Cumulus seems to have dropped the minus sign briefly

[daj]

Cumulus is reporting my highest temp today as 5.8C at 16:49. This struck me as odd as it seems to have been below zero all day. If I look at the graphs I can see that it was -5.8(ish) at 17:00 today. In fact the highest it has actually got to all day is -4, the lowest -7.6

I had a look at the December log file and can see -5.8C at 16:50. It looks as if the negative sign has disappeared somehow during Cumulus processing (or part of it).

Steve, before I fix things, is there anything you wish to see?

I'm on version 1.8.9 build 888

thanks

[daj]

here are some screen shots showing that it recorded -5.8, plotted -5.8 but shows +5.8 in Cumulus

dayfile.png

graph.png

Cumulus.png

[steve]

As you say, almost certainly the sign got lost; I think the location must have read 5.8 briefly when Cumulus read it, i.e. the sign bit got flipped. There is no CRC checking on the data whatsoever, so if a bit gets corrupted somewhere, there's no way of telling. There's always the 'EWtempdiff' setting, which would catch things like this, set to a suitable value.

Does the time of that 5.8 max temp correspond to a period when the temp was actually -5.8?

[daj]

Does the time of that 5.8 max temp correspond to a period when the temp was actually -5.8?

Yes, at 16:49

I was just slightly confused because the was logged as -5.8 in the day file, but 5.8 in today.ini and onscreen

I fixed the today.ini and put in the correct highest of -4. I've kept a copy of all the necessary files, including the debug log just in case, but as you say it is probably just one of those little oddities we live with now and again

[steve]

I was just slightly confused because the was logged as -5.8 in the day file, but 5.8 in today.ini and onscreen

You don't mean in the dayfile, do you? I'd be very surprised if that was the case. It got logged in dec09log.txt because that is just a snapshot of the data every N minutes, but the daily highs and lows take note of every value received and update accordingly. If it's a one-off error, then it becomes the daily max, but if the error doesn't occur on the read before the log entry, then it doesn't appear in the log.

I fixed the today.ini and put in the correct highest of -4. I've kept a copy of all the necessary files, including the debug log just in case, but as you say it is probably just one of those little oddities we live with now and again

If you had the debug log turned on, then the raw data as read from the station should be in there, which might be interesting to look at; but almost certainly would just confirm that the sign bit got flipped.

[daj]

You don't mean in the dayfile, do you? I'd be very surprised if that was the case.

No, December Log file, sorry!

If you had the debug log turned on, then the raw data as read from the station should be in there, which might be interesting to look at; but almost certainly would just confirm that the sign bit got flipped.

Debug file attached -- just the minute of 16:49, sorry I can't understand the raw data but I assume in there is the data missing the minus sign

Anyway, don't waste much time on this -- it was really just for interest



extract of today.ini

Code: Select all

[Temp]
Low=-7.59999990463257
LTime=07:24
High=5.80000019073486
HTime=16:49
Total=-8707.236328125
Samples=1376

extract of Dec09Log

Code: Select all

22/12/09,16:40,-5.8,87,-7.6,0.0,0.0,0,0.0,0.0,981.4,108.9,17.9,40,0.0,-5.8,-5.8
22/12/09,16:50,-5.8,88,-7.4,0.0,0.0,0,0.0,0.0,981.5,108.9,17.9,40,0.0,-5.8,-5.8
22/12/09,17:00,-5.7,88,-7.3,0.0,0.0,0,0.0,0.0,981.5,108.9,18.0,40,0.0,-5.7,-5.7

[steve]

Here's the data with the sign bit set:

1533.23141 : 16:49:07 Data: 03 28 B3 00 58 3A 80 FB 24 00 00 00 0B 6B 01 00

then it gets unset:

1533.33141 : 16:49:17 Data: 04 28 B3 00 58 3A 00 FB 24 00 00 00 08 6B 01 00

then set again:

1533.43125 : 16:49:27 Data: 04 28 B3 00 58 3A 80 FB 24 00 00 00 08 6B 01 00