We are using the SFP-10G-T on an NXP LX2160A platform.
The module successfully auto-negotiates with a 1G laptop and traffic passes correctly. However, the host always reports:
Port: FIBRE
Supported link modes: 10000baseSR/Full
Speed: 10000Mb/s
The SFP EEPROM also appears to advertise only 10G optical-style host capabilities. Is the SFP-10G-T designed to expose a fixed 10G host-side interface while performing copper-side auto-negotiation internally?
If so, is there any mechanism (MDIO, vendor-specific EEPROM page, diagnostics page, etc.) that allows a host system to retrieve the actual negotiated RJ45 speed (100M/1G/2.5G/5G/10G)?
Thank you.
Hi,
Yes, that behavior is expected for this class of 10GBASE-T SFP+ module.
The SFP-10G-T presents a fixed 10G host-side serial interface to the switch/MAC, while the internal copper PHY handles RJ45-side auto-negotiation independently. Because of that, the host reports the SFP port as a 10G fibre-style interface, for example 10000baseSR/Full or 10G full-duplex, even when the RJ45 side has negotiated 1G with the attached copper device.
The EEPROM advertisement reflects the host-side SFP+ interface rather than the copper-side link partner capability. This is normal for many 10GBASE-T SFP+ RJ45 modules and does not necessarily mean the copper side is running at 10G.
Regarding the actual negotiated RJ45 speed: there is no universal standards-based mechanism that all hosts can use to retrieve this from a 10GBASE-T SFP+ module.
Some modules expose copper-side status through vendor-specific I2C/EEPROM pages, diagnostics registers, or module-specific firmware behavior, but this is not guaranteed or portable across platforms. On many systems, ethtool will only show the host-side SFP+ link state, not the copper PHY’s negotiated RJ45 speed.
So in this case, the LX2160A reporting 10G/FIBRE while traffic passes correctly to a 1G laptop is consistent with the module operating as intended.
Best,
Josh