Turning Impedance into Frequency: An LC Experiment

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

In earlier versions of this LC meter (up to V6), calibration was handled automatically by a relay. The relay would insert the reference capacitor into the tank circuit at the start of each measurement session. Simple, automatic — but not without problems.

Problems with the relay approach

Cost: small signal relays suitable for ~1.5 MHz switching are not cheap

Availability: sourcing a consistent relay across builds was annoying

The V7 solution: a 6P DPDT switch

I replaced the relay with a readily available 6-pin double-pole double-throw (DPDT) switch. The switch physically inserts or removes the reference capacitor from the LC tank — the same function, but done by hand.

This makes calibration a conscious 3-step manual operation:

Set switch to CAL position

Press the calibration button — the sketch measures f1 and f2, solves for L and C of the tank

Set switch to MEASURE position — ready to use

"Manual calibration is a feature, not a bug"

Once you understand that calibration only needs to be done once per power-on (or when ambient temperature changes significantly), the manual switch becomes perfectly acceptable. In fact, it makes the measurement process more transparent — you always know exactly what state the circuit is in.

The sketch for V7 is available on GitHub, along with the KiCad PCB files and assembly manual (English and Japanese):

👉 https://github.com/Nobcha/ARD_LCM_MANUAL

A detailed build log is also on my blog:

👉 https://chitose6thplant.fc2.page/lc-meter-v7-dpdt-calibration-3-steps-layers-pcb-model/

One note on the PCB

The first version of the V7 PCB had three errors (OLED header pinout, R6/R7 silkscreen, ICSP header). These are documented in the repository as a "Mistake prevention list." A corrected schematic is also included.

Next up: V8, where the sketch was restructured to reduce flash usage from 95% down to 80%, making room for future improvements.

kpa radio ©nobcha 

The LCD is connected via I2C, so I replaced it with an OLED 128x64 and rewrote the sketch. I uploaded it to GITHUB as V6.4. 

I used KiCAD and uploaded the Gerber files on GITHUB.

I used LTSpice analysis to verify that the Franklin oscillator circuit works with an LC tank circuit and an inverting amplifier using a 74HCU04.

I found the history for this meter. I'm interested in the principle how to measure the values.

At first, the tank parameters (C1 and L1) are derived by adding a known standard capacitor to the tank. Please refer the equation note.

Why a Franklin Oscillator, Not a Precision LC Meter.