Page 1 of 1
Temperature spikes
Posted: Thu 07 May 2009 5:28 pm
by merlin617
Hi Steve,
Twice now quite suddenly Cumulus has recorded a sudden rise in temperature. The first time it happened a couple of days ago Cumulus recorded a sudden rise of about four degrees then immediately returned to the correct temp. unfortunately when this occurred it was enough of a spike to set a new record. I have managed to remove the incorrect data from the log

and all was well until today when the same thing happened again. I thought it might be a fault with my station so to check, ran Easy Weather to see whether it picked up the spike, but it did not. It just showed the correct temperature for the time in question. I have been running Cumulus for several months now without problems, apart from the ones of my own making

and am at a loss as to why this has happened twice in quick succession. Apart from these two incidents, all appears to be fine. I wondered if you could shed any light on it and maybe advise me if I can do anything to prevent it.
Many Thanks,
Dave
Re: Temperature spikes
Posted: Thu 07 May 2009 6:30 pm
by steve
Whenever I've seen this before it has been a case of data corruption - occasionally a bit gets flipped, and Fine Offset haven't implemented a checksum which would detect this. I don't understand how you can say Easyweather didn't register it when you can't run them at the same time. Did Cumulus get the data from the logger or during normal running? Either way, if it's a simple case of a bit being flipped in transmission, then this is random so not repeatable anyway. Or can you actually repeat it in Cumulus?
One way to work around this is to use the cumulus.ini setting which makes Cumulus ignore temperature differences greater than a value you set, which I've mentioned recently in another thread. Edit cumulus.ini, and in the [station] section, add a line
EWtempdiff=x.y
Where x.y is the maximum temperature difference you want to allow (in Celsius). As it's very unlikely that the temperature would change very much over a 48 second period, you can probably set the the value quite low.
Re: Temperature spikes
Posted: Thu 07 May 2009 6:44 pm
by merlin617
Hi Steve,
Thank you for the prompt reply. Sorry I am not very good at explaining. Cumulus was running when the error occurred.To check on easy weather I shut cumulus down. What I meant by Easy weather not registering it was that it did not show up in the temperatures recorded from the station in the easy weather log. I'm afraid I don't understand what you mean by not repeatable. Anyway thank you for the advise.
Dave.
Re: Temperature spikes
Posted: Thu 07 May 2009 6:56 pm
by steve
There's a difference between data that is recorded when Cumulus (or Easyweather) is running, and the data that gets read when you start up and download the data from the logger. The logger only contains a subset of the data values, so any spikes stand a good chance of not being present in the logger. If you 'rewound' Cumulus from the backup data and thus caused it to download the data from the logger, you would probably find that it didn't record the spike either. So you can't really compare the two.
I don't really know what to suggest other than the EWtempdiff setting. Apart from doing its usual discarding of stupidly high or low values, there's no way for Cumulus to know that the data it has read is corrupt.
Re: Temperature spikes
Posted: Thu 07 May 2009 7:05 pm
by merlin617
Right I understand now

. I have entered the EWtempdiff line so that should stop my graphs being spoilt

.
Thanks again for your help.
Dave
Re: Temperature spikes
Posted: Thu 07 May 2009 7:34 pm
by steve
If the ini-file setting doesn't help, or causes you problems, or you start to see other invalid data, let me know.
Steve