Well, I'm happy to know the workaround works not only for me...
However I noticed that sometimes (let me say about once in a couple of week) I still see a strange reading.
(At the moment I had a couple of -30°C temperatures in two months)
I tried to extend time parameters of the usb driver to the limits but it seems to be no real differences.
Anyway it's important to know that the problem of (seldom) strange readings it's present also on heavyweather, which means that the problem is not mainly related to the software.
I also noticed that strange readings are present in the station itself so they're not totally related on rs232-usb communication conversion.
So if we can have the same error rate on cumulus we can be happy.
The final real problem is how to purge the strange readings, which is not very difficult using file editor (which is present in the cumulus toolbox 3rd party application).
Using cumulus 1.9.1 beta today, after the second -30°C spike, I set the spike removal for temperature to 20°C (so if there's a step of +/- 20°C in temperature it will be filtered out, maybe this is enough)
Another curious thing is that strange readings seems to be to the extremes values of the lacrosse, so they could be easy to filter out (even without spike removal features). Is it the same for all of us?
P.S. EDIT
I also would like to know if the problems disappears when connecting the station to a PCI-SERIAL interface card, without using USB...
This is interesting because I found a link (at moment I don't remember it) where the author has changed all the wires of the WS2300 with a shielded ones and he doesn't get strange readings anymore.
The author says that it seems to be a station problem related to electromagnetic interference grabbed by the wind and thermohygrometer sensors...
I'm guessing if maybe USB creates that sort of interference, while old embedded motherboard serial connections (or internal PCI-Serial cards on newer motherboards) doesn't ... ?!?!
I'm also guessing if everyone of us has its own set of "black sheep" readings... so that in my case I seldom read -30°C while other of you read -20°C, while other ones read +180km/h or 300mm/hr of rain... and so they're not full-scale-value spikes but random-fixed-value spikes related to the same electromagnetic interference.
(So it would be good to have a set of user-customizable values to filter out)
Please let me know what case is it
