Hi fellow engineers,
we are developing a board with RT1050 that uses the RNA physical chip instead than RNB version as in the EVKB.
So far we have had no success in making our board work. I hope someone has experience with this.
This is the scenario.
1) we modified the connections between the PHY and the MCU as shown in the image.
2) we are using the lwip_ping demo firmware that comes in MCUXpresso
3) the only change we did is setting BOARD_ENET0_PHY_ADDRESS to zero instead than 2. By doing so, the PHY recognizes a connection and negotiates a 100 Mbps link. The MCU correctly reads the status of the PHY.
We get traffic on the RX[0..1] lines, but TX_EN is low, and there is no traffic on TX[0..1].
We connected ETH.RST and ETH.INT to GPIOs. When the board starts, there is a cycle reset.
REFCLOCK gets a 25 MHz from the MCU (pin B13 configured as REF_CLK)
CRS_DV is connected to pin G2, configured as RX_EN like in the EVK board.
Any hints?
:-)
In the end I discovered the problem.
The hardware is perfectly OK, and you can use versione RNA of the physical with zero changes.
The problem was a settings in the reference clock pin. You have to enable the "Software Input On" flag in the pin configuration.
By default, this is set to Disabled. SDK projects override the value to Enabled, but when you relocate the signal to a different pin (this was our case) you lose the override.
So, caveat emptor!
Once the flag is set, the lwip examples work with no problems.
Giuseppe
In the end I discovered the problem.
The hardware is perfectly OK, and you can use versione RNA of the physical with zero changes.
The problem was a settings in the reference clock pin. You have to enable the "Software Input On" flag in the pin configuration.
By default, this is set to Disabled. SDK projects override the value to Enabled, but when you relocate the signal to a different pin (this was our case) you lose the override.
So, caveat emptor!
Once the flag is set, the lwip examples work with no problems.
Giuseppe
