TLK106L: SMI does not work reliably

Dear TI-Team!

We have a problem with the TLK106 similar to these forum requests:

1. (1) TLK105 / TLK106 in Ethercat-network doesn't start at temperatures over 35°C - Interface forum - Interface - TI E2E support forums
2. https://e2e.ti.com/support/interface-group/interface/f/interface-forum/1102675/tlk106-some-phys-fail-to-boot-at-temperatures-above-50-c?tisearch=e2e-sitesearch&keymatch=TLK106#

We would like to initialize the phy via SMI. Sometimes it works and the phy answers, sometimes the phy does not answer and the MDIO remains high. It already happens at roomtemperature.

Our schematic looks like this:

We have noticed that PFBOUT and RBIAS are at low voltage level, when SMI-communication does not work. When there is no communication, we reset and restart the phy (as described in the forum request: (1) TLK105 / TLK106 in Ethercat-network doesn't start at temperatures over 35°C - Interface forum - Interface - TI E2E support forums), but in our case it does not work.

This is the scope, when SMI-communication does not work:

When SMI-communication works, both pins (RBIAS and PFBOUT) remain at high voltage level and the scope looks like this:

Do you have an advice for us how we can make the TLK106 work properly?

Thanks a lot and best regards

Anna Kaiser