I originally assumed the nanoVNA would be the clear winner for inductance measurement. It's a vector network analyzer after all. So I ran a direct comparison using the same set of sample inductors across three methods: LC meter V2.4, LC meter V8, and nanoVNA with a 50Ω adapter (S11 measurement).

The setup
For nanoVNA measurements, I used a simple 50Ω adapter — a 100Ω chip resistor in parallel, placed in series with the inductor under test at the terminal. This shifts the Smith chart operating point toward the center for typical inductance values, improving readability.

The three instruments compared:

LC meter V2.4 — Franklin oscillator (74HCU04) + Arduino Nano, relay-switched Cref
LC meter V8 — Franklin oscillator (74HCU04) + ATmega328P, DPDT-switch Cref, LIPO powered
nanoVNA — standalone S11 reading (no PC connection)

Results summary

Several things stood out immediately:

The 10mH inductor could not be measured by the nanoVNA with this adapter configuration. The LC meters handled it without issue.
The 100μH inductor showed anomalous readings on the nanoVNA, likely due to self-resonance effects pushing the measurement point far from 50Ω — the adapter is simply not optimized for that impedance range.
Both LC meter versions agreed well with each other across most of the range.

A key insight: readability vs. generality
The nanoVNA's Smith chart representation is powerful for distributed-circuit thinking — useful for antenna and transmission line work. But when designing at HF or VHF where coils are treated as lumped elements, a direct inductance reading in µH or nH is simply more actionable. You don't want to interpret a Smith chart position every time you're winding a toroidal coil for an ATU-100 or a band-pass filter.
In this sense, the LC meter wins on readability for everyday coil-winding and verification work.

An honest limitation
Stray inductance in the measurement terminal wiring is not cancelled by the calibration procedure. This means accuracy degrades for very small inductances — the kind you'd encounter with a few turns of heavy wire on a small former. PCB layout optimization at the measurement terminals would help here.

What this suggests for future development

There is one area where the current Franklin oscillator LC meter falls short: the oscillation frequency is fixed in the 1–2 MHz range by the tank circuit. Real-world air-core coils used at VHF (50MHz+) have frequency-dependent inductance — so a measurement at 1.5 MHz may not reflect the actual inductance at operating frequency.
A worthwhile improvement would be to offer a high-frequency tank circuit option — a separate terminal or switch position optimized for air-core coil measurement at higher oscillation frequencies (e.g., 10–30 MHz). This would make the instrument more relevant for VHF coil work. Something to explore in V9, perhaps.

Full details

GitHub (V2.4): https://github.com/Nobcha/ArduinoLCM
http://chitose6thplant.web.fc2.com/LCM/Arduino_LCM_EXP.htm

GitHub (V7/V8): https://github.com/Nobcha/ARD_LCM_MANUAL
https://chitose6thplant.fc2.page/lc-meter-v7-dpdt-calibration-3-steps-layers-pcb-model/

Blog (Japanese, with measurement table): https://nobcha23.hatenablog.com/entry/2026/03/09/220241