LS1028A: MDC-MDIO interface

[Alexey_Stepuk]

Hello.

We are busy with development of a product based on the LS1028A processor and Ethernet PHY VSC8514XMK-03. We have a problem with initialization of Ethernet PHY. In the document LS1028ARDBE(Reference Design Board Errata) we found the following information:

Due to a silicon bug, the MDC-MDIO interface is always busy after first few transactions. This happens due to non-monotonic MDC at source (DUT). The workaround makes the MDC signal monotonic at the LS1028A MDC pads.  Add a 0402 capacitor of 100 pF on MDC trace as near as possible to DUT pin .

Ethernet PHY VSC8514 is used on the reference design board. We added a capacitor of 100 pF and now we don't have problems with Ethernet PHY.

My question: If we use a different Ethernet PHY, will the problem with the MDC signal reoccur?            Or this issue only occurs when using the Ethernet PHY VSC8514.

[ufedor · Accepted solution]

To determine whether the capacitor will be needed please use the PHY and the processor IBIS models to simulate the MDC behaviour in HyperLynx.

[jincaisong]

Hello，

I met this problem also when I designed one board with LS1028. In our board, we had added capacity 220pf near the DUT, it does take effect comparing the case without capacity. But it still happened in the bootup phase.
We have tested reboot and power cycle, we found it happped once per 1000, but never happen during running.  I would like to confirm whether the non-monotonic MDC happen only when boot up and not happen after bootup?

Look forwarding to your reply.

BR,