rftools

Python client for rftools.io — 241 RF & electronics calculators, plus 13 async simulation job types (FDTD, antenna NEC-2, filter Monte Carlo, and more), accessible from Python and the command line.

Installation

pip install rftools-io

Quick Start

Get a free API key (5 calls/month) without copying it anywhere:

rftools login

This prints a page and a code. Open the page, sign in or sign up, and approve the code. The key is saved to ~/.config/rftools/credentials.json (%APPDATA%\rftools\ on Windows, readable by you only), and every later rftools.Client() and rftools command reads it from there. The key is never printed. From Python, the same thing is rftools.Client().login().

Or set a key you already have. The variable takes precedence over the saved file:

export RFTOOLS_API_KEY=rfc_your_key_here

import rftools

result = rftools.calculate("vswr-return-loss", {"vswr": 2.5})
print(result["returnLoss"])  # 9.54 dB
print(result.values)         # {"returnLoss": 9.54, "reflectionCoeff": 0.333, ...}

Solving for an Input

Most calculators go forward: inputs to outputs. solve() goes the other way — given every other input and a target value for one output, it finds the value of one input that reaches it, in one metered call.

result = rftools.Client(api_key="rfc_live_xxx").solve(
    "microstrip-impedance",
    {"substrateHeight": 1.6, "dielectricConstant": 4.2, "copperThickness": 35},
    "traceWidth", "impedance", 50.0,
    grid=0.001,   # round to a manufacturing grid; omit for the unrounded solution
)
print(result.value, result.reached)   # 3.052 True
print(result.result["impedance"])    # the forward result at exactly that width
print(result.result.provenance.formula_ref)

grid and range (a (low, high) pair narrowing the search) are each sent only when given. When no value in the search range reaches the target, the call still succeeds: .reached is False and .value is the closest point found, with its own forward .result. See Error Handling below — solve() raises the same typed errors as calculate().

Authenticated Usage (API tier)

Get an API key at rftools.io/pricing ($19/mo, 10,000 calls/month).

import rftools

client = rftools.Client(api_key="rfc_live_xxx")
# or set environment variable: export RFTOOLS_API_KEY=rfc_live_xxx

result = client.calculate("free-space-path-loss", {"frequency": 2400, "distance": 100})
print(result["pathLoss"])  # dB

A key is sent as Authorization: Bearer <key> and as X-API-Key. The service accepts either on every route.

Usage

The monthly allowance belongs to your account. Every key on the account draws on it.

usage = client.usage()             # GET /v1/usage, which is never metered
print(usage.used, "of", usage.allowance, "used;", usage.remaining, "left")
print("resets", usage.reset_at)
for key in usage.keys:
    print(key.key_id, key.label, key.used)

result = client.calculate("vswr-return-loss", {"vswr": 2.5})
print(result.usage.remaining)      # from this response's X-Usage-* headers
print(client.last_usage)           # the latest metered response's, refusals included

batch() results carry .usage too, and so does a job submitted with a key (job.usage).

Provenance

A calculator result says how it was computed. provenance is None until the service sends it.

p = result.provenance
print(p.method, p.version)          # calculator:microstrip-impedance api@4ce2c2e0a1b2
print(p.formula_ref)                # Hammerstad & Jensen (1980); Wadell, ...
for a in p.assumptions:
    print(a.code, a.text)
print(p.valid_range.status)         # inside | outside | unknown
print(p.valid_range.outside)        # the inputs outside their stated bounds

BatchResultItem.provenance is the same per item. A member missing from an older result reads as None and is listed in p.missing. The contract is vendored as rftools/schemas/result-provenance.schema.json.

Batch Calculations

Run up to 50 calculations in a single HTTP request (API tier only):

results = client.batch([
    ("vswr-return-loss", {"vswr": 1.5}),
    ("vswr-return-loss", {"vswr": 2.0}),
    ("free-space-path-loss", {"frequency": 2400, "distance": 50}),
])
for r in results:
    if r.ok:
        print(r.values)
    else:
        print(f"Error: {r.error}")

Typed Category Stubs

IDE-friendly typed functions with parameter defaults matching the web UI:

from rftools.calculators import rf, pcb, antenna, power

# RF
result = rf.vswr_return_loss(vswr=2.5)
result = rf.free_space_path_loss(frequency=2400.0, distance=100.0)

# Antenna
result = antenna.dipole_antenna(frequency=433.0)
result = antenna.parabolic_dish_antenna(frequency=10000.0, diameter=0.6)

# PCB
result = pcb.trace_width_current(current=2.0, tempRise=10.0, thickness=1.0)

# Power
result = power.voltage_divider(vin=12.0, r1=10000.0, r2=10000.0)

All 13 categories available: rf, pcb, power, signal, antenna, general, motor, protocol, emc, thermal, sensor, unit_conversion, audio.

CLI

# Run a calculation
rftools calc vswr-return-loss --vswr 2.5

# With API key
RFTOOLS_API_KEY=rfc_xxx rftools calc free-space-path-loss --frequency 2400 --distance 100

# JSON output (pipe to jq)
rftools calc vswr-return-loss --vswr 2.5 --json | jq '.values.returnLoss'

# Solve one input for a target output (one metered call)
rftools solve microstrip-impedance --for traceWidth --target impedance=50 \
    --grid 0.001 --param substrateHeight=1.6 --param dielectricConstant=4.2 \
    --param copperThickness=35

# Narrow the search range instead of (or as well as) rounding to a grid
rftools solve microstrip-impedance --for traceWidth --target impedance=50 \
    --range 0.1 10.0 --param substrateHeight=1.6

# Submit an async simulation job, wait for it, print the result summary
rftools sim eye_diagram --param dataRate=10e9 --file trace.s2p

# Submit and return immediately (poll later with the printed job id)
rftools sim fdtd_sparam --param solveMode=express --no-wait

# List all calculators
rftools list

# Filter by category
rftools list --category rf

# Show calculator inputs/outputs
rftools info free-space-path-loss

# Get an API key by approving a code on rftools.io (saved; never printed)
rftools login
rftools login --client-id my-tool --no-browser

# Library version
rftools version

Error Handling

Errors are classified by HTTP status code (and, for a finished job, by the service's errorKind) — never by matching text in a message.

from rftools.exceptions import (
    AuthError, QuotaError, RateLimitError, ValidationError, NotFound, JobFailed,
)

try:
    result = client.calculate("vswr-return-loss", {"vswr": 2.5})
except QuotaError as e:
    print(f"{e.used} of {e.limit} calls used; resets {e.reset_at}. Plans: {e.upgrade_url}")
except RateLimitError as e:
    print(f"More than {e.limit} requests in {e.window_seconds}s. Retry after {e.retry_after}s")
except AuthError as e:
    print(f"Invalid or missing API key. Get one: {e.key_url or 'rftools login'}")
except NotFound:
    print("Unknown calculator slug")
except ValidationError as e:
    print(f"Bad inputs: {e.failures}")

RftoolsError is the base class for all of the above. Every exception raised for an HTTP response carries .status_code and .error_kind (the service's errorKind). Pre-0.2 names (RFToolsError, NotFoundError) still work as aliases.

Before the API's metering release, a spent allowance on POST /v1/calculate came back as HTTP 401. It is now 402 on every metered route, so it raises QuotaError.

Simulation jobs

Async simulation job types — FDTD transmission-line, antenna NEC-2, filter Monte Carlo, eye diagram, and 9 more — run through the same Client. Parameters and input files are validated locally against the job type's contract before anything is uploaded or sent.

import rftools

client = rftools.Client(api_key="rfc_live_xxx")

# Submit and wait for it to finish, then fetch a summary of the result
job = client.submit_job(
    "eye_diagram",
    {"dataRate": 10e9, "prbs": "prbs15"},
    files=["trace.s2p"],   # paths, or (filename, bytes) tuples
    wait=True,
    on_progress=lambda j: print(j.status, j.progress, j.stage),
)
print(job.result())          # summary: headline values, warnings, provenance
print(job.result(full=True)) # the whole result payload

# Or submit now, poll later
job = client.submit_job("fdtd_sparam", {"solveMode": "express"})
print(job.id, job.status)
...
job = client.get_job(job.id)
job = job.wait(timeout=60)   # returns even if not finished by the deadline

A job that ends in failed raises JobFailed from wait() (or result()), carrying .kind (one of the service's errorKind values) and the error message.

wait() polls the way the web app does: the first check happens immediately (no fixed initial delay), then every 2s for the first 30s, every 5s until 5 minutes, and every 15s after that, until the job reaches completed or failed.

Uploads need a key

Uploading a file needs an API key; set RFTOOLS_API_KEY or run rftools login. The service refuses an anonymous upload, so upload() and submit_job(files=...) raise AuthError with that sentence locally — before the file is read and before any request is made. A job type that takes no file still runs without a key, on the free lane, as before.

The uploaded object is recorded against the key's account, and only that account may submit it: a key obtained from a log or a shared link is refused as though it did not exist.

Uploaded files are capped at 10 MB (the API's limit) — upload()/submit_job() raise ValidationError locally if a file is over that, before any request is made.

Async Support

import asyncio
import rftools

async def main():
    async with rftools.AsyncClient(api_key="rfc_live_xxx") as client:
        result = await client.calculate("vswr-return-loss", {"vswr": 2.5})
        print(result["returnLoss"])

asyncio.run(main())

Ideal for running many calculations concurrently in FastAPI or async scripts.

Browse the Catalog

# List all calculators
calcs = rftools.list_calculators()
print(f"{len(calcs)} calculators available")

# Filter by category
rf_calcs = rftools.list_calculators(category="rf")
for c in rf_calcs:
    print(f"{c.slug}: {c.title}")

# Get a single calculator's metadata
info = rftools.get_calculator("vswr-return-loss")
print(info.inputs)   # tuple of InputField
print(info.outputs)  # tuple of OutputField

All Calculator Categories

The typed stub catalog (rftools/calculators/, rftools.list_calculators()) is generated from a snapshot of the registry and is due for a refresh to the current 241; run python scripts/generate_stubs.py --frontend-dir /path/to/rfhub/frontend to regenerate it. The counts below describe that snapshot.

Contributing

Regenerate stubs after calculator changes:

python scripts/generate_stubs.py --frontend-dir /path/to/rfhub/frontend

Then bump the version in pyproject.toml and publish a new release.

License

MIT — see LICENSE.