2
\$\begingroup\$

I am having trouble with a randomly floating digital signal. Transmission is between a microprocessor and a CPLD unit and is used to run a half bridge. Measurement was taken after the output resistor of the processor and appeared after about 5 to 50min of running time for multiple times. Do you have any idea what could have caused this? From my point of view a mechanical loose connection didnt cause this, because its first to fast and also the system does not respond to mechanical forces.

Here you can see the problem (green signal), red is the other signal sent towards the CPLD. Blue can be ignored, this was the error output by the CPLD because of the signal received.enter image description here enter image description here

\$\endgroup\$
9
  • 1
    \$\begingroup\$ Is there any possibility that something in the MCU code is switching your PWM output signal from an output to an input? \$\endgroup\$
    – jwh20
    Commented Mar 16, 2021 at 12:40
  • \$\begingroup\$ I have had this before. I assume green is coming from the CPLD? Try to add a flipflop at the output, registering green. \$\endgroup\$ Commented Mar 16, 2021 at 12:42
  • 1
    \$\begingroup\$ It looks like it could be +ve ground shift from a flyback effect on a switch causing a negative differential slope, but without details of your grounds + physical layout, power and logical interface, that’s a SWAG \$\endgroup\$ Commented Mar 16, 2021 at 12:55
  • \$\begingroup\$ They are both from the MCU to the CPLD. CPLD is used to check the dead times in between the two incoming signals. What really is bringing questions that it appears only after some time running. Software wise there is no change made in the code. Also other boards are just running fine with the same code. So I dont think the issue is coming from there. \$\endgroup\$
    – Julez
    Commented Mar 16, 2021 at 13:07
  • 1
    \$\begingroup\$ It's possible there's some mechanical issue. The different time scales to recover are suggestive of something non-electronic. Inspect soldering on the relevant pin, if possible, and try heating/cooling the board and MCU. \$\endgroup\$ Commented Mar 16, 2021 at 14:51

0