ADC to ADAT feasibility

[Dimitree]

hello everyone

during my work lunch break, I like to do some audio DIY,
so since I got a V1401 chip (clone of the AL1401 ADAT transmitter) I'd like to build a 8 channel digital converter (with line inputs) that I can use with my Focusrite Scarlett 18i6 interface that has got an ADAT input.
I know I could just buy one if I'd need, but I really want to build it, that's my hobby 
I already successfully built a DAC from scratch (schematic, smt PCB) around PCM5122, also I have large experience with building analog/digital synthesizers, but I still don't know much about ADAT and digital audio transmission.

So first I'm trying to understand if it can be done like this:
4x PCM4220 (or CS5381) ADCs connected to the AL1401, one of them set as master, the other three as slave.
Master clock generated on the board (no externarl word clock input needed), using a pair of low jitter TXOs, manually switched depending on the desidered sample rate.
I'm wondering if this master clock can be supplied to both the ADCs and the ADAT chip, and if that would create problems since on the Focusrite interface, you should set the sample rate on the software (so the master must be the Focusrite?).

many thanks to all!

 

[mhelin]

Should be fine, AL1401 (or V1401) doesn't need other clocks than WDCLK. Just set the PCM4220 to use master mode and left justified, 24-bit audio format. The datasheet has nice differential input stage circuit using OPA1632 which is perfect.  For 48 khz sampling using "normal" sampling rate you need just a single 12.2880 MHz (256x 48kHz) oscillator. Tentlabs have some ready modules and a bare chip: http://www.tentlabsshop.com/DetailServlet?detailID=1671
You connect this to all ADCs, better use some buffering between (not sure how many adcs single oscillator can source) and avoid long traces. When ready select ADAT sync source in audio interface control panel.

 

[Dimitree]

many thanks for the reply.
Any reason why I should I prefer left justified instead of other modes?

 

[mhelin]

Sorry no reason other than just don't use I2S as it's not supported by 1401. Any common format for the converter and V1401 is fine.

 

[Dimitree]

thank you,
any suggestion about buffering between each slave PCM4222? also about buffering the clock lines

 

[Rochey]

A) Use PCM4220 - there's no difference for PCM modes.
B) Look on page 3 of the datasheet: DIGITAL I/O CHARACTERISTICS. The capacitance is given, you should be able to work out each digital pins impedance at their respective carrier frequency. Providing your source can supply that much current, without compromising clock edges etc, you should be fine.
c) Where possible, follow Bob Martin's Eval kit for the PCM4222. It's a bit haphazard in how it's drawn, but Bob was always able to get the performance from the board.


/R

 

[Dimitree]

) Where possible, follow Bob Martin's Eval kit for the PCM4222. It's a bit haphazard in how it's drawn, but Bob was always able to get the performance from the board.


/R

Do you have any link about that? Thank you

 

[Rochey]

First result on google for “pcm4222 evm”

http://www.ti.com/lit/ug/sbau124/sbau124.pdf

 

[Dimitree]

sorry I thought you were referring to a particular implementation of the PCM4222EVM and not the TI one.
Mostly I'm interested in how Martin's design implements the led bargraph. I see the PCM4222 has got a digital output for the overlflow indicator, but what do you think is the best way to add a "sub 0dbfs" level bargraph too? I don't need extreme precision (10 green leds are more than enough, + red for the overflow).

 

[Rochey]

Doing it in the digital domain is way too hard. Better to use an analog input and calibrate it on the bench.

Lm3915/6 or one of the clones is your best bet.

 

[Dimitree]

And where in the signal chain (assuming I’ll use the EVM schematic) would you put it, so to not overload/degrade the analog input signal?

 

[Rochey]

hang it off the differential input, into a high impedance JFET input buffer.
The input impedance of the PCM4220 is quite low, so you'll need something like an OPA1632 to drive the bugger anyway.

Safest bet would be a TL072 hanging off the output of the OPA1632, output going to an LM391x circuit.
The input impedance of the TL072 is 10^12Ohms. Any capacitance on the inputs of the TL072 should be easily driven by the OPA1632 - it's a BEAST.


/R

 

[Dimitree]

thank you,

so do you suggest a simple TL072, and then direct to the LM3915 without a rectifier?
should I need the TL072 used ad differential buffer?

at the moment I draw this schematic based from the datasheet