Other Parts Discussed in Thread: TPS23755, TPS23758
Tool/software:
Hello all,
Recently we have had an unexpected communication issue between the Phy and ethernet. We noticed that by replacing the part we can fix it.
The schematic design is identical to: https://www.ti.com/lit/df/snlr011a/snlr011a.pdf?ts=1721333043067
The only difference is that there is a 2kohm resistor between 3V3 and MDC pin on ours. I can remove it if you acquire.
We have a POE 15W on one side and MCU with a 3V3 coming from a battery through an LDO on the other side.
After some tests, we found out that by plugging/unplugging the POE or battery we can simply kill the phy.
When the failure happens, sometimes by forcing a crossover on the MDIX we can make the phy still communicate. But this one is not guarantied, either.
One of our assumptions was high temperature during the reflow due to having large components on our POE section on the board. However, this case also happened on a Phy that was soldered manually and gently.
I would like to know how dangerous the unwanted noise/voltage-spike over the Power-rail/ground caused by plugging/unplugging the Battery/POE can be.
We have had this phy seating on some other boards and had no problem among a high number of production. The major difference between this board and the other ones is having the POE 15W.
Any help and guidance would be appreciated.
Sincerely yours,
Arash