Weather Station Choice

For some time I had be thinking of buying and automatic weather station (aws). I'd looked at the consoles available in places like Maplin, seen the ads in those gadget magazines that seem to drop through the door around about Xmas and even had a quick look around the web. Problem was many of the stations seemed to be just glorified alarm clocks (just what use in weather forecasting is indoor temperature anyway), and the thought of running cables to outdoor sensors put me off. But late in 2005 I finally decided to take the plunge. 

Never one to spend too much time doing research and always a bit of a sucker for a special offer, after very little time looking around on the web I had half decided on a La Crosse WS2300, available (on special offer of course!) from  Skyview Systems. Taking a closer look it had most of the functions I wanted (outdoor temperature, humidity, wind direction, wind speed, air pressure and rain fall), it could be hooked up to a PC, the sensors could be connected by a radio link, the console could be mains powered. A look around the web showed that it had some open source support and not too many problems (there was some talk of noisy cables but surely that would have been fixed by now - yeah right). All of this and the price looked about right, not a really cheap toy and not very expensive just in case it turned out to be a waste of time. So out with the credit card and make the order online (even better the station subsequently turned into an early Xmas present!).

As you will see I have had a few problems, this may seem to some people that I made a poor choice. However you have to know that I was looking for a new project here, not a finished packaged plug in and go sort of thing. I didn't want something that had lots of polished software support, with hardware that worked right out of the box. Well I got most of what I wished for! I've had lots of fun so far with this product, learned loads about the weather, used it to become familiar with a new area of software development and got to climb up and down ladders a lot, what more could you ask for?

Initial installation

The station turned up in an impressively large box just a few days later (great service from the Skview people thanks). Opening the box you get the major components shown here, the console, a wind direction/speed sensor, a combined temperature/humidity sensor with radio transmitter and the rain gauge. Also in the box are a bunch of US telephone style cables to link the sensors together, a cable to connect the console to a PC serial port and a "brick" to power the console from the mains. The package turned up just as I was about to go away for a long weekend so I grabbed the instructions to read while I was away.

On getting back home a quick rush around the local DIY stores to buy TV aerial mounting brackets and pole plus a bunch of other odds and ends and I was all set. I had decided that to check things out I would mount the sensors on my workshop rather than on the house (not ideal for the wind sensor, but no ladder required). This would give me a chance to check out how things worked and make sure that everything was OK. Only one problem it was November, dark and raining! What my neighbour made of me drilling and banging to install a six foot pole on my shed late on a Sunday night I'm not sure, but he is probably used to this sort of thing. Anyway a few (wet) hours later and we were ready to go.

Radio problems

The sensors had been temporarily installed on my workshop which is at the rear of my house. I wanted to have the console close to my computers (I normally run two, one on all the time running Linux, the other a Windows XP workstation), these are in my office which is a small bedroom at the front of the house. I made space,  plugged the console in and waited. After a few minutes it had synchronized the clock against the DCF77 transmitter, so far so good, and then the "receiving data" light started to blink and the console began to display temperature, and wind data. A quick trip out to the garden with a jug of water confirmed that it was also picking up rain data! Great! Fire up my Windows PC hook-up the serial cable and install the Heavy Weather software that came with the unit, yep I could download data from the console. Wonderful! Play around for a few minutes then retire downstairs for tea and to plan what to do next.

Back up to my office an hour or so later and oh dear the console is displaying "---" and the "receiving data" symbol is on all the time. The instruction books tells me this means the console has lost contact with the sensors. Reset the unit, no good. Back outside to take the batteries out of the sensor, still no good. Back inside try moving the position of the console, no improvement. Move it to another room works fine. Hmm, well after trying lots of things (including changing the location of the temperature, humidity, transmitter unit at about 2am - sorry again neighbours!), I found a combination that worked. The console was now balanced high up on a bookcase not easy to see but at least it works.

Back home later that day I pulled down a copy of the open source software and began to work on hooking the station up to my Linux system (see the software page for more details). A little later I noticed that the "receiving data" indicator was staying on for increasingly long times, then finally the dreaded "---" appears. Well to cut a (very) long story short, my office environment was just too electrically noisy, the combination of computers, printers, scanners, micro development systems, various bits of home brew electronics, cables and a dect phone are just too much! I tried everything, by disconnecting various things and using ferrite beads and the like I could just about make it work, but the slightest change (like rain) would mean a loss of signal. Oh well maybe a wired connection would be best after all!

Actually I was lucky, the La Crosse station can be used either wireless or wired. Also in the process of trying to sort out the radio problems I had discovered a couple of useful web sites, the Yahoo La Crosse forum and the UK Weather World forums both of which had some interesting information on the WS2300. Firstly I discovered that using a wired connection would give me a much better update rate (every 8 second rather than 128 seconds for wireless), also the wired mode allowed the external sensors to be powered from the console rather than batteries. Since I had already discovered that with the 128 second update you really couldn't gather wind gust data and external batteries are often a pain, the wired option was beginning to look more attractive. So armed with this new knowledge I began to plan the "proper" installation of the sensors.

Sensor installation

Having read all of the advice on the above forums I knew that I needed to get the wind vane as high as possible, the rain gauge out in the open and the temperature/humidity sensor in the shade. In addition I had decided to wire the sensors back to the console so I had to work out how to route the cables. My plan called for the wind sensors to be mounted on a pole on the side of the house positioning them about a metre or so above the apex, the rain gauge would be on a small wall roughly in the centre of my back garden and the temperature unit would by fixed to the back wall of the house which is out of the sun for most of the day. None of these locations are ideal but they are the best I could come up with without moving house!

The first problem was mounting the bracket to hold the wind sensors. I have a ladder, the problem is that used sensibly (i.e. at the leaning on the house at the recommended angle), it would only reach about two thirds of the height of the house. Luckily one of my other interests is climbing, so the cunning combination of a climbing rope, an open window and the banister at the top of the stairs let me position the ladder vertically against the house. This in combination with my climbing harness and a couple of slings allowed me to climb to the very top few rungs of the ladder to install the bracket. Not something that the Health and Safety Executive would approve of but loads of fun!

Next came the cables. The first was to the console. The fun part here was drilling a hole right through the wall of the house. After loads of careful measuring I gave up and drilled the thing and to my great surprise the hole ended up in exactly the right place. Threading the cable through this hole proved frustrating but I eventually managed it. Next I needed to route a cable between the temperature unit and the rain gauge. Nothing for it but to lift some of the slabs and bury the cable under them. As usual with a job like this getting the slabs up was easy getting them back no so good. Still a few hours later and the job was done.

The rain gauge was carefully levelled (see guys I did read all those posts on the correct way to do this), then secured to the wall. Then the temperatures/humidity unit was screwed to the wall of the house. Finally, tidy away a few wires, connect everything up,  sit back and work on the software.... If only!

Cable problems

After the installation, everything seemed to be working fine. I was getting much better wind speed readings thanks to the more frequent updates and the better position of the sensors. Everything was looking great. By now I had my station hooked up to the web and I could view graphs of the data. Then I began to notice rather high wind speed readings. Oh dear, I was suffering from the dreaded 91.8km/h problem. When I had been looking around I had seen the mention of a problem with the WS2300 unit that resulted in spurious wind speed readings. However in testing my unit I had not seen any sort of problem so I had assumed that La Crosse had fix it. But now with the wind sensor mounted on a pole high above the ground I was getting those same readings. Good old Murphy strikes again!

The problem lies with the interface between the wind sensor and the temperature/humidity unit. Basically a combination of the interface design and the cable supplied by La Crosse means that in some installations the cable will pick up electrical noise resulting in a wind speed reading 91.8kmh or 25.5m/s (this actually comes from a byte in the data-stream from the device having a value of 255 which will be familiar to hardware/software engineers!). In my case when testing I had not seen any of these errors, but now at times over a third of the wind speed readings had this value. Something had to be done about it.

The details of this problem have been explained rather well by Kenneth Lavrsen on his web site (Kenneth is the main author of some of the open source software for the WS2300 and many owners including me owe him thanks). This site also gives details of how to fix the problem. Unfortunately the fix required the replacing of the cable to the unit that was now a metre above the roof of my house. So on a very cold December morning it was out with the ladders again, then the soldering iron then back up the ladder to replace the unit and run the new cable. The end result of all of this was no more errors hooray!

Rain gauge modification

After running the station for a little while I noticed that all was not as I would like with the rain gauge. At times I would look out of the window and see that it was raining, but no rainfall would be logged by the device. At this stage the sensor was working correctly because if I dripped water into it then it registered fine. The problem was the resolution of the gauge. The WS2300 like most other automatic weather stations uses a tipping bucket gauge. In the case of the WS2300 each tip of the bucket is 0.518mm (or 0.02") of rain. This may seem a small amount, but the ground can seem very wet with smaller amounts of rain. Stations seem to to vary a lot in this area. The Davis range for instance offer a resolution of 0.01" while the popular WMR928 has a resolution of only 1mm (0.04"). 

An easy way to improve on the resolution of the sensor is to increase the surface area of the collector. I did this buy mounting a larger funnel to feed rain water into the sensor. The only down side of the modification is that the display on the console is no longer correct. Since I don't really use the console display this was not a problem for me. With these modifications the rain gauge now has a resolution of 0.115mm (0.005"), which means I can record much smaller rain showers. You may get a few odd looks when shopping for a funnel using a ruler though!

Stevenson Screen

After running the station for a week or two I noticed that there was often a sharp peak in the outdoor temperature readings in the middle of the afternoon. Investigating further I discovered that this coincided with the sun shining directly on the WS2300 temperature sensor. Although I'd tried to mount the sensor in the shade I hadn't been able to completely avoid the sun and the screen supplied with the WS2300 was obviously not working very well. I knew that the ideal sensor location was over grass inside a Stevenson Screen. Well I couldn't really locate the sensor over grass, but perhaps I could do something about the screen. 

A quick search of the web and I found a design using plant pot saucers which I liked the look of. So one quick visit to my local garden centre and a bunch of drilling and sawing later and I had made the screen you can see here. It is not perfect (The plastic is very soft and hard to drill cleanly), and I probably need to make some changes before the Summer (like an insulated base). But for now this seems to have removed the mid afternoon temperature spikes, so it must be doing some good.

Rain gauge reed switch problems

After the station had been operating for almost two months I noticed that it was no longer recording rainfall correctly. I checked the sensor to make sure it was not blocked. Then tried dripping water into the gauge. I could hear the bucket tip but no rainfall was registered on the console. I tried resetting the sensor unit, unplugging the rain gauge and finally came to the conclusion that either there was a problem with the cable or the rain sensor. I removed the cover of the gauge and used a short section of wire to short out the reed switch that is used to detect the tipping of the bucket. This time the console registered rain fall so the cable was probably OK (just as well as I really didn't want to have to dig it up again). I used a meter to check the resistance of the reed switch in the gauge and although it did drop when the bucket moved it was still very high. It looked like there was some sort of problem with the contacts of the switch, Perhaps there was a crack in the glass envelope and the contacts have become corroded . Whatever the problem was, I needed to replace the switch.

I guess at this point I could have returned the unit, but it seemed like a small problem and maybe I could improve things at the same time. A quick trip to my local Maplin store for a few miniature reed switches, a couple of snips to disconnect the sensor board and I could start work. I quickly removed the old reed and again tested it. It was still showing a high resistance when closed (over 50KOhms). At this point I decided to make a few modifications. Rather than just replace the single reed I decided to install two switches next to each other. That way if one fails (or perhaps fails to detect the bucket tipping), the other can take over (I've seen that some professional rain gauges have two reeds to provide this backup). I also decided to install a waterproof plug and socket between the cable and the rain gauge to make it easier to work on the gauge if this ever happens again. Installing the two reeds was a very fiddly job the position and orientation is very critical to ensure that they both open and close as the bucket tips. I spent ages adjusting them before I was happy. Then it was just a case of adding the plug and socket and reinstalling the gauge. Since the changes the gauge seems to have been working perfectly.

Conclusions

So what do I think of the WS2300 and would I buy another. Well for me the unit has been interesting! Clearly there have been a number of faults with the device, but I've been able to overcome them. This has made the entire process rather fun. Would I recommend one to anyone else. That depends. If you like fiddling with electronics and understanding how things work then this may be for you. If you want a reliable weather station then perhaps not! Would I buy one again? Maybe. You do get a lot for your money, but so far it has been rather unreliable and I do worry what will happen if anything goes wrong with the actual electronics, or with the wind sensors (yet another trip up the ladder would not be fun). I guess I have to wait and see!

Copyright 1998-2011 Andy Shaw