Page 1 of 1

Data flatlining

Posted: Wed 15 Nov 2017 3:40 pm
by devil
At irregular intervals from a few days to about three weeks, the data to Cumulus (or WD) flatline for no apparent reason. It is easy to bring them back on: close down Cumulus, disconnect the console, remove the batteries, after a few seconds replace the batteries, reconnect the console, press the Down key on the console for a few seconds, reload Cumulus and, after a couple of minutes, everything works again until the next glitch.

Is there any way to avoid this flatining?

Re: Data flatlining

Posted: Wed 15 Nov 2017 4:15 pm
by steve
Are you saying this happens with Weather Display too?

What weather station? You have two in your profile. It sounds like you mean the Fine Offset station - these are well known for the USB connection locking up. I assume you are using the settings in Cumulus to synchronise reads with the station?

Re: Data flatlining

Posted: Thu 16 Nov 2017 9:31 am
by ConligWX
have you any power management set on your windows 10 box to user power Management on USB connections?

Which Weather station console connection is locking up? Davis or aercus?

here are some tips on stopping Windows from preventing power management.

https://helpdeskgeek.com/how-to/prevent ... sb-device/

please note even when Windows Power Options Scheme is set to High Performance, USB ports can still be powered down!

https://helpdesk.flexradio.com/hc/en-us ... ed-Devices

Also installing Ferrite Cores onto the USB cable of the Fine Offset console or using a USB powered hub can help with lockups.

Re: Data flatlining

Posted: Thu 16 Nov 2017 2:41 pm
by devil
steve wrote:Are you saying this happens with Weather Display too?

What weather station? You have two in your profile. It sounds like you mean the Fine Offset station - these are well known for the USB connection locking up. I assume you are using the settings in Cumulus to synchronise reads with the station?
Steve, thanks for your reply. It happens with both WD and Aercus at the same time. Sorry, I don't know what 'Fine Offset' means. The back of the console reads:
Aercus Instruments
Weather Station (receiver)
Model WS 3083...

Yes, settings in Cumulus are synchronised.

I realise that this problem is Aercus and not specifically Cumulus, but I thought I'd try for help here, Ta much!

Re: Data flatlining

Posted: Thu 16 Nov 2017 3:08 pm
by devil
Toxic17 wrote:have you any power management set on your windows 10 box to user power Management on USB connections?

Which Weather station console connection is locking up? Davis or aercus?

here are some tips on stopping Windows from preventing power management.

https://helpdeskgeek.com/how-to/prevent ... sb-device/

please note even when Windows Power Options Scheme is set to High Performance, USB ports can still be powered down!

https://helpdesk.flexradio.com/hc/en-us ... ed-Devices

Also installing Ferrite Cores onto the USB cable of the Fine Offset console or using a USB powered hub can help with lockups.
Thanks for your help. Both the Davis and Aercus systems stop working at the same time; both flatline. As I said to Steve, I know this is not specifically a Cumulus problem but probably within the Aercus console as rebooting it by USB connector and battery removal is a workaround for both weather systems at the same time. Your idea of ferrite cores may be a help.

Re: Data flatlining

Posted: Thu 16 Nov 2017 3:50 pm
by ConligWX
devil wrote:Your idea of ferrite cores may be a help.
A Powered USB Hub is usually a better option though slight more expense or try a better USB cable with Ferrite chokes built in may possibly help.

Re: Data flatlining

Posted: Thu 16 Nov 2017 9:17 pm
by Super-T
WH-1081 Weather Station.
Slightly different problem but possible the same cause. I get just the outside temperature flat-lining at around 7am each day. It was intermittent but is turning into very regular now. Strange thing is that it is only the Outside Temperature that flat-lines.
Easy to fix by going back to the previous backup and copying the data to the data folder. Then all good for another day.
First pic shows the outside temperature flat. Second pic shows the Pressure as NOT flat-lined (Wind Direction & strength both NOT flat lined as well).
Third pic shows all OK after replacing the data from backup.
This suggests to me that the problem is in the Console and only affects the outside temperature data as the data is actually still in the console but just not displaying correctly. Possibly a problem with Cumulus but unlikely as nothing has changed there.
This Weather Station has been up and running for quite a while. Connected to an "always On" laptop that doesn't sleep. The USB Cable has a couple of loops through a Ferrite.
One suspicion could be that somehow the latest MS updates have upset things.

Ignore the temperature rapid climbs.....sun-strike which I am working on.

Any logical explanation welcomed as I can't see a pattern or a reason yet.
2017-11-17_092849.jpg
2017-11-17_092943.jpg
2017-11-17_094959.jpg

Re: Data flatlining

Posted: Fri 17 Nov 2017 10:21 am
by AllyCat
Hi,
Super-T wrote:Slightly different problem but possible the same cause. I get just the outside temperature flat-lining at around 7am each day. It was intermittent but is turning into very regular now. Strange thing is that it is only the Outside Temperature that flat-lines.
You have checked the spike removal setting, haven't you ?

Cheers, Alan.

Re: Data flatlining

Posted: Fri 17 Nov 2017 9:04 pm
by Super-T
Yes, set at 0.3C and that hasn't changed.
Mind you I have had to remove spikes larger than that at times so perhaps the spike removal is not working.
Also interesting is that the temperature spike removal is the only one operational under "Spike Removal"
SpikeRemoval.png

Re: Data flatlining

Posted: Fri 17 Nov 2017 9:16 pm
by steve
That’s a very low setting for spike removal, a Fine Offset “spike” is typically very much higher than that. Your flat-lining is almost certainly caused by your spike removal setting (i.e. your station is recording an increase in temperature of more than 0.3 degrees over a 48 second period, this is perfectly reasonable and is not a spike).

Re: Data flatlining

Posted: Sat 18 Nov 2017 9:39 am
by Super-T
Bumped it up to 1 degree and will see what happens. Will take it higher if required.
Why would a low setting make the data flat-line. I thought it was to limit the height of the spike when there was a spike?
What do other people have the temperature spike removal set to?

Re: Data flatlining

Posted: Sat 18 Nov 2017 10:48 am
by steve
All the true spikes I’ve seen generated by Fine Offset stations have been several degrees, so I I’m assuming that’s what most people have it set to. The typical spikes which Cumulus tries to ignore exist for one reading only, and the reading returns to normal. So the spike removal setting makes Cumulus ignore a reading which suddenly increases by more than the setting, on the assumption that it will return to normal on the next reading.

If it’s not actually a spike, just a valid increase, the reading won’t revert, so Cumulus continues ignoring the reading.

This isn’t a sophisticated facility, it does a reasonable job of filtering out the sudden, very large, one-off readings which Fine Offset stations are known to produce from time to time.

Re: Data flatlining

Posted: Sat 18 Nov 2017 11:08 am
by sfws
steve wrote:All the true spikes I’ve seen generated by Fine Offset stations have been several degrees
My spike setting has been 20 degrees Celsius, and I don't recall any problems ever with that figure in half a decade.

Re: Data flatlining

Posted: Sat 18 Nov 2017 11:25 am
by AllyCat
Hi,

"0.3 degrees rise in 48 seconds" might sound like a reasonable limit, but it isn't.

Firstly, several "messages" from the transmitter might be "lost" due to radio noise or interference of some type. The Console only assumes a "problem" if about 7 consecutive transmissions are lost (> 5 minutes). Also, the Real Time Clock versions (not in Australasia) actually stop transmitting (weather) data for around 5 minutes after each hour. So users may need to consider the greatest possible rise that might ever occur over say 5 minutes or more.

Also, there have been several reports (from Australia I believe) that at least some of the WH308x (solar) versions generate quite large "steps" (approaching one degree C) in the scale at certain temperatures. This is presumably a "sample" calibration fault, but hasn't been explained or disproved. Personally, I don't use a spike threshold, but if I did, it would be at least several degrees C.

Note that the spike detection doesn't apply to Logger data, which explains why downloading from the Console fixed the flatline problem.

Cheers, Alan.

Re: Data flatlining

Posted: Sat 18 Nov 2017 10:09 pm
by Super-T
Thanks all. Much clearer now.