[sdiy] STM32 vs WM8731

[Richie Burnett]

No worries, thanks for clarifying Eric :-)  It just seemed a bit strange to 
me, to be filtering the high-frequency MCLK signal in order to prevent 
crosstalk onto the lower frequency SCL clock.  Thought it seemed more 
logical to low-pass filter the slower SCL clock to remove the 
higher-frequency MCLK feedthrough.  But, hey if it works...

The MCLK from my dsPICs to Crystal CODEC chips certainly seems to couple 
onto everything nearby, so I will give your method a try in future!

-Richie,

-----Original Message----- 
From: Eric Brombaugh
Sent: Tuesday, February 16, 2016 4:03 PM
To: Richie Burnett ; synth-diy at dropmix.xs4all.nl
Subject: Re: [sdiy] STM32 vs WM8731

Richie,

Yes - on MCLK. The MCLK input to the WM8731 is going into the crystal
oscillator input pin (XI) and works fine with a nearly sinusoidal
waveform. RC filtering the higher harmonics allows the critical
fundamental to get through to the oscillator input while reducing the
high-frequency aggressor that bothers the I2C SCL input pin.

I used that approach on this circuit - worked great. See R103/C117:

http://ebrombaugh.studionebula.com/synth/dsPIC_sp/dsPIC_sp_pg1-2.pdf

Note that this usually isn't necessary with a good layout. It did help
while prototyping on a solderless breadboard and I carried it over to
the PCB design just because.

Eric

On 02/16/2016 08:53 AM, Richie Burnett wrote:
> Eric wrote:
>
>> If you have a scope with high impedance probes, look at the SCL input
>> pin - if you see MCLK there at a low level, it's a good bet your
>> WM8731 can't hear the I2C bus due to the crosstalk. Reroute the
>> signals or put a simple RC lowpass filter on the MCLK to mitigate that.
>
> On MCLK?  Or lowpass filter the I2C SCL input to remove the MCLK clock
> feedthrough?
>
> -Richie,
>

_______________________________________________
Synth-diy mailing list
Synth-diy at dropmix.xs4all.nl
http://dropmix.xs4all.nl/mailman/listinfo/synth-diy