Cumulus Support

werribee_au wrote:Now the only thing I can put this down to is that at about that time I plugged my iPhone into the station notebook's usb to charge it and maybe at the time cumulus was reading the station console.

That sounds possible to me. I think that in addition to the spikes caused by RF interference (or whatever else causes them) I think it is also possible for data to be corrupted over the USB connection.

I now have 48 second logging set up (courtesy of Jim Easterbrook and his pywws software) and nicely displaying spikes on the temperature plot. Once I get a few hours worth, I'll post the plot, but if anyone wants a preview, the Testing Page available on my main weather site from the Home page Extras Menu, shows the results to date. Note that 48 sec logging has only been turned on from about 10:45 this morning - previous data is at a 5 min sampling rate. (Or use the direct link to this page http://ginad.org.uk/weathergd/DevelopmentTestPage.html )

Been running for a few hours now and lots of spikes! In fact 16 spikes in just over 4 hours!!

A full 6 hours at 48 second sampling :- Tomorrow I'm hoping to add some interference suppression. I'm going to make up a plug and swappable circuit board so that I can have 2 boards. Then I can alter one while the other is in use and swapping over will only take a few seconds rather than having to dismantle things and bring the lot indoors to work on.

Test circuits

This is just simple linking through as now with earthed screen on wind sensor cable. Next step might be to add a capacitor across the anemometer connection. This is probably the ultimate RFI filtering using LC filters on all four data wires from the wind sensors.

Gina,

Have you tried running the 48s logging software on both your receivers concurrently just to establish, beyond all doubt, that the spikes are in the data sent by the sensors?

No not yet but it's a very good idea, thanks Jim I'll set that up tomorrow.

I now have both consoles/computers running the same pywws 48 second data logging software, uploading to separate data directories on the web server and now displayed side by side in my "testing page". But because I'm not currently running the normal logging and data processing, the main weather station site is not being updated. This will be the case until I write code to link the two together, hopefully tomorrow (I'm not up to thinking it out tonight - busy day). I would like to run parallel consoles for a reasonable period (a day or two) to confirm that all the spikes originate from the transmission from the external sensors but don't really want to have the weather site out of action for too long.

Looks like the two consoles produce the same results...

Console1 Console2

So its definitely being introduced in the transmitter or by RF when in-flight then - maybe!

Certainly looks like it. I think I'll run a direct comparison between the raw data logged by the two systems. I wouldn't expect any difference in any of the data but we shall see.

I'm thinking of running as now overnight then return one system to the previous 5 min logging and 10 min uploading while I use the other to test interference filters and further develop the software. I would like to provide 48 sec updating of the wind gauges (as in Cumulus) and simplify the plot image processing.

On the hardware side and particularly relevant to this thread, I have prepared connectors and circuit boards to experiment with interference filtering. So tomorrow I hope to start the experiments again.

This morning I have switched the software to 5 minute logging and normal uploads so the weather site is back to normal (as may be seen from the banner in my sig). The second system is still running 48 second logging and can be seen on the "Testing page". I have just (as of 10:30) changed the filter circuit board to one with a 100nF capacitor across the anemometer sensor.

Here is the plot of temperature over the last 24 hours (ie. without the capacitor) :-

Well that didn't work as this plot shows - the slight warmness at 10:30 was where I held the transmitter unit while changing circuit boards. Next I'll up the capacitor to 470nF and see what happens. EDIT - 470nF ceramic capacitor connected across anemometer at 15:40 GMT on 10/11/10.

Seems worse if anything Now then... what next... Maybe I need to check if it's the anemometer or the wind vane - just to be sure.

Still getting lots of spikes. I'll disconnect the anemometer and leave the wind vane circuit connected (when the rain stops!), I'll report results when I have them.

EDIT - Anemometer disconnected from 10:30 today, 11/11/2010.