RPi GPCLK0 and GPi LED driver instability problem does not go away

When I shifted the scope's probes to display GPCLK0 and digit enable, what I hope not happening happened - both the clock and the digit enable signals went dead!

I did expect the circuit unstable.  So I executed the led_demo program again, and another thing I hope not happening happened - the clock restarted, but the digit enable did not.  This means the RPi clock circuit is restarted, but the GPi still got deadlocked.

I have experience in resetting the GPi, by powering off and on again, and the CPLD works again, without restarting the led demo program.

So I jumped to the following conclusion.

1.  The RPi's clock circuit is not stable.  If the clock hangs, just run the program again.  No need to reset the RPi.

2.  The GPi's circuit is not stable,  If the CPLD hangs, just power off and on to reset the CPLD.  It is not necessary to run the program again.

3.  Since I am using another PSU, RPi, and GPi, so the RPi clock and GPi LED driver circuit is not stable because of the nature of the RPi and GPi, not related to any independent devices.

.END