I am a hobbyist, designing DIY DACs based on old, R2R chips.
Some time ago I finished a project with two mono "DIP16" DACs. Here you can find the schematic: [Schematic] and below, the top layer of the PCB.
AD1896 works in slave mode, with BCLK = 13.5 MHz, LRCLK = 210.9375 kHz. Shift registers "translate" I2S to feed the D/A chips. Optional jumper allows to choose between 16 and 18 bit DACs. Logic works on 3V3, +/-5V on Vl pins and +/-12V on Vs pins.
When using PCM56 (16bit) and PCM61(18bit), everything works fine. There are many AD ICs which are, in theory, pin compatible - but the reality is different! I have tested AD1860N, AD1860N-K and AD1856N-K. The sound is heavily distorted. Lowering sample rate to 192 kHz did not help.
There is a similiar project, on which my project is based on, this time using AD1865. The designer included a charge pump in his desing to generate -5V, which is required by AD1864 to operate correctly. The AD1864/AD1865 are pin compatible, with only difference being the negative logic supply. While AD1865 works fine, AD1864 has the same problem as my project - heavily distorted sound. Interesting thing is that when the negative voltage fed to -Vl pin of the AD1864 is raised to ~-1.9V, it plays fine. I tried the same for AD1860N and the results are similiar - one DAC started playing properly, but the other went silent. I used a potentiometer and tried to find the optimal voltage, but with no success.
Here you can find the schematic: [AD1865]
Given that all AD DACs I have tested suffer the same problem, while old Burr Brown chips work fine, I assume there is something different in the project of the internal logic in AD ICs. So, here comes my question - do you have any idea why these chips don't work?
Also - AD1851/AD1861, similarly to AD1865, don't need negative logic supply. If my theory is correct, these should work fine in my desing. Do you have any idea where could I source them? I suppose there are not laying in some drawer, hidden in an old AD warehouse?
Well, in case anyone was interested, shift registers caused slight delay on data lines. 470R/100pF on LE lines was enough to solve the problem.