w8681 / Cumulus misbehaving ! - corrupted data ?

[dc1500]

I must have one of the lucky units (touch wood). No errors so far and have about four other wireless transmitters and wireless PC nearby not to mention the home phones.. Only about 20m away though. As a general tip AA lithium batteries perform more consistently, especially in sub zero temperatures.
Dave

[adrian5750]

HI again

Thanks for the comments...

I'm wondering if the problem could be the base station, rather than the remote unit....

Next time it flatlines, I must remember to reset the base station first - and see if that sorts out the problem...
it's as though the base station loses communication with the remote sender, and then isn't able to reinstate it -
it would be an elegant thing the base station was able to tell Cumulus 'whoah - we have a problem' - but then I guess Cumulus would need some way of dealing with / not displaying the false data..

Reminds me of the old programmers' adage 'Never check for an error condition you can't handle!' <g>

Thanks
Adrian

[harrym1byt]

I wonder if Steve might be persuaded to add a little routine into the FO section to indicate to us when there is data missed, or his software has had to compensate for data corruption?

[steve]

I wonder if Steve might be persuaded to add a little routine into the FO section to indicate to us when there is data missed, or his software has had to compensate for data corruption?

It logs to the debug log when it detects an error in the data, so it's already there. But I don't recommend leaving the debug log turned on, it will cause problems after a while. I should change to a different logging component (like the one I use at startup), but like everything else it's a question of finding the time.

Steve

[steve]

it would be an elegant thing the base station was able to tell Cumulus 'whoah - we have a problem' - but then I guess Cumulus would need some way of dealing with / not displaying the false data..

There's a flag in the status byte that means that contact with the sensors has been lost; Cumulus logs this in the debug log and ignores the data. But of course because of the way Cumulus works it continues to log the last good data it received.

Steve

[adrian5750]

HI Steve

Ah - so Cumulus does know that the comms to the sensor is broken.....
...but then there's the question of 'what do we do with this information...?' <g>

Many years ago, working on automatic-guided fork-trucks - the hardware designer proudly
gave me details of the new 'processor-fail' detection circuitry - and suggested that, when
it indicated a problem we could display a suitable message..... on the display screen
that was driven by said processor..... !

You almost need a flag within each stored reading in Cumulus that says 'duff one' - and then tweak
'everything' that uses that data to show 'duff' - but I can imagine it's a heck of a lot of work -
I'll try moving the location of the base station first ! <g>

All trotting along nicely at the moment.... claiming 7c and 3.4c after the wind-chill - brrrr!

Thanks Steve
Adrian - West Cork, Ireland

[steve]

Yes - two design flaws in Cumulus are that it has no way of indicating bad or missing data, so it just ignores it, and that it continues to log and plot the last good data that it received. These are the sort of things that made me realise it was time to start again from scratch...

Steve

[adrian5750]

Hi Steve

Yes - two design flaws in Cumulus are that it has no way of indicating bad or missing data, so it just ignores it, and that it continues to log and plot the last good data that it received. These are the sort of things that made me realise it was time to start again from scratch...

Steve

No problem - just think how many 'goes' Microsoft have had at getting it right <g> - and it's so often the case that you can't see the 'obvious' flaws in your design until you've finished it !

Regards
Adrian - West Cork, Ireland
www.ambquality.co.uk/weather/index.htm

[steve]

And Microsoft don't have the excuse that they didn't know from the start they were writing something for other people to use. Cumulus was only ever going to be for me to use on my weather station. How did all this happen?

Steve

[harrym1byt]

"You almost need a flag within each stored reading in Cumulus that says 'duff one' - and then tweak
'everything' that uses that data to show 'duff' - but I can imagine it's a heck of a lot of work -
I'll try moving the location of the base station first ! <g>"

Just a count maintained of the amount of dud data received which had to be ignored - and the count (optinally) displayed in one corner of the display. Count cleared at every restart of Cumulus - Just enough to give some indication of how good or bad the RF link is.

[harrym1byt]

Cumulus was only ever going to be for me to use on my weather station. How did all this happen?

Steve

Obvious innit - You were far too obliging, made too good a job of it and recommendation works much better than advertising