INDI weather station

[Gina]

I'll certainly be looking at your weather station implementation Wim as I want to make it INDI compatible and use the RPi with WiFi rather than Arduino.  I'm using the Melexis MLX90316 rotary encoder chip for the wind vane with Arduino code and don't know if there's RPi code for it.  OTOH I'm not entirely happy with this chip and considering other ways of measuring the wind vane angle.

[Gina]

Read through the thread and I see you're not implementing a full weather station - just enough for controlling your roof.  I expect to add cloud sensor and SQM meter in time but currently I'm looking at a pretty much standard weather station with wind speed and direction, temperature, humidity, atmospheric pressure and rain measuring.

Actually, I'm not so sure adding a weather station to my KStars/Ekos/INDI system is really needed.  Think I'll look into other ways of using WiFi.

Edited June 21, 2020 by Gina

[wimvb]

14 hours ago, Gina said:

Read through the thread and I see you're not implementing a full weather station - just enough for controlling your roof.  I expect to add cloud sensor and SQM meter in time but currently I'm looking at a pretty much standard weather station with wind speed and direction, temperature, humidity, atmospheric pressure and rain measuring.

Actually, I'm not so sure adding a weather station to my KStars/Ekos/INDI system is really needed.  Think I'll look into other ways of using WiFi.

It's not a full station yet. I have a rain sensor which I haven't implemented yet, and am struggling with an anemometer. Most of the sensors are very basic. They all use i2c, which is easy to implement. The rain sensor is analog, also easy. The anemometer uses rs485/modbus, which is a completely new protocol for me. Information on the internet suggests that it's straightforward, but not so. I can read values from the anemometer with a windows modbus master. I can also communicate with arduino code and a modbus slave emulator on windows. So I know the various parts are working as they should. But I can't get the anemometer to talk to the arduino or esp32. I've tried everything, including various ttl-rs485 interface boards and chips, but no success. I will try connecting it to a rpi with the usb/rs485 adapter, but if I can't get that to work, I'm considering hacking the anemometer and use its analog signal, bypassing the modbus logic alltogether.

This is my hardware

https://www.amazon.co.uk/gp/product/B07WD8LZD6/ref=ppx_yo_dt_b_asin_title_o02_s00?ie=UTF8&psc=1

https://www.amazon.co.uk/gp/product/B078X5H8H7/ref=ppx_yo_dt_b_asin_title_o03_s00?ie=UTF8&psc=1

https://www.amazon.co.uk/dp/B0721BB8PQ/ref=dp_cerb_1

Max485 chip

https://www.amazon.co.uk/gp/product/B07G1867ZX/ref=ppx_yo_dt_b_asin_title_o00_s00?ie=UTF8&psc=1

Edited June 22, 2020 by wimvb

[Gina]

I just used a Hall effect switch and tiny magnets on the anemometer to count revolutions for wind speed. That's with my own design 3D printed anemometer.

Edited June 22, 2020 by Gina

[jiberjaber]

I think most anemometers use read switch pulses, my original AAG one did and it's replacement Davies one does to (though I am just using their instruments attached to my own Arduino ESP32), I almost went down the route of an ultrasonic based anemometer but it looked a bit too difficult to homebrew at the time and the Davis kit managed to swing the vote for it's easy of mounting. I coupled that with their rain bucket as well.

http://gillinstruments.com/products/anemometer/principleofoperation2.html

One word of experience is make sure you have a robust humidity sensor, the BME style sensors are not reliable overtime even in a screen box as they are not condensation resistant. I went for a solution based on a one-wire battery sensor IC and a more robust humidity sensor.

https://taaralabs.eu/1-wire-humidity-temperature-sensor/

 

[JamesF]

16 minutes ago, jiberjaber said:

I think most anemometers use read switch pulses, my original AAG one did and it's replacement Davies one does to (though I am just using their instruments attached to my own Arduino ESP32)

That's also my understanding.  I think some of them actually have two reed switches placed next to each other, though I'm not sure why.

James

[wimvb]

The cheap chinese anemometers, such as the one I have, come in a variety of flavours, from modbus to pulse to analog output. If I can’t get mine to work, I’ll have a look at the electronics inside. What appealed to me when I bought it, was that the rs485 version was already calibrated.

[Gina]

Good point.  Mind you no anemometer is going to read the true wind speed unless it's high up and well away from buildings or trees.  Few people will have a suitable place to erect a tall mast for the wind instruments.  There's a fight for my best location!!

[jiberjaber]

42 minutes ago, Gina said:

Good point.  Mind you no anemometer is going to read the true wind speed unless it's high up and well away from buildings or trees.  Few people will have a suitable place to erect a tall mast for the wind instruments.  There's a fight for my best location!!

There's a met office standard about the height of the equipment (10m I think) - mines on a scaffold pole and still suffers from some shelter in a couple of directions, less so in the winter due to tree foliage. 

https://www.metoffice.gov.uk/binaries/content/assets/metofficegovuk/pdf/weather/guides/10_0230_fs_17_observations.pdf

 

[Gina]

Yes, 10m above ground is the standard.  I had mine on a mast 6.8m high but it had to go because it interfered with my astro imaging.  Planning regs limit the height and it has to be self supporting.

[wimvb]

Location isn ’t going to be much of a problem; I live well away from any neighbours, my house is on a ridge, and the few high trees around are mine, and planned to go down in the not so distant future. But the anemometer will definitely not be placed on a 10 m high mast. Just above the obsy will do for me.

[wimvb]

On 22/06/2020 at 10:53, Gina said:

I just used a Hall effect switch and tiny magnets on the anemometer to count revolutions for wind speed. That's with my own design 3D printed anemometer.

 

On 22/06/2020 at 12:40, jiberjaber said:

I think most anemometers use read switch pulses, my original AAG one did and it's replacement Davies one does to (though I am just using their instruments attached to my own Arduino ESP32), I almost went down the route of an ultrasonic based anemometer but it looked a bit too difficult to homebrew at the time and the Davis kit managed to swing the vote for it's easy of mounting. I coupled that with their rain bucket as well.

http://gillinstruments.com/products/anemometer/principleofoperation2.html

One word of experience is make sure you have a robust humidity sensor, the BME style sensors are not reliable overtime even in a screen box as they are not condensation resistant. I went for a solution based on a one-wire battery sensor IC and a more robust humidity sensor.

https://taaralabs.eu/1-wire-humidity-temperature-sensor/

 

 

On 22/06/2020 at 12:58, JamesF said:

That's also my understanding.  I think some of them actually have two reed switches placed next to each other, though I'm not sure why.

James

I opened my anemometer just the other day. It contains 1 reed switch in the static, bottom part, and 8 (!) neodyme magnets in the rotating top. Three wires lead to/from the reed switch: Red, Black, Yellow. The whole sensor part is covered with silicone, so i can't see how it's connected. My guess is that the red wire (which supplies 3.3 V to the sensor) is connected to a resistor which is in series with the switch. From the resistor/switch, the yellow wire gets the signal out. And the other side of the switch is connected to ground (black). I verified that the voltage on the yellow wire goes low whenever a magnet passes the reed switch. The pulses are fed into a microcontroller for processing, and then into a MAX485 type chip for RS-485 signal conversion.

The MCU is this one (overkill if you ask me)

https://www.st.com/en/microcontrollers-microprocessors/stm32f030f4.html

I still haven't got the anemometer to work with either an Arduino or an ESP32. But it does work with a Windows based Modbus test tool. And my Arduino/ESP32 software works with a windows based modbus slave emulator.

I'm leaning more and more towards butchering the anemometer and just get the pulse output of the reed switch. If I remove the mcu, I will also lose the calibration. Hacking the anemometer isn't much of an issue, but I would need another wind speed meter to calibrate the hacked device. Otoh, I could then turn the anemometer into an I2C device, as are all the other sensors in my weather station.

[wimvb]

On 22/06/2020 at 10:53, Gina said:

I just used a Hall effect switch and tiny magnets on the anemometer to count revolutions for wind speed. That's with my own design 3D printed anemometer.

Did you do any calibration to get wind speed in m/s, or does your anemometer just report revolutions/minute?

[Gina]

I seem to remember that somehow I estimated that a third of a revolution per second represented 1mph but I'll need to go through my Arduino weather station thread to see if I can find it.

[Gina]

Found the post but it doesn't say how I got the calibration info.

[JamesF]

I seem to recall reading somewhere that with anemometer cups of the commonly used sizes you can assume a given wind speed gives you one revolution per second.  I can't find the reference at the moment though.

James

[JamesF]

Aha!  I was forward-thinking enough to document it in my code.  The figures I have are that you divide the number of revolutions per second by:

• 1.096 to give m/s
• 3.9477 to give km/h
• 2.453 to give mph
• 2.130 to give knots

James

[JamesF]

Just checked the OWW code and that uses 2.453 for mph, too.

James

[JamesF]

This paper looks like it might provide a reasonable basis for guesstimating wind speed from a spherical cup anemometer of one's own design:

https://core.ac.uk/download/pdf/148670067.pdf

James

[Gina]

37 minutes ago, JamesF said:

Just checked the OWW code and that uses 2.453 for mph, too.

James

Ah, so the value of 3 that I assumed wasn't that far off, but the PDF may give a better value.  I shall be reading it.  Thanks James.

[wimvb]

2 hours ago, JamesF said:

Aha!  I was forward-thinking enough to document it in my code.  The figures I have are that you divide the number of revolutions per second by:

• 1.096 to give m/s
• 3.9477 to give km/h
• 2.453 to give mph
• 2.130 to give knots

James

That should depend on the diameter  of the anemometer head (more than the cups). I’ll read that article you referred to. Thanks.

[Gina]

PHEW - Ma bwain 'urts Bwian!!!  Just been "reading" the PDF.  I shall need to try to read this again to get the calibration factor for my anemometer.  One thing stood out though - the size of cup made little difference but the radius of cup centre from axle did.  Intuition (or common sense) tells me that bigger buckets/cups should be better at overcoming friction in the bearing.  I wonder if anyone has read this article and boiled it down to basics.  How bucket centre radius relates to calibration.  May go and have a lie down!

[JamesF]

I only skimmed it, but I think you may be able to get away with just reading the conclusion (about page 10 onward?)

James

[Gina]

I've found

Quote

Ar=20.7Rrc

With Rrc in mm but not found what Ar is in.

[Gina]

Hmm...  A is in metres/sec per Hz so presumably Ar is too.

My Rrc is about 90mm, giving Ar = 1800.  That doesn't make any sense!!  I would expect a value around 1.