Topic: Automated Weather Station - Ongoing Build

[TheBushPilot]

Hi all,

I've decided to create a thread to document the progress on my new AWS.

Thus far I have acquired the tripod, enclosure boxes, and have been working to get instruments talking with the data logger. The 61302L and HMT337 communicate via serial RS-232. What's cool with the Vaisala is that you can manually set the pressure to compensate for more accurate water vapor measurements like dew point. An Arduino will read and upload data to a database on my webserver since it is cheap and incredibly low power.

Wind speed/direction, temperature, relative humidity, and dew point, as well as pressure, rainfall, and solar radiation will be measured.

Here's the tripod and the power enclosure:

 [ You are not allowed to view attachments ]

And the DAQ:

 [ You are not allowed to view attachments ]

I would have liked to use a CR23X but space did not allow using the HMT337. CR10X will suffice despite having a slower baud rate, since this will force a slower program execution hopefully negating the gradient dampening of the PRT a little bit. This particular variant of the 337 has the warmed dew point probe like ASOS which is a little higher power draw but shouldn't be a problem with a 60Ah or so SLA battery.


Cheers

[TheBushPilot]

Update - Near Completion:

 [ You are not allowed to view attachments ]

I did end up going with the CR23X, however the HMT337 required slight modification to accommodate how large both devices were. And beside the Vaisala is the 61302L for pressure measurements. The CR10X severely limited what I was able to do data transfer wise from the logger to Arduino so I went for the 23X.

 [ You are not allowed to view attachments ]

Wire management still needs to be done but for the time being it works. I have a GPS16-HVS setting the time automatically at 00Z every day so there's no need to intervene.

Here's the station rundown from top to bottom:

R. M. Young 05103L Wind Monitor for wind speed and direction, quad plate pressure port for increasing barometric pressure measurement accuracy in wind. R. M. young 41003 Multi-Plate Radiation Shields for both the warmed HMT337 dew point probe and external PRT. The enclosures are 2 old style Campbell Scientific ENC12/14 which house a 55Ah SLA battery charged with a Campbell Scientific CH100 and Amazon 20 watt solar panel. Campbell Scientific CR23X and Vaisala HMT337 Humidity and Temperature Transmitter and R. M. Young 61302L Barometric Pressure Sensor housed together. All mounted on a Campbell Scientific CM110 tripod.

The HMT337 required the detachment of the wire glands for both T/Td probe wires so that the box could fit. Since the mounting holes were offset by 0.75", I opted to zip tie the enclosure to the back plate which seems to be working fine.

All that remains is acquiring an Arduino with a U.FL antenna socket that will be connected to a high gain directional yagi antenna as the stock MKR 1010 Wifi did not suffice range wise. The Arduino will push data every 5 seconds to my SQL database with which will allow my site to display minutely data in current and graphical displays.

Hopefully will have live data fed to my website in the coming days.


Cheers

[TheBushPilot]

Final update until move to final location:

After trial and tribulation the AWS is finally online as of yesterday night 02/02/2024.

Yagi antenna installed for "long range" (160ft from the building) WiFi connectivity - just some $30 one I found on amazon. Was originally to use a Sparkfun ESP32 Thing Plus but the chipset was probably defective and would run a brownout boot cycle when trying to execute the WiFi communication with the antenna. Did a little research and opted for the QuinLED-ESP32. This board is incredibly high quality and should have no problem operating for long term for the station. This board did also have issues with power consumption so I put a smoothing capacitor between power and ground and that fixed the issue. As with all different boards, the programming was nuanced, and this one was no exception. After troubleshooting the issues and getting everything correctly outputting, inputting, and transmitting online, all that was left to do was plug n play the board into the system.

I made a point to design it for ease of modularity, this being no exception. Took 5 minutes to get the station online and uploading to my webserver. A simple swap of board and program updates and she was good to go.

 [ You are not allowed to view attachments ]

Next steps?

The site is a little disheveled with labels as the station location and ID have changed to TRX001/New Truxton, MO where it will permanently reside on a friends property. From there it will remain logging away until we install the infrastructure for a more robust platform in the coming year or so.

This next iteration will be a little more involved. Upon designing it I actually indirectly came to a similar conclusion to a well known AWOS manufacturer Mesotech. Their stand alone DCP design is one I look to adopt and adapt.

- 10 meter wind both mechanical and heated ultrasonic for redundancy in winter.
- 2/9 meter FARS temperature and co-located dew point probes. (Inversion/lapse rate measurements possible.)
- DCP located 2m temp/dew point HMT337 for published thermodynamic obs. (Cables will be extended so the use of FARS for PRT will be possible.)
- 3 pressure transducer redundancy.
- Rainfall (Heated in winter.)
- Solar radiation with class 1 pyranometer.
- 3m Visibility & Present weather.

DCP will utilize a CR1000 and serial to ethernet server for IoT. CR1000 for analog measurements and USR server for serial devices. These will connect to a modem with which an Arduino or some IoT device will scrap from and send to my webserver.

The sort of hierarchy I'm going for follows:

Analog sensors -> CR1000

• Wind Monitor
• Thies Clima Ultrasonic
• Fast Response Thermistors
• Setra Transducers
• Pyranometer
• Rain Gauge

Serial Devices -> USR Serial Server (4 Port RS-232/RS-485/RS-422)

• HMP155
• HMP155
• HMT337

Not sure if the CR1000 will input to the server or if it will connect to the modem directly - up in the air at this point.

What's nice about the HMT is it can output a multitude of parameters including but not limited to Temperature, Relative Humidity, Dew Point, Wet Bulb Temperature, Mixing Ratio, etc. Simplifies processing requirements since the sensor itself does the processing prior to data input.

(May integrate 7 meter UVW wind measurement at some point as well for "research purposes" but that might end up being used for the radiation shield intercomparison down the road.)

Here's an example of a Mesotech AWOS:



I have big plans for the station in the coming years and look forward to documenting its progress. Incredibly blessed to be taking on one of my childhood dreams.


You can find the live station data here:
https://www.met-instruments-project.com/


Cheers

[TheBushPilot]

Well we haven't moved to the new location but I figured this was update worthy:

I've made a bit of an upgrade - acquired a CR1000 with the NL121 network peripheral. Changed up the placement of everything such that the data logger resides in the same enclosure as the battery and charge controller. I plan on adding a USR-W630 for ethernet to WiFi connectivity with the data logger and Arduino. The CR1000 is configured with an API of sorts through a webpage accessible via browser on your LAN. This page displays your table of choice with which the latest values for the measurement period are output. The Arduino will scrape these values from the page and construct and send a HTTPS query to a PHP script on my web server. This PHP script is in one form or another a crude API from the external device to the SQL database acting as the net between external traffic and the internal database.

Data table from CR1000:

 [ You are not allowed to view attachments ]

Both CR1000 and Arduino will be connected via ethernet to the USR device. The USR will have an external WiFi yagi antenna (probably maxrad or similar) to connect to the LAN on the property a couple hundred feet away. I figure doing away with the physical serial communications would significantly reduce data loss issues and increase overall robustness in the system, simplifying implementation.

I've sort of laid out the plan of execution of everything as a sort of two part thing:

Installation 1 consists of the 3 meter tripod with applicable battery powered instrumentation allowing for a complete suite of measurements short term. This setup will include all required instrumentation for the final DCP install with the second part of the plan. The primary goal for this will be integrating all instrumentation for the final DCP into a single cabinet that will be transferred via backplate from one designated enclosure to the other. As each specified enclosure has either 1 or 2 openings. The idea is to simply unscrew the backplate, take it out, and put it back in with all the instruments still attached to eliminate complexity. This will be done for both DCP and instrument enclosure.

Installation 2 consists of the permanent install. 10 meter tower, dedicated fixture for mounting enclosures and applicable instrumentation as well as AC available power to the site. Should be significantly less effort put into mounting everything or we hope so.

Here's the packed cabinet with data logger, charger, battery, and soon to be included ethernet to WiFi converter:

 [ You are not allowed to view attachments ]

And the instrumentation cabinet, soon to house 3 barometric pressure transducers in addition to the already HMT337:

 [ You are not allowed to view attachments ]


And a bit of a side note - I plan on temporarily deploying this station in the center of the path of totality for the April 8th solar eclipse. Think that would be really cool to archive.


Cheers

[TheBushPilot]

The AWS is nearing completion for the April 8th, 2024 total solar eclipse. Finally acquired the pressure transducers for barometric pressure measurement, and figured out how I am mounting the pyranometer to the tripod. I opted to utilize an Apogee Instruments ST-110 thermistor temperature sensor in R. M. Young 43408 FARS for fast response temperature to supplement the lagging response of the HTM337's PRT in the Gill shield. Given this setup is temporary I am not worried about power as I will not be using the FARS operationally when semi-permanently installed.

After this event, as previously mentioned I intend on installing it on a friends property in a small town in east central Missouri where it will reside for a few years until I am ready to move forward with the permanent installation. Turns out I will be adding an addition enclosure due to the need for more battery power. The 55 Ah SLA I originally had planned on using and putting in the second polyester fiberglass reinforced enclosure will not cut it. I've opted for another of the same battery and am putting them in an aluminum enclosure I acquired a while ago. This way I will have a combined total of 110 Ah available for the station and required IoT equipment. This will get me my ~2 weeks of reserve. Power will be maintained with a 100 watt panel and adequately matching charge controller.

Here's how the pyranometer mounts - I went this route as the sensor itself weighs around 7 lbs. I did not want this thing to get any ideas. The base to unistrut mount was the perfect size for the Eppley PSP incase anyone needs to mount theirs. Lol. There were already pre-tapped 1/4-20 holes for bolts that I threaded and drilled out holes a little larger than the diameter of the bolt such that the angle of the device could be adjusted based on the position of the nuts. It's actually resting on the box the batteries are going to reside in.

 [ You are not allowed to view attachments ]

Here are the transducers mounted to the back plate, and within the enclosure:

 [ You are not allowed to view attachments ]

 [ You are not allowed to view attachments ]

And R. M. Young FARS with the fast response thermistor:

 [ You are not allowed to view attachments ]

I was originally going to use the sonic temperature from the 81000 3D anemometer. However, the accuracy and general characteristics of how it behaved in low wind was discouraging so I opted for the physical sensor. Too much variation and from what I recall the accuracy was ±2°C which was far greater than what I wanted.

The next update will be an in depth discussion over components used and installation. Hoping to have that completed by the end of April assuming all goes as planned...


Cheers

[countzero]

Are you using a thermopile pyranometer? if so, what model?

I've been cobbling a station on and off using rs485 MODBUS sensors connected to a rs485 to wifi serial bridge device that cost about $20 No local logging, just having my server pull it straight off the wire as it were

[TheBushPilot]

Are you using a thermopile pyranometer? if so, what model?

I've been cobbling a station on and off using rs485 MODBUS sensors connected to a rs485 to wifi serial bridge device that cost about $20 No local logging, just having my server pull it straight off the wire as it were

You'd be correct, it is a thermopile - Eppley Lab precision spectral pyranometer.

https://solardata.uoregon.edu/EppleyPSP.html

You can read the output as single ended or differential voltage. Not sure how you would get that to work with an RS-485 converter. I've got everything read via data logger.

[countzero]

I've seen a lot of thermopile pyranometers with various industrial standard outputs like 4-20mA 0-5v and RS485. Campbell Sci sells a secondary standard Kipp & Zonen that outputs via MODBUS. I'm not sure how the ADC affects the data though to be honest.

https://www.campbellsci.com.au/smp10-l

You seem to be operating on another level here of measurement fidelity here which is quite fascinating to see. Have you investigated plain Lora trancievers (not LoraWAN) for data transmission of packets back to base? If you are using Lora just point to point you can efficiently send a lot of sensor data back to base with more flexibility for distance and less hassle with directional wifi antennas

[CW7491]

How does the ST-110 do in the FARS? Is it too responsive/noisy? Is the ST-110 a simple 2 wire thermistor?

[TheBushPilot]

I've seen a lot of thermopile pyranometers with various industrial standard outputs like 4-20mA 0-5v and RS485. Campbell Sci sells a secondary standard Kipp & Zonen that outputs via MODBUS. I'm not sure how the ADC affects the data though to be honest.

https://www.campbellsci.com.au/smp10-l

You seem to be operating on another level here of measurement fidelity here which is quite fascinating to see. Have you investigated plain Lora trancievers (not LoraWAN) for data transmission of packets back to base? If you are using Lora just point to point you can efficiently send a lot of sensor data back to base with more flexibility for distance and less hassle with directional wifi antennas

Interesting. Full disclosure I haven’t looked very deep into solar radiation measurements, I just found the one I have for a good deal and went with it. I’m not sure how the conversion would effect the measurement but I would assume it’s negligible. Furthermore assuming it’s calibrated through that mode of communication it would be compensated for if I had to guess.

As far as connectivity goes, I’m using a NL121 Ethernet interface for the CR1000 with which is connected to a PUSR USR-W630 2 port Ethernet/1 port serial to WiFi converter that connects to my LAN. The converter has an external antenna with SMA connector that will have a 2.4 GHz maxrad directional yagi pointed at the building. Figure that was the easiest way to get wired data wireless. The CR1000 can be configured to output a sort of API HTML web page. I have an ESP32 arduino scraping that data and sending it to a SQL database on my web server every 10 seconds.

How does the ST-110 do in the FARS? Is it too responsive/noisy? Is the ST-110 a simple 2 wire thermistor?

I actually haven’t tested it yet. That being said for the eclipse I want very rapid response temperature measurements so noise will more than likely be encountered. But nothing that can’t be modified in post analysis. And I will also have the HMT337 as a sort of reference with the PRT for sanity check.


Cheers

[countzero]

As far as connectivity goes, I’m using a NL121 Ethernet interface for the CR1000 with which is connected to a PUSR USR-W630 2 port Ethernet/1 port serial to WiFi converter that connects to my LAN. The converter has an external antenna with SMA connector that will have a 2.4 GHz maxrad directional yagi pointed at the building. Figure that was the easiest way to get wired data wireless. The CR1000 can be configured to output a sort of API HTML web page. I have an ESP32 arduino scraping that data and sending it to a SQL database on my web server every 10 seconds.

Wifi really is the easiest and what I'm using too, but if you ever need more range or flexibility in siting you could basically drop in replace your NL121 and PUSR USR-W630 with something like this plugged into the serial port of your logger and another back into a monitoring device on your lan/laptop in the feild:

https://www.amazon.com.au/USR-LG206-H-P-convertisseur-dispositif-lagriculture-intelligente/dp/B083K7RDMD

I'm only going on about it because I've done a bit of Lora and particularly LoraWAN IoT infrastructure deployment particularly with standalone LoraWAN sensors and stations for agriculture/smart cities stuff. Sadly a lot of it is getting edged out by cellular solutions like NB-IOT and CAT-M which I find utterly painful technologies

[countzero]

Also if you don't mind me asking what is the most expensive single sensor you're using on this outfit, and where do you think you got the best deal and bang for your buck putting it together? If one (like me) on a smaller budget, wanted to put together something in a similar modular fashion of yours, where would the greatest areas of economising on sensors and accuracy while still delivering something pretty nice?

While Hi-Fi audio is obvious a completely different market (and mostly aesthetic with questionable technical claims) its an area where there is basically no limit to the money you can spend. You can get a cheap amplifier that might be acceptable, a great one for some real expense, or you could get one true acolytes fawn over priced in the kind area usually only reserved for realestate. I kind of want something professional, but not something priced for a research institution or state meteorology organisation. Is it possible?

[TheBushPilot]

As far as connectivity goes, I’m using a NL121 Ethernet interface for the CR1000 with which is connected to a PUSR USR-W630 2 port Ethernet/1 port serial to WiFi converter that connects to my LAN. The converter has an external antenna with SMA connector that will have a 2.4 GHz maxrad directional yagi pointed at the building. Figure that was the easiest way to get wired data wireless. The CR1000 can be configured to output a sort of API HTML web page. I have an ESP32 arduino scraping that data and sending it to a SQL database on my web server every 10 seconds.

Wifi really is the easiest and what I'm using too, but if you ever need more range or flexibility in siting you could basically drop in replace your NL121 and PUSR USR-W630 with something like this plugged into the serial port of your logger and another back into a monitoring device on your lan/laptop in the feild:

https://www.amazon.com.au/USR-LG206-H-P-convertisseur-dispositif-lagriculture-intelligente/dp/B083K7RDMD

I'm only going on about it because I've done a bit of Lora and particularly LoraWAN IoT infrastructure deployment particularly with standalone LoraWAN sensors and stations for agriculture/smart cities stuff. Sadly a lot of it is getting edged out by cellular solutions like NB-IOT and CAT-M which I find utterly painful technologies

I've never put thought into LoRa but that is certainly an option. I am fortunate enough that where this station will be sited is on a friends property with clear line of site to his house, so WiFi connectivity is guaranteed. That being said for others I cannot say the same. The other consideration is getting my data online, simply routing the data through serial is far less robust than having an external device grab the data from a page that is not going to have potential intermittent issues with packet loss causing data cleanliness issues if you will.

Interesting bit about the cell solution. I wouldn't even try to bother with that given you'd have to pay for data. No thanks, just one more thing to worry about.

Also if you don't mind me asking what is the most expensive single sensor you're using on this outfit, and where do you think you got the best deal and bang for your buck putting it together? If one (like me) on a smaller budget, wanted to put together something in a similar modular fashion of yours, where would the greatest areas of economising on sensors and accuracy while still delivering something pretty nice?

While Hi-Fi audio is obvious a completely different market (and mostly aesthetic with questionable technical claims) its an area where there is basically no limit to the money you can spend. You can get a cheap amplifier that might be acceptable, a great one for some real expense, or you could get one true acolytes fawn over priced in the kind area usually only reserved for realestate. I kind of want something professional, but not something priced for a research institution or state meteorology organisation. Is it possible?

Most expensive thing in this system over all was actually the data logger. The Campbell loggers hold their value to a fault, but to their credit will probably last longer than we will. I think the data logger will always be the most expensive component in an automated weather station given how well built they are plus the inherent supply/demand. Now with that being said it will be interesting given CSI has seemingly retired their main lineup of CRBasic loggers in the last 5 or so years (CR3000/1000/800). As such government agencies and educational institutions will inevitably upgrade due to the need for latest hardware with support/calibration. I would wager we see them more available on the market for a lower price in the coming years. And with that being said, you can find good deals on instruments - given you already know what to look for/have experience with the nuances - which just takes trial and error and experience. A lot of the time I see listings on eBay for equipment new or lightly used being sold for seemingly unknown reasons. That or retirement as with the aforementioned reason. A few years ago there were loads of listings for a particular kind of temperature and relative humidity probe (HMP35C) that were listed for pennies on the dollar. Looked to have finally been a stock pile of sorts that finally got auctioned off. I got a few and most looked to have had their RH chip replaced and probably had another couple years left in them. But that is just one example of many instances you may encounters be it on eBay or surplus sites.

You can put something quite impressive together for less or equal to that of a fully loaded Davis station. I actually did this for a friend of mine down in Florida. I assembled and calibrated all of the instruments and shipped it down to him where he put it together on his tripod. He managed to get away with a fully loaded IoT capable station that allows him to log locally every second and send data to my website every 10.

I don't know what your specific budget is or what capabilities you need it to perform such as but not limited to internet connectivity or what have you, but it's not too difficult to pull something off for less than some of the prosumer stations on the market.

I could go on for hours about the price/performance/deals of sensors and their benefits and drawbacks. Ideally going in you know your critical measurements for which sensors you need to pay premium for versus those that you could live with accuracies on the order of ±5% ° etc. Also what you want your station to do, that is essentially the driving factor in cost.

If you want professional but not something receiving an accolade for it's overall existence, you can certainly pull it off. I don't think it is for here nor there for offering services but I would be more than happy to consult a station for you if you wanted to venture into a more substantial system.


Cheers