Click to expand...

78xx/79xx are not all the same ... the best experience are those from onSemi: MC78xx/MC79xx

there are different approaches on how to connect LM317/337 ... try the connection without TL431, just with simple resistors

what sounds better is hard to say for me ... I like both :D

as @diyiggy said, do a comparison and tell us your results :D
 
diyiggy

diyiggy

Disabled Account
Joined 2019
So it's maybe the Rpi despite the Fifo after, I thought it was not proof at 388 sample rate, but maybe even 192Kz is too much ? If someone has a Rpi with the board and could add an inputt about that ?


@ har917 (the LM317 is well known to have very low noise, less than the 78 family, and easy to use, there are of course tons of good better regs nowadays, but damn for the price of it they are hard to beat and enough in many analog applications, but notice I'm not electronician as Miro1360 is....
Something defintly cool to read is what Nazar member wrote in him blog here at Diyaudio about fast regs with the LM317/337 : could be also a good cheap enough reg for the aop with only 1 to 2 uf with 100 nF with the fastest oaps on their legs of the discussed list... for the digital side Miro said the AD1862 had very low noise regs inside so a proper decoupled 79/78xx regs from On Semi is certainly good enough -as NJR ones?- .... want to improve ? just buy a R-Core traffo instead a torroid; there is in India a good supplier of R-Core but I don't remember the name , sorry)
 
Last edited:
M

manniraj

Member
Joined 2012
miro1360 said:
Ok guys, I already tested AD1862 with the same glue-logic on jlsounds (32b/384kHz) and diyinhk xmos (24b/192kHz). easyDAC ... it has the same schematic for AD1862 and it’s data input.
I used 10cm data wires. R1=R2=R3=0R (it should work with other values too)

I will test this PCB in the future because I haven't soldered it yet :D
Click to expand...

Thanks Miro, I have set R1=R2=R3=0R resistor and tested this via my Pi3 running Ropieeexl as Roon transport. I get the distortion when I play 24bit/192khz file as soon as it starts. Did you try the above check via Pi?
 
R

Ripster

Member
Joined 2018
Robin De Wolf said:
Interesting! I will try that later, after setting up roon account. What settings did you use in Roon so that I can replicate a similar setup?
Click to expand...


In roon I have DSD over PCM max sample rate 192kHz, max bits per sample 24 this is max for hifiberry and I have no other higher choices only lower. I can than play some audiophile speaker setup album that has 192kHz and 24bit tracks... and on ropieee it is selected raspberry pi DAC i2s...
 
You must log in or register to reply here.
Top Bottom