Cumulus Support

Steve - I have been noticing, from my equipment, sudden spikes in temperatures (mainly humidity readings, dew point etc).

Now these are most likely due to failing sensors and, unless I either replace the sensor or the whole system, there isn;t much I can do.

But - because the readings are so abnormally above previous readins, is there something the software in Cumulus can do to prevent this faulty/erroneous data being recorded.

For instance - this is the log for one of these spikes:
Notice the 1851 line with the reading of 154.0 whereas the previous and latter readings are 5.0. 154.0 is also well above any possible genuine readings.

So - would it be possible (in whatever version of Cumulus) to have a 'filter' or cut off applied to data?

As an example - FreeWx, when it encounters dsata such as this, actually records 'Err' (think that is the wording he uses) and, because if is non-numeric, doesn't get processed by other software utilising the data.

I also know, for example, that in about 4 weeks time, I am going to have a period of about 6 weeks where the outside temperature sensor will, at times, start gooing from a temperature reading of 35 C plus and suddenly show a reading of -6 C ie it appears to go open circuit on extremely hot days for a short perid before re-setting itself. Again - no spares available so nothing I can do about it.

Being about to put limits to the data would help alleviate these problems (fo example, there is no way in the world a gauge here (opr anywhere else for that matter) will ever record a temperature above 60 Celsius and, for me at least, never go below -3 or -4 C where I live).

Try Configuration/Calibration on your Cumulus screen.

Might have to update your software to Beta 191 version.

See also https://cumulus.hosiene.co.uk/viewtopic.php?f=13&t=3425

Oops

You need to discover the source of those spikes for sure

I've found the Beta versions to be superior in functionality, except for the odd hiccup. If you stay a few days behind the latest Beta then any problems will have been remedied hopefully.

Hmmm beta - at this stage will have to forego that solution (sorry as a software devloper myself I won't put beta packages onto production machines).

That link - did you mean another one as it comes straight back to this thread.

How will the configuration/calibration help though? These either add an offset (can't use that as these are sprikes) or a multiplier (99% of the data is correct, so I can;t apply this either).

I guess what I am trying to say is that temperature should NEVER go above 100C anyway (realistically, 60C would be tops) so there could be a point at which it is clipped and prevented from going into the data.

Edited the post to correct my whoopsie

My station is a WM-918 - wored so I know it isn;t interference from other stations close by (there is one over the road that is wireless).

The temperature sensor that gives -6 is, from what I have been able to discover, a known issue with the sensor and I have tried all the fixes suggested for it. Last summer I alpmost beat it - it only happened twice in 4 months and then on 40+ degree days. it also only happens when there is a possibility of sun affecting the sensor (it is located on the south side of the house but at the solstice for about 2 days it can actually get sunligh on the bottom part of the mounting braket).

The new spikes though are my concern - humidity of 154 is suggesting a failing sensor and, as I mentioned, I cannot get replacements any more and can't afford to replace the entire system (wouldn;t anyway if wioreless is the only way to go as it appears!)

Beta software - Cumlus is running on a production Windows Server which houses the majority of 3 of my web sites. Whilst I am not casting any doubt over Steve or Cumulus, I have a very simple rule - absolutely no beta or test software makes its way onto the server under any conditions. When it comes out of beta, I will upgrade, but until then (there was something else Steve said was fixed that was affecting my station a while ago - I think he took it the wrong way when I mentikoned I wouldn;t use the beta software...meant no offence then or now to him about this).

The 'spike removal' settings (which are on the calibration settings screen) in the beta, and other existing 'stupid value' filters only apply to those stations which have known 'design issues' with bad data, i.e. Fine Offset and La Crosse. It may be possible to make them also apply to other stations, but I don't know when I'll be able to look at doing that.

If I could suggest maybe putting it on the C2 wish list then? As I said - in this case I believe it is faulty sensors (and I will have to deal with this issue later). But if the erroneous data can be kept out of the data files it would be a big help - especially in the case of temperatures where we can have known min and max's.

So the implementation of EWTempdiff in cumulus.ini doesn't work with Oregon stations to control temperature spikes?

Ned wrote:So the implementation of EWTempdiff in cumulus.ini doesn't work with Oregon stations to control temperature spikes?

That's correct. The "EW" part means "EasyWeather"; it was originally just for Easyweather stations (i.e. Fine Offset), but I extended it to La Crosse stations when it became clear that these could also produce 'spikes'. In the three years that I had my Oregon station, I had no problems at all of that type.

Maybe time to get an "El Cheapo" Fine offset to replace your wm-918?

I've been considering it but whilst 99% of functionality is working I need to keep it (to buy anything you need money, to get monmey you need to be working and to be working these days you need to be under 45 and remember when asked 'Any health issues' you do NOT say 'No - had prostate cancer 18 months ago but that appears OK now!')

Self employed works for a 62 year old....not too much work but enough to stay off the Benefit
It certainly gets harder to be employed the older you get.
I think that people seem to think that a 62 year old can't possibly know anything about computers even having been in the industry from 1965, originally mechanical computing but came up through the first electronic stuff the size of a barn.