Here (preliminary) design decisions are documented, rationale are given as footnotes

Table of Contents Model compiler ICIM-CI Open Questions Rationale

• Don't make conversion calculations in the compiler tool ^[1]

(These remarks are non-normative, refer to IEC standards for production use.)

• An ICIM-CI shall consist of exactly one PDN. ^[2]
• The PDN may be sparse, i.e. in an S-parameter matrix, not all incident-reflected pairs need to be defined. ^[3]
• The frequency range of each disturbance input of the IB must be included in the frequency range of the PDN. ^[4]
• An ICIM-CI shall consist of one or more IBs. Every IB (^[5]

• Should we specify the frequency range of validity of the ICIM-CI?

1. ^ Because the impedance (e.g. S11) may be sampled at different frequencies than the immunity (e.g. DPI) This moves the interpolation problem to the builder tool, but at least we don't lose information in the compiler phase.
2. ^ Because different IBs may share the same PDN.
3. ^ The model author may know that the crosstalk between different parts of the IC is negligible, so he needn't measure them.
4. ^ For each frequency point for each disturbance input of the IB, the impedance must be known.
5. ^ ...