Loose RJ45 Connectors

[ali987]

I have two separate PCB boards communicating data with each other. I soldered a female RJ45 PCB mount connector (unshielded) on each board so they can be connected by a standard ethernet cable. The problem is that the connection is lost whenever I wiggle the cable. Connectors and cables are brand new from digikey an they are not among the cheapest items. Any experience or advice on solving this would be highly appreciated.

 

[KlausST]

Hi,

there could be many issues.

If it would be my task I´d surely spend some time to find out what causes the problem.
* Replacing the cable
* wiggeling one side only .. then the other only
* using heat gun / cold spray
* use lense / microscope to inspect cable, connectors, solder joints, PCB traces

....
We can´t really to something for you, sinc you provide no information we can validate. (Connector type and brand, cable, photos of the whole setup, or solder or contact details...
Links to datasheets..

We even don´t have a proof that there is a problem at all.

Klaus

 

[barry]

if the connectors and cable are new and not defective, i would investigate the pcb soldering of the connector

 

[spike45]

Hi,

It looks like there could be a loose connection or a faulty cable causing the issue. Try these steps to fix it:

1. Replace the cable – Use a different one to check if the problem persists.
2. Check connectors – Inspect both ends carefully for dust, damage, or loose connections.
3. Test continuity – Use a multimeter to see if the cable is transmitting signals properly.
4. Secure the connections – If there’s any movement, try stabilizing the connector with proper support.
5. Inspect for damage – Look at the PCB, solder joints, and traces under a magnifying glass or microscope.
6. Use a heat gun or cold spray – This can help identify temperature-related faults in the components.

If none of these work, providing details like connector type, brand, and setup photos would make troubleshooting easier.

[Moderator action: deleted text with non-contributing link]

 

[dick_freebird]

The connectors "should" pretty snug. Then the gold plated
spring contacts "should" maintain adequate low contact
resistance.

Now some of this behavior may not be totally the
connectors' fault (the hoser doing the tugging, maybe
should be given something else to do). It might point
to the value of strain reliefs or zip-ties. Why does
"line noise" result in a connection-drop rather than a
"resend packet" in this system? That's bad above-PHY
design choices (like assuming all environmental aspects
are benign) in my partially-informed opinion.

But robust connector jack selection and cable strain
management is a PHY-layer design choice, albeit
unglamorous. How high does your purview go?

Pushing in too much noise (which discontinuity will
do) can really mess with clock / data recovery which is
key to many high speed serial standards, and clock
recovery PLL has its own loop time, which could have
been set super slow to get BER minimized (perhaps way
more than necessary) and you're just stonewalled until
lock-detect comes back.