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

Re: XAUI experts: Another XAUI issue that crossed my mind - help




Benny,
      One of the key reasons for using 8B:10B encoding is to ensure DC
balance.  The encoder should ensure that the scenario you described
never occurs.  Each 8-bit data value has two possible 10-bit symbols,
one with a majority of 1s, the other with a majority of zeros.  The
algorithm choses which to use in order to keep the total number of ones
and zeros transmitted equal.


"Christensen, Benny" wrote:
> 
> Hi XAUI people
> 
> Here is another problem that came across my mind yesterday.
> 
> I was looking at Mysticom's eye diagram simulations and all see quite
> realistic to me with the given set of parameters. However, I  assume that
> this has be done with a 50 % transition density signal. (i.e. overall DC
> balanced)
> 
> I wonder if anybody have taken this example into consideration and if my
> assumptions are valid:
> 
> I have been told that the worst case relative DC voltage (unbalance) could
> be 0.3. I.e. let's assume a 8B10B character (I don't know if it is a valid
> character, but I guess one can find a similar valid example): 01000 11000
> 
> If this character is  repeated infinitely (or at least characters with the 3
> to 7 mark/space ratio) and when the signal is AC coupled (as defined by the
> XAUI channel), this will cause the differential signal to look like this:
> 
>   _   __
> == ===  ===
> ==_===__===
> 
> 00100011000
> 
> ie. the differential swing after the AC coupling has a reduced swing for the
> '0' symbols. In other words: the '0' has a single-ended relative amplitude
> of -0.3 ant the '1' has +0.7 (ie. relative to the DC level of the signal and
> due to the AC coupling). Now the complementary(inverted) signal has the same
> properties ie. its DC values is -.3 and the amplitude of the inverted '0' is
> +0.3 and the amplitude of the '1' is -0.7. So the differential AC coupled
> signal swing at the input of the CDR (or LIA) is  0.6 for the '0' and 1.4
> for '1' symbols. Hey, that is quite amplitude modulation in relation to a DC
> balanced or DC coupled signal.
> 
> This is all due to the AC coupling. For DC coupling the amplitude of all
> signals are 0.5 and the differential swing is 1 for both symbols.
> 
> What is wrong with my conclusion (if it is wrong - I don't think so).
> 
> Benny
> --------------------------------------------
>   llllllll   ii     llllllll     llllll
> ll                ll           ll      ll
> ll    llll   ll   ll    llll   llllllllll
> ll      ll   ll   ll      ll   ll      ll
>   llllllll   ll     llllllll   ll      ll
> 
> GIGA, an Intel company
> Benny Christensen, M.Sc.E.E, Ph.D.
> Mileparken 22, DK-2740 Skovlunde, Denmark
> Tel: +45 7010 1062, Fax: +45 7010 1063
> e-mail: benny.christensen@intel.com, http://www.giga.dk

-- 
-----------------------------------------------------------------------
Iain Robertson SMTS (mailto:i-robertson@ti.com)
UK Design Centre, Texas Instruments, Northampton, UK
Phone: 3412; Fax: 3456; Prefix: +44 1604 66 (Speed: +8 447)