tskew and BIRD 6.X

Hello Fellow IBISans --

     While reading over the mail on BIRD 6.1 and the tskew specification
it occurs to me that we don't spec the much more general parameter of
skew between outputs. In fact, for an IC designed for clock driving, this
specification is CRITICAL. For this and other reasons (see below) I
belive a lot more more thought is required before lumping
in a tskew spec with BIRD 6 (I agree, however, that BIRD 6 is still needed for
specifing differential output associations). At the least, I propose a
separate bird that addresses the tskew specification in general; both for
differential outputs and output to output skew for clock drivers. I'd like to
kick off a separate discussion of tskew independent of BIRD 6. Some points to
address:

     1. Do others feel the output-to-output skew is important?

     1. Is an IBIS specification the appropriate place for including a
         parameter that goes beyond the basic I/O buffer and package
         characteristics? (Is this just like specifying the propagation
         delay of a device?)

     3. The tskew spec is dependent on the specific interaction of the
         packaging, internal xsistor delays, etc. Can this really be
         modeled accurately? For example, suppose the tskew spec for an
         IC is 1ns. Would simply delaying the xsistion of one waveform
         by 1ns result in the output waveforms crossing the logic threshold
         1ns apart after going thru the packaging LRCs? To put it another
         way, how does a simulator know when to start the two transitions
         so that they end up 'tskew'ns apart?

     Best Regards,
     Stephen Peters
     Intel Corp.
Received on Tue Feb 1 18:00:35 1994