Weather Station Ideas

[Gina]

Successfully cancelled the delivery of the ESP8266s and will order an ESP32.  Looks a much better bet - everything in one board.

[Gina]

ESP32 ordered.

[Gina]

That would seem to be the ideal solution.  A genuine Arduino Nano costs around £30 - 4 times the cost of the ESP32.  I've found the cheap Chinese Nano clones to be unreliable - I have a large pile of dud ones.  No problem with genuine ones except the cost!  The ESP32 is a bigger than the Nano but not too big.

I shall probably order another ESP32 for the other weather station location viz. the observatory.  One in the wind sensor casing will cover wind measurements and can also read light level.  The observatory will have temperature and humidity (maybe pressure) in the Stevenson Screen and temperature and humidity in the scope room.  Not yet sure where to put the rain gauge but one possibility is on the wind sensors mast.  It used to be on a framework on the grass a couple of yards north of the observatory but that wasn't very convenient.

My plan is to just take simple readings with the ESPs and collate all the data on my main PC indoors.  I shall need to see how I program that.  Maybe Python.  I may request to "borrow" other members ideas and code.  I had lots of displays written in Python and using GnuPlot for graphs, rose and pie charts etc. some years ago but that's long since lost!

Edited June 27, 2020 by Gina

[skybadger]

For me, the esp8266-01 is tiny which is the benefit over the -12 or -32. I have many of each to hand...

[Gina]

Yes, that's certainly a point.

[7170]

@Gina I've had a ESP32 taking light, temp and humidity readings and sending the data to a server for about five months now as a test for a longer term project. Very reliable, but i'm not sure how strong the WiFi is as mine is currently close to the router.

I found WiFi on my ESP8266 to not be that sensitive as it couldn't connect from the garage at my parents which is close to the router. It is currently there taking outside temperature (via a 1-wire sensor) plus pressure readings and submitting to WeatherUnderground every few minutes.

[skybadger]

I get 20m out of mine, across the garden from a wall plug AP.

[Gina]

That would be well adequate

[Gina]

The ESP32 seems like overkill but the ESP8266 is not enough so overkill it is!  OTOH I'll look at ESP8266 versions with more pins and see if one would serve for the wind sensors.

Edited June 28, 2020 by Gina

[7170]

What sensors are you planning to use @Gina? 
 

I have to start thinking about what to do with my station when things start to wear out. Currently I have a combination of Oregon and Davis sensors, plus a rain wise tipping bucket, one wire temperature sensors and home made made soil, solar, leaf wetness and lightning sensors.

In terms of receiver two arduinos (one for each protocol) intercepting the wireless transmissions connected to a pi2 running a python script to collect/process the data (calculate things like 10 min dominate wind direction etc) and saves to a mysql DB every minute. For all my custom sensors they are wireless using off the shelf 433mhz modules and barebones ATmega328s to save power, and I emulate the oregon v3 wireless protocol which the receiver then picks up.

No off the shelf weather station is perfect and if you look at some of the more expensive ones they are still using off the shelf sensors that you can buy fairly cheaply.

[Gina]

Haven't decided on the sensors yet - I have some experimenting to do.

[Gina]

This is an ESP8266 with lots of pins but smaller than the ESP32.

[Gina]

Reading the instructions, how to use on the Amazon site it mention Win32 and Win64 so may need windows and I have Linux only so I'm doubtful of this version.

Quote

How to use?
1. Download the Arduino IDE, the latest version. 
2. Install the IDE
3. Set up your Arduino IDE as: Go to File->Preferences and copy the URL below to get the ESP board manager extensions: http://arduino.esp8266.com/stable/package_esp8266com_index.json. Placing the http:// before the URL lets the Arduino IDE use it...otherwise it gives you a protocol error.
4. Go to Tools > Board > Board Manager> Type "esp8266" and download the Community esp8266 and install. 
5. Set up your chip as:
Tools -> Board -> NodeMCU 1.0 (ESP-12E Module)
Tools -> Flash Size -> 4M (3M SPIFFS)
Tools -> CPU Frequency -> 80 Mhz
Tools -> Upload Speed -> 921600
Tools -> Port -> (whatever it is)
6. Download and run the 32 bit flasher exe at Github (Search for nodemcu/nodemcu-flasher/tree/master/ at Github)
github.com/nodemcu/nodemcu-flasher/tree/master/Win32/Release
Or download and run the 64 bit flasher exe at:
github.com/nodemcu/nodemcu-flasher/tree/master/Win64/Release
7. In Arduino IDE, look for the old fashioned Blink program. Load, compile and upload. 
8. Go to FILE> EXAMPLES> ESP8266> BLINK, it will start blinking. 

 

Edited June 28, 2020 by Gina

[7170]

For my ESP boards (32 and 8266) alI ever needed to do was add them in board manager and nothing further. The only funny with the ESP32 is I have to hold reset until the upload starts then I can let go. Never had to use any separate flashing app. 

[skybadger]

That will work fine under Linux, using the Linux Arduino ode or the esp tools which are python.... setup the board library and you are off running. 

That is a -12f version. 

[skybadger]

Once you have the boot loaderi n there, you can update over the air using the updater library.

Edited June 28, 2020 by skybadger

[Gina]

Thanks everyone - good news.  I like the smaller size for mounting atop the wind sensors mast.  Think I'll order one - should arrive tomorrow.

The ESP32 is due to arrive around 1pm by DPD, the Hall devices and optical sensors by Amazon currently saying between 3pm and 6pm.

[Gina]

BTW, does the orientation of the ESPs matter or are the aerials omnidirectional.

[Gina]

As for sensors, the anemometer has a single Hall effect switch with 3 miniature magnets on the rotor.  I estimated this as giving 1pps per mpg of wind speed.

The wind vane will probably use 4 reflective optical sensors with Gray code disc giving 16 directions and 4 bits of data.  This needs analogue inputs as the phototransistor current varies.  OTOH I will be trying 8 Hall switches with one magnet on the rotor with overlapping operation like the cheap wind vanes use reed switches.  This would then use 8 digital inputs.  Both systems have their pros and cons.

If anyone has any comments on these ideas I would be glad to see them.

Edited June 28, 2020 by Gina

[Gina]

Looking at the GPIO pins I see 10 Capacitive sensing GPIOs.  Must investigate.  Wonder if these could be used with the wind vane.

[Gina]

Found this

Quote

Capacitive sensors

 

Capacitive sensors are capable of detecting plastic, wood, and other raw materials including metal. An inductive sensor can detect only metal. A common application is the detection of liquids, plastics, and grains. Capacitive liquid detection is used for level and presence detection.

If these pins measure capacitance as seems the case, I think I could make a Gray code plate as a variable capacitor in conjunction with small metal plates as the sensors.

Edited June 28, 2020 by Gina

[Gina]

I've ordered the IZOKEE NodeMCU Module ESP8266 ESP-12E from Amazon.  Due to arrive tomorrow.

Edited June 28, 2020 by Gina

[7170]

28 minutes ago, Gina said:

Found this

If these pins measure capacitance as seems the case, I think I could make a Gray code plate as a variable capacitor in conjunction with small metal plates as the sensors.

I would guess this would be in the low pf range, if so this is a tough call to consistently read the exact value without external hardware and that is before you factor stray capacitance and the fact the capacitance of the wires/ESP input may be too close to your actual capacitor. I found this while making capacitive leaf wetness sensors, had to use a 555 and then read the frequency change. The other problem will be keeping it dry and consistent humidity as the value will change depending on humidity. Personally I would use a magnet on the shaft and number of reed relays and a resistor network if I were building one. A lot of ones you can buy actually just use a linear potentiometer, hence why they don't last forever. The idea of a rotating card and optical sensors will work. The hardest part is calculating the dominant direction (via vectors) as you obviously can't use just an average.

Regarding wind speed I would only have one magnet so you get one trigger per full rotation. You will want to have as long a debounce time as you can based on the max windspeed you would want to measure. Three magnets could result in too short a debounce time.

Hope the above makes sense.

[Gina]

I agree about the capacitance - it was just a thought but I can see it wouldn't work.

I haven't found debounce needed with Hall devices but I can see a problem if I used Hall switches with the wind vane as that can "wobble" due to wind turbulence.  The Hall devices have a Schmitt trigger circuit built in so don't suffer bounce with a fairly steady increase or decrease in magnetic flux.  I have examined the output on an oscilloscope and there was no "bounce" - just a clean, square pulse shape.

I used reed switches in previous wind vanes but did find the the magnet was attracted to the reeds and would "cog" as it moved in a light breeze and although the magnet was spaced to give an overlap representing 16 directions, statistical analysis showed a distinct preference for the 8 main compass points.  I did a lot of statistical analysis on my weather stations years ago.

With the wind vane data, it is first necessary to decode the Gray code to binary.  Then the sampled data can either be added into a Rose Chart or converted to an average direction using Consensus Averaging.

Edited June 28, 2020 by Gina

[Gina]

The ESP32 has arrived and it isn't all that big - just over 1" wide and 2" long (28mm x 50mm).  Now I'm waiting for the optical and Hall sensors.