Re: C_COMP etc.

Ibisiens:

C_comp brings up some issues between measurements and spice generated data.

1. There are two main types of buffer modelling with different requirements.
One is the buffer designer who is designing buffers " in to out". This is the
type of person who may be extracting buffer info from spice type circuits. The second type is the person who extracts buffer model from measurements.

2. If the v-t is primarily performed by measurement, then the effective parasitics should include both the package and the die parasitic. In this case it makes sense to lump them into smallest set of efffective parameters like c_dut, l_dut etc. Then the v-t data is explicit with all the conditions of its generation specified. There are no implicit quantities like c_comp to account for.

3. Like John Powell points out implicit quantities like c_comp need to be considered only if the measurements are spicetype. In that case it can be excluded or can be explicitly specified by setting c_dut = c_comp for the measurement.

- kumar
Received on Thu Dec 8 07:31:13 1994