Strange graphs - take 2

[malc-c]

Yesterday the graph data for windspeed showed that the sensor possibly froze up over night, but now last nights pressure graph showed strange drop for several hours yesterday evening.

Not sure if it's my sensor array that is playing up or Cumulus recording incorrect data from the files ?

[steve]

Do the times of the sudden drop and the sudden increase correspond with restarting Cumulus? If so, this FAQ probably applies:

http://wiki.sandaysoft.com/a/FAQ#My_pre ... dropped.21

[malc-c]

Not sure Steve,

I can't remember if I kept Cumulus running through out the evening or it was closed whilst I did other things on the PC last night. I think it was closed

I've certainly not changed any values or settings in the software

[steve]

Please zip and attach your diags folder.

[malc-c]

Steve,

I've checked the month log file and sure enough there are drops at 16:50 (went from 992.2 to 979.4) and then at 00:25 it went back up from 980.9 to 993.3

diags folder zipped and attached.

I've also checked the settings and offset has been set.

[steve]

Yes, the times correspond to starting Cumulus, so the FAQ applies.

You stopped Cumulus at 1654. For that run, Cumulus had been using a pressure offset of 7.5:

2010-01-30 16:13:42 : EWUSB: Read pressure values: rel=992.0 abs=984.5 offset =7.5

When you started Cumulus again at 0024, the pressure offset was read as minus 5.3:

2010-01-31 00:24:15 : EWUSB: Read pressure values: rel=981.1 abs=986.4 offset =-5.3

You then stopped Cumulus and started it again at 1008, when the offset was back to 7.5:

2010-01-31 10:08:09 : EWUSB: Read pressure values: rel=994.2 abs=986.7 offset =7.5

[malc-c]

Thanks for the diagnostics Steve,

What could cause the offset to change ?

I've read the FAQ, but it's hasn't sunk in. What would be the best / simple way to correct the off set for those readings in that period ?

[steve]

What could cause the offset to change ?

Some sort of corruption. Probably a single bit flipped - there are no checksums on the data, this is one reason for the garbage data that is sometimes read.

What would be the best / simple way to correct the off set for those readings in that period ?

You can either edit the log file and add 12.8 to all of the wrong values, or restart from the 0024 backup.

[malc-c]

Steve,

Restored a backup prior to 16:50... which then gave me the incorrect rain fall that was discussed and corrected yesterday. I then re-edited the log file as before and everything is now showing corrected readings.

Thanks again for the support

[steve]

Restored a backup prior to 16:50... which then gave me the incorrect rain fall that was discussed and corrected yesterday.

Why didn't you use the 0024 backup as I suggested, which is when the problem happened?

[malc-c]

Sorry.. my logic was that as the readings changed between 16:50 and 00:25, using the 00:24 backup would not resolve the issue as it would still have the incorrect readings ??

[steve]

The data with the incorrect readings was downloaded at 0024 - you didn't have Cumulus running for the period with the incorrrect data, so it came from the logger.

Anyway, it doesn't matter now as you've got it sorted. It would just have been easier, that's all.