TRENDnet TFC-1000MSC pair shows no Link LED after reversing TX and RX at both ends
Putting a fibre run between two comms rooms in the same building and using a pair of TRENDnet TFC-1000MSC converters (1000BASE-T to 1000BASE-SX) to bridge it. Copper side goes straight into the switches we already have, nothing unusual in the config.
- TRENDnet TFC-1000MSC at each end, each on its own power supply
- multimode ST to SC patcch cords onto the house fibre
- Catalyst 2960 in one room, a Netgear ProSafe in the other
- two runs available in the same bundle
The instruction sheet says to reverse TX and RX, so I swapped the strands at both ends. What I get:
Converter A: PWR on, TP Link on, FX Link off
Converter B: PWR on, TP Link on, FX Link offNo Link/Act on the fibre side of either unit. Copper side is up on both, the switch ports come up and forward into nothing.
Already done: swapped the SC patch cords for spares, moved everything onto the second pair of converters we bought, reseated both ends twice, and tried the second run in the bundle - same result there.
What the front panel offers on the fibre side is one FX lamp per unit, and both are dark. That tells me the optical side never came up and nothing beyond it - not whether it is my own transmitter that is silent or the one at the far end, not which direction is broken. Where should I be looking - the fibre, the converters, or my own wiring?
Comments 4
Reversing at both ends is your fault right there. The crossover belongs at one end only - TX on one converter has to land on RX at the other. Swap the strands at both ends and the mapping is back to straight-through, transmitter facing transmitter, so no receiver ever sees light and neither Link LED comes up. Put one end back the way it was and leave the other crossed.
If it stays dark after that, two more things bit people with these exact units. DIP switches: the recommendation from TRENDnet support was switch 1 on for forced mode and switch 2 off to disable link loss return, and both converters need to be set the same way before you conclude anything from the LEDs. And the fibre itself: on one of two runs in a case like yours the termination turned out to be bad and had to be tested and redone properly. Since your second run behaves identically, I would still bet on the TX/RX mapping first, but once the mapping is right and it will not link, that is a job for test gear rather than another guess.
Two side notes. A dirty ferrule on its own will take a gigabit link down, so clean the ends before you start swapping hardware around. And never look into a fibre end to check for light - the source is an infrared laser and you would not see it anyway.
You say you reversed at both ends, and that sentence is doing a lot of work. Spell it out: take one converter and tell us which strand goes into its TX port and which into RX, then do the same for the far end. Draw it if it helps.
Two more things while you are down there. Read out the DIP switches on both unitts, position by position, and say whether either of those runs has ever carried a link before or whether the whole bundle is a fresh install. If it is fresh, the fibre is unproven and stays a suspect until somebody tests it properly.
Seconding the complaint about that FX lamp, because it is the real reason this kind of fault drags on. A lamp that is merely lit or dark cannot tell you which direction is dead, so stop reading the front panel as soon as there is anything in the path that can be asked directly - switch port counters, or a CLI if the box has one. On a carrier handoff I stopped guessing from the LEDs entirely and read the operational speed instead:
and looked at the Oper column. That also catches the trap where a 100 Mbps service feeds a converter that hands out 1000BASE-T on the copper side: everything behind it negotiates a gigabit quite happily and nothing on the panel hints at the real rate. Different box, same lesson - the converter is the least informative thing in the chain.
Adjacent case, in case your lamps ever start flickering instead of staying dark. Mine was an SG300-28PP with a Cisco MGBSX1 in SFP port 27, multimode across to a TP-Link MC200CM. The converter's FX indicator would come on for a few seconds and then break into a fat flicker; the machine behind it pulled an address, dropped off again within seconds, and the throughput left over could be measured with a ruler. I went through the hardware piece by piece - the second MGBSX1, the other SFP port, the spare converter on its own power brick, a handful of strands, fresh copper patch cords - and none of it moved the needle. Meanwhile SG300 to SG300 over fibre held, SG300 to an HP held, and the two converters back to back held, so on its own every component looked innocent.
What came back form the vendor was to put the link mode on auto and reboot, plus a warning that an MC200CM only ever drives its copper side at 1000 Mbps, so anything slower behind it is a non-starter. Nobody spelled out whether auto meant the converter or the switch port, and the discussion died before anyone confirmed a result - a direction to look in, not a fix.