Wild excursions in readings from WS2300

[John-G1EUH]

Hi

I'm in the process of setting up Cumulus with my WS2300.

From time to time i'm seeing some wild readings, for example at about 5 am this morning Cumulus reports that we had 47 inches of rain. At the same time the outside temp shot up to 50 deg. C. It didn't rain, and the temp has remained more or less constant at about 8 degrees.

I think its safe to say that the WS2300 is working fine, as it has been running for the past 9 months with various weather software without problem. The WS2300 and its sensors is powered from a mains adapter, not from batteries.

Anyone come across this before, is it something i'm missing or not doing?

Thanks in advance - John

[steve]

Which version of Cumulus are you using? Do the false readings occur at times when Cumulus is running, or are they downloaded from the station logger at startup?

Steve

[John-G1EUH]

Hi Steve

I'm running 1.8.4, build 603. It's running on XP Pro.

This system is on 24 hours a day. I've just checked the graphs and found that it has happened again at about 12:45. today, humidity has gone down to -15, and outside temp has gone up to 50 deg.

Hope this helps -

Regards - John.

[steve]

That simplifies it somewhat - Cumulus is presumably getting those values back when it reads the data. Could you please turn on the debug log and send me the debug.log file the next time you get some bad data (wait at least 5 minutes after the bad data). Then I can see exactly what is being read.

Thanks,
Steve

[John-G1EUH]

Hi Steve

Ok, - log files attached, i've had the log running from about 9 this morning. I've also put a copy of dec08log.txt in the archive.

If you need anything else please let me know.

Regards - have a good Xmas - John.

[steve]

Thanks, John. Your station is supplying the invalid data. Here's the outside temperature reading that went wrong at 13:05

5350.8125 : WS2300: Reading outdoor temp
5350.8407 : WS2300 data: 90 79

'90 79' decodes as 79.9, and then you subtract 30 to get the actual value in degrees C (I don't know why it works this way) = 49.9C

Here's the next reading with the correct value:

5351.5750 : WS2300: Reading outdoor temp
5351.5891 : WS2300 data: 90 37

'90 37' decodes as 37.9, subtract 30 = 7.9C

It looks to me like the station encoded the 7.9 incorrectly in the first one, or maybe that's a coincidence. Maybe the byte just got corrupted.

I don't know what I can do about this. I suppose I could put more checks in for obviously incorrect values, but even that isn't going to catch everything. I'm already discarding values over 60C and below -40C.

Steve

[John-G1EUH]

Hi Steve

Thanks for taking a look at this, i'll try a couple of things here, maybe install Cumulus on another PC as there are other applications running on the APRS / WX machine, but this doesn't seem to have had an adverse effect until now. The WS2300 is using an RS232 to USB converter, I might try a different brand of serial converter. If I find anything hardware related i'll report back.

Thanks again - regards

<edit> - Just been hunting round for info on the WS2300, seems it's not that bright at understanding RS232 requests, and often fails to recognize a request for data. Found some interesting info on this site:

http://www.lavrsen.dk/twiki/bin/view/Open2300/OpenWSAPI

(BTW - I'm definitely NOT trying to teach my grandmother how to suck eggs here!).

Have a good Xmas

John.

[steve]

Yes, I couldn't have added support for the WS2300 without the information on that site. Cumulus retries requests for data until it gets a response, but in your case (and in others I have seen) the response is coming back with the wrong information.

It occurs to me that while it's not so easy to decide what is bad data in absolute terms (e.g. how high a temperature should you ignore?), it's easier to ignore large changes in values between readings. I've done a similar thing for Fine Offset stations, where you can configure how large a change in outside temperature you regard as bogus. I could easily add the same for the WS2300, and perhaps look into doing similar things for parameters other than the temperature...

Steve

[Budgie]

I know that the wired WS2300 has been known to suffer from spikes in the wind data when using Weather Display and this is caused by the unshielded cable between the sensor pack & the anemometer. Changing this cable for CAT5 will normally cure it.
I'm wondering if Cumulus is getting the same sort of problem from the temp sensor?
Is the cable from the temp/hum sensors to the console loose, able to blow around in the wind, at all or running close to a possible source of interference like a plug socket, light switch or wireless phone?

Just an idea.

[John-G1EUH]

Ok - thanks for your replies and info

Budgie - yes, ok, thanks for the info, I'm aware that the WS2300 can suffer from interference, I operate an amateur radio station here, and soon found that RF interference was "getting in" to the electronics in the WS2300. I solved this one by wrapping the leads connecting the various component parts on the weather station around ferrite rings, practically used up my entire stock of ferrite rings - but it worked - no more RF getting into things when I'm transmitting. Also see below, data corruption only appears to occur when I'm running Cumulus.

Steve - Ok - yes, one solution would be to look for large excursions in the data and trap them. I've just swapped back to Weather display lite, and there is no trace of data corruption in the logs from previous months, only for the time I've been running Cumulus. Maybe Weather display lite is already doing what you suggest, and trapping any rogue data.

I'll run "Weather Display lite" for a day or so to confirm that data corruption only appears when I'm running Cumulus.

I hope this will be of help to you - I'm not out to prove a point or anything here - I'd really like to be able to run Cumulus with my WS2300.

Regards - John

[gottl49129]

Can confirm the behavior with my 2316..

But I also have the Heavy Weather software installed and it doesn't have any such fluctuations.. are you sure it's not a misread of the serial output somewhere?

Also Cumulus doesn't seem to be reading my daily rainfall total..

Thanks!

[gottl49129]

Also trying Weather Display and no wild readings.. Thanks! Would love to use Cumulus, it's a little..... easier to read... than Weather Display

[steve]

Well, yes, the evidence would seem to point to Cumulus doing something wrong. But I'm following the documentation on the Open2300 site, and I don't know what else I can do. The author of Heavy Weather will have had access to the protocol spec (and probably the author of WD too), so it's a little easier for them. It would be easier if I actually had one of the stations to experiment with.

If anyone feels like installing a serial sniffer (there are some free ones out there) and comparing the trace from Heavy Weather and Cumulus, that might prove useful.

Steve

[John-G1EUH]

Hi Steve

Yes, i'll have a go at logging some comparisons for you once todays festivities are out of the way, do you have a preferred serial sniffer you would like me to use?

Regards - John.

[steve]

Hi Steve

Yes, i'll have a go at logging some comparisons for you once todays festivities are out of the way, do you have a preferred serial sniffer you would like me to use?

Regards - John.

Thanks, John, that would be great. I don't mind what you use as long as I can understand the output

I've used one called 'HHD free serial port monitor' myself, it has various 'views', I found one of the 'views' to be particularly good, but I can't remember which one.

Steve

(Just popped in for a quick look, couldn't resist)