Cumulus Support

Hello to all on the Forum. I'm completely new to Cumulus and the Forum having struggled along with standard EasyWeather software for about a year before discovering what appears to be a great product. I have only had it running for 48 hours and downloaded the latest version. All seems to be well apart from the temperature spikes experienced by so many of you on this Forum. I put this down to the 10m extended cables I am using to get the temperature sensor in a suitable location.

The spikes I see occur several times a day and are usually just one data point around +/- 3 degrees C in amplitude. I can see them on both the EasyWeather graphs and also the Cumulus graphs (I have not removed the EasyWeather software yet but have not run the programs at the same time). The log interval is set to 10 minutes on both. I have tried to use the spike removal setting in Cumulus but no matter what setting I try from 0.1 to 1 degree C it does not remove the spikes. I have read all the posts and advice I can find in Help and on the Forum but I figure I must be missing something here?

Any advice much appreciated.

Skyline

I'm guessing that these values are read from the logger, particularly as you say EW sees them too. The 'spike removal' facility in Cumulus does not operate on the logger data; generally those figures are OK, and it is only transient corruptions in the live data which are usually a problem with these stations. Typically the sort of spikes experienced are much larger than the three degrees that you are seeing.

Thanks for the reply. I have seen the spikes on the graphs in both sets of software but also watched the temperature jump up on the WH1081 base station display (is this the logger?). Forgive me for not being familiar with the technical talk but are you saying that this is a different type of problem and that 3 degrees is normal for this type of system? I have not fully understood what you mean by this functionality not applying to the logger data and if not this then under what circumstances Cumulus would remove the spikes?

Thanks,

Skyline

When I say 'logger' I mean 'historical' data stored while Cumulus isn't running and then downloaded when Cumulus starts up. The 'spikes' which Cumulus attempts to remove are typically caused by electrical interference, or some other corruption, and this works only while Cumulus is running and reading the data 'live', which changes every 48 seconds. A three degree change over a 10-minute period is not unrealistic, so it is hard to filter these out. I am no expert, but it sounds to me like your temperature sensor is actually generating these values. Perhaps one of our hardware experts can comment.

I see these spikes too which when taken in context with the temperature trend are clearly anomalies. I guess Cumulus would need some trend analysis to be able to smooth these spikes.

Thanks for the explanations, now I understand better how it works. All my data is collected while Cumulus is offline so that is probably why. Maybe I will have to look into removing those extension cables instead.

Thanks,
Skyline

Skyline wrote:Thanks for the explanations, now I understand better how it works. All my data is collected while Cumulus is offline so that is probably why. Maybe I will have to look into removing those extension cables instead.

Thanks,
Skyline

Happy New Year, Skyline:

Maybe you don´t need to remove your cabling extensions, first try to implement the spike removal feature built in Cumulus.

I have a LaCrosse WS-2310/2315 at home and I was having the same problem a while ago, (was harder to correct my data files since this station has been running since July 2008), but with the spike filter solved this headache.

Maybe using some toroidals or shielded cable help to solve the problem.

Below is my actual Cumulus.ini section related to this issue, hope my figures can help you as starting point.

Simple telephone extension cables are susceptible to spikes. I found using a Cat5E (twisted pair) cable reduced the frequency and severity of temp spikes, although not completely. I used Cat5 (6 way) sockets to connect the cables, which accept the original RJ45 phone connectors without a problem.

Most of my data is collected with Cumulus live, and the spike filter works well. A spike in my overnight data (from the logger) is extremely rare...

WirelessCH wrote:I see these spikes too which when taken in context with the temperature trend are clearly anomalies. I guess Cumulus would need some trend analysis to be able to smooth these spikes.

Yes, the 'spike removal' facility is very basic and ideally it would be rather more sophisticated. It does seem to suffice for the majority of cases.

Thanks all, Cat 5 here I come. I guess the use of inline connectors is also a bad practice? I.e. best to remove the covers and refix the longer cables direct to the device. I didnt think it was so important so I just used RJ45 straight in line female - female connectors when I installed it.

Thanks.

when I set up my WH3081 I used the standard leads left on but for the winds extension its cat5e from the standard cable onwards with waterproofed modular joiners, the method has been succesful, not a single spike since install *touches wood*

at the other end it terminates back into a short length of RJ12 cable, works well and is a simple setup, aim for about 5 metres of Cat5e and your woes should disappear

Just thought people might be interested in progress on this one.

I have now fitted aluminium shielded Cat5e cabling from the wind direction sensor all the way to the transmitter. I decided not to risk keeping the old cables so I removed them altogether and soldered the Cat5e directly into the direction sensor circuit board. At the transmitter end I fitted an RJ11 directly onto the Cat5e and plugged into the transmitter. The total run is about 6m.

I am disappointed to report this does not seem to have helped eliminate the outdoor temperature spikes. I am stuggling to claim even a marginal improvement and given the fact that the weather has fallen relatively calm since the New Year gales I think I'm heading up the wrong track but at least its eliminated the cable. What I have proved is that there is no such problem when the wind sensor is disconnected from the transmitter. So this implies that either there is a strong source of interference which only affects the wind section/cable or that the signals entering the transmitter affect the temperature data. Given that the transmitter only transmits a signal every 48s or so, presumably whatever causes this problem would have to coincide with the transmission time unless it related to the wind equipment. I'm guessing upon transmission a simple snapshot is taken of wind direction and speed as well as temperature. By overlaying the wind and temperature graphs you can also see that the temperature spikes sometimes, but not always coincide with high wind gusts.

So it remains a mystery but if I discover anything new I will let you know.

Skyline

This sounds like the sort of thing that Gina found (see her long thread(s) in this forum).

Hi,

It's probably because the wind vane wire is permanently connected to the input of the same Analogue-Digital conveter* as is used for the external temperature measurement. The micro selects which conversion is done by applying a voltage to the opposite end of either the thermistor or the vane resistors. *(for want of a better description - it only times the ramp on a capacitor connected to one of the micro's pins. I'm amazed that it works at all ! )

You might be able to devise a suitable filter to put at the lower end of the cable but I think Gina failed and eventually used an optocoupler for her "electric fence" problem. I'm considering some type of "input gate" circuit (together with other associated enhancements) but sorry, at present it's still just a "gleam in my eye".

Cheers, Alan.

I get temp spikes and use the spike removal tool. I have just gone from a calibration of 1.0C to 1.1C. 1.0 still had a spike removed but never recovered. I got the flat line! I believe there is a request in to make it come with a type of error light notification.
Anyway, my spikes only appeared after moving the temp sensor to a location to provide better readings, unfortunately that has gone from being 3 meters away with 1 brick wall in between, to being 13 meters and 3 brick walls. This does have extensions for both rain and wind gauges. Off memory, I think it has spikes during wind gusts, but the latest one was due to nothing else, unless of course, we did have a temp rise of 1C or more within 48 seconds! (unlikely). I only make it this high because sometimes it doesn't recover from the spike removal.