Hi,

As part of a design that I'm investigating, I am looking at connecting
a number of FPGA based MACs to a dedicated broadcom switch chip.  The
difficulty with this is that to interface both chips, I need to have
two closely coupled and redundant PHY .

Is it possible to connect two MAC (SGMII) directly point-to-point,
bypassing the PHY completely? Or is it necessary to have at least some
PHY functionality, even if it is a point to point link?

Kind regards,

Stephen

Steve wrote:
> Hi,
> 
> As part of a design that I'm investigating, I am looking at connecting
> a number of FPGA based MACs to a dedicated broadcom switch chip.  The
> difficulty with this is that to interface both chips, I need to have
> two closely coupled and redundant PHY .
> 
> Is it possible to connect two MAC (SGMII) directly point-to-point,
> bypassing the PHY completely? Or is it necessary to have at least some
> PHY functionality, even if it is a point to point link?

I suppose it works.

Cross connect RXD[] to TXD[]. Cross connect RX_DV to TX_EN, and provide 
a 125 MHz clock (could be supplied from another FPGA pin)

On Feb 12, 9:01=A0pm, Steve <stephe...@gmail.com> wrote:
> Hi,
>
> As part of a design that I'm investigating, I am looking at connecting
> a number of FPGA based MACs to a dedicated broadcom switch chip. =A0The
> difficulty with this is that to interface both chips, I need to have
> two closely coupled and redundant PHY .
>
> Is it possible to connect two MAC (SGMII) directly point-to-point,
> bypassing the PHY completely? Or is it necessary to have at least some
> PHY functionality, even if it is a point to point link?
>
Generally,
It should be possible if you have your MAC with some basic PHY within
it.
Some PHYs are not replacable and theire requirement is a must.
I think, PHYs cannot be bypassed so easily.

Karthik Balaguru