after running PoE-powered for more than 2 years, I recently noticed that my hap ax2 is no longer able to link at 1 Gbps on the PoE port (ether1). All the other ethernet ports are still able to achieve gigabit speed. Apart from the reduced link speed on ether1, everything is still functional, including PoE-in.
Disabling auto negotiating and setting the port to 1Gbps results in no link at all. Configuration reset and rebuild from export does not change things. Tested with several cables, ports and devices, so I am confident that the hap ax2 ether1 port is the one to blame. Power via poe or barrel jack makes no difference.
Could this be related to recent RouterOS updates (currently at 7.17.1), or is this most likely a hardware failure? There is no visible damage to the port. I decided to run it with power brick for now and reconfigured ether5 to take the role of ether1, but it sucks to lose the ability to run gigabit ethernet and PoE over the same cable.
Unfortunately “cable test” tells me everything is ok, from this view it looks like the opposite side does not advertise 1G which is not true:
Plugging the same cable into ether5 results in gigabit speed:
I’m not sure if it shows broken pairs on a working link l. Maybe worth a try plugging in a cable that’s not connected on the other end?
But I’d guess it would tell you if the cable was broken even if a 100mbit link is possible
I already did this before opening this thread, I spent hours trying to find the error elsewhere. Result: Everything connected to my hap ax2 on port ether1 is limited to 100 Mbps, everything else works at gigabit speed.
IIRC if pins 4,5,7 or 8 aren’t properly connected to the peer, then switch will show only speeds up to 100Mbps as advertised by peer … even if peer advertises faster speeds.
My guess is some (electrical?) damage to ether1. Does PoE-in still work? The passive PoE-in uses same cable pairs as are needed for 1Gbps, it could be that some PoE elements shorted a pair … but if pair is shorted between them, PoE would still work while data wouldn’t. And cable test should show short on that cable pair.
On cable test screenshots … can you click the ‘+’ sign next to CablePairs … it should show state of each cable pair individually.
Other device connected through different cable, now it does not even link at 100Mbps…
I guess this means that somewhere inside my hap ax2 there is a broken connection?
It is not particularly clear how the pairs are called/to which pin they correspond, I have to assume that:
1st pair=1,2
2nd pair=3,6
3rd pair: 4,5
4th pair: 7,8
mkx’s theory, that makes a lot of sense, should have resulted in either the 3rd or 4th pair shorted (provided that cable-test results are reliable).
Your results seem to lead to another possibility , one of the two wires on 3rd pair, either pin 4 or pin 5 doesn’t make contact at all (in the socket of the hap ax2 or however internally) so there is no link on Gb (only 7/8 wires connected) and PoE positive voltage flows through only one wire (as an example 4+,7, instead of the normal 4,5+,7,8-).
I wonder how it can be checked if this is what is happening.
Do you have a couple of keystone/RJ45 sockets and a multimeter? Or can you sacrify a (good) patch cable?
I guess that part of PoE-in, there are capacitors on each line between PoE-in power “ejector” and ethernet transformers. And broken capacitor (not shorted but burned) would effectively isolate that particular line.
Or some soldered point simply developed a crack.
I don’t think it’s easy to identify actually broken element though.
Well, at least this rules out firmware issues.
I currently do not feel like removing the heatsink with all the thermal compound and trying to solder tiny smd components without even knowing which one failed. So I guess this is where this topic ends for now. Thank you for your support in verifying that it is unfortunately a hardware issue.
Maybe I will revisit this when I have more spare time and a replacement device. I guess you could use a functional port to identify the affected pair and then try to trace the signal through the pcb…
Or maybe a diode, to make PoE in one way only .
The problem should however be near the ether1, very likely no need to remove the heatsink, but surely re-soldering or replacing one of these tiny components Is anyway a challenge.
You are right, without a ready replacement it is risky.
even with the other device connected, what does the cable test pairs show then?
, one of the two wires on 3rd pair, either pin 4 or pin 5 doesn’t make contact at all (in the socket of the hap ax2 or however internally) so there is no link on Gb (only 7/8 wires connected) and PoE positive voltage flows through only one wire (as an example 4+,7,
instead of the normal 4,5+,7,8-).