[10GBASE-CX4] RE: Useful Spreadsheets (Signal Detect-IPG & pre-emphasis-conversion)
One at a time; the second...
Peter Bradshaw
> -----Original Message-----
> From: Peter Bradshaw
> Sent: Thursday, February 13, 2003 6:50 PM
> To: '10GBASE-CX4 (E-mail)'
> Subject: Useful Spreadsheets (Signal Detect-IPG &
> pre-emphasis-conversion)
>
>
> Apparently original did not get loaded on the reflector (too big)
>
> Am trying again
>
> > -----Original Message-----
> > From: Peter Bradshaw
> > Sent: Tuesday, February 11, 2003 8:29 PM
> > To: 'DOVE,DANIEL J (HP-Roseville,ex1)'; 10GBASE-CX4 (E-mail)
> > Subject: Useful Spreadsheets (Signal Detect-IPG &
> > pre-emphasis-conversion)
> >
> >
> > I am attaching the latest update of the 10-bit patterns to be
> > expected, w.r.t. the SIGNAL_DETECT function, as I offered Dan
> > today. Beside each three-symbol set I have added the number
> > of run lengths of >2 (i.e. 3, 4 & 5 bits), of 2 (bits), and
> > of <2 (i.e. 1 bit) in each pattern. It is assumed that the
> > bits on either side are different from those at the end of
> > each pattern, which is true roughly 50% of the time. The
> > counts were done by hand, so feel free (encouraged) to check
> > the values.
> >
> > The patterns covered are : those at the beginning of each
> > IPG; other patterns within IDLE in an IPG; the ||LF|| and
> > ||RF|| Sequence Ordered_Sets (the Signal Ordered_Sets are
> > similar); the patterns at beginning of a frame (the first
> > data byte is assumed to be B5, worst case in this situation
> > for SIGNAL_DETECT, though fairly unlikely as a destination
> > address, since it would be a group or broadcast address). The
> > conclusion is that any IPG must start with at least 4 runs
> > (and they are all of 4 or more) in three symbols, and any
> > group of three symbols within an IPG must have at least 3
> > runs of 3 or more, EXCEPT a repeated ||RF||, which will have
> > just one run of 4, but 7 runs of 2. Since no more than 2
> > ||RF||s may be generated, and ||RF||s may only be deleted for
> > clock compensation (like ||R|| they are disparity-neutral),
> > not added, adding 2 characters to the maximum allowed packet
> > that SIGNAL_DETECT must survive without above-threshold
> > signals would cover the ||A|| ||RF|| ||RF|| ||S|| ||Dp/Ds||
> > ||B5h||*n worst case scenario. (Excel only version; will
> upload PDF later)
> >
> > I have also appended an updated/improved version of the
> > spreadsheet I offered once before, allowing easy conversion
> > between the various definitions in use for pre-emphasis. I
> > have created a calculation sheet, where one may enter a value
> > in any one form (alpha, CX4-spec, Vhi/Vlo, etc) and get the
> > corresponding value in any other form. The sheet has three
> > rows for each calculation type, so that one may put in a min,
> > typ, and max set, for example, and look at all three at once.
> > (If you use it, be careful not to enter any values into
> > non-green cells, many of them contain formulae; some
> > calculated cells depend on other calculated cells for the
> > correct result, so stepping on one cell may wreck much more).
> > The currently-entered values reflect some plausible ranges,
> > but only the 'From CX4-Spec Value' section reflects any
> > proposed spec range. I also added some more charts between
> > the various definitions. (Excel version).
> >
> > Peter Bradshaw
> > Applications Manager
> > BitBlitz Communications
> > +1 408-597-8222 Direct
> > +1 408-480-0142 Cell
> > pbradshaw@bitblitz.com e-mail
> > pdbradshaw@tmomail.net SMS messages (<140 chars)
> > www.bitblitz.com
> >
> >
>
Preemph_Defn.xls