I've developed a custom carrier board for the CM4 and i have found that i couldn't detect carrier or operstate on the on ethernet port. I have two CM4 modules (eMMC without wifi) which have both been on this custom board and both show no carrier, i then took this CM4s off and placed them in a rpi3 adapter board and booted them up and found that on this adapter board the ethernet was also not working.
I then took a brand new CM4 module (this one with eMMC and wifi) and tried it in the same rpi3 adapter board and i was able get carrier/operstate on the ethernet port. I have not tried this CM4 on my custom board out of fear its killing the phy.
My fear is that something on my custom board has fried the ethernet phys on the CM4s (note that everything else on those CM4s appears to be working)
Attached is the schematic of my ethernet circuits
-RJ45 jack = HC-WK88-H16-1X2-DB
-On board ethernet magnetics = CND-tek G2415S
I can't find a datasheet for the phy on the CM4 (and at the time of design couldnt get the integrated mag/rj45 jack) so i did my best to recreate it with discrete components as per the CM4IO design.
Some other observations
-The two CM4s which i think have fried ethernet phys draw less power (100-200ma) then the fresh CM4 when used on the same pi3 adapter board (they all have the same OS image).
-On one of the CM4s which isn't detecting ethernet the broadcom ethernet phy on the CM4 gets very hot
-I noticed that the ofther CM4 which isn't detecting ethernet got very hot and drew more power when i first used it (first boot up) however over time (a few hours) dropped down.
-In addition to the on board ethernet i also have a usb-ethernet adapter on my custom board, this port can detect carrier/operstate without issue. It also has the same POE connections (thou POE has never been enabled while testing)
Apologies for the less then systematic analysis but my carrier board has quite alot on it and i thought the extra draw was from other parts of the ciruict so didn't notice the hot cm4 straight away. Any suggestions or ideas would be most appreciated
I then took a brand new CM4 module (this one with eMMC and wifi) and tried it in the same rpi3 adapter board and i was able get carrier/operstate on the ethernet port. I have not tried this CM4 on my custom board out of fear its killing the phy.
My fear is that something on my custom board has fried the ethernet phys on the CM4s (note that everything else on those CM4s appears to be working)
Attached is the schematic of my ethernet circuits
-RJ45 jack = HC-WK88-H16-1X2-DB
-On board ethernet magnetics = CND-tek G2415S
I can't find a datasheet for the phy on the CM4 (and at the time of design couldnt get the integrated mag/rj45 jack) so i did my best to recreate it with discrete components as per the CM4IO design.
Some other observations
-The two CM4s which i think have fried ethernet phys draw less power (100-200ma) then the fresh CM4 when used on the same pi3 adapter board (they all have the same OS image).
-On one of the CM4s which isn't detecting ethernet the broadcom ethernet phy on the CM4 gets very hot
-I noticed that the ofther CM4 which isn't detecting ethernet got very hot and drew more power when i first used it (first boot up) however over time (a few hours) dropped down.
-In addition to the on board ethernet i also have a usb-ethernet adapter on my custom board, this port can detect carrier/operstate without issue. It also has the same POE connections (thou POE has never been enabled while testing)
Apologies for the less then systematic analysis but my carrier board has quite alot on it and i thought the extra draw was from other parts of the ciruict so didn't notice the hot cm4 straight away. Any suggestions or ideas would be most appreciated
Statistics: Posted by dbomb — Fri Apr 19, 2024 1:27 am