Hi there,
I am currently evaluating a design that utilizes the NXP platform #i.MX8Mini processor. However, I have encountered a problem with the Gigabit Ethernet after a few months of usage with prototype boards. Please refer to the following pictures for a better understanding of the connection.
The issue is that although Gigabit communication appears to be established between my board and the Ethernet switch, it refuses to function. In contrast, 10/100-T base communication appears to be working fine. Upon measuring the net_rgmii_txc signal, it is evident that the signal has a 64% duty cycle, which falls outside the acceptable range specified in other specifications. I would like to clarify that the measurement was taken on the pad (AG24), and the trace is entirely cut off from the PHY.
Currently, we suspect that the problem may be related to a broken pad/CPU peripheral block/clock source, or an unoptimized firmware setting. I would appreciate any support you can provide regarding this issue.
Thank you.
Hi Ritesh, @riteshmpatel
Thank you very much for the quick reply.
- we found 4 of 40 boards that have a similar Gigabit ethernet problem. all these four boards have similar behavior.
the boards worked well for periods of more than a month before they became problems. and we have clock signals measured in the picture below on the boards working well.
- we have verified that the Tx clock is problem, by a test providing an external patched 125MHz clock signal to the PHY, then ethernet worked as intended. we even now that clock signal is corrupted at point on the Pad of the MAC@CPU. due to this we need answer for many questions like
- why is it problem with the clock signal? does some thing wrong happen on the iMX8MIni overtime? what happens to the clock sources? is there only one clock source in on-chip that can be used for gigabit ethernet? or can we utilize an other clock source for ethernet by changing setup in SW?
