Easy weather wild results

[KJM]

Getting some crazy results with EasyWeather like 900mm of rain in the last 24 hours when it's dry as a bone outside. Bogus values not just confined to rainfall but other parameters too, including the inside sensors. This makes me think it's not a radio comms problem or a fault with the outside equipment.
Anyone experienced this before?

[steve]

Do you get the same with Cumulus?

[KJM]

No I haven't activated cumulus yet. I wanted to test the weather station hardware first as a stand alone to be sure it was reliable before I went to the next step of web enabling it.

[KJM]

I've tried a couple of fine/offset stations now & both give wild results for rainfall & outdoor temperature, I'm assuming there is no error correction in the radio comms? Can anyone recommend a brand that does not suffer from this problem? I'm keen to move to getting weather on to the web but want a reliable hardware platform to start with.

[Super-T]

As a matter of interest, did you use different batteries and USB cable on each of the Stations you tried?
Does the Console itself show the erroneous readings?
My Fine Offset has been running for 22 months and the only problem that I have really had was the bad soldering on the USB socket in the desktop Console. Changed the external sensor batteries last week and that was the first time I had changed them.
I've also sold 15 of these stations and only had 1 come back for exchange.

[tjaliwalpa]

I've tried a couple of fine/offset stations now & both give wild results for rainfall & outdoor temperature, I'm assuming there is no error correction in the radio comms? Can anyone recommend a brand that does not suffer from this problem? I'm keen to move to getting weather on to the web but want a reliable hardware platform to start with.

Not certain of why you are getting wild temperature results. However, as to the wild (unexplained) rainfall with the FineOffset, I experienced the same. I have isolated the rainfall guage from the tower and have placed it on a solid object (metal box). This prevents wind and/or movement from tipping the bucket see-saw device and thereby registering flase readings.

In addition, I added a funnel that is larger in area to the built-in collector. The only way I could recalibrate the new rainfall was to use the facility in Cumulus to apply a multipier to the figures reported by the station. I calculated the multiplier by using both a manual rainguage mounted alongside the station collector and figures from a nearby official station. That is the big advantage of the Cumulus software, it allows you to make these adjustments. I very rarely refer to the stations screen readouts.

Bob

[KJM]

Yeah I changed the whole shootin' match Terry, outside station, base console & the USB cable. The rainfall line graph isn't just showing unbelievable values it becomes a wall of solid colour. (see attached).

[KJM]

OK, you've heard of the Chinese airbrush approach to history? That's where out-of-favor party officials have been removed from photos (& sometimes worse) to expunge the record. Well guess what? You can do it with easy weather too!
There are 2 buttons at the bottom of the HistoryData page. The ClearData button erases the EasyWeather.dat file on the PC, while the ClearMemory button erases the eeprom in the weather station.
In my case a lot of the corrupted data seems to be generated in the PC, either in the history table or during the drawing of the graphs. If I hit the ClearData button only, the PC software uploads a new copy of the data from the station, often the anomalies are gone in the new file.
My hunch is that raw radio packets are stored in the stations eeprom. A routine in the weather station microprocessor uses these packets to generate the local weather station display. Another routine in the PC software uses the same packets to generate a history table &, from that, the graphs. This means that you can get different results on the PC to the station.
The other thing I've learnt is not to ask the software to draw graphs while it's getting data from the weather station or during it's subsequent compute of rainfall stats (it announces these activities in the bottom left hand corner of the main screen).
I guess all this is irrelevant to Cumulus users, I just needed to satisfy myself where things were going wrong.

[Gina]

You are correct You have also found what so many of us with these stations have found - EasyWeather has many failings. Cumulus is a vast improvement. That is why we are here in the Sandaysoft Cumulus support forums, of course Now you know what to do next... follow the flock

[DanielF]

Cumulus is a vast improvement.

I agree, but not immune from such erratic behaviour problems (which I believe originate in the hardware, not Cumulus).

Here's my recent experience with temperature readings displayed in Cumulus from my WH1081PC station. These wild temperature fluctuations continued for about 48 hours, but now things seem to have returned to normal without me taking any action.

I have no idea what the cause might be. The station is about 12 months old. I forgot to look at the base unit's display during this time (occupied elsewhere), so don't know if the hardware also displayed the wild behaviour.

[Gina]

Two thoughts... It looks like severe interference such as something electrical arcing nearby or it could be a bad contact in your equipment somewhere. There's nothing you can do really unless or until it happens again. Keep your fingers crossed!

[steve]

You could try the 'EWTempdiff' setting in cumulus.ini - http://wiki.sandaysoft.com/a/Cumulus.ini - to filter out the temperature spikes; it might help, depending on the pattern of the readings. But that doesn't help find the underlying cause, of course...

[DanielF]

...or it could be a bad contact in your equipment somewhere.

Gina,
Bad contacts are an ever-present possibility when living right on the ocean edge! If it happens again I might go up on the roof and re-plug everything and 'roll' the batteries, to make sure.

Steve,
Thanks for the tip about EWTempdiff - I might play with that if there is a 'next time'. Presumably when the diff exceeds the set value, Cumulus just retains the last 'valid' value until it receives another value within EWTempdiff of that? Do you do a running average on temperature readings (given that they are normally slow-changing, it would help filter out the odd outlier)?

[steve]

Presumably when the diff exceeds the set value, Cumulus just retains the last 'valid' value until it receives another value within EWTempdiff of that?

Yes, that's right. It's a bit basic, but it helps in many cases.

Do you do a running average on temperature readings (given that they are normally slow-changing, it would help filter out the odd outlier)?

I'm hoping to put some more sophisticated filtering in Cumulus 2.

[KJM]

I wish I knew why the Fine offset logger is downloaded in total every time. My problem with corrupted rainfall appears in the easy weather PC graph at a time well before current (rainfall will always show up corrupted data anywhere in a record because it's cumulative). This means that the easy weather software must download the full contents of the logger each time. Presumably something goes wrong on the USB interface & the umteenth download of old data corrupts the new. When ever I clear the pc data in easy weather & reload the logger the new info is always accurate. If the damn thing confined itself in normal operation to just getting new data from the logger since the last download then the corruptions wouldn't be an issue?

I'm worried that cumulus has the same problem with corrupted data as easy weather. Witness this gem from the cumulus help file

Every time Cumulus starts, it takes a backup of the data files and puts them in the backups folder. Each backup is inside a folder whose name is the timestamp of the backup time. If you find at any time that bad data has been downloaded from your station, you can stop Cumulus and 'rewind' by using the appropriate backup files to over-write the ones in the Data folder. Cumulus will then re-download the data from your station (assuming it is a type which has a data logger), and this may well correct the problem

this is just another way of patching corrupted files except that it's done from backups on the pc rather than from the logger.

The one thing that puzzles me is why a data corruption is not self correcting? Assuming most transfers are accurate & that the logger is downloaded in total every time you would think that subsequent reads would over write corrupted data with good stuff? It never does, I always have to clear the pc data in easy weather then let easy weather get it all again from the logger.