I have a generic question about the IBIS model C_Comp value compared to the data sheet C_In value. I'm asking because I've found your IBIS models to be of consistently high quality, but one thing I've noticed (from many vendors, and across many ICs, even from TI) is that the data sheet C_In value generally seems to be as much as twice as large as the IBIS model C_Comp (plus C_pkg or C_pin as appropriate) value, and I'm trying to understand this.
For example, on the SN74LV1T34 buffer, the data sheet gives C_in as 2pF typ, 10pF max for VCC=3.3V. But the IBIS model for the same VCC gives C_Comp as 0.677pF typ, 0.6725pF min, and 0.6814pF max, with C_pkg of 0.26256pF typ, 0.24997pF min, 0.27952pF max, and C_pin of 0.2582pF for the DBV package, with the DCK package being slightly lower. So that adds up to less than 1pF, either typ or max. I understand that data sheets can be guard-banded, but a 10x factor seems a little much. (And this is just one example, most every other LVCMOS part I've looked at has a similar discrepancy.)
The reason that I ask, is sometimes I have to create an IBIS model for an input when the vendor does not provide one. I can use package parasitics from a similar package, as these are usually small and don't have a significant effect, but the C_Comp value? How does it relate to the data sheet C_in value? So, again for example, if you didn't provide an IBIS model for the LV1T34, and I had to create one from the data sheet, I would guess at C_pkg of 0.25pF, and C_comp of 1.5pF type, 1.0pF min, and perhaps 5pF max. Which would be wildly high compared to the less-than-0.75pF values in the IBIS model you supply.
Can you help me understand this? Given the wide discrepancy that I see across multiple parts from multiple vendors, I think I'm overlooking something. Thanks.
