The Weather with Raspberry Pi – unsettled but brightening up later

9 06 2012

Well, my Raspberry Pi weather station project has been stumbling along. So, first, a little update on where I am at with it.

pywws doesn’t work with my weather station, ws2350, not unexpected and a quick chat with its author confirmed it. There are a few other weather enthusiasts out thee who are trying to get it working on the RPi so check on the forums. I will be keeping an eye on progress as getting basic comms working is a common problem, as you will see….

Open2300 has been most successful at this point, its a simple set of programs that will obtain and decode the weather station data and help you use it, there is even a program that will update Weather Underground for you. Compilation on RPi did kick out some warnigns, but I had similar warnings when I compiled it on Ubuntu. The bad news, I am still not able to confirm this works on RPI, because…

Serial/USB converters – as you would expect, take two well defined standards, the latter of which was ushered in as the replacement for the former and connect them together, yeah, it was never going to just work was it! Lets be clear this is NOT a RPi issue, it’s bigger than that.

So my weather station, an entry level consumer product, about a hundred quid, has a serial link. It also comes with a serial to USB adapter as very few modern computers come with an old RS232 serial port. The USB adapter is well known as a source of endless data trouble with even the most reliable connections producing corrupt data from time to time.

Plugging this adapter, which identifies itself as FTDI, seems fine, its recognised but trying to run Open2300 with it hangs the RPi. hard to know which bit is at fault or if all of it is contributing. Running the same adapter and Open2300 under Ubuntu (on a PC not RPi) works fine though, the weather station data is retrieved and decoded.. Using a powered USB hub makes no difference in this case, the adapter is drawing around 25ma at most so power shouldn’t be the issue anyway.

To see if it was the weather station on the end of the wire causing the problem I dusted off a Psion Series 5 and serial cable, plugged it in to the RPi via said adapter and tried the terminal program, nothing, dead, hung RPi.

A second adapter arrived this morning, this time using the Prolifc PL2303 chipset (or rip off of), first off, as it was still sat on the desk, I plugged in the Psion Series 5 and, LIFE! The terminal app on the Psion communicated with the RPi.

Encouraged by some communications success I plugged in the weather station and tried Open2300, nothing for a few minutes then a time out of sorts. Moving the test to an Ubuntu PC produced the same disappointing result, it seems that although RPi is happy with the PL2303 the weather station is not.

So two different types (chipset) of Serial/USB converters and different failures. At this point I started to look for any other adapter types but on such cheap items expecting a datasheet is a little too much.

Clearly the combination of things is a key and finding an adapter that both the RPi and the WS2350 are happy with is critical, it could also be like looking for a very specific bit of hay in a haystack (I always thought that looking for a needle in a haystack sounded quite do abel to me, big magnet, little needle, problem solved)

It was time to start exploring alternative options and RPi is not without some, namely the GPIO header.

Next step, obtain or build an RS232 interface for the GPIO header, the good news is others are way ahead of me and it looks like it should be fairly straight forward, though I cant find anything pre-made for the RPi, yet.

After some hours trying to find out more about the issue via the RPi forums, lots of folks with similar issues, and more generally looking for Debian, Arm, FTDI and so on it would seem that the FTDI drivers are not ARM compatible, so if this is correct the FTDI based adapters are not going to work. I think them being identified by RPi when plugged in and checked means nothing at this point as that’s not using the driver it’s just listing the USB data, but I don’t know.

Redwood Lake Weather Station

2 04 2010

Over the last month of so I’ve been making a number of incremental changes to the web site for the Redwood Lake Weather Station. I still would love to build my own site from the ground up and a few prototypes have even gone from white board to web site and back again.

My biggest issue is instrumentation or gauges, Ive tried various offerings the most successful being the free gauges offered by Bindows but even these have left a lot to be desired, wide browser compatibility being top of a list of niggles.

So I decided, while I continue my search for the tools to build my own site, to apply a lot of the other tricks, tips and technology to my Cumulus based site from SandaySoft?

My weather station, a WS2350, is connected to an old server here in my office and via the excellent Cumulus software from Steve at  SandaySoft keeps the Redwood web site up to date.

Redwood Lake Weather Station Web Site

Redwood Lake Weather Station Web Site

So what are the key differences between the standard Cumulus web page and the Redwood site?

The old multi page standard site has been replaced by a single main page incorporating some of the data found on the old additional pages and driven by AJAX based dynamic updates based on a live data feed from the weather station. You can find details of the AJAX scripts on the Cumulus forum pages, implementing it was very simple.

The page has been reformatted a little with some minor tweaks to the CSS and the addition of a banner image (a picture of Redwood Lake, of course).

I have also implemented a data log to drive some graphs. The graphs are based on JPGraph package using the logging and data generation scripts by TNET Services, these are really good and very simple to get working. However they dont seem to deal with data gaps as I would like, I would prefer to see time on the graphs with no data rather than the period omitted. I’ve been looking at Open Flash Chart but have not really had the free time to define scripts to feed data in the JSON format required.

The navigation bar has been changed to allow me to include some great external content from Weather Underground and from Meteorologica, the standard Cumulus instrumentation gauges have been retained but I’m no fan of Silverlight and at best they are troublesome, which is a shame as I think Steve from SandaySoft has actually done a great job in building them.

One thing I always disliked about the standard Cumulus web site was the Moon phase graphic, so I’ve replaced it with some new images.

Overall I’m fairly pleased with the results, I think there is still a lot that could be done (time permitting of course). I would like to implement a better forecast, currently I’m using the simple (Zambretti based) forecast produced by Cumulus, the WS2350 doesn’t produce anything better itself as some weather stations do.

I think a forecast graphic and, if I can work out the scripting, a ‘current’ conditions graphic are next on the TO-DO list.

#uksnow – Snow and the Weather Station

8 01 2010

As every one not living under a rock will know the UK has been hit by quite a lot of snow in the last week. Here at Redwood we still have 8 to 10 inches laying on the ground, now frozen as we have not see a temperature above freezing for a few days.

I’ve been watching the weather station sensors of my WS2350 out on the roof outside my office windows and its stood remarkably well to its second winter in the snow. The wind speed anemometer has not frozen and has continued to provide good readings, The temperature, pressure and humidity have provided readings as accurate as they are ever going to be and well with in the margin of error when cross checked with other local weather stations. In fact only the rain sensor lets the side down a little by having no way to distinguish between snow melt and rain fall, hardly the sensors fault I know, but it does raise the question of how one avoids measuring snow melt in the rain fall stats. Its also bound to suffer from frozen water on the tipper which when it melts will add to rain fall stats too, less of a snow issue, more a freezing temperature issue.

I was also thinking about how to deal with recording snow fall, both depth and rate. Obviously there is no sensor to record this automatically (that I am aware of) so one would need to take manual readings and record these. The excellent Cumulus software ( I use as my data logger and web site feeder has provision for a weather diary in which snow conditions can be recored but it does not seem like the standard Cumulus site presents this diary. Of course snow fall is one thing, sunshine is another and lightning strike, both of which can be measured by the amateur using available sensors and software. Snow, it would seem, is not so easy to measure with automation.

There was an excellent series of weather programs on the BBC Four earlier this year, The Weather (, in four parts. One of which covered snow in some detail. I was hoping there may be a repeat showing but I cant find them.

The excellent #uksnow twitter tag mashup by Ben Marsh ( has been incredibly popular again, showing a simple picture of snowfall across the UK based on a tweet comprising your postcode then a 1 to 10 scale of snowfall. But it only records snow fall not depth (in fact it only records what you tell it so its very subjective). I think there are definitely a few changes that would take it forward and make it even more popular.

The Weather Underground ( has a fantastic mashup map, based on personal weather station feeds, such as the one here at Redwood, but again, snow is obvious by omission only.

I’m sure in places where snow fall is greater that recording and publishing methods are much further advanced than they are for the typical personal weather station, but I am curious what else I can do. Its likely that these winter conditions are indicative of future winters and that snow fall will be a useful addition to the recording going on here at Redwood.

You can follow the weather here at Redwood on the web site ( and on twitter @RedwoodWeather

There is more snow expected this weekend

Finally, here a few pictures from my office window out over the frozen garden Redwood lake

Posted via email from Steve’s Blog