Testing and fixing the cabling of a camera system at an industrial site
Live video minutes late, cameras dropping out. Wiring, instrument and software tests, a measurement report, then a cost-effective fix: a few cables replaced and the network redesigned.
Why it was needed
The live video of the site’s camera system lagged by minutes, and camera images disappeared at random. The question was whether the cabling met the requirements of a computer network, or the fault was somewhere else.
Testing the whole system would have taken several days, and that was not part of the job. So I picked three sections and we examined those in detail. The method and the measurement plan were mine. I can measure alone, but I usually don’t: there are many points to walk, and two people are much faster.
What we measured and how
Four questions: what cable is in the wall, how much the network lets through, how good the terminations are, and where the faults are.
- visual inspection and photos: cable type from the jacket print, terminations against the standard
- instrument test: CableMaster CM600, time-domain reflectometry (TDR) and capacitance, PoE (802.3af/at)
- software test: iperf between two laptops with gigabit cards, over UDP (cameras send video over UDP, so this is what counts)
- data collection: we watched the traffic on the cables over a longer period
What we found
There were termination faults in all three places: damaged cables on the patch panel, badly cut cable ends, pairs punched down with the wrong tool or method, and badly crimped plugs at the camera end.
Data traffic was fine on two sections (except one port), with minimal packet loss. On the third section communication was “hit and miss”: noise got onto the cable and disturbed the data stream.
Nearby metal objects and unshielded 230 V cables also came up. By the standard (EIA/TIA-569) there should be at least 12.7 cm between an unshielded power cable and UTP.
The result was a 20-page measurement report with the method, the results per section and a proposal.
The fix
I did not have the whole thing re-cabled. I planned the fix to be cost-effective:
- a few cables had to be pulled again, because unshielded, poor-quality cables ran next to high-power machines
- the rest came from redesigning the system; the key step was replacing the main distribution switch with a gigabit one
- I went through the branches, calculated the bandwidth each branch needs, and worked on the other branches based on that