Page 1 of 1
The 48 seconds rule for the FineOffset
Posted: Tue 09 Feb 2010 4:14 pm
by daj
Can someone remind me about the 48 second rule for the FineOffset.....is it that the station only updates data every 48 seconds, therefore Cumulus is potentially waiting 48 seconds for a new packet of data?
I know Cumulus checks every few seconds for data but where does the 48 seconds come in to it
thanks
Re: The 48 seconds rule for the FineOffset
Posted: Tue 09 Feb 2010 4:29 pm
by steve
Yes, the station updates every 48 seconds; I think that probably corresponds to how often the remote sensor transmits. Cumulus reads the data every 10 seconds because it doesn't know when the station is going to update.
Re: The 48 seconds rule for the FineOffset
Posted: Tue 09 Feb 2010 4:34 pm
by daj
Thanks
It got me thinking about the real-time thing again -- updating my site every 5 seconds, whereas at some point in the 48 second window I am getting one real update
Re: The 48 seconds rule for the FineOffset
Posted: Tue 09 Feb 2010 5:14 pm
by steve
Yes - a 40 to 50 second interval would make more sense - provided you are willing to accept the fact that your 'real time' data may actually be from some time in the past. Take the case of a wind gust. In the worst case, the gust occurs, 48 seconds later the station updates, 10 seconds later Cumulus reads the data, 40 seconds later (say) the real time update occurs, and the data is already nearly 100 seconds old. Your web site may then have another delay before it reads the new data and updates (e.g. the Flash or Silverlight gauges). But does it matter if the data is a minute or two old? Who knows or cares?
Re: The 48 seconds rule for the FineOffset
Posted: Tue 09 Feb 2010 5:36 pm
by daj
Yeah, I think it is all about managing your visitors expectations. If I am updating every 5 seconds they are probably expecting something to change. Whereas if I am updating in 30 seconds something will possibly change on the next refresh

Re: The 48 seconds rule for the FineOffset
Posted: Tue 09 Feb 2010 5:38 pm
by Repairman77
But does it matter if the data is a minute or two old? Who knows or cares?
Absolutely Steve; unless you are one of those people who sits in front of the console watching it all day, LOL.
Mike.
Re: The 48 seconds rule for the FineOffset
Posted: Tue 09 Feb 2010 5:39 pm
by daj
I think I will put a note on the Wiki under the realtime section about the 48 second rule for the FineOffset stations
Re: The 48 seconds rule for the FineOffset
Posted: Tue 09 Feb 2010 10:43 pm
by Gina
I have a strange delay with my weathercam. Something like 20 minutes!! I haven't been able to pin it down other than it's in the
webcam software. I'm planning to include the webcam in my weather station software rather than running it as a separate entity with separate FTP etc. Maybe then I can sort this out. Not really of any great consequence but it's bugging me a bit and it's strange around dusk, to look out the window and it's dark but the webcam is still showing quite a good picture. The reverse happens at dawn, of course. I know this delay exists because I set up a large faced clock on a pole in view of the webcam. Oddly, it makes no difference if I use a fast or slow computer to handle it. Nor does the capture rate make a difference. Most peculiar.
I was thinking of a fast update for the weather data but decided that data being 10 or 15 mins out of date is of no consequence. The only thing that would look nice to have updated more frequently are the wind dials, as in Cumulus. But OTOH instantaneous wind speed and direction are really rather meaningless anyway. I think it's really just that we are used to seeing movement and a moving gauge looks more interesting than a static one.
So I've decided to go for 5 min data sampling with FTP uploads every 10 or 15 mins, though I may leave the webcam on 5 min intervals - not sure yet.
One thing I would rather like to implement is a time-lapse view of the weathercam showing clouds racing across the sky and rain showers etc. too. I think that would be rather neat

For that, a capture rate faster than 5mins would be preferred. My trials with a frame every 5m produce a very fast video

Think I'm fantasising again....
Re: The 48 seconds rule for the FineOffset
Posted: Tue 09 Feb 2010 11:14 pm
by hills
I used to use Cumulus to upload my webcam image every 5 minutes, so with yawcam taking a snapshot every 5 minutes I could potentially be 10 minutes behind, however I have now changed over to yawcam uploading it every minute and so far it has been 100% reliable and my webcam image is never more than about a minute old. Luckily uploads and downloads to/from my ISP are in the freezone so don't affect my monthly quota.