A DIY Stevenson screen

[JamesF]

Now the observatory is up and running my mind has turned to wiring up my weather station kit again and getting it all working as I can now connect it all to the PC in the observatory and have some records of conditions when I've been observing/imaging.  I dismantled everything some time back when my last bodged-together screen gave up the ghost, so this time I thought I'd do a bit better job though most of the materials will be stuff I have sitting around the place.

Initially I had considered 3D printing the louvred sides, but I discovered that Screwfix sell vents about 200mm square with built-in insect screens (I'm not desperately keen on it becoming a home for bees/wasps/hornets) for about £1.40 each and anything else looked rather too much effort at that point, so I bought four of them, one for each side and two for the door:

A hunt about in the stack of wood "that may come in useful one day" turned up some 30mm square softwood, a bit of T&G and a few offcuts of ply and I also found some galvanised mesh I could use for a floor, so I set to with the tools and made up a box and frame for the door:

The wooden floor section is to allow me to drill holes for feeding cables in.

The next steps are to make up a second roof (apparently a second roof with an air gap underneath is desirable, presumably to reduce the effect of the Sun) and to paint everything so I can finish assembling it all.  I was contemplating covering the second roof with an offcut of EPDM, but I'm not sure at the moment.  It might well defeat the object of having a second roof in the first place.  On the other hand, this will be going on the north east wall of the observatory, so shouldn't really get any significant heating from the Sun anyhow.

James

[Gina]

My latest weather station will use all electronic components for measuring temperature, humidity etc. using an Arduino with shield so I'll be using a simple small 3D printed Stevenson screen.

[Oldfort]

Looking good.  Just one thing about the vents, you will probably need to vacuum clean quite frequently.  Once the spiders find it, it will be heaven for them.

[JamesF]

25 minutes ago, Oldfort said:

Looking good.  Just one thing about the vents, you will probably need to vacuum clean quite frequently.  Once the spiders find it, it will be heaven for them.

Perhaps they'll move out of the observatory then

James

[JamesF]

51 minutes ago, Gina said:

My latest weather station will use all electronic components for measuring temperature, humidity etc. using an Arduino with shield so I'll be using a simple small 3D printed Stevenson screen.

Mine is going to be electronic too, but I want enough space inside to be able to get my hands in and fiddle about with stuff, especially as I might also use it as the main connection point for external sensors such as a rain gauge, anemometer and perhaps a sky quality meter.

James

[JamesF]

I didn't have much time today, but managed to get a fair bit of the painting done.  Hopefully I can finish it during the week ready for installation next weekend.

James

[JamesF]

The painting I did yesterday didn't feel particularly dry today (perhaps not surprising given the temperatures), so I thought I'd dig out some of my old 1-wire kit and have a tinker this evening.  What a bundle of joy that has been!

First I found that my 1-wire hub wouldn't power up.  It looks as though the 15V PSU has died.  I reckon I might be able to feed it from the 12V supply for the observatory though, so perhaps all is not yet lost (or at least waiting for me to buy another PSU) there.

I downloaded the last release of OWW, the software I used using to read the 1-wire devices and just couldn't get it to build properly.  After a lot of faffing about I tried the previous version and that seems to be fine.

The next step was to plug in the 1-wire controller device and a few sensors and see what happened.  Well, nothing much happened at all because it seems the way the controller interfaces with Linux has been completely changed in the intervening years and OWW can't find any of the devices.

I discovered that I might need to load some additional driver modules to get the devices to be recognised, so I tested with a few of those.  I've crashed my desktop twice so far (the first time I've managed that in years).

Everything is now plugged into my laptop instead so I don't need to reboot my desktop again...

James

[Gina]

Oh dear!!!  That doesn't sound very good news for when I resurrect my weather station project.  Hope you can get it sorted out, James.  Good luck.

Actually, thinking about it, I have a slightly different setup with an Arduino Uno and shield.  The Arduino acts as the 1-wire controller.

Edited October 29, 2019 by Gina

[JamesF]

I didn't have a lot more time left last night, but out of lack of other ideas tried it out on an RPi where it seems to work much more reliably.  I wonder if the drivers aren't stable on 64-bit environments or something like that.  As I was quite probably going to end up with an RPi in the screen anyhow, that may not be the end of the world and it does have the advantage that all the 1-wire cabling can be short.

However, I think I might be able to get it to work a different way on my desktop and laptop if I blacklist the standard 1-wire drivers.  It looks as though OWW can talk to the USB device directly and control everything from there.

James

[JamesF]

I'm making progress.

After blacklisting all the 1-wire modules on my desktop and making a udev rules script that allows non-root users access to the USB device, I've plugged in the circuit board of my AAG weather station, a pressure sensor and a combined solar/temperature sensor and fired up oww.  It lives, Igor!

The software is whining about an error reading the wind direction ADC, but I think that's because I dismantled the weather station so I could remove one of the 6P6C plugs (the entire unit was blown down in a storm a few years back and the cable tore out of the plug) and replace it.  I've not reassembled it yet.  I'm actually half-tempted to try to 3d print an entire new housing as the original has somewhat degraded in the sunlight.

Getting this bit of software working is actually more than I need in fact.  There's a non-GUI version that can just log the data to a database which means it can be pulled out to be used however I want.

I also have a combined temperature/RH sensor that I want to try out (need to make up a cable for that, but it shouldn't take long) and some loose DS18B20(?) temperature sensors I can test.

James

[JamesF]

Oh!  I nearly forgot!  I also have a rain gauge that needs a bit of a clean-up.  Given that it was only rain falling inside it's incredibly filthy.

James

[JamesF]

Progress is slow, largely due to the colder weather meaning paint drying times have increased, but also I was waiting for some stainless screws to arrive.  Today however I fitted the door and all the vents.

The upper roof section is also in progress, but I'm quite tempted to try to fit this in position tomorrow anyhow and fit the additional roof section when it is ready (probably next weekend).  It's not like it needs much protection from the Sun at the moment, after all.  I just need to add two or three rails across the back so I can fix sensors etc. in place.

James

Edited November 9, 2019 by JamesF

[Kev M]

I can see why you are making your own.....

https://www.metcheck.co.uk/products/metcheck-met-m11-standard-size-stevenson-screen

How much.... £785 .... !

 

[JamesF]

32 minutes ago, Kev M said:

I can see why you are making your own.....

https://www.metcheck.co.uk/products/metcheck-met-m11-standard-size-stevenson-screen

How much.... £785 .... !

That's quite horrifying.  I imagine they don't get many personal customers

James

[JamesF]

This week I have completed the painting and finished fitting the roof.  I also nipped into Screwfix on the way to a meeting at school and picked up some elbows and clips for the 20mm conduit that I found stashed in the ceiling of the workshop.

Today I fitted the screen to the north-east wall of the warm room and ran three cat5e cables (one 1-wire, one ethernet, one spare) and a 12V power connection through the conduit into the warm room, which meant I could plug in the 1-wire hub/power injector and a couple of the 1-wire devices that I've cleaned up since their last outing.  They're "casually abandoned" inside the housing for the time being.  I'll fit them more neatly when I have the other devices ready to go.

Quite pleased with  the project so far.

The computer end of the 1-wire network is plugged into a DS2490 (I think that's the one) USB interface and I'm working on getting the software side of that up and running at the moment.  I also need to work out where and how I'm going to fit the rain gauge and the original AAG weather station that has the anemometer and wind direction meter. (both of which need some repair work first).

Which reminds me... Today I found my copy of a paper documenting the design of a solid-state combined anemometer and wind direction meter using some ultrasonic transceivers and a single-board computer.  It was published in 2006, so before the advent of the RPi and Ardiuno, but I imagine either would be suitable these days.  In fact I bet someone's already done it.  I quite fancy having a go at building one at some point.

James

[Gina]

That's interesting though I already have a magnetic transducer arranged for the wind vane and simple Hall device for the anemometer.  1-wire instruments connect to an Arduino Uno plus data logging shield which also has the pressure sensor on.  It all works basically - I just have to get round to finishing it.

[JamesF]

The AAG weather station uses magnets on the end of a spindle to trip reed switches as a means of calculating wind speed (count the number of times the reed switch is tripped per minute, I guess) and direction (eight more reed switches arranged at N, NE, E, SE, S, SW, W, NW).  It's actually a bit of a mess now -- I guess UV has taken a bit of a toll on the plastic.  It's just occurred to me that I could paint it all gloss white and perhaps get a bit more life out of it whilst I design a replacement to 3D print.  The vane for the direction is dead though -- all the actual vane has gone.

I could start work on the painting and whilst that is drying get the rain gauge up and running again.  It really requires a new base to allow the water to drain out, but I could 3D print something suitable pretty quickly I reckon.  A longer cable would be good, too.

James

[Gina]

Mine is all home built.  The wind vane uses a magnetic field direction sensor in an SMD chip.  Measures in degrees - not that it needs anything like that level of accuracy.  I must take the thead (assuming I can find it) and convert it to a Blog.

Edited November 16, 2019 by Gina

[Gina]

Found my weather station thread HERE.

[Gina]

I think it's high time I resurrected this project though I might finish my ASC first.  BTW the rotary encoder chip is MLX90316.

[JamesF]

I'm still working on this, though like everything else in my life it's taking a little longer than I hoped

Today I thought I'd take a look at the rain gauge which isn't working.  I opened it up and had to drill out one of the screws to get at the electronics as it had rusted too much to come undone.  It appears that the unit uses a reed switch that is activated by a magnet on the tipping system that collects the rain:

The battery has clearly expired, so I thought I'd pop it out and replace it, but that's proving a little more awkward than I expected.  It seems the battery is somehow welded to the contacts.  Not ideal

I guess I have a few options at this point.

I could attempt to unsolder the contacts from the PCB, add a new battery holder either on the PCB or on the end of some wires, and then fit a new battery.  But that still leaves me having to replace the battery every so often.

Although there are only two wires connecting to the board to connect it to the 1-wire hub, the main cable has four cores and the hub supplies 5V on one of the "spares".  I'm wondering if I dare attempt to re-wire it to provide power from that instead of the battery (and whether I actually need to step the voltage down as it might run quite happily on 5V anyhow).  Given the amount of hot-melt used it's quite hard to make out what the other components are.  Ultimately though I think it uses the same 1-wire counter components as the weather station anemometer (which also works using a reed switch) and that runs off 5V.

I don't feel I have much to loose at this point though.  If I unsolder the wires from the board so I can remove it I can perhaps get a better look at it.

Unfortunately I've not yet been able to find much information about the rain gauge on the web.

James

[JamesF]

I think I have most of my other sensors up and running now.  I have owfs running to get the sensor values with a script that displays them every thirty seconds or so.  Clearly there's some tweaking to be done.  I have four temperature sensors, one in the AAG anemometer, one in the solar sensor, one in the RH sensor and one in the barometer.  This is the data I'm seeing:

21:17:22 AAG: -0.19, SOL: 1.62, BAR: 1.54, RH: -0.53 degC, 1024.0 mb, 102.2% RH
21:17:58 AAG: -0.19, SOL: 1.56, BAR: 1.53, RH: -0.62 degC, 1023.9 mb, 102.5% RH
21:18:31 AAG: -0.25, SOL: 1.56, BAR: 1.53, RH: -0.66 degC, 1023.9 mb, 102.7% RH
21:19:07 AAG: -0.19, SOL: 1.56, BAR: 1.52, RH: -0.56 degC, 1023.9 mb, 102.5% RH
21:19:42 AAG: -0.25, SOL: 1.56, BAR: 1.51, RH: -0.53 degC, 1024.0 mb, 102.8% RH

Obviously they can't all be right.  As the dew is already freezing, I'm inclined to think that the AAG and RH values are probably nearer the mark.  Atmospheric pressure looks about right.  Perhaps a little optimisitic, but not too bad.  I assume it's telling me that the RH is over 100% because we're below  the dew point.  I'll have to work that one out.  owfs gives me access to the raw data from the sensor so I can recalculate it myself to check.

James

[JamesF]

As mentioned in another thread the RH reading got up to 118% the other day, so I checked the datasheet for the HIH4000 that's inside the RH sensor unit and found that the voltage reading for 100% RH (the HIH4000 outputs a voltage proportional to RH) should be 4.07V with a supply voltage of 5V at 0C (which was very close to the temperature when I checked).  Mine was allegedly producing 4.11V from a supply of 4.83V.  Using the formula from the datasheet in reverse I think it should have been closer to 3.93V.

So I visited the Bay and bought a new HIH4000 which arrived today.  A little work with a soldering iron and some braid removed the old sensor and left nice clean holes to fit the new one which is now back online and reading about 82% RH though it's still rising slightly.  Less than two hours ago before I disconnected it the old sensor was reading over 102%!  The built-in temperature sensor looks as though it might be more accurate, too.  At the moment it's pretty much in agreement with the DS18B20 sensors that I have installed.

James

[JamesF]

Over the last couple of days I've printed a couple of blocks:

and a post mount

to allow me to remount my AAG weather station on a pole and connect it up to the rest of the weather monitoring kit.  I had a few niggles printing the second one.  It's deliberately quite chunky (because last time I was using it the wind ripped it off the pole in a storm) and takes a while to print.  About two thirds of the way through my first attempt we had a power cut   I shall be moving the printer to a UPS as soon as it is convenient to do so.

The weather station has sensors for wind direction and speed and also a temperature sensor, though the latter is often inaccurate when the Sun is out because the weather station housing gets warmer than the surroundings so I'll be relying on other temperature sensors inside the screen instead.

After spending much of the afternoon dodging showers I actually have it all mounted up now, though the cabling isn't completely finished because I ran out of daylight having decided to replace the standard cable with UV-stable cat5e.  I'm sure I can find time to do that tomorrow -- it should only take ten minutes.

When I set up the Stevenson screen I ran 12V and ethernet to it and I'm now considering putting an RPi inside to run all the instruments rather than using the desktop in the warm room.  I have a splitter for the power and one output will continue to power the 1-wire hub as it does now.  I'm going to wire the other output into a socket that's intended to replace a cigarette lighter socket and has two USB charging ports on the front.  That should power the RPi.

Then I have the rain gauge to get operational again.  I've been doing some testing and I think it works ok now, but I'm not completely convinced that it isn't double-counting some bucket tips.  I may need to look into that a bit more.  I'm also going to replace the original cabling with the same cable I used for the weather station, and make up some kind of spiky edge for the top of the funnel to stop birds perching on the edge and using it as a toilet.

After that I think the last thing I'm going to do is to try to calibrate the RH sensor a bit better.  As I posted before, the datasheet suggests that the output voltage at 100% RH should be 4.07V with a 5V supply voltage.  The original sensor was way out, but even the replacement manages to get up to 4.11V with a supply voltage of 4.84V so obviously something is not right there.

James

[JamesF]

Last night I also printed a couple of these to make a sleeve where the cable for the anemometer etc. enters the screen.

This morning I sneaked out for a few minutes, drilled the hole and cabled everything up properly.  Hopefully now it has a few coats of gloss paint it will last a while before the plastic of the housing finally gives up and I have to build a new one.  I don't think these are available any more.  At least, last time I checked the company that made them was no longer in business.

When I had the circuit board inside the screen the temperature sensor was reading about 0.3C below the average of the sensors I'm intending to use long-term.  Now it's out in full sunlight in the housing despite being on the underside of the circuit board it is reading 1.5C above the average, so it's clearly not to be relied upon.

James