Protecting Capacitive Soil Moisture Sensors

24 min readApr 12, 2021

Looking to measure the moisture in the soil of some pot plants, I initially purchased a couple of resistive soil moisture sensors. However, the probe legs, electronic components, and in-and-around the connector soon corroded. It was dead. I then started looking at capacitive sensors, as I had seen them marketed as “corrosion resistant”, and purchased a cheap bunch of them. Unfortunately they had similar problems. In this article I write about the damage the capacitive soil moisture sensor sustained and my attempt to remedy and prevent it.

1. The Sensor

The capacitive soil moisture sensors I purchased appear to be functionally equivalent to DFRobot’s Analog Capacitive Soil Moisture Sensor. How exactly the sensor works is outside my current knowledge. But in any case, electronicclinic.com has a great write-up about it.

An important note before progressing any further: some sensors are faulty. This article assumes a sensor’s 1MΩ resistor(R4)is properly grounded, and the chip in the centre is a TLC555. Please read here for more information.

2. The Damage

Before placing the sensor in a pot plant, it seemed obvious the sensor would need some kind of protection from the environment (and my watering of the plant) given the bare electronics on the PCB. I did some light reading on protecting them and, with what I had on hand, settled on covering the electronic components and connections on the rear with glue from a hot-glue gun, and placing some plain heat-shrink tubing over the glue. The sensor then went into the soil and worked as expected.

After approximately six months, I noticed the readings seemed a little abnormal. The sensor was indicating a much higher moisture level than it did before, even when the soil was bone-dry. I took out the sensor, peeled off the heat-shrink and glue, and observed there to be corrosion around the lower third of the electronic components. Particularly around capacitors C1, C2, C3, and C4.

I also observed the solder mask covering the ground plate which runs around the outer edge of the sensor to be wrinkled. I posited this to be the result of water somehow working its way underneath. There were also very fine breaks on the outside of the solder mask where the ground plate was now visible.

I decided to connect it to an Arduino Uno to get some readings of where the sensor’s capability was at. I used the following pin connections and program below. No other components (ie. resistors, capacitors, etc…) were added.

SENSOR                  ARDUINO
-------------------------------
VCC (RED)               5V
GND (BLACK)             GND
AOUT (YELLOW)           A0

The result was a roughly consistent value of 23. Placing it in water or soil had no effect. It looked to be dead.

I opted to try cleaning it by first disconnecting it and then dripping some WD-40 over the electronic components in the hope it would remove some of the corrosion, thereby fixing it. I came back after a day and wiped off what I could with some paper towel.

Connecting it back to the Arduino initially yielded promising results. The values being read were around 530. However, those values quickly and progressively dropped to around 120 despite no environmental changes. Promisingly however, there was a small change in the correct direction when the sensor was placed in water.

I decided to clean the sensor again. This time I dripped WD-40 over the entire sensor to cover it from top-to-bottom and let it soak for a day. I wiped off what I could with paper towel, but unfortunately I damaged the solder mask covering the ground plate which peeled off leaving the copper exposed.

damaged solder mask exposing ground plate

Testing with the Arduino yielded much better results than before. The sensor was now reading roughly 624 with some consistency. I hypothesised that “soaking” the sensor in WD-40 had displaced water which had been trapped under the solder mask where I observed the wrinkling.

3. Test

With the damaged sensor now at least somewhat working, I wanted to see what kind of state it was in. I decided to test the following five environments:

dry (sitting in the open air on my desk)
dry soil (dry soil in 500g glass jar)
moist soil (dry soil mixed with 125ml of tap water in 500g glass jar)
wet soil (dry soil mixed with of 250ml of tap water in 500g glass jar)
water (tap water in 500g glass jar)

However, I needed to get a baseline measurement to compare it to something. I opened up a new sensor from a an ESD-safe bag and tested that as well. I also included DF Robot’s “Value 1” and “Value 2” values as dry and water respectively. Here are the results:

They do appear to be roughly the same, so it seems the damaged sensor was/could be repaired to at least allow it to read again. But some things worth noting:

Both sensors were connected to the same Arduino simultaneously. I am unsure whether this had any impact on the results.
I placed both sensors in the same glass jar simultaneously when obtaining the results. The exception to this was the water test where I held them in-place (by the connector) to avoid the water damaging the electronic components.
When placing the sensors in water or soil with water, I allowed the sensors’ readings to stabilise whereupon they were rarely fluctuating. I did not measure how long this took as I did not anticipate it, but it is an interesting metric that could have implications.
The new sensor took much longer to reach a stabilised reading than the damaged sensor. I would estimate two to three times as long as the damaged sensor.
Assuming there is expected to be a downward trend in the readings as the sensor detects higher levels of moisture, the new sensor’s wet soil reading could be an anomaly.
DFRobot’s wiki lists some value ranges for “dry”, “wet”, and “water”. If DFRobot’s “dry” is taken to be comparable with dry soil from this experiment, “wet” with wet soil, and “water” with water, only the damaged sensor’s wet soil measurement is out of range at 312, putting it in DFRobot’s “water” category. It is possible that the damaged solder mask exposing the ground plate on the sensor makes it more sensitive to moisture. This theory is supported by my initial observation that the sensor seemed to be erroneously giving much higher moisture levels than before.
A more robust experiment might be conducted by reading the value of a sensor until it stabilises, increasing the water content of the soil, and repeating.

4. Proposed Fixes

My hypothesis is that the sensor became damaged by:

corrosion of the electronic components; and
water becoming trapped under the solder mask covering the sensor.

My aim is therefore to address both of these problems.

4a. Corrosion of Electronic Components

I attempted to prevent corrosion using glue from a hot-glue gun as I described earlier. But I am inclined to believe this did not adequately protect the components. When I peeled the glue off, it appeared not to have dried uniformly across the PCB. This may have been why the corrosion appeared worse on one side than the other. A better application of glue against the PCB may have been more effective, especially if it were less viscous, but I would not rely on it again.

Fortunately, preventing capacitive moisture sensor corrosion is not a new issue. And it appears that the general approach is to coat and/or cover the electronic components and connector. Compounds and materials used appear to include nail polish, urethane, epoxy, and glue-lined and non-glue-lined heat-shrink tubing.

Colorado State University Professor Jay Ham describes in his instructables.com article as having coated all the electronic components with nail polish — front and back — followed by covering the top of the sensor with glue-lined heat-shrink tubing.

On switchdoc.com, an article describes covering the same with only silicon caulking (ie. no heat-shrink). It also notes that conformal coating is specifically designed to address this issue.

Adosia specifically makes and sells “ruggedized” capacitive moisture sensors and provides a video on their process. Note their use of urethane to coat the PCB.

An article on thecavepearlproject.org describes using heat-shrink as a container, filling it with epoxy, and heating it from the bottom to cover the components.

The article also states:

While the heat shrink/epoxy method is our gold standard for sensor encapsulation, adhesive lined 3:1 heat-shrink can do a reasonable job on these sensors if you make sure the surfaces are super-clean with isopropyl alcohol & take time to carefully push out any air bubbles. (use gloves so you don’t burn your fingers!) A third alternative is to use hot glue inside regular heat shrink tubing — squashing it into full contact with the circuits while the glue is still warm & pliable. You still have to treat the edges of the PCB, but that can also be done with nail polish. Both the heat-shrink & hot glue methods can ‘pull away’ from the smooth sensor surface over time — epoxy encapsulation is more robust.

While thecavepearlproject.org article states the use of glue-lined heat-shrink will do a “reasonable job”, this is presumably using only heat-shrink. The question is therefore whether the epoxy is a more or less effective protectant than the urethane in addition to heat-shrink in this situation. But unfortunately I do not know the answer to that question. It may simply be that at some point, with the amount of protection being applied, it really does not matter any more.

However, from my brief research, it appears that “hard” materials — such as cured epoxy — may begin to cause electrical failures when the temperature of the environment is cycled between extreme temperatures (-40°C to +85°C in the experiment). The article goes on to note that in “extreme or harsh environments” it is preferable to use a softer material.

I would expect soil moisture sensors to be able to be placed anywhere a plant could grow, which is virtually everywhere on Earth. It goes without saying that includes a myriad of variable, hostile environments. Given its soft property, I have therefore decided to use urethane and follow Adosia’s process described above. However, there are some modifications I will be making.

First, there is a discrepancy about how many coats of urethane should be used. In Adosia’s video, two coats of urethane are applied to the PCB, while their product description states that it is “triple-coated”. Given the need to protect the electronic components in potentially adverse conditions — that is, outdoors — for long periods of time, I will err on the side of caution and assume three coats is recommended.

Second, from the instructables.com article earlier, it was recommended to file down the top edges of the sensor to avoid the heat-shrink from being pierced.

I have therefore opted to do the same, although I will try to round them.

Third, Adosia did not appear to add any other protection after applying the heat-shrink. In the instructables.com article, nail polish was applied to the bottom of the heat-shrink “to provide extra waterproofing protection”.

Regardless of whether the nail polish here actually makes any difference, applying it is not inconvenient and the premise makes sense.

Finally, and most prominently, is that Adosia did not remove the connector, whereas in the instructables.com article it was removed to allow different wire to be attached. It may have been that the Adosia tutorial video did not remove it simply to show the “ruggedization” process. Regardless, my concern with the connector still being present is how much area the the heat-shrink needs to cover and any resulting “pockets” which may be susceptible to punctures. In the image from Adosia below, there are sections of the heat-shrink around the connector which do not sit flat against the PCB.

I cannot envisage cutting open the heat-shrink simply to attach another cable and then having to apply more heat-shrink. I will remove it and solder wires directly onto the PCB.

4b. Water Trapped Under Solder Mask

I first want to discuss some methods to avoid damaging the solder mask.

First, scraping the sensor against anything rough when inserting or removing it from the soil (or just in general) should be avoided. The rough edge of a rock may be all that is needed to cause a tear in the solder mask. If possible, a small trench should be dug, the sensor placed inside, and then soil packed around it.

Second, if, for whatever reason, the sensor needs to be cleaned (use isopropyl alcohol), avoid wiping motions or using anything which could tear at or become caught on the solder mask. Lightly dabbing it with something non-abrasive instead, such as a soft cloth, is safer. This is especially true if there is any wrinkling of the solder mask as in the image above.

Third, if water is suspected to be trapped under the solder mask, it needs to be removed for normal functionality to be restored. It also needs to be removed in the most non-destructive way possible. While I used WD-40, there may be alternatives. A method to evaporate the water using heat could work, but I would be concerned with how the solder mask — which may already be damaged — could react. It may retract from the ground plate. Whatever the method, be sure to disconnect the sensor first.

Fourth, the protection of the edges of the PCB does not appear to have been given much consideration. However, in the thecavepearlproject.com article cited earlier, it notes the application of epoxy to the edges because “[c]ut PCBs can absorb several % of water if edges are left exposed”. Additionally, Adosia responded to a comment on YouTube stating they tested “sealing the entire board”, but that it resulted in a “reduced analog operating range”.

While Adosia apparently tested sealing the entire board, there was no indication regarding merely sealing the edges. I will attempt to seal the edges with urethane. Failing that, I will use nail polish. Others have recommended superglue and PlastiDip as alternatives.

Fifth, I am inclined to additionally coat the solder mask covering the ground and copper plate with a protectant to prevent the breaks at the edges of the solder mask I observed. That is to say, to help stop the solder mask from breaking away and leaving it exposed for water to enter. It may be possible to limit the application of nail polish, urethane, epoxy, or some other suitable substance while still maintaining a sufficient range of operation. Rather than coating the entire PCB as Adosia states they did, merely coating the portions which are susceptible to damage may offer an acceptable trade-off.

There may also be an alternate solution. If the solder mask becomes damaged without any prior protectant having been applied, that same protectant could be used to repair the solder mask as a substitute. For example, in the image near the beginning of this article showing the exposed ground plate, a coating of urethane across it could be an acceptable fix. Paper may be used to section-off the exposed part of the sensor to avoid urethane being applied to the rest of the sensor. That is what Adosia did in their video. I would avoid using tape — including electrical tape — simply because, if the solder mask is already loose, removing the tape may also remove the solder mask and further damage the sensor.

I will therefore apply a very small amount of nail polish directly to the inner and outer edges of the ground plate to avoid the tearing I observed. Separately, I will attempt to fix the damaged sensor by coating the exposed ground plate with urethane.

5. Protection and Procedure

5a. Summary of Chosen Protections

I plan to:

Apply three coats of urethane over the electronic components. I purchased CRC Red Urethane Seal Coat; the colour should not matter as it will be hidden by the heat-shrink.
Apply 3/4" (19.1mm) 3:1 ratio glue-lined heat-shrink tubing over the electronic components.
Attempt to apply three coats of urethane to the entire PCB’s edge. If, for whatever reason, this becomes infeasible, apply three coats of nail polish to the entire PCB’s edge. I purchased Sally Hansen Hard As Nails nail polish.
Apply one coat of nail polish over the inner and outer edges of the ground plate.

5b. Proposed Procedure

Removal of the connector.
Filing down of the top corners of the sensor to round them.
Soldering of wires to GND, VCC, and AOUT.
Application of heat-shrink to the wires at the connection to the sensor (although I did not previously mention this, I am applying it merely to keep the separate wires bundled together).
Application of three coats of urethane or nail polish to the PCB’s edge.
Application of one coat of nail polish over the inner and outer edges of the ground plate.
Application of three coats of urethane to the front and back of the electronic components.
Application of heat-shrink over the electronic components and urethane.
Application of one coat of nail polish to the bottom of the heat-shrink where it joins to the sensor.

Additionally, for the damaged sensor:

Sectioning-off of exposed copper plate with paper and application of one coat of urethane to the exposed copper plate.

6. Results and Observations

I have opted to protect the damaged sensor I described earlier.

6a. Removal of Connector

First, to protect the solder mask from damage in the third-hand’s alligator clips, I cut a couple of pieces of cardboard and sandwiched the sensor between them. I then de-soldered and pulled the connector off the PCB, then used a solder sucker to clean up the remaining solder.

I noted the discolouration of the PCB’s edge which I had previously missed. It was also curious to see bubbles forming from the centre of the top edge of the PCB as the soldering iron was near the area. The bubbling occurred after I removed the JST connector. Unfortunately at this point I also damaged the sensor even further, with more solder mask peeling off the ground plate.

6b. Filing of the Corners

I had planned to mark the PCB or attach a small piece of paper to the corners of the sensor as a guide to know how much I needed to file down. But the amount to remove is actually so minimal that I realised I could judge it by eye instead. The image below shows the corner of a credit card sitting on top of the corner of the sensor. I used this as a guide for how much to file off the sensor.

I placed the sensor in a bench vice — again sandwiched between plenty of cardboard — and gently and slowly filed the corners.

6c. Solder the Wires

I have opted to make use of the cable included with the sensor (see the images at the beginning of this article). For the purposes of this article it does not matter which end is removed; I could either cut off the female JST (white) or female Dupont (black) connector. I have chosen to remove the Dupont connector merely so I can connect the female JST connector to the male JST header I earlier removed from the sensor. For reference, the wires are 24 AWG. Also note the heat-shrink in preparation for the next step.

One small note at this point. I cut off the excess wire and did some small filing down of what was left on the rear of the sensor where it was soldered-in. I was concerned about the remaining sharp pieces of solder/wire piercing the heat-shrink.

6d. Application of Heat-Shrink to Wires

I used a small diameter heat-shrink to combine the wires from the sensor. It came from a pack of heat-shrink I had; I measured it to be approximately 3mm (1/25") in diameter. I cut a 1.5cm (0.6") length and threaded it over the wires before I soldered them to the sensor as I did in the previous step.

I did not have a heat gun to heat the heat-shrink so I used a hair dryer instead.

note how tightly packed the wires are together

6e. Urethane or Nail Polish to PCB’s Edge

After some consideration, I decided to use some closely packed cardboard to again sandwich the sensor and only expose the edges. The idea was that by only exposing the edge, I could limit the spray of urethane to only the edge. Unfortunately it proved far too finicky and would have required moving pieces of cardboard all over the sensor. In fact, while trying to do this, more solder mask became dislodged. I resorted to using nail polish.

I applied three coats of nail polish to the edges. After each coat, I left it to dry for 30 minutes in a well-ventilated area. I found it easier to apply the nail polish by holding the sensor vertically by the cable and coat the edges from top to bottom. I also used the point at the bottom of the sensor to rest it against my workbench. The result after three coats was a mostly smooth edge, as shown below.

6f. Nail Polish to Edges of Ground Plate

This step was a little more difficult than I had predicted. The size of the nail polish brush has meant that more nail polish has ended up on the PCB than I had intended. But I do not know whether it will make any significant difference to the analog signal at this stage. Again, the purpose of this application of nail polish was to help protect the solder mask. So perhaps the trade-off will be negligible. In any case, I only applied one light coating to the edges of the ground plate then left it to dry for 30 minutes in a well-ventilated area.

superimposed green lines show edges of ground plate

Two issues also worth noting. First, I needed to wipe excess nail polish off inside the bottle to avoid runny “blobs” of it on the PCB. Second, I needed to get very close to to the sensor to try to apply the nail polish as directly and minimally as possible only to the edges of the ground plate. I strongly recommend wearing a mask to avoid breathing in the fumes, and/or applying it in a well-ventilated area.

6g. Urethane to Front and Back of Electronic Components

When adding paper over the top of the sensor to prepare it for the urethane spray, I discovered I could use a layer of brown packaging paper. I cut a piece wide enough to wrap around the sensor and then pinned it on the back with paper clips. The clips have the added benefit of propping the sensor up off the ground.

I then applied three layers of urethane to the front of the sensor and let it dry for at least 30 minutes between each coat. I very strongly recommend doing this in a well-ventilated area and using a mask and gloves. Do not do this indoors without adequate ventilation. To test whether the urethane was dry, I lightly pressed the PCB while wearing a glove. If the glove remained clean, I considered it dry. If the glove remained clean, I considered it dry. As a reminder of what I noted earlier, I purchased red urethane rather than clear. It will not make a difference after the heat-shrink is applied.

front of sensor before applying urethane

front of sensor after first coat of urethane

front of sensor after second coat of urethane

front of sensor after third coat of urethane

I then flipped the paper over and applied three layers of urethane to the rear of the sensor and let it dry for at least 30 minutes between each coat.

rear of sensor after first coat of urethane

rear of sensor after second coat of urethane

rear of sensor after third coat of urethane

Here is a look at the front and back.

6h. Heat-Shrink Over Electronic Components

Around eighteen hours had passed since the last coat of urethane was applied. I cut a 40mm (1.6") length of 3/4" (19.1mm) 3:1 ratio glue-lined heat-shrink tubing and fit it over the top of the sensor.

I then used a hair dryer to heat up the heat-shrink and, using an oven mitt, compressed it while it was still hot. Here is the result.

Unfortunately, when I compared this with Adosia’s, it appeared as though the heat-shrink had not properly shrunk. I hypothesised that the hair dryer could not produce enough heat, especially when compared with a heat gun. I used the barrel of my soldering iron to shrink it further, although this was a very tedious process which resulted in a couple of burn marks on the heat-shrink.

6i. Nail Polish to Heat-Shrink-to-Sensor Join

After the heat-shrink had completely cooled to room temperature, I inverted the sensor and applied one coat of nail-polish completely around the join. I then left it to dry for 30 minutes in a well-ventilated area.

6j. Urethane to Exposed Copper Plate [DAMAGED SENSOR REPAIR STEP]

After having used the urethane spray, I decided against attempting to apply it to the exposed copper on the sensor. I considered it too difficult to apply given how indiscriminate and uncontrollable the spray was. I may attempt it at some point in the future, however. I have therefore opted to use nail polish in its place.

I applied one coat of nail polish to the exposed copper plate and left it to dry in a well-ventilated area for 30 minutes. As before, I expected to need to get close to the sensor, so I chose to wear a mask to avoid the fumes.

note the outline of the nail polish over the copper

As this was the last part I needed to do, I sprayed some isopropyl alcohol onto the front and back of the sensor, dabbed it clean with a microfibre cloth, and stored it in an ESD-safe bag. I did this in a well-ventilated area and wore a mask and gloves.

7. Retest

I retested the now-protected sensor under the same conditions as before although with one caveat. Because some time had passed between the original test, I had disposed of the soil I used. For this test I used mulch, though I kept the amount the same. Here are the results for the protected sensor overlayed with the previous results.

8. Conclusions and Remarks

I want to be careful about drawing conclusions from the results. I would like to re-run them in the future and keep the soil, mulch, or whatever the medium, consistent throughout all the tests. I also want to take into account the notes I made at the time of the initial tests in order to design a more effective methodology.

In any case, just looking at the dry and water results (which would not be affected by the use of soil or mulch as neither was used), they appear to be very consistent with the new and damaged sensor results. This is very interesting because it demonstrates that the protections applied to the sensor had little, if any, effect on the resulting analog signal. But perhaps even more relevant in the context of this article — repairing a damaged capacitive soil moisture sensor — is the addition of the coats of nail polish to the edges of the sensor, the edge of the ground plate, and the exposed copper plate: the nail polish does not appear to have any effect either.

So with the exception of some issues — particularly the dislodging of additional solder mask — I am generally pleased with how the sensor has turned out. Given that it was initially defunct and would almost certainly have been discarded, I have high expectations for its longevity now. I am particularly interested to see the effectiveness of the protections I applied to the damaged sensor after some time has passed. I aim to provide an update to my efforts in the future. I intend to apply these protections to any existing and new sensors I will use. There are a number of further points I will now expand on.

It may well be worth giving the sensor a light clean with isopropyl alcohol prior to some of the protective steps. For example, consider that applying urethane and nail polish is effectively sealing-in whatever is already there. That might include hair, dust, oil (i.e., from skin), and other foreign bodies and artefacts which could affect the sensor and its operation.

Instead of using cardboard as I did in many steps to “sandwich” the sensor, it may be less abrasive to use a soft cloth. A soft, thick cloth would probably have been easier in the bench vice, too. Thin polyethylene foam sheets typically found in parcels may be a good choice. It has a soft waxy feel that would hopefully glide over the solder mask rather than pull on it. If cardboard or something similar to it is used, be sure to gently press it against the sensor rather than moving it into place to avoid tearing at the solder mask.

The discolouration of the PCB I noted earlier is particularly noticeable, especially when compared with the edge of a new sensor.

damaged sensor on the left, new sensor on the right

I am curious to know whether the discolouration indicates a PCB which has absorbed water. This may have been why I observed bubbles to be forming when heat from the soldering iron was present. I would hope the nail polish applied to the edges will help to prevent this occurring.

I considered using a Dremel with a router attachment to cut the corners of the PCB. But only a small amount of the PCB needed to be removed to round it. Filing it by hand was very easy.

The heat-shrink I applied to the wires at the top of the sensor was not glue-lined. This may (strong emphasis on “may”) reduce the effectiveness of the outer glue-lined heat-shrink, as it effectively creates a “tunnel” inside it. For this reason I chose to cut a length of glue-lined heat-shrink which encapsulated the end of the “tunnel”. Be aware of this is if non-glue-lined heat-shrink is used around the wires. Either:

do not use any inner heat-shrink;
use a short length to allow the outer heat-shrink to encapsulate it; or
use glue-lined heat-shrink.

For future sensors I have opted to use an inner length of glue-lined heat-shrink which extends beyond the top of the outer heat-shrink. I would hope doing so would allow a better outer seal to form around something larger than just the insulated wires.

If the glue inside the heat-shrink is not properly heated until it melts as I was unable to do using a hair dryer, a seal will not properly form around the sensor. If this occurs, its ability to keep water and moisture away will be severely compromised. I highly recommend using a heat gun instead.

I wonder if some kind of “cap” can be fashioned and placed over the top of the heat-shrink atop the sensor. Perhaps something 3D-printed. It would likely be more aesthetically pleasing than leaving it as-is and have the added benefit of more protection. I can envisage a sensor sitting outside in a pot-plant while it is raining and water hitting the open top of the heat-shrink. I would be concerned with how this would affect the protective function of the heat-shrink (particularly the glue) over time. It would also be interesting if something similarly could be created to house other electronic components, such as a radio transmitter.

I still have reservations about the use of nail polish where the heat-shrink joins with the sensor. I wonder whether more urethane could have been used instead (although I would not use red coloured urethane here). But I also wonder whether any protection here would ultimately be superfluous. There is already urethane underneath and behind the join, and glue from the heat-shrink. When I thought about it some more however, I realised this join is where the soil is closest to the electronic components. The added protection may well be worth it, even if just for peace of mind.

Unless it were economically infeasible, I would not use red urethane again. The “spillage” seen along the right edge of the sensor is rather obvious and not aesthetically pleasing. I would use clear urethane instead. Of course, this is predicated on the ability to effectively apply it only to the area(s) needed to. If this can be done, the heat-shrink will be able to completely hide the urethane, regardless of its colour.