Tse Mac Sgmii Pcs

Altera_Forum · ‎07-23-2009

Has anyone used just the 1000BASE-X/SGMII PCS variant (no MAC) of the TSE megafunction? What I'm trying to figure out is how to actually connect the GMII/MII interface to a seperate MAC.

On the surface it looks wonderful. The only problem is that the MII interface of the PCS function only provides a 125MHz rx_clk and tx_clk along with rx_clkena and tx_clkena signals. A real MII MAC requires a 25MHz or 2.5MHz clock. My assumption is that the clock enable signals are constantly asserted during GMII operation and asserted every 5th or 50th clock during MII operation.

So it would seem that you'd have to derive your own clock if you want to connect it to any other standard MAC. Seems pretty useless to have a non-compliant MII/GMII interface. I suppose you could create a gated clock.

Any ideas?

Jake

Altera_Forum · ‎09-24-2009

Is usually the standard MAC come with the clock enable signal? If yes, then I suppose that you don’t need to derive the clocks. It looks like the benefit of doing this is to save the clock network resources in the FPGA (e.g. using the clock enable without deriving the clocks).

Btw, instead of using gated clock to derive the clock, perhaps using PLL or some other synchronous methods that will be better. My 2 cents.