nobcha
While researching the origin of this LC meter approach, I found that the core idea is much older than microcontrollers.
The oscillator itself traces back to the 1920s, when Mr. Franklin at Marconi described what is now known as the Franklin oscillator: an LC resonator combined with a phase-inverted amplifier.
‐The history
Much later, in 1988, Dr. Hegewald in East Germany applied this principle to an LC meter concept using a microcontroller. This idea was later popularized by Neil Heckt through AADE, where he sold LC meter kits and openly shared both the circuit and the PIC firmware. The design was also described in detail by Phil Rice (VK3BHR) in an Australian amateur radio magazine.
What fascinated me is not absolute accuracy, but how naturally this circuit maps an analog impedance problem into a frequency domain that a microcontroller can handle very well. By focusing on differential measurements and a known reference capacitor, the system becomes simple, transparent, and easy to verify.
I’ve noticed that many builds closely replicate the original AADE design, which is perfectly valid. However, fewer people seem to change both the analog front end and the microcontroller itself.
By simplifying the implementation and using Arduino, my goal was to make such experimentation easier. In one case, this approach even led someone to re-implement the design again using a PIC, after first studying both versions.
This project is my attempt to revisit that idea with modern low-cost MCUs, while keeping the same spirit of simplicity and openness.
‐The principle
I think the principle behind this meter is very clever.
An additional element is temporarily added to the LC resonant circuit, and the oscillation frequency is measured. By comparing this frequency with that of the original LC circuit, the value of the added element can be derived by solving a simple set of simultaneous equations.
The measurement is divided into two steps. In the calibration step, a high-precision reference capacitor is added to the LC tank to determine the base LC parameters. In the measurement step, the element under test is added instead, the frequency is measured again, and its value is calculated from the difference.
By relying on frequency differences rather than absolute values, this approach naturally reduces errors caused by stray inductance and parasitic effects.
‐LTSpice simulation
First, 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 imported the .asy and .lib files related to the 74HC and performed a simulation as follows:
(1) Download the schematic from the link below and place it in the LTspice sym folder under the file name 74hcu04.asy.
http://ltwiki.org/files/LTspiceIV/lib/sym/Dig_Add/74HC/74hcu04.asy
This will become the circuit diagram data, and it will appear when you select it in the Component menu.
(2) Import the simulation data from the following location into the 74HC-related directory using 74HC.lib:
http://ltwiki.org/files/LTspiceIV/lib/sub/74hc.lib
Specify it as available for calling via Edit → SPICE Directive. (This is referenced when 74hcu04.asy calls the 74hcu04 simulation data.)
.lib 74hc.lib
(3) 74hcu04.asy requires an external power supply of 5V via "Vcc." Add a DC power supply and specify Vcc.
(4) Oscillation will not start unless initial operating conditions are set, so I set .ic V(n003) = 1V in Edit → SPICE Directive.
The circuit diagram for simulation is as follows.
LCM_FRANCLIN_HCU04_HC.asc
- The sketch overview
To count the frequencies: start measurement
→ wait for stable oscillation
→ count pulses for gate time "To refer FreqCount Library"
→ average N times until the adjacent frequencies are within 0.5%
→ output frequency
Calibration flow
→ measure freq1
→ insert Cref by turning on the relay
→ measure freq2
→ solve L,C of base tank
→ store parameters
Measuring flow
→ insert DUT
→ measure...
The residual inductance at the measurement terminals can actually be measured directly — just insert a shorting plug and read the value. In my setup it comes out to around several nH, which is consistent with what the nanoVNA adapter measurement shows as well.
That said, this is also within the natural variation range of the Franklin oscillator for this circuit configuration, so sub-µH measurements are indeed at the noise floor. For the intended use case — verifying hand-wound coils in the µH range for HF/VHF work — this is acceptable. The limitation is real, but the instrument is not useless.