Course Plan

The course runs as a set-up week and fourteen weeks of about six hours each: one lecture-sized block of reading, a lab, the experiments at the end of the chapter, and a checkpoint that reproduces a number of the chapter, by hand or with your own run. Every week designs or checks a piece of one radio, the course’s 2.4 GHz transceiver, and the last two weeks put the pieces together against the budget of Chapter 5. Instructors can use the plan as a syllabus; self-learners can use it as a pace.

Before you start. The course assumes the analog design background of SkillSurf’s The Art of Analog IC Design: transistor models, gm/ID sizing, amplifiers, feedback and noise. The Preface lists the rest, and Part IV continues where that course’s PLL chapter stops.

Taking it as a class. The course is planned to run as a class in 2027. The draft is open now, so you can study it on your own at this pace in the meantime.

How a week works. Lab lines name the hands-on labs. Practice lines name the experiments in the chapter’s Try it box, extra work run with the chapter’s own script; weeks 2, 3 and 10 have no lab and use them alone. Checkpoint lines name a number or a table of the chapter that your run must reproduce within the stated tolerance. Every script lives in the course files’ tools/figs/ folder and writes the numbers it quotes to tools/figs/chNN_numbers.json, so a checkpoint can be compared figure by figure. Most scripts redraw from cached simulations in a minute; --resim reruns the simulations, which takes from seconds to an hour (each Try it box gives the time).

Week 0 · Set-up and a first S-parameter

Read: the Preface, Chapter 1 and Chapter 2: what makes RF different, the course radio, and what each simulator does.

Lab: Lab 0: install IIC-OSIC-TOOLS 2026.10 (the Preface and Chapter 2 describe the image), put the course files in place, run Chapter 1’s script and a first S-parameter simulation.

Practice: the experiments of Chapter 1 (tools/figs/ch01_why_rf.py: the nMOS at 0.1 and 0.4 mA/µm, an eight-emitter HBT, SKY130 gates contacted on one side) and Chapter 2 (tools/figs/ch02_rfsim.py: a 2 kΩ bias resistor, a 50 ps transient step, three harmonics in harmonic balance); the ngspice and Xyce sections of Appendix B.

Checkpoint: python3 tools/figs/ch01_why_rf.py gain --resim reproduces the fT of Table 1.3 within 1 % (80.8 GHz for the SG13G2 nMOS, 353 GHz for the npn13G2, 63.5 GHz for the SKY130 nMOS), and Chapter 2’s small part gives Table 2.2’s \(\vert S_{21}\vert\) = 18.355 dB and NF = 3.569 dB at 2.44 GHz within 0.01 dB.

Week 1 · Transmission lines, the Smith chart and S-parameters

Read: Chapter 3.

Lab: Lab 1: lines, the Smith chart and de-embedding; its matching part follows Chapter 8 in Week 5.

Practice: Chapter 3’s experiments with tools/figs/ch03_sparams.py: a higher-Q load (0.2 − j1.0), a 500 Ω stabilizing resistor, and pads without the cd capacitance.

Checkpoint: the SG13G2 nMOS of Tables 3.3 and 3.4, each number within 1 %: K = 0.108, μ = 0.151 and MSG = 16.7 dB at 2.44 GHz; fT = 45.8 GHz bare, 26.4 GHz through the pads and 45.9 GHz after open–short de-embedding.

Week 2 · Noise and linearity

Read: Chapter 4.

Practice: Chapter 4’s experiments with tools/figs/ch04_noise_linearity.py: RD = 1000 Ω in the Friis cascade, a 100 MHz tone spacing for the HBT’s IIP3 (hb --resim, about half an hour), and a 20 dB LNA in the receiver budget; the Xyce section of Appendix B for harmonic balance.

Checkpoint: Table 4.2’s cascade by hand with Eqs. (4.11) and (4.24): NF 6.06 dB within 0.01 dB and IIP3 −12.7 dBm within 0.1 dB; and the simulated NF of the CS stage in Table 4.1, 6.22 dB, within 0.05 dB.

Week 3 · Modulation and transceiver architectures

Read: Chapter 5.

Practice: Chapter 5’s experiments with tools/figs/ch05_architectures.py: an LO of −100 dBc/Hz at 1 MHz in the BLE tests, a second-order complex filter, and a 5 % error in the Hartley resistor (--resim reruns everything in about ten minutes).

Checkpoint: the receiver row of Table 5.3 by hand, each within 0.1 dB: NF 6.06 dB gives a sensitivity of −95.6 dBm, and 0.2 dB of gain error with 2° of phase error give an IRR of 33.6 dB (Eq. 5.14).

Week 4 · Transistors at radio frequencies

Read: Chapter 6, with Appendix C as a reference.

Lab: Lab 2: fT, fmax, stability and noise parameters of the SG13G2 transistors.

Practice: Chapter 6’s experiments with tools/figs/ch06_rf_transistors.py: the nMOS drawn as one cell, SKY130 without its contact resistance, and the npn13G2 at VCE = 1.0 V.

Checkpoint: rerun mos hbt --resim: the peaks of Table 6.2 within 1 %, fT = 75.7 GHz at 0.336 mA/µm and fmax = 89.7 GHz for the SG13G2 lv nMOS, fT = 353 GHz and fmax = 456 GHz for the npn13G2.

Week 5 · On-chip passives and impedance matching

Read: Chapter 7 and Chapter 8.

Lab: Lab 3: an SG13G2 spiral in Palace, its Q and its self-resonance (the vector-fitting step waits for Chapter 18, in Week 12); then the matching part of Lab 1.

Practice: the experiments of Chapter 7 (tools/figs/ch07_passives.py and the Palace runs in tools/em/ch07_inductors/: a 20 µm capacitor feed, a wider spiral, a halved mesh) and Chapter 8 (tools/figs/ch08_matching.py: VCC = 1.0 V, the trap moved to the drain, a third resonator against the Bode–Fano limit); the openEMS and Palace section of Appendix B.

Checkpoint: inductor L2 of Table 7.2 from your own Palace run, 0.9 nH at 1 GHz and Q = 8.91 at 2.44 GHz, within 5 %; and the loss of Table 8.1’s PA L-section (10.6 → 50 Ω with SG13G2 spirals), 0.892 dB, within 0.05 dB.

Week 6 · Low-noise amplifiers

Read: Chapter 9.

Lab: Lab 4: the course LNA at 2.4 GHz.

Practice: Chapter 9’s experiments with tools/figs/ch09_lna.py: a Q = 25 gate inductor, a 192 µm input device, and rf = 400 Ω in the wideband LNA.

Checkpoint: rerun design hb corners --resim: Table 9.4’s band-minimum gain (17.43 dB), band-maximum NF (1.18 dB) and band-maximum S11 (−15.39 dB) within 0.05 dB, IIP3 (−4.9 dBm) within 0.2 dB, and every corner of Table 9.3 inside the specification.

Week 7 · Mixers

Read: Chapter 10.

Lab: Lab 5: the course I/Q mixer.

Practice: Chapter 10’s experiments with tools/figs/ch10_mixers.py and its periodic small-signal module tools/figs/_ch10_lptv.py, among them four-cell switches, a 2 % switch mismatch, and lptv() swept over RF and the LO edge time.

Checkpoint: the course mixer at RF = 220 Ω against its Table 5.3 budget (Section 10.8): conversion gain +8.02 dB in Xyce within 0.1 dB, SSB NF with the behavioural op-amp 11.25 dB from lptv() within 0.1 dB, and IIP3 +9.9 dBm within 0.5 dB. Lab 5 runs this behavioural design; Chapter 10’s final design, with the transistor-level OTA in both TIAs and RF = 223 Ω, which Chapter 19 uses, gives +8.05 dB, 11.88 dB (channel average) and +10.6 dBm (Table 10.2).

Week 8 · LC oscillators and phase noise

Read: Chapter 11.

Lab: Lab 6: the VCO.

Practice: Chapter 11’s experiments with tools/figs/ch11_lc_vco.py: a tail-filter capacitor four times larger, half the tail current, and 10 fF more fixed capacitance, checked in the slow and fast corners.

Checkpoint: rerun design isfdev frozen dcinj --resim: Table 11.2’s frequency at code 8 and 0.6 V (4.868 GHz) within 0.2 % and phase noise at 1 MHz (−116.7 dBc/Hz) within 0.5 dB, with the PMOS pair as the largest contributor (Table 11.1).

Week 9 · PLL dynamics and frequency dividers

Read: Chapter 12 and Chapter 13.

Lab: Lab 6, continued: the PLL loop around the VCO.

Practice: the experiments of Chapter 12 (tools/figs/ch12_pll_noise.py: a crystal floor of −160 dBc/Hz, a 2 ns charge-pump on-time, RT3 = 0.3) and Chapter 13 (tools/figs/ch13_dividers.py: vcm = 0.70 V in the nominal CML design, an 8/9 prescaler, a quality factor of 20 in the injection-locked divider).

Checkpoint: Table 12.1’s 100 µA row (R1 = 24.5 kΩ, Ctot = 63.5 pF) within 1 %, and Table 12.2’s total, −113.0 dBc/Hz at 1 MHz within 0.1 dB and 451 fs within 1 %; Table 13.1’s nominal divider, 2.01 mW and fso = 4.46 GHz, within 2 %.

Week 10 · Fractional-N synthesizers and digital PLLs

Read: Chapter 14 and Chapter 15.

Practice: the experiments of Chapter 14 (tools/figs/ch14_fractional_n.py, under a minute: the cancellation residual, the α = 0.5 channel, a MASH 1-1 with seed=0, a worse DTC through seg_sigma_ps and pwl) and Chapter 15 (tools/figs/ch15_digital_pll.py: narrower TDC inverters, a wide loop, FCW = 122.0 and 122.25).

Checkpoint: Table 14.2’s channel word for 2442 MHz by hand (Nint = 61, K = 838861); Table 14.4’s total, −112.7 dBc/Hz at 1 MHz and −126.9 dBc/Hz at 3 MHz within 0.2 dB and 467 fs within 2 %; and Table 15.1’s TDC floor, −84.3 dBc/Hz at the DCO, from Eq. (15.3) and the 43.5 ps inverter delay, within 0.1 dB.

Week 11 · Power amplifiers

Read: Chapter 16.

Lab: Lab 7: the power amplifier.

Practice: Chapter 16’s experiments with tools/figs/ch16_pa.py: VPA = 2.5 V, a relaxed second-harmonic rejection, a 5 Ω class-E switch, and a 1 nH collector-trap inductor.

Checkpoint: rerun pa --resim (about seven minutes): Table 16.3’s PA with the 3f0 trap at the collector, +10.4 dBm saturated within 0.1 dB, 31.0 % PAE at +10 dBm within 0.5 points, and a third harmonic of −69.8 dBc within 1 dB.

Week 12 · Millimetre waves and RF layout

Read: Chapter 17 and Chapter 18.

Lab: finish Lab 3: bring the spiral into ngspice by vector fitting (Section 18.5).

Practice: the experiments of Chapter 17 (tools/figs/ch17_mmwave_pa.py and tools/figs/_ch17_redesign.py: another stabilizing R‖C, a self-resonant DC block, 8 pH of emitter inductance) and Chapter 18 (tools/em/ch18_pa/ch18_layout.py, tools/em/ch18_pa/ch18_palace.py and tools/figs/ch18_rf_layout.py: a finer first-order mesh, the 60 µm pad, a vector fit with fewer poles); the Touchstone section of Appendix B.

Checkpoint: the device match parts of tools/figs/ch17_mmwave_pa.py give Table 17.3’s gain at 50 GHz with TopMetal2 and cap_rfcmim, 9.07 dB, within 0.05 dB; from the Palace Touchstone files, tools/figs/ch18_rf_layout.py gives the 80 µm pad’s 37.2 fF at 2.44 GHz (Table 18.1) within 1 % and the after-layout gain of 4.82 dB at 50 GHz (Table 18.4) within 0.1 dB.

Week 13 · The complete transceiver

Read: Chapter 19.

Lab: start the Lab 8 capstone, the transceiver budget: choose the block, or the complete receiver, and collect its numbers.

Practice: Chapter 19’s experiments with tools/figs/ch19_transceiver.py and its receiver model tools/figs/_ch19_rx.py: 1 dB less loss in the band filter and switch, a 1.2 MHz channel filter, and a 0.5 µs PA ramp.

Checkpoint: Table 19.2’s cascade from the designed blocks, with Chapter 10’s transistor-level OTA mixer: NF 4.62 dB within 0.01 dB, IIP3 −11.6 dBm within 0.1 dB, and the sensitivity of −97.1 dBm that follows from them.

Week 14 · Capstone

Read: Section 19.1 of Chapter 19 again, with Chapter 5’s Table 5.3 beside it.

Lab: finish Lab 8: check one designed block, or the complete receiver, against Table 5.3 and the BLE tests, as Chapter 19 does. Give every specification a row with its target, the designed value, the margin and how the value was obtained (Table 19.1), then put the numbers into the behavioural receiver and read the margins of Table 19.3. A transmitter capstone uses Table 16.1 and the transmitter tests of Table 19.4 instead.

Checkpoint: the capstone report.

Assessment suggestions

For a credit-bearing course we suggest: the weekly checkpoints (40 %), marked against the checkpoint table of each lab and, in weeks 2, 3 and 10, against the numbers above; the Try it experiments (20 %), each answered in a few sentences with the number that answers it; and the capstone (40 %), assessed on a short report and the scripts that produced its numbers. The report follows Table 19.1 of Chapter 19, and its “How obtained” column carries much of the mark:

Criterion Weight Full marks for
Completeness 20 % a row for every specification of Table 5.3 (or Table 16.1) that applies, with target, designed value, margin and how it was obtained
How obtained 30 % margins from your own simulations (ngspice, Xyce, Palace) at the typical corner and at the worst corner the chapter studies; every number that no simulation produced marked “hand” or “assumed”, as Table 19.1 marks them
Doubtful numbers 20 % each hand or assumed number tested for how far it may move before the system fails, as Section 19.2 does for the mixer’s noise figure
System check 20 % the block’s numbers in the behavioural receiver and the BLE tests of Table 19.3 (Table 19.4 for a transmitter), with the tightest margin explained
Judgement 10 % the thin margins named (Table 19.1 prints those below 1 dB in bold) and a change proposed for each

Score each criterion from 0 to 4 and weight it, so that two markers reach the same mark:

Score The criterion’s items are…
4 all present and correct, each with its evidence beside it: the script that produced it, a plot, or the entry in a numbers file
3 all present, but one is unexplained or one number does not match its evidence
2 mostly present, but several are asserted rather than shown
1 a few present, mostly asserted
0 missing

In a large class, peer review scales well: before the capstone is handed in, each student reviews two classmates’ reports against the same two tables and writes, for each criterion, the score and the one change that would raise it. Students then revise their own reports before submission. Mark the reviews too, by how many of the problems the instructor later finds were already flagged in them.

Shorter versions

Eight weeks. Keep Week 0, then merge the weeks: 1–2, 3, 4–5, 6, 7, 8–9, 10–12 and 13–14. Read Chapters 13, 15, 17 and 18 for their objectives and key takeaways only, and make the capstone a single block rather than the complete receiver.

Two days. Chapters 1 and 2 with Lab 0 on the first morning; Chapter 4’s sections on noise figure, the Friis formula and sensitivity, and Chapter 5’s link budget (Table 5.3), in the afternoon. Chapter 9 with a guided version of Lab 4 on the second day, closed by checking the LNA against Table 5.3 as the LNA rows of Table 19.1 do.


Back to top

SkillSurf · RFIC Design with Open-Source Tools, draft v0.1 (November 2026). Every source is listed on the References page.

This course was written with the help of AI tools and may contain mistakes. If you find one, please tell us through SkillSurf or email skillsurfedu@gmail.com, and we will correct it.

This site uses Just the Docs, a documentation theme for Jekyll.