RJ45 LED Debug issue

Hi,
How to debug the RJ45 LED issue? The LED should be Green at 100M and Orange at 1G, but it's always shown kind of yellow at any speed (Both LEDs could light up together). LED nets are individuals from PHY through an inverter to a connector. The schematic should be correct. So, it might PHY's issue(MDIO's issue?)?

From PHY
Through an inverter

ETH_PHY_LED1_N and ETH_PHY_LED2_N connect to a LED in a connector.

Bothe LEDs should not share a common series resistor. Typically the PHY can drive LEDs directly without an inverter. Without the PHY datasheet we can't check for circuit correctness.

FvM said:

Bothe LEDs should not share a common series resistor. Typically the PHY can drive LEDs directly without an inverter. Without the PHY datasheet we can't check for circuit correctness.

Click to expand...

Thanks for your reply. The datasheet https://datasheet.lcsc.com/szlcsc/Realtek-Semicon-RTL8211EG-VB-CG_C69264.pdf
It seems the circuit is no problem, since it can work at any speed. It could be a driver issue that needs to modify the MDIO on BMC (controller).

If you read the datasheet thorougly, you'll realize that the LED ports need pins strap resistors that are sampled on power on reset to set the LED actice state. No resistors shown in your schematic snippets.

FvM said:

If you read the datasheet thorougly, you'll realize that the LED ports need pins strap resistors that are sampled on power on reset to set the LED actice state. No resistors shown in your schematic snippets.

Click to expand...

Thanks for your guidance. There are pull-down resistors SR6151 and SR6191 (sorry for missing), but there is another pull-up resistor R3181. Not sure if R3181 causes the issue and what the function of inverter is.

Enshuo said:

Thanks for your guidance. There are pull-down resistors SR6151 and SR6191 (sorry for the missing), but there is another pull-up resistor R3181. Not sure if R3181 causes the issue and what the function of inverter is.
View attachment 163151

View attachment 163152

Click to expand...

Studied the datasheet again. Found the answers. It looks the inverter should be for changing to active low, but not sure why ETH_PHY_LED1/2 just pull-up at first place.