Soil Moisture Sensors

June 26th, 2022, last changes on July 4th, 2022

note this post is still being improved: additional measurements are to be done

Each time I buy these capacitive soil moisture sensors, I get a slightly different thing. Most of the simply do not work and here is the answer why.

Forget about any confidence that by ordering a version, say 1.2, you get the version 1.2. It's really random.

The principle of the sensor is very simple:

Soil Moisture Sensor Circuit Diagram

It consists of four function blocks:

Component v.1.2 with LDO v.2.0 with LDO v.1.2 without LDO
R1 (LPF) 10 kΩ 10 kΩ 10 kΩ
R2 (Rb*) 2.2 kΩ 1.6 kΩ 1.6 kΩ
R3 (Ra*) 470 Ω 330 Ω 330 Ω
R4 (Smoothing) 1 MΩ 1 MΩ 1 MΩ
C1 (LDO out) . . .
C2 (LDO out, chip protection) 89-93 nF . .
C3 (Ct*) 120 pF 13-17 pF 680-830 pF, maybe 770 pF
C4 (Smoothing) . . .
C5 (LDO in) . 98 nF 7.3-8.6 μF
C6 (LDO in) 94 nF 100-101 nF 89-103 nF
calculated frequency** 2.5 MHz 27 MHz*** 530 kHz
measured frequency 263 kHz**** 1.5 - 1.6 MHz (avg 1.6 MHz) 359 - 788 kHz (avg 423 kHz)

Please note, I don't have a reliable way of measuring the capacitors at the moment, maybe will update it one other day.

* The Ra, Rb and Ct values determine the wave's frequency and ratio.

** Based on the measured C3 value, which may be quite off.

*** This value is way beyond what a 555-chip is able to deliver. Additional reason of the faulty operation?

**** The only explanation for this value being so much off the calculated one is that the only C3 I've got to get measured has been partially damaged while desoldering.

f = 1.44 / ( (Ra + 2 * Rb) * Ct )

=> Ct = 1.44 / ( f * (Ra + 2 * Rb) )

Calculated (as I don't trust my multimeter) values of the Ct capacitors for two common frequencies (see below to understand why the value) in accordance to the resistors found on the PCBs:

370 kHz 1.5 MHz
Ra = 470 Ω and Rb = 2.2 kΩ ~ 800 pF ~ 200 pF
Ra = 330 Ω and Rb = 1.6 kΩ ~ 1 nF ~ 270 pF

Output of v.1.2 with LDO at 5V

555-timer's output of v.1.2 with LDO at 5V

Output of v.2.0 with LDO at 5V

555-timer's output of v.1.2 with LDO at 5V

Output of v.1.2 without LDO at 5V

555-timer's output of v.1.2 with LDO at 5V

What are the problems?

C f
7 pF 2.274 MHz
30 pF 530 kHz
400 pF 40 kHz
1260 pF 13 kHz

It means, the generated signal should be somewhere in this range to make it work (probably close to the frequency for the dry soil). And guess what: yes, they don't match! The original design of the sensor (as the Internet says) was to produce 370kHz. The new design v.1.2 changed it to like 1.5MHz (my own measurements confirm it to be about 1.6MHz for the v1.2 with LDO PCB). However, the latest v2.0 version delivers low values just by a few drops of water and almost no difference as the amount of water changes. They increased the frequency of the signal to improve stability (for as far as I may guess, in order to limit the influence of salts in water). And guess what: yes, the filter has not been adapted for that! This setup leads to more noise than real data.

Low Pass RC-Filter cut-off frequency Wet-vs-Dry-vs-Frequency

v2.0 Noise

The above graph shows readouts of two v.2.0 sensors and watering (both) at 6:50 p.m. As you see - hard to distinguish the readouts from the noise.

How to make it work?

Well, there are a few ways to go. My first approach was to take the new PCBs marked as v.2.0, because I work with 3.3V and they contain the TL55C low-voltage timer, and do the following:

Update: second approach was to exchange the LM555 to TL555C on the PCB v.1.2 without an LDO (same reason: to be able to work with 3.3V). I have removed C5 and C6 - however same story: they don't hurt and may stay there. I wasn't sure what the value of the C3 is, so used the opportunity to desolder it temporarily and measure it (my multimeter is very bad at measuring capacities in-circuit). It seems, the frequency should be in the 500-kHz-area. The exact measurement comes later...

As you can see below, the response of the latter modification is much better: significant difference between different water levels.

Measurements

In the table below you may see measured voltage levels depending on the water level for some of the PCB variants, as well as the ratio to the previous voltage level to better see the issues.

Capacitive Soil Moisture Sensor v1.2 with LDO

Version 1.2 including the 3.3V-LDO and working in kHz-range.

Capacitive Soil Moisture Sensor v2.0

Version 2.0 including the 3.3V-LDO and working in MHz-range (mismatch with the filter)

Capacitive Soil Moisture Sensor v1.2 without LDO

Version 1.2 without an LDO, seems to be working in MHz-range (not yet measured)

My Mod

Modified (first approach) version 2.0 from above (removed LDO and modified working frequency)

My Mod 2

Modified (second approach) version 1.2 without an LDO (exchanged LM555 to TL555C)

Water level measurement setup Water levels reference point

How the water levels have been measured

Water level v.1.2 LDO on 3.3V v.1.2 LDO on 5V v.2.0 on 3.3V v.2.0 on 5V v.1.2 no-LDO on 5V My Mod 1 on 3.3V My Mod 2 on 3.3V
5.7 cm 1.232V 1.262V 1.024V 1.057V 1.859V 1.360V 1.292V
5 cm 1.250V (+1.5%) 1.278V (+1.3%) 1.027V (+0.3%) 1.059V (+0.2%) 1.876V (+0.9%) 1.373V (+1%) 1.328V (+2.8%)
4 cm 1.279V (+2.3%) 1.302V (+1.0%) 1.034V (+0.7%) 1.068V (+0.8%) 1.906V (+1.6%) 1.396V (+1.7%) 1.389V (+4.6%)
3 cm 1.337V (+4.5%) 1.362V (+4.6%) 1.044V (+1.0%) 1.081V (+1.2%) 1.963V (+3.0%) 1.430V (+2.4%) 1.484V (+6.8%)
2 cm 1.432V (+7.1%) 1.456V (+6.9%) 1.068V (+2.3%) 1.105V (+2.2%) 2.066V (+5.2%) 1.491V (+4.3%) 1.66V (+12%)
1 cm 1.662V (+16%) 1.687V (+16%) 1.169V (+9.5%) 1.208V (+9.3%) 2.301V (+11%) 1.684V (+13%) 2.04V (+23%)
0 cm 2.532V (+52%) 2.539V (+51%) 2.346V (+101%) 2.16V (+79%) 3.511V (+52%) 2.276V (+35%) 2.815V (+38%)
air 2.61V (+3.1%) 2.58V (+1.6%) 2.73V (+16%) 2.73V (+26%) 3.9V (+11%) 3.01V (+32%) 2.873V (+2.1%)

Voltage-vs-Water-Level Graph


Next: Secure DNS service

Previous: Digital Watering System

Main Menu