Another TDA1541A based dac (dual differential)

[Tazzz]

I changed the HEF4517 for a CPLD (XC9572XL) to split the i2s into left and right now 96K is not a problem but for 192k the frames need to be 32f or data input in the simultaneous format, I have seen some reports of the latter being superior but I need to figure out how to best invert the msb in a serial data stream before I'll attempt it.

I have also tested an amanero async usb to i2s module in the converter with Jtest (using arta to generate the testfile) and attached here are the measurements,

I also tried the amanero with an improved low noise regulator (simple LTC6655 + pass transistor)

 

[Tazzz]

Here are a pair of extra measurements, apears as the noisefloor I see is what I get when the amanero is actively recieving a data stream.

 

[regal]

What soundcard are you using? Those are impressive measurements!

 

[Tazzz]

They were taken with a juli@.

 

[Tazzz]

I did some mods to the amanero module I improved the decoupling by adding some 10uf/0603/6.3V in parallel with the original 100nF caps this extends the flatness of the decoupling impedance without to much effort. (see attached LT simulation based on models from muratas simsurfing web resource) notice the scales !

I also did a couple of slight trace cut jobs to separate the 3.3V plane between the atmel and the xilinx cpld. Im feeding the usb inteface backwards through the 3.3V ports by TL431 based shunt regulators as Im using it as a replacement for a CS8412.

Also I found why the noisefloor was degraded when fed through the amanero I had to export the Jtest signal as a .wav file from arta and play it back through foobar with wasapi plug in.

Then the noisefloor drops see attached image for jtest result.