ESP8266 Community Forum

I had a standard Nodemcu board, BME280, solar panel, Lipo battery and charger last year and it worked for about 3 months unattended until I had a problem with the Raspberry Pi I was using to save and display the data. I could not leave over the winter anyway due to the battery charging below 0 deg and there probably would not have been enough sun anyway.

I brought another single Lipo and picked up a couple of four Lipo cell battery packs that were being thrown away and decided to revamp the setup and add a couple more sensors for the practice.
I now have a board with:
Solar panel
Battery charger
Single Lipo connected to the battery charger – I am not sure if connecting it to a four cell pack may damage the charger
Four cell lipo pack
Auto source switch over IC – I later found there is a demo breakout board for a different version of the IC
Mini Nodemcu
2 x Soil moisture sensors
LDR for light
BME280
ADC
Battery level recording
Note a couple of capacitors and the P MOSFET are surface mount and soldered on the back.

Over the winter I plan on removing the solar panel and replacing the single cell with the second battery pack. I am not sure how long the packs will last but I suspect about 3 weeks each.

I am currently not using deep sleep mode as it was causing my website to refuse the connection after two or three connections. When using delay there are not a problem, which is strange. I do not want to test any more as the server refuses my connection for two hours. I have a P type MOSFET which turns all the sensors off when not required.

The soil sensors are a bit random which I suspected they would be. It might be a problem with my soil as it is mostly light compost. Just had a thought and they probably need to "settle" after powering up; I noticed that when testing them indoors?

Currently I am sending the data to my website, saving it to a database and displaying the data using a combination of php and JavaScript. I also get warning emails when the batteries reach a certain level.

It has been an interesting project and I have learnt a bit about electronics along the way.
I should have added some connectors to the box to enable easier changing of the batteries. I intended on having them inside the enclosure and the single cell fits but the four cell packs are a bit to big.

I am still setting everything up and have some data here: http://rubblewebs.co.uk/IOT/
I am using Dygraphs to display the data and there is some smoothing as indicated in the bottom left corner of the graphs. This can be changed and you can zoom into an area of the graph by clicking and dragging over the area of interest. The bottom gauges are using some Google visualisation code.
I also have a long term stats page in progress which will take the highest/lowest value from each day and display them. The main graphs will probably be limited to a month although I will still have the data and so can recall any month I like if/when I add the code for that option.

Interesting to see on the 24 March ( apart from I did not originally have the battery pack installed ) the system is now running on the battery pack and the single battery is being charged. This should switch back over at about 3.5V

Looks like a very nice design, very neat layout, nice!

Sometimes people add small heat sources inside projects to keep it over 0c outdoors in the winter, I know in your case it seems there is not enough power available, but its always a thought... I am sure a 5W resistor would could do it and some insulation...

RichardS

Thank you for the comments RichardS.

Good point about the heater but I think there would not be enough sunlight to charge the battery anyway; I will see how it looks over the year.

I made a bit of a mistake with a couple of environmental testing units I have on the roof at work. I put the PSU in the box with the sensor and it keeps the sensor warm which was partially defeating the object of the test. When it was -6C the internal thermometer only went down to -1C Although it still found a problem with the sensor firmware - when it has a negative reading it messed up some of the output data.

I have just noticed another potential problem. The single battery monitor is going through a potential divider and there is another potential divider to Vsense. So even if the circuit is not powered by the battery it will still discharging. Looks like I will need another component - a Low Power Auto Reset! But I can not isolate Vsense otherwise it will not switch back over when the single battery is charged.

I suppose I can monitor this and perhaps increase the switch over voltage from say 3.5V to 3.6V. That would mean modifying the potential divider to Vsense.