Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [8023-10GEPON] Question about dual MII solution



Title: [8023-10GEPON] Question about dual MII solution
Marek,
My guess (and only a guess) is that a dual-rate TIA could be designed that was optimized for 10G and which is sub-optimal by one or two dBs in 1G operation.
  Thanks, Brian H.
 

_______________________________________________
Brian Holden        PMC-Sierra, Inc.
3975 Freedom Circle, Santa Clara  CA  USA
+1.408.239.8123   Fax +1.408.492.9462 
brian_holden@pmc-sierra.com   http://www.pmc-sierra.com   

-----Original Message-----
From: Hajduczenia, Marek [mailto:marek.hajduczenia@siemens.com]
Sent: Tuesday, January 16, 2007 6:41 AM
To: Brian Holden; STDS-802-3-10GEPON@listserv.ieee.org
Subject: RE: [8023-10GEPON] Question about dual MII solution

Dear Brian,
the real question is whether the solution 2a is viable without the need to go into two discrete TIAs and having to switch between them in the anticipation of the incoming burst, since then the same problem as in point 1 arises I believe. You would have to switch off one of the TIAs in advance. Additionally, there is a problem of reduced signal amplitude versus noise background that Frank was addressed if the electrical signal needs to be separated into two paths before TIA.
I do not know whether anybody in our group has sufficient expertise in the design of TIAs to state clearly that such dual data rate TIA design is feasible at first and can be cost effective as a second requirement. What do You think about it ?
 
Best wishes
Marek


De: Brian Holden [mailto:Brian_Holden@PMC-SIERRA.COM]
Enviada: seg 1/15/2007 7:27
Para: STDS-802-3-10GEPON@listserv.ieee.org
Assunto: [8023-10GEPON] Question about dual MII solution

All,
During the break, I was suggesting the following to several folks:

Question - Is the assertion that Frank E. made that the initial Trans-Impedance Amplifier needs to know what rate is coming to be configured properly true?

   1)  if true, then one possible solution may be to
        a) set up a schedule for the receiver via the XGMII management interface (the encoding may need to be clever to fit into the unused registers)
            b) the latency properties of that management interface may need to be constrained to allow the schedule setup in real-time

   2) if false, then the dual MII solution is likely workable
            a)  presumably the solution would involve some sort of "dual Trans-Impedance Amplifier"
            b)  a likely compromise is that some additional loss over what would be needed for bandwidth optimized designs may need to be budgeted for

  Brian Holden
  PMC-Sierra

_______________________________________________
Brian Holden        PMC-Sierra, Inc.
3975 Freedom Circle, Santa Clara  CA  USA
+1.408.239.8123   Fax +1.408.492.9462 
brian_holden@pmc-sierra.com   http://www.pmc-sierra.com