Hi,
I wanted to share my experience with a hardware failure on my OpenWrt One after a thunderstorm, including a detailed diagnosis down to component level.
What happened
After a nearby lightning strike, I found my LAN port completely dead the next morning – no LEDs on either side, no link. The WAN port and WiFi were still working fine.
Diagnosis
Via serial console I confirmed that eth1 (the internal MT7981B GPHY) was still recognized by the kernel and TX-VCM calibration completed successfully, but carrier: 0, speed: -1, duplex: unknown – no link negotiation at all.
Using the official schematics from one.openwrt.org/hardware/ I traced the signal path and measured with a multimeter on the PHY side of the SG24002G transformer. I found a GND short on pins 2/3 (TD1+/TD1-). The transformer itself was intact (RJ45 side all high-impedance).
After desoldering R89 and R91 (the 1R series resistors), the short on the transformer side disappeared completely – confirming the fault is at the MT7981B SoC pins GBE_TXVP_A_P0 / GBE_TXVN_A_P0. The internal analog PHY driver has failed.
Additionally, the USB-A port is also dead – 5V present but no data (R57/R58 shorted to GND), which points to the USB PHY driver inside the same MT7981B SoC being damaged as well.
Root cause
The overvoltage likely entered via a LAN cable from the internal network. The critical finding: TVD5–TVD8 footprints (NC\PAM02SD2303C) on the board are not populated. The schematic clearly shows these TVS diodes were designed in but omitted, presumably for cost reasons. With these diodes populated, the damage would likely have been prevented.
The WAN port survived because it uses an external Airoha EN8811H PHY chip with its own protection structures, rather than the exposed internal GPHY.
Recommendation for other OpenWrt One users
If you live in a lightning-prone area, consider:
- Soldering PAM02SD2303C TVS diodes onto the TVD5–TVD8 footprints – the pads are there, just unpopulated
- Adding an Ethernet surge protector on any cables coming from outside
Hope this helps someone avoid the same fate.
What are my options?
I'm quite attached to this device and sad to see it partially dead. The digital parts still work fine – kernel recognizes the PHY, calibration runs, WiFi works, WAN works. Only the internal GPHY analog driver and USB PHY are gone.
I see a few options but would love to hear the community's thoughts:
- Buy a replacement board – I couldn't find the bare PCB available anywhere. If anyone knows a reliable source I would be very grateful
- USB-to-Ethernet adapter – Would restore LAN functionality, but the USB port is also dead, so this doesn't work in my case
- Keep it as a WiFi-only access point – WAN and WiFi still work, so it could serve as a dumb AP, but I also had my UPS connected via USB for NUT monitoring which was a nice extra feature – without USB that is no longer possible at this location
- Professional BGA rework – Replacing the MT7981B SoC would fix everything, but likely costs more than the device is worth
Has anyone else experienced similar damage on the OpenWrt One? And does anyone know of a reliable source for the bare board?
Thanks for reading this far – it was quite a journey tracing this fault today! 