part1.md

The first step I had to take was design a DRC clean inductor in the SKY130 process. Rather than doing it by hand I used openPCells to generate the design. It has a simple interface and ready made examples for spiral inductors of different shapes. I chose the square inductor with conductor width 6um, spacing 6um and three turns. A small modification of example to place the ports where I wanted them and a run through Klayout to fix a polygon bug resulted in the following design:

./opc -T skywater130 -I gds -C spiral_inductor --turns=3 --width=6 --spacing=6 --innerradius=25 --extension=25

git clone https://github.com/ediloren/FastHenry2.git
cd FastHenry2
make all

export PATH=$(pwd)/bin:$PATH

To generate the FH2 input files, thankfully OPC has a debug interface that can be (ab)used to output an FH2 description.

vizualization

fasthenry square_spiral_inductor_6_6_25.fh -f refined && zbuf zbuffile2 && evince zbuffile2.ps

simplified analytical calculation

set expectation to check them in the tools

calculating R from length to first order

calculating L using many of the available formulae

single port extraction of a toy ind

(universe with only an inductor) graph with matplotlib in svg?

fasthenry square_spiral_inductor_6_6_25.fh && ReadOutput Zc.mat

recovering skin effect with logarithmic subdivision

compare with graph above draw trendline for skin effect dependence

introducing uniform ground plain (starting to make sense)

describe how slow it is with this many elements and introduce nonuniform ground

two port extraction wrt ground

show the contribution of low silicon conductivity

model export

show how to turn the output file into a spice netlist with mutual coupling setup the intuitive explanation for self resonance frequency and need for fastercap

automation

show iterate_fh.py