[leavesofgrass]

PR 2 of 3. This PR is stacked on #13 and targets that branch, so the diff shows only the exposure work. Merge #13 first; this PR will then retarget to main.

Why

US amateurs must evaluate RF exposure under 47 CFR § 97.13(c) and § 1.1307(b). abax had no tools for this. The requirement was to use only rule text verified from primary sources, so every number built in here comes from rule text I could read:

Built in	Source
MPE limits: E, H and S for both tiers; 6 min / 30 min averaging	47 CFR § 1.1310(e)(1) Table 1
SAR limits: whole-body, 1 g, 10 g extremity	§ 1.1310(b), (c)
Exemption-threshold ERP, valid only at R ≥ λ/2π	§ 1.1307(b)(3)(i)(C) Table 1
Household members may be evaluated against controlled limits	§ 97.13(c)(1)
Sites with multiple transmitters: shared responsibility above 5 %	§ 1.1307(b)(5)

What

11 formula functions:

• MPELIMIT, MPEEFIELD, MPEHFIELD
• PWRDENSITY, MPEDIST, MPEPERCENT
• AVGPOWER
• EXEMPTERP, EXEMPTMINDIST
• SARLIMIT, SAR

Frequencies are in Hz, as in all abax RF functions. Power density is in mW/cm², the FCC table unit.

Ground reflection is a power-density factor F in S = F·EIRP/(4πR²):

• 1 = free space.
• 4 = full in-phase reflection: the reflected wave doubles the field.
• Any other value you enter yourself.

Two independent checks on the table transcription, both in the tests:

1. Wherever § 1.1310 gives both columns, E²/377 Ω equals the power-density limit.
2. Every row of the § 1.1307 exemption table equals the general-population limit × πR² / 1.64, which is the ×4 model applied to ERP, to the rule's rounding.

Tools → Radio → RF exposure estimate (MPE)

• Inputs: frequency, PEP, feed-line loss, gain (dBi), duty cycle, transmit time in the averaging period, reflection, and distance to the nearest person.
• Outputs:
  • average power, EIRP and ERP
  • both tiers' limits and compliance distances
  • percent of each limit at your distance, stated in words ("below the limit" / "ABOVE the limit")
  • the § 1.1307 exemption check
• The log-log density-versus-distance graph has these accessibility features:
  • it is keyboard-focusable
  • its accessible description is written from the numbers and updates on every change
  • limits are labeled with text, not color alone
  • it comes with a data table and Data → new sheet
• A plain-language caveat is on the dialog: this is an estimate and not a substitute for a station evaluation.

Deliberately not included

These could not be verified from primary text; the proxy blocked fcc.gov, ecfr.gov, arrl.org and icnirp.org:

• The P_th SAR-based threshold for 0.5–40 cm (§ 1.1307(b)(3)(i)(B)). It is an image in the eCFR.
• The OET-65 Supplement B per-mode duty factors. Sources conflict, for example 40 % versus 50 % for processed SSB.
• The OET-65 ground-reflection factor.
• ICNIRP 2020 reference levels.

The duty cycle is therefore an input (default 100 %, the worst case), and no reflection factor between 1 and 4 is preset. The docs say this and list these items as future work.

Also in this PR

• GraphPanel / rows_to_new_sheet move from the circuit dialog into a shared dialogs/_graphpanel.py.
• Vertical reference-line labels move to the bottom of the plot. They had collided with the threshold labels, which I found by rendering the chart and looking at it.

Verification

• ruff check . passes.
• pytest -n auto: 4744 passed, 196 skipped. The skip count is unchanged.
• mkdocs build --strict passes.
• python make_pyz.py builds, and abax.pyz --version runs.
• Coverage ratchet (CI command): core 84.54% (floor 83), engine 63.26% (floor 62).
• All 77 =formula → value examples in the formula reference's RF section were evaluated against the engine, and all match. A test also pins the worked exposure example in rf-toolkit.md; while writing it I found and corrected two wrong numbers in my first draft of that example.

🤖 Generated with Claude Code

https://claude.ai/code/session_0157FDmhHjtSvT3pWABUTvmX

---

Generated by Claude Code