Modification for: Nvarcher R2R DAC PCM61 Quad-Parallel Differential Design can it use PCM56P?

[Krokotam_]

Product chars :

Input method: SPDIF & I2S

Support sampling rate: SPDIF 44.1khz 88.2khz 96khz 192khz

I2S:44.1khz~384khz

I`m trying to modify this board to accept pcm56, but without any success, as the datasheet states both dacs are pin compatible, but when trying direct swap, the pcm56p is playing music but with noise. Should I assume that the WM8805 is the main issue in this configuration or something else? Another question, can I use the I/Out of the pcm`s in parallel? Removing the single op amps and feeding to valve output?

 

[abraxalito]

PCM56 only accepts 16bits and PCM61 takes in 18bits. This could be why you get noise as the LE signal won't be in the right place relative to the data with PCM56. WM8805 doesn't generate LE directly there will be some logic to convert from its output format (say I2S) to the format the DACs need. Perhaps there's a 74HC164 or two there inserting the necessary delays.

You certainly can parallel current-out DACs, I have made a habit of doing that on my designs. You'll need to watch the DAC output compliance if you want to go passive I/V (no opamp) and the DSs don't normally mention this crucial spec.

 

[Krokotam_]

I was thinking the same that the noise is from the bits, but I saw a member of this forum schematic, which is somewhere near the board from above, and

second 74hct164d is taking the bits from the pin 12 which is 16 bit ?

 

[abraxalito]

That looks like a schematic from Miro - you might check that pin10 is connected to Data for PCM61 (18bit) and you'll need pin12 for 16bits.

 

[Krokotam_]

Pin 12 from the second 74hct164D is exactly 16 bits, I will check with the board for pin 10 for the data. But there can be some differences because Miro is using 6 I think 74hct164 ... here i have only 2.

 

[abraxalito]

The reason for the extra shift regs is to hold the earlier sample (L channel) until the R channel arrives so both DACs get updated at the same time. Without them the L and R channels have a time delay between them of half a sample time.