When USB is not connected, I measure the stability of the 32KHz crystal oscillator at better than 0.1ppm. When I connect USB, the stability is worse than 100ppm. I see this on our custom PCA and on the TI demo board. Using MSP430F663x ICs and measuring the ACLK output with a 53131A Universal Counter. Experimenting with trigger sensitivity and the 100KHz filter lead me to believe this is not a measurement noise problem but actual fluctuations in the oscillator output.
Has anyone else seen this?
Has anyone checked 32KHz crystal oscillator stability with and without the on-chip USB active and NOT seen this?
Does anyone know of a fix or workaround?