From the time I began working to gain my Boy Scout Weather merit badge in the mid 60's I wanted to build a weather station for myself. some 45-50 years later, technology has simplified the task and various companies offer some pretty nifty "personal weather stations" to the average consumer.

Prices range from less than $200 to over $1,000. One would expect vast differences in quality and accuracy between them. Having bought one of the least expensive stations and not having had the opportunity to closely examine one of the higer-priced I'm reluctant to comment on quality. But as to the subject of accuracy, I'll offer some facts, ideas and tweaks on what I've done to get the best bang for the least buck out of mine.

The KJ4AJP station is the low-cost TP1080WC "Professional Weather Station" from Tycon Power. When it arrived, I wasn't too surprised to find the generic box marked "made in China" and since I had seen strikingly similar units sold as both the Zephyr PWS-1000TD-X and National Geographic 265NU I also wasn't surprised to find the only references to Tycon Power were on the literature inside and not on the outside of the rather generic box. (Penny-pincher's note: Around April 2010 Ambient Weather started selling a version they call the WS-1080, with no mention of the atomic clock feature but for 2/3 what I paid for the Tycon.)

Installation of the sensor units per the instructions was fairly simple, I ended up putting a 10' section of 2" diameter PVC pipe over a T-post on an open spot in the fence line and then wire-tying the array to the pipe mast. (Click picture to enlarge, use your "back" button to return to story). The receiver mounted on the wall in front of my computer desk, and the supplied USB cable ran between that and the PC. With the included "Easyweather" software installed I was able to set up the parameters and log data files of the history of my station. Without carefully reading the documentation, it's easy to overlook the section on entering the barometric pressure. Without filling in that blank with the correct value from a reliable reference on a day when the pressure is relatively stable, you'll wonder why your readings are so different from the NOAA website for your area. For what it's worth, the Tycon/proweatherstation.com website appears to be the best place to find the latest version of the Easyweather software, as well as spec sheets and the propriatary set-up guide included with the station (zephyrinstrument.com has no documentation links, and vendors for the National Geographic version seem to only have a blurry photo of their offering, something to think about if considering where to purchase). The more detailed manual for the Easyweather software is available here That set-up kept me moderately satisfied for around 6 months, but I still wanted to put my station online so it could be accessed from a web browser. At the time of initial setup I had come across software such as Weatherview 32, Virtual Weather Station and Weather Display, but either compatibility issues or the price of between $70 - $100 kept me away.

During time off for the 2009 Christmas holiday, I went back to the Tycon website and saw they now had a link to Cumulus. I found it was "donationware", and the "support" for the software was a forum of users I could join. I also discovered that in Europe my little station was sold under names like Fine Offset and Watson, and around the world weather enthusiasts and Cumulus users had taken these various brandings and modified them for the best possible accuracy. So here was my software and a community of folks who had already posted the cure for most of the configuration problems I would run into, and were instrumental in offering assistance with the ones I would encounter. The Cumulus software is the key to getting a station online. Once configured, it creates its own webpage and FTPs it to your webspace. It will also FTP to Weather Underground and also to PWS Weather, the host of Weather For You. Likewise, Cumulus will send data to the Citizen Weather Observer Program CWOP/APRSWXNET server, for inclusion in observations by MADIS and NOAA. Cumulus not only works with the generic stations like mine that are initially configured with Easyweather, but also with stations manufactured by Davis, Oregon Scientific and La Crosse. In addition, Cumulus generates the "wxnow.txt" file needed by UI-View, a popular Windows APRS program used by Amateur Radio.

Siting of the station is an important factor for accuracy, and while mounting all the sensors together as described in the instruction manual is convenient, it's not the optimum solution. Campbell Scientific has published a siting manual that describes the recommended placement criteria for the various sensors.

One of the first things I discovered was that the rain gauge needs to be solidly mounted. One day I was standing beside the station looking up at it and the neighbor's Golden Retriever came running up looking for a handout. She hopped up and laid her front paws on the fence, creating a vibration that caused the tipping bucket (teeter-totter) sensor in the gauge to record over 5 inches of rain on a perfectly sunny day. Aside from the stability problem, the rain gauge should be at 30cm or 1.0m depending on the recommendations of whatever agency you care to follow.

With the "central cluster" configuration the temperature/humidity sensor was also over the recommended 1.25m to 2m height. Since this sensor also contains the batteries, the need to lower this part of the station was actually good news. So I set about to develop a new design for my mounting system. My "best effort" puts the temperature/humidity sensor at around 1.5m and the wind speed/direction sensors at 6m. My rain gauge is around 1m. By the way, the clear rain gauge next to the station's is the Stratus RG-202 aka the "CoCoRaHS 4" Rain Gauge". After a few rains and comparing the readings between the station and the Stratus, I noted that on days when between .10" and 1.15" of rain fell, the readings differed by no more than .01", with the station reading high. So I'm fairly pleased with the "out of the box" accuracy of the rain guage section of the station. In addition to the height specification, temperature readings are supposed to be made "in the shade" and the "radiation shield" of the sensor supplied with the station is woefully inadequate. Although at night and on cloudy days my temperature reading was close enough to the NOAA website, when the sun was out my temperature reading could sometimes be up to 10 degrees higher.

Some of the folks on the Cumulus Forum had websites showing how they used planter bases or bowls, painted white and attached with threaded rods, to construct "Aspiration Shields" for their temperature sensors rather than pay the $40 for a factory-made one. Still other forum members insisted that the classic Stevenson Screen was the way to go. After $8 worth of plastic bowls from the Dollar Store cracked and split on me when attacked by a dull bit in my drill press, I decided to give my woodworking skills a try and see if the Stevenson Screen was the better option. My version is made of treated 2x4s, some halved to 2X2 size, and painted white. It uses white sheet metal louvered panels with integral wire mesh screen available from a builder's supply for a few dollars to make up for my lack of craftsmanship and to expedite construction. The roof and floor are double layers of left over 1/16" fiberglass sheeting. Inside the screen the sensor is clamped to a short piece of 1/2" CPVC mounted to drop-eared elbow secured to the floor.

For aspiration, a few forum members had put 3"-4" muffin fans in their bowl shields, others had done the same with their screens. Some powered the fans with solar cells, some were using batteries or hardwired to a power supply inside their house or garage. After a modicum of pondering and a check of eBay I decided on a couple of solar cell fans, delivered direct from China for $4.95 each (the entertainment value of the "Engrish" printed on the back of the card was worth the price). I bought a few spares in case their longevity wasn't mil-spec. Although originally designed to clip on the brim of baseball caps, they disassembled easily enough into two component parts, the solar cell and the fan. I sandwiched the solar cell portion of each between two pieces of clear plexiglass, and caulked around the perimeter to aid in water resistance. The top is drilled for mounting brackets, one mounts to the roof of the Stevenson Screen slanting toward the east, and the other on the station's 4X4 mast pointing west. The fan units slide off the hat clip, and the clip is then dissassembled. The spring is removed and discarded along with one half of the clip, and the half the fan attaches to is used as a mounting base inside the Stevenson Screen.

Here's a pic of the carriage bolt in the lower portion of the 1" EMT conduit used as the mast for the air speed/direction sensors. That bolt comes in handy to stop the pipe from slipping down through the bottom clamp during initial installation of the other three clamps, and when all clamps are mounted but still loose, the bolt can be used as a handle for fine adjustment of the sensor array to true (not magnetic) north.

The station uses standard "RJ-11" 6-position 4-contact modular cordage (like from your landline phone to the wall jack) for cabling, but is constructed in such a way that one end stays attached to a sensor. In other words, the anemometer's tail plugs into a jack in the wind direction sensor, the wind direction sensor's tail plugs into a jack in the temperature gauge/base. The rain guage's tail plugs into another jack in the temperature gauge/base. Therefore, you have to use an "inline coupler" and a line cord (the kind used by a landline phone to connect to the wall jack) if you want to extend either the wind sensor array or rain guage further from the temperature gauge/base than their supplied tails. I'm not sure if polarity matters here, so be aware that some inline couplers and some line cords are wired "crossed" per the telephone convention and some are "straight" per the data. So to be safe, you need to verify that your coupler is the same as your cord... two straights stay straight, two crossed end up straight. If you don't know what I'm talking about here, or don't have some kind of simple LAN tester that can verify the wiring you're going to attach to the end of the sensor's tails, find someone who does! And since the cables have 4 conductors, I'd say make sure anything you add has 4 too, since some line cords only have 2 conductors.

The coupler for my wind sensor array just fits in tne 1" EMT. I used coax sealing tape to plug the top of the pipe to try to keep rain out as best as I could.