TP-Link ER7412-M2 SFP WAN shows 0.235 mW RX on a BiDi optic but the link never comes up
Small office, we are dropping the ISP-supplied box and terminating the fibre straight on the router. The handoff is AON, the provider only asks for a VLAN tag, no PPPoE and no ONT registration. What is in the rack:
- TP-Link ER7412-M2 V1, firmware 1.0.1, fibre into port 11 (SFP WAN)
- a TP-Link TL-SM321B first, then an FS generic 1000BASE-BX-U (TX 1310, RX 1490, 10 km, DOM)
- a FRITZ!Box 5530 Fiber with its FRITZ!SFP AON module (1000BASE-BX10, TX 1310 nm, RX 1490/1550 nm) as the reference that works
- the same LC patch cord for every test
Both modules are seen by the router and the DDM page is not empty:
Port 11 (SFP WAN): link down
DDM: RX power 0.235 mWSo light is arriving with the right wavelength pairing, and the port still will not come up. What I have done so far:
- replaced the TL-SM321B with the FS BX-U precisely because its wavelengths match what the provider sends
- moved that FS module and the same patch cord into the FRITZ!Box, where the link comes up immediately - fibre, module and provider port are therefore all fine
- re-checked the VLAN tag on the WAN interface, although that should have nothing to do with the link itself
What else on the router side can hold a port down when RX power is clearly present?
Also available in: English
العربية
Deutsch
Español
Français
हिन्दी
Bahasa Indonesia
Italiano
日本語
한국어
Nederlands
Polski
Português
Русский
Türkçe
Tiếng Việt
简体中文
Comments 3
Take auto-negotiation out of the picture on that port. Set the SFP WAN port to a fixed 1000M instead of auto, save it, let the port re-initialise. That is the change TP-Link support points people to on this box and it is normally the entire fix: the port comes up and the WAN interface pulls an address from the provider straight away.
Mechanically what you are looking at is 1000BASE-X negotiation that never completes between the router port and the provider's AON side. Nothing is wrong with your optics, and the DDM reading is the proof of that - RX power at the expected wavelength means the fibre path and the BX pairing are healthy, so the failure lives on the electrical and negotiation side, not the optical one. The FRITZ!Box linking with the very same module says the same thing from the other direction.
Two caveats worth having in writing.
Fixing the speed on one end only helps when the far end is not still trying to negotiate. This is the same trap people hit with a 1G SFP sitting in an SFP+ cage, where the module is detected, no rx loss, no tx fault, and the interface stubbornly shows no link until speed and duplex are pinned by hand on both ends rather than just on the router.
And do not turn this into a universal rule for fibre ports. The opposite case exists: a Huawei AR2220 fibre interface that stayed down against a third-party switch until it was put back to
negotiation auto, because the peer expected negotiation and the router had it forced. Match the far end, do not guess at it.What is port 11 set to, auto or a fixed speed? And do you know how the far side is configured - is that provider port pinned at 1000M or left negotiating? Those two answers together decide whether 1000BASE-X negotiation has anybody to talk to at all.
One more sanity check while you are in the DDM page: does the RX reading fall to nothing when you pull the fibre out of the module? I have seen enough stale web UI values to want that confirmed before building a theory on 0.235 mW.
Port 11 was on auto. How the far side is configured I honestly cannot tell you - nobody on the provider's support line will discuss port settings with a small business customer, so pinned or negotiating is a guess either way. The DDM value is at least real: pull the fibre and the RX reading falls to nothing, put it back and 0.235 mW returns.
Fixed port 11 to 1000M, saved, and the link came up within seconds. The WAN interface pulled its address from the provider on the first attempt. Nothing else changed: same FS BX-U, same patch cord, same VLAN tag, firmware untouched.
So the optics were never the problem, the port was simply waiting on a negotiation that was never going to finish. Slightly annoyed that a DDM page showing healthy RX power sat next to a link-down status for that long without anything hinting at the speed setting.