rf-bench-drivers-nanovna 0.1.0

NanoVNA / NanoVNA-H / NanoVNA-H4 driver for Python bench automation — ASCII shell over USB CDC

rf-bench-drivers-nanovna

Python driver for the NanoVNA family of low-cost vector network analyzers — the original edy555 NanoVNA, hugen79 NanoVNA-H / NanoVNA-H4, DiSlord NanoVNA-H4 builds, and the Deepelec NanoVNA-F (verified 2026-06-30 against an F-series HW2.3/HW3.1 running firmware 0.2.1).

Speaks the ASCII shell protocol over the on-board USB CDC serial port (no NanoVNA-Saver / nanovna-python dependency). Returns numpy arrays.

Status

🧪 First release. Verified against the user's NanoVNA-F: identify, setup_sweep, get_frequencies, get_s11, get_s21, get_trace_db, get_trace_phase, set_parameter, set_marker, is_correction_on, average_s_data, single_sweep, pause, resume, get_trace_db_at, refusal of S12/S22 and set_power(dbm). Calibration flow (cal_open/short/load/isoln/thru/done/save) is wired but has not been exercised end-to-end with the user's cal kit.

Not yet supported:

• NanoVNA-V2 / S-A-A-2 (binary protocol — a different driver class).
• LiteVNA / NanoVNA-F V2 (also binary protocol).

Supported hardware

Variant	Protocol	Tested?
NanoVNA (edy555 original)	ASCII shell over CDC	✅ same protocol family
NanoVNA-H / H4 (hugen79)	ASCII shell over CDC	✅ same protocol family
NanoVNA-H4 (DiSlord)	ASCII shell over CDC (superset)	✅ extras via .raw()
NanoVNA-F (Deepelec)	ASCII shell over CDC	✅ tested 2026-06-30
NanoVNA-V2 / S-A-A-2	binary protocol	❌ not supported
LiteVNA	binary protocol	❌ not supported

Install

pip install rf-bench-drivers-nanovna

or development:

cd drivers/nanovna
pip install -e . --break-system-packages

Requires pyserial>=3.5 and numpy>=1.20.

Quick start

from rf_bench.nanovna import NanoVNA
import numpy as np

with NanoVNA("/dev/ttyACM1") as vna:        # autodetect: NanoVNA()
    print(vna.identify())
    vna.setup_sweep(start_hz=1e6, stop_hz=900e6, points=101)
    freqs, s11, s21 = vna.get_s_data_full() # all three at once
    print(f"S11 dB: {20*np.log10(np.abs(s11)+1e-30).min():.1f}")

Swappable API with HP 8712B

This driver implements the same core method names as rf_bench.hp.HP8712B, so projects can switch between them by changing only the construction call:

def run(vna):
    vna.setup_sweep(1e6, 900e6, points=101)
    vna.set_parameter("S11")
    vna.single_sweep()
    freqs = vna.get_frequencies()
    s_data = vna.get_s_data()
    db = vna.get_trace_db()
    phase = vna.get_trace_phase()
    return freqs, s_data, db, phase

# Works with either driver:
from rf_bench.nanovna import NanoVNA
from rf_bench.hp import HP8712B
run(NanoVNA())
run(HP8712B("10.1.1.70"))

Methods available on both drivers:

Method	NanoVNA	HP 8712B
identify() -> str	✅	✅
setup_sweep(start, stop, points)	✅	✅
set_parameter("S11" | "S21" | "S12" | "S22")	✅ (S11/S21 only)	✅
get_parameter() -> str	✅	✅
set_format("MLOG" | "PHAS" | ...)	✅	✅
single_sweep() -> bool	✅	✅
pause() / resume() / hold() / continuous()	✅	✅
get_frequencies() -> ndarray	✅	✅
get_s_data() -> ndarray (complex)	✅ (selected param)	✅ (selected param)
get_trace_db() -> ndarray	✅	✅
get_trace_phase() -> ndarray	✅	✅
get_trace_db_at(freq_hz) -> float	✅	✅
get_s11() -> ndarray	✅	✅
get_s21() -> ndarray	✅	✅
set_marker(freq_hz, marker=1)	✅	✅
get_marker_value(marker=1) -> float	✅	✅
marker_off(marker=1)	✅	✅
correction_on() / correction_off() / cal_on() / cal_off()	✅	✅
is_correction_on() -> bool	✅	✅
average_s_data(n) -> ndarray	✅	✅
close() / context manager	✅	✅

Methods NanoVNA-only (HP raises AttributeError):

• get_s_data_full() -> (freqs, s11, s21) — exploits the NanoVNA's simultaneous S11+S21 measurement (HP must re-sweep per parameter)
• cal_reset/open/short/load/isoln/thru/done — SOLT walkthrough
• save_cal(slot) / recall_cal(slot) — flash slots
• iter_segments(start, stop, seg_points) — span > 401 points across segments
• set_marker_index(marker, point_index) — direct index addressing
• raw_power_index(0..3) — coarse hardware power index
• raw(cmd) -> str — escape hatch to the shell

Methods HP-only (NanoVNA raises NotImplementedError):

• set_power(dbm) — NanoVNA exposes only a coarse index (raw_power_index)
• set_averaging(count) — NanoVNA firmware has no host-side averaging control; use average_s_data(n) for software averaging

S-parameter scope: NanoVNA hardware is forward-only (1.5-port VNA). set_parameter("S12") / set_parameter("S22") raise NotImplementedError. To measure S22 / S12 on a DUT, reverse it physically and re-measure as S11 / S21 from the new orientation.

Calibration

SOLT walkthrough for the standard 12-term error model:

with NanoVNA() as vna:
    vna.setup_sweep(1e6, 900e6, points=101)
    vna.cal_reset()
    input("Attach OPEN to port 0,  press Enter…"); vna.cal_open()
    input("Attach SHORT to port 0, press Enter…"); vna.cal_short()
    input("Attach LOAD  to port 0, press Enter…"); vna.cal_load()
    input("Attach LOAD  to port 1, press Enter…"); vna.cal_isoln()
    input("Connect THRU port 0→1,   press Enter…"); vna.cal_thru()
    vna.cal_done()
    vna.save_cal(0)            # save to flash slot 0

To recall later:

with NanoVNA() as vna:
    vna.setup_sweep(1e6, 900e6, points=101)
    vna.recall_cal(0)

Calibration applies only over the swept range it was measured at. Change the sweep, recalibrate.

Building & publishing

cd drivers/nanovna
python -m build
twine upload dist/* --skip-existing

License

GPL-3.0-or-later