Going through the logs, I feel there could a problem with the EMCA itself.
According to specification, there should not be any problem with sourcing VCONN indefinitely, except that it will waste power if there is no need to communicate with SOP'/SOP"
So I believe the notebooks are fine, the cable might be behaving strangely. Can you try with different cables? Do you see differences in behaviour between different EMCA's?
Thanks for getting back. We have tried with multiple cables so not sure that is the problem. Most likely our problem is somewhere else, just double check so to remove one possible cause.
But the reason for some laptop starting VCONN, can that be changed by modifing some of the paramters in the PD-configuration file? When we use default Cypress configuration file for CCG4 host we don't see this behaviour, but when we reconfigure the SVIDs to become a docking station (DP sink) it starts to happen. Isn't continously powered VCONN only there for certain dongles that doesn't want to use VBUS for power or for active cables where the chip needs power all the time?
The problem was cornered to signal integrity on the DP main link. The PS8742 output was too strong leading to over-equalization at the reciever. Finally Aux interception had to be disabled to get the pre-emphasis and swing to the lowest values. The EQ settings will have to be altered by changing the parameters in datamux_ctrl.c file.