# Capacitance to Voltage converter - Water level sensor

I have to design by myself a capacitive level sensor. To create my "capacitor", I use two copper strips that I take from a copper tape and put it on the tank where I have to take the measurement.

However, the measurement must be a voltage or a current as my PID controller can only "read" this kind of value. Therefore, it means that I need something to change my measured capacitance into a V/I and I do not find an easy way to achieve the conversion. By easy way, I mean that I am in a city where there are not a lot of electronic devices that are available so easy means "without too many components hard to find".

The specifications of my system are the following ones:

Measured Capacitance : [5pF(low level) --> 35pF(High level)]

This is the full range but I could use a small range inside this one

Input of Controller: 0-10V or 4-20mA

I hope that someone can resolve my problem, or at least give me an idea. Thank you in advance.

Sincerely.

An oscillator can be fashioned from the capacitor, and it is simple to either count the frequency, or convert to a controlled current. By DC-blocking the electrodes with C2, corrosion should be minimal.

simulate this circuit – Schematic created using CircuitLab

The NOT inverters are 74HC14 Schmitt inverter sections, intended to be powered either from 5V, or the emitter of Q1. If the ammeter isn't of the moving-needle type, some filtering will be required. R2 * Csense should be larger than R3 * C3.

I've used this circuit with a shielded cable from the Q1 emitter, so the oscillator can be adjacent to the sense capacitor, but the power supply and meter at a more convenient location.

First, you probes should use stainless steel. Copper will corrode.
There is a “simple” capacitance to voltage converter below. It’s been around a long time. You can make a 1kHz osc. with an op amp or a 555. The waveform is not critical. Active rectifier circuits abound (instead of an AC meter). R is for bias current and should be high for best linearity. I've used an LF356 because I had some laying around but I'm sure there are better modern choices. That means you want a very low bias current op amp. Cgain sets the gain, if Ctest=Cgain the AC gain will be 1.

Since you live in a city where components are hard to find, how about starting with a simple LC oscillator with a single transistor than can run from two AA cells?

If you choose the right frequency range, you should be able to detect the radiated carrier on any nearby AM radio.

It does not deliver the 4-20mA you ultimately require, but perhaps it could stimulate some ideas!

The circuit below does just that.

Here is an on-line simulation of the circuit. Almost any general purpose NPN transistor would work.

The hardest parts to find are probably the 1mH inductors, but you can absolutely make your own - a 1mH inductor is around 200 turns of thin magnet wire on a 1.5 inch diameter coil form (e.g. cardboard) that is also 1.5 inches long.

You can use two AA cells or even a 3V coin cell for the 3V power. The current draw is less than 2mA.

Your calibration could be water level versus where on the AM dial the carrier is detected. (You will not hear a tone, just a distinct "quieting" of the AM noise").

Note that most RF emissions are subject to laws in your country, but based on your project description I'm guessing you are not going to manufacture this for sale.

If you have an oscilloscope (or Arduino) to measure the frequency, even better!

Finally, don't build the circuit on "proto board" because there is a lot of capacitance between adjacent rows of contacts - lay down soome of the copper strips you are already using an insulator and solder everything together while keeping all interconnects as short as practically possible.