What causes the +999 and -500 Temp errors?

[mattsimis]

I asked here before and was told that Cumulas automatically filters these out.. but it doesnt!
http://weather.netcessible.com/trends.htm

Totally messes up my Records and Trend pages.

[steve]

They're not filtered out because they're not generated by the station, so whoever told you that those values are filtered out was mistaken. They are the default values which Cumulus uses at start of day, so that they get overwritten as soon as the first real data is read.

So, on whatever day it was that those values occurred (I can't tell exactly which day due to the 3D graphs), you had some kind of problem which resulted in those values erroneously being written as the end of day values into dayfile.txt. The error could have been caused by your station stopping sending data (but this would have had to have been for a prolonged period) or by some problem with your PC clock. If you have Windows automatically change your clock at an inconvenient time, i.e. within a hour or so of midnight, this can cause a spurious day 'rollover'.

You should be able to correct the problem by editing dayfile.txt, with Windows stopped. If you want me to look to find the cause of the problem, please zip up the diags folder and attach it - assuming the relevant file is still there, it depends on how many times you have started Cumulus since then.

[Synewave]

You should be able to correct the problem by editing dayfile.txt, with Windows stopped.

Windows? Did you mean Cumulus?

[steve]

You should be able to correct the problem by editing dayfile.txt, with Windows stopped.

Windows? Did you mean Cumulus?

It makes it much more interesting if you do it with Windows stopped

But, yes, you're right of course; thanks.

[mattsimis]

So, on whatever day it was that those values occurred (I can't tell exactly which day due to the 3D graphs), you had some kind of problem which resulted in those values erroneously being written as the end of day values into dayfile.txt. The error could have been caused by your station stopping sending data (but this would have had to have been for a prolonged period) or by some problem with your PC clock.

Well thats likely the issue then, every now and then (once every 2.5wks maybe) the Weather Station stops responding from say 22:00 to 04:00 the next day. Why does Cumulus intentionally write such obviously incorrect figures though? Is there anyway of changing this (even to just use the last known good figure)? This would only introduce a minor error instead of a graph breaking massive one.

[steve]

Well thats likely the issue then, every now and then (once every 2.5wks maybe) the Weather Station stops responding from say 22:00 to 04:00 the next day.

That wouldn't cause this problem, because a real reading would then be taken at 0400 (assuming you spot the problem and restart Cumulus) and the defaults would disappear.

Why does Cumulus intentionally write such obviously incorrect figures though? Is there anyway of changing this (even to just use the last known good figure)?

It doesn't "intentionally" write the figures; it doesn't check each and every value before it writes it to the file. The assumption is that in a period of 24 hours you will get at least one good reading. Not an unreasonable assumption, I think. There is no "last known good" figure, because this is a new day and all of yesterday's figures have been reset. I suppose it could check that the figures haven't changed from the default before writing the dayfile.txt entry, and not write the entry, but this is such an unlikely scenario on all but completely broken systems, that it's not worth writing the code. The solution is to fix the problem, not to hide the symptoms.

[mattsimis]

Well thats likely the issue then, every now and then (once every 2.5wks maybe) the Weather Station stops responding from say 22:00 to 04:00 the next day.

That wouldn't cause this problem, because a real reading would then be taken at 0400 (assuming you spot the problem and restart Cumulus) and the defaults would disappear.

Well thats not quite what I said. The WS occasionally (and for no apparent reason) stops sending data (or the Receiver stops picking it up), then resumes. I do not restart Cumulus, I just note the -500 +990 error in the recorded temps and the "no data" error in the time log. Obviously restarting the App hours after it has started to record data does nothing.

The solution is to fix the problem, not to hide the symptoms.

Well obviously that would be best, but my Im not going to buy another weather station to fix such an intermittent problem. The Receiver and Weather station are as close as humanly possible (given one is outdoors on a 5m mast and one in the house on the same side), Ive moved the WS to its own dedicated switched PSU (instead of batteries) etc.

It just seems such a "dumb" problem that can be worked around by simple text editing would be nice if the user was given a more automated way of implementing this kludge.

[steve]

Well thats not quite what I said. The WS occasionally (and for no apparent reason) stops sending data (or the Receiver stops picking it up), then resumes. I do not restart Cumulus, I just note the -500 +990 error in the recorded temps and the "no data" error in the time log. Obviously restarting the App hours after it has started to record data does nothing.

So what you actually mean is that the console loses contact with the sensor. Again, as long as this condition does not persist for a period of more than 24 hours, it shouldn't cause the default values to be written to the log.

Well obviously that would be best, but my Im not going to buy another weather station to fix such an intermittent problem.

That's not what I said.

It just seems such a "dumb" problem that can be worked around by simple text editing would be nice if the user was given a more automated way of implementing this kludge.

It would be masking the problem. Better to fix the problem.

[Gina]

I've been suffering from spikes but never had those values when running Cumulus. Also, I can't remember seeing anyone else mention it. You may have a faulty unit.

[6719jason]

Hi All,

I am afraid for reasons unknown to me, I had this +900'C and -500'C issue back in Mid October/Early November. The issue seemed to clear after replacing the batteries with new ones and ensuring that it was a clear path between the transmitter and receiver units.

The console with often report and sound a 'Low Temp' or 'High Temp' Alarm and show a clearly incorrect figure but Cumulus fortunately does not take any notice of these weird values and within a few mins, the values return back to normal on the Console. It normally occurs just after or before a 'Lost Sensor Contact' message.

[mattsimis]

Well obviously that would be best, but my Im not going to buy another weather station to fix such an intermittent problem.

That's not what I said.

Lol, no but its the only option if the unit is a little iffy.

It would be masking the problem. Better to fix the problem.

I have no problem masking the problem, but how would one fix the problem in your opinion?

[daj]

I had these problems and it appeared to be a poor signal/interference between the transmitter and the LCD console. I moved the console closer, and stood it upright rather than on its back

[steve]

how would one fix the problem in your opinion?

I can't say yet, because I don't know what the problem is.

Discounting for now an as yet undiscovered bug, there are two ways that I know of that this problem can occur. The first is that your PC clock goes backwards at around midnight or 0100 (normally because Windows is setting the clock). The solution to this is obvious, but we can discount it because I've already mentioned it, and if it were the problem we wouldn't still be having this discussion.

The second way is that your station can stop supplying valid data, whether because the console loses contact with the sensors, or for some other reason, for a period of more than 24 hours. Are you saying that is what happens, because that's not clear from what you've said so far.

Another possibility: For some reason you're still using a beta version of 1.9.0. Perhaps there was a bug that I fixed in the three months between that version and the final release.

[steve]

I've changed the latest code so it checks when it's about to write dayfile.txt that the high/low temp values aren't still at the default, and if so, it logs an error instead of adding an entry to dayfile.txt.

[mattsimis]

Sorry went AWOL there due to other weather related issues (ie no central heating for 4wks now!), but will update to the latest version ASAP and try it out, thanks.