[En-Nut-Discussion] PHY for ethernut boards

[Ulrich Prinz]

Hi William or Bill?,

the needed basis set of registers needed for communication is standard
in all PHY chips.
If it works out of the box depends on two things:
1) Correct resistors on off-cpu config pins at power on / reset
2) Correct resistors on pins connected to CPU.

The problem is, that some pins of these chips work at config inputs
during power-on or reset. If these are connected to the CPU and the
GPIOs haven't been initialized the config result of the PHY will be
random. You have two options, add the correct pull-ups / pull-downs to
these pins or you connect the reset line of the PHY to a separate GPIO
line ( there is an option to use the reset input of some ARMS as a reset
output too).

You can see examples of both ways in the driver of the sam7x EMAC.
There are many PHYs supported that do not require any extra setup as
they are used on boards where correctly placed resistors work together
with a correct reset sequence.
Then you have the AT91SAM7X-EK where it doesn't work and GPIOs have to
be configured an the reset is pulled.

Unfortunately PHY and EMAC is mixed in the code. So I started a rework
of that driver part to separate PHY and EMAC. So you can add a simple
standard EMAC driver to your setup and you can add a special init
routine to your PHY. To bad that I had a really big project going on in
the last months so I couldn't finish that part of the driver.

If you have time and a CortexM3 like STM32F1xx or STM32F2xx you are
welcomed to join and put the things straight.

If not, you're still welcomed and I try to support as good as I can.

Ulrich

Am 24.03.2011 18:36, schrieb William Basser:
> Has anyone looked at using a NSC DP83848 PHY in place of the Davicom PHY?
> How much modification would be required to support this?
> Or is there a supplier of Davicom parts, I have not been able to locate on
> in the US.
> 
> Bill
> _______________________________________________
> http://lists.egnite.de/mailman/listinfo/en-nut-discussion