[U-Boot] ITEAD Board (A20) fails to communicate on boot unless UART is connected

[U-Boot] ITEAD Board (A20) fails to communicate on boot unless UART is connected

[Kevin Welsh]

I'm trying to resolve a strange networking issue with the ITEAD Core EVB + AW2041 module.

If I connect all 3 UART wires (GND, RCV & TX) between the board and a USB UART device: I get an IP 100% of the time.
If I connect only TX & RCV: I get an IP 100%.
If I connect TX & GND, I get an IP 100%.

But, if I only connect RCV and GND... I do not get an IP.

Using WireShark on the DHCP server, I see no traffic is generated by the device.

If I connect the UART after booting, it will not work, even if I send DHCPCD commands, or reboot.

The only thing that seems to work is if I apply power while the UART is connected.

[U-Boot] ITEAD Board (A20) fails to communicate on boot unless UART is connected

[Hans de Goede]

Hi,

On 08-04-16 16:54, Kevin Welsh wrote:
> I'm trying to resolve a strange networking issue with the ITEAD Core EVB + AW2041 module.
>
> If I connect all 3 UART wires (GND, RCV & TX) between the board and a USB UART device: I get an IP 100% of the time.
> If I connect only TX & RCV: I get an IP 100%.
> If I connect TX & GND, I get an IP 100%.
>
> But, if I only connect RCV and GND... I do not get an IP.
>
> Using WireShark on the DHCP server, I see no traffic is generated by the device.
>
> If I connect the UART after booting, it will not work, even if I send DHCPCD commands, or reboot.
>
> The only thing that seems to work is if I apply power while the UART is connected.

If you connect the uart after boot, do you get a console ? Since you say:

"it will not work, even if I send DHCPCD commands, or reboot"

This suggests that you do get a serial console, is that correct ?

Are you in u-boot at that time, or did the board actually boot all the way
into Linux ?

What seems to be happening is that the board is drawing a trickle of power
from the tx line, this is know to happen on various boards. Usually this
may result in the axp209 communication not working (the axp209's i2c
communication dies until a full power cycle is done).

What you seem to have is the problem the other way around, that the axp209
will get stuck if it is not "powered" via the tx first.

Assuming the problem is the axp209 getting in this stuck state, you
should see an error:

"Failed to set core voltage! Can't set CPU frequency"

Really early on (from u-boot) when you reboot. If the axp209 is in this
state the only known way to get it out of it is a full power cycle
(often including unplugging the usb uart lines).

But as said first see if you get the:

"Failed to set core voltage! Can't set CPU frequency"

Error (right after the SPL prints the memory size). If you do this is
the problem, if you do not then we've some other problem.

Regards,

Hans