Plant watering Fiasco

As more technical people wake up to the folly of trying to use simple resistive DC plant moisture sensors, it seems the Chinese have latched onto the idea of capacitive sensing…

Capacitive sensor

And here it is, the sensor that is all over Ebay right now at am amazingly low £2 inc free shipping, that is until the UK rip-off artists get hold of it and at least double that price, which they’ve already done. If you are happy to sit over your plants waiting for a quiet “beep” – not quite so practical then, not to mention the constant supply of new battery pairs.

And definitely not to mention that the easily corroded batteries and active end are only an inch or two away from the waterproof, non-corrosive sensor end. Perhaps where the designers live the soil doesn’t move and there is no rain or snow.

tmpF5A2Sadly, the only half-decent Chinese sensor that was around a couple of years ago has all but disappeared and definitely had a price-hike. Amazon will even sell you the same thing WITHOUT batteries or beeper for £12.99

However, the idea here is part way to being sound. Sadly we still see electronics at the complete  mercy of creeping soil and the environment – utterly impractical as it stands so the user is left with DIY resin potting solutions or hope and prayer.

Capacitive sensor

I just LOVE this variation: Bog standard 4-way cable is supposed to be “moisture resistant” ignoring the electronics which is openly exposed to the elements. Still, cheap enough and heading the right way but no indication of the output. “Impress your friends with real-time soil moisture data” – what? For maybe a couple of weeks until the rot sets in?

Take the above cheapy, take the connector off,put on a properly weatherproof cable and pot the top end in decent resin, THEN you might have something… however, lack of information in the ads leaves me wondering if this has some kind of digital output or are we needing a processor at the other end able to handle real time oscillating signals. Needs clarification but the chip doesn’t look sophisticated enough to me to do any kind of processing.

Ideally a digital signal indicating percentage dryness – bit from this circuit, I don’t think so,

These circuits are in a different world altogether to the crude DC PCB-strip-based sensors so very popular on EBAY but sadly still not in the right world for practical, unattended garden moisture monitoring and control.

Mi FloraWe really need something better. The Mi Flora sensor is the ONLY thing I’ve seen that is even remotely practical and I got mine for £8 a couple of years ago. They have since seen a price hike to £9.53 from AliExpress, or, he says, laughing out loud, £20 from Amazon. Far from perfect but at least they don’t rot after a few weeks.

These devices are Bluetooth-based (which immediately raises range alarms and is why I don’t think, with 2 tears of experience that they are the end solution) and they give out sensible digital messages for moisture, light level and battery state. They do need additional treatment to make them properly waterproof. Potting is no good as you will need to change batteries at some point. I ended up using cling film, which was good for a year or so. I did not feel I could trust the battery readings. 

Now, getting an internal Raspberry Pi to RELIABLY read a pair of external Bluetooth garden sensors, month after month in the winter – well, there’s another discussion.

Comments and especially ideas absolutely welcome. Some lateral thinking needed.

Facebooktwitterpinterestlinkedin

34 thoughts on “Plant watering Fiasco

  1. Some use these here in Iowa. For the beans fields not so much for the corn.
    Actually measures soil “suction” as the soil dries it pulls moisture from the prepared sensor or a gypsum block (which are sometimes custom made at low cost). So moisture is measured indirectly.

    Fodder for school science projects. Probably a little much for a home garden.

    A LoRa network of sensors would be fun.

    https://www.amazon.com/Watermark-Soil-Moisture-Sensor-Cable/dp/B00IAWEHSQ

  2. Hi Pete, there is another capacitive soil moisture sensor on the market from Tindie at $13 for the standard version. It is most easily found by typing Catnip Electronics into Google. It was developed by the guy who owns Catnip Electonics. It looks remarkably like the one you showed so I guess the Catnip original has been cloned.
    Catnip Electonics are very honest and allude to the issues of making the thing completely water proof. I have three of these units but I haven’t tried them yet.
    The issue of making them completely waterproof would need resolving but the upper part which is out of the soil could easily be sealed in small plastic box.
    Quality of product is good too.
    Alternatively there is a Vegetronix device at $39.00, I understand this device is completely waterproof. However, in my application I would need several and so whilst I don’t mind paying a reasonable amount for reliability I baulked at $39.00 + delivery and no doubt UK VAT etc.
    I have read a lot about soil moisture sensing and completely agree capacitive is the way to go. I think Martins set up from an earlier blog is probably the lowest cost approach with any credibility.

    1. The Catnip man has quite detailed Github repositories (handle Miceuz) about the I2C and RS485 versions of this project. There are some schematics which help to understand the capacitive sensing was done.

  3. Personally I have found the combination of the Miflora’s and an esp32/lora board to work very well and have been giving regular feeds for several months. Miflora’s seem to go on sale regularly on bangood for approx £8 and the esp32/lora boards including case and battery are approx £26 per pair. I get over 500m range even using the cheap whip antenna supplied with the boards. My setup wakes up every hour listens for the polling data from any local Miflora using builtin esp32 bluetooth and then transmits data over Lora – battery life has been good and a little clingfilm on the miflora has so far proved an effective rain barrier. Because the mac address is also sent I can eliminate any duplicates caused by overlap within node red.

      1. I’m just revising it at the moment – the Arduino IDE bluetooth libraries had an issue writing to the old Miflora devices in order to initiate the transfer of readings – so I had workarounds for this. The latest firmware on the Miflora devices now publishes the readings in its polling data thus removing the need to write to them – you just capture the polling string and parse it. However there is still an issue with versions of the Miflora intended for the chinese market – these will not sync to the app and therefore will not easily upgrade to the latest firmware. Unfortunately it’s not always possible to know whether you are being sold the Chinese or the International version although all the latest ones I have sourced have all been international. The Chinese ones are still readable it’s just not as straightforward and I am still working on ways to upgrade the firmware other than through the app. Having said all of this if you have international versions then the process is simple – Listen for bluetooth polling – parse data – send via Lora – sleep. At the receiving end I pass every message received through to Node Red and display it through an extension to one of Csongor Varga’s excellent dashboards you’ll find on youtube. I don’t have any automated watering devices just 3 battery powered esp32/lora boards monitoring the house and gardens plus an additional receiving device which receives over lora then retransmits to the pi

  4. Peter, one idea might be to use the capacitive sensing in the ESP32. Just run some wires from an enclosure to the pot as probes and if you want absolute readings, calibrate the touch values it reads using one of the other sensors you have as a reference. I haven’t tried it, but I’d bet they are sensitive enough.

    It’ll probably end up being more expensive per unit, but the ESP32 has up to 10 touch inputs so in theory one unit could serve several plants.

  5. Hi Pete

    I built my own based off a similair prinicpal to the cheap chinese capacitive sensors that use 555 timers. It sends out a voltage value you can read with the esp8266 adc (with a voltage devider) and control everything through node red through your script you provide for esp8266. as for making them actually waterproof i have a 3d printed housing which i then fill with silicon that covers the esp8266 and the top control electronics of my sensor. I will add all my sensors are in inside plants and we dont havae cold winters here in Brisbane anyway so its unlikely they would be exposed to as harsh as conditions as over in europe.

      1. It’s actually based of the schematic dfrobot supply and the analogue electronics use the principles found in the diagram. I believe you have to use a high frequency because the effect of different minerals in the soil is lessened at higher frequencies. I believe you can actually get the esp to output at 80mhz. So you could presumably get the moisture readings through a few external components and the esp clock.

  6. oh, i like this concept! And i have A LOT of these “little bottles” (they’re called “preforma” – pre-shape – in Italy: they’re heated up and inflated into a model that force them into the bottle shape, and they become full PET bottles, in various sizes)

    1. Lucky to have them Antonio, I have seen them before in process called blow moulding where a mould is closed around the ‘bottle’ and then air is injected to blow the bottle out to the mould shape. I used to make those moulds years ago.
      What sensor is that you have in the picture?

  7. I’ve solved this problem.

    I use an old analog solution where I have 10 sensors built in to the ends of my hands. I call them “fingers”

    I take one of “finger” sensors and put it into the soil, if it detects moisture I don’t need to water the plants, if it feels dry, I water the plant. The best part is the “fingers” are completely waterproof!

    Sorry for being silly 😉

    1. We have a slight bug with “fingers” here.

      When we insert “fingers” into the soil, “funnel web spiders” give them a little nip. Shortly thereafter “fingers” stop working.

  8. This might be of interest, though a bit more costly, it seems a better idea to keep the sensor on the end of a cable and the electronics in a IP69 rated enclosure well away from the elements:

    (URL shortened vanilla link to AliExpress)
    https://goo.gl/Pr8X9a
    or just do a search for ” FS200-SHT10″

    1. That’s an Sensiron SHT10 sensor in a sintered metal housing not recommended for prolonged submersion in liquid , > 1 hour, so best not to over water! Arduino Sensiron library should run it OK. Adafruit have a similar module. Module above is available from HK for approx. £14.00 in the UK.
      sintered metal, no ferrous filters are common in pneumatic systems so it might be possible to find a suitable filter and install a SHT10 sensor in it for a DIY job.

  9. I built resistive as well as capacitive sensors. The former, in spite of all their concerns are performing quite acceptable for hobby farming and when you only apply a voltage for the few microseconds you are taking a reading, there is practically no electrolysis. Resistive also has the advantage of not needing electronics close by: 2 nails and a long wire is all you need to have near the soil.
    Tried the gypsum kind as well. See no advantage in it. The readings are always lagging behind

  10. “The guy with the Swiss accent” just did a YouTube blog about soil sensors:
    https://www.youtube.com/watch?v=udmJyncDvw0&feature=push-u&attr_tag=dyXapvmZ306wfQTU-6

    I hadn’t realised that the capacitive sensors would work even if the bottom part of the sensor is inside a plastic bag. It’s also worth noting his comment about the sides of the sensors not being waterproofed and needing some protection adding to stop them failing over time.

    Pete.

  11. DrFragle and MrShark posted the github link for the original “Chirp” software and hardware but I thought it might be interesting to also include the original link with the overall concept and design. i saw this one long before the chinese clones started popping up

    https://wemakethings.net/chirp/

    I am not a big fan of exposed/corroding electrodes especially when combined with high nutrient levels. If the capacitance approach works in principle I would welcome a design that replaces the exposed and beeping back-end with more robust and useful one including wireless. Might have to make it myself

    J.D.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.