Lost Sensor Contact!!

[EvilV]

That isn't the first complaint about a bad USB cable I have seen. In another case a small scale manufacturer of software defined radios (using software and a computer to do a lot of digital processing of signals) had sent out a batch of hardware to purchasers only to have about four of them returned as faulty out of the twenty he had made, only to find it was the cheapo cables he had used, and not the main equipment at all.



software defined radio on the web. Scroll down and click in waterfall displays to see what it is about. http://websdr.ewi.utwente.nl:8901/

[Gina]

I see. Yes the supplied cable was certainly rubbish. The replacement is well screened with a layer if silver paper and then braid. The one supplied had no screening whatsoever.

Back in action! All working again. Now all I need to do is correct my data for the down period. It seems to think we had 3.5 metres of rain, at the time it went wrong!!!

[EvilV]

Hope all the glitches were related to the same cause.

[Gina]

Hope all the glitches were related to the same cause.

They weren't!! Would've been nice if they'd disappeared but there's been one since I got it going again. I've cleared the false data from when the station was off-line. Result below...

24hrs-Temperature-5.png

EDIT... The console in now on the floor next to the computer. I think the signal strength is alright - I don't think that was the problem.

[Gina]

Here is a temperature plot over the last 5 days.

120hrs-Temperature.png

I thought there was a pattern but perhaps not.

[bruce45]

could this have anything to do with the operating system that you are using doing other things causing the problem. never used linux so could be talking out of the top of my head. maybe Steve could confirm if it is possable?

[Gina]

could this have anything to do with the operating system that you are using doing other things causing the problem. never used linux so could be talking out of the top of my head. maybe Steve could confirm if it is possable?

Thanks for your suggestion but the only scheduled event is the weather station process so while this could be possible when I'm doing other things on this computer, there is nothing that could do it when I'm not.

[steve]

I'm pretty sure it's nothing to do with Linux. Cumulus users get this problem under Windows just the same as Gina gets it with her own software under Linux, so it's pretty certain that it's some sort of design problem with the station.

[Super-T]

Gina
I had to re-solder the socket on to the pcb as there wasn't enough heat supplied on the original soldering.
Had same intermittent connection problems but not spikes.

[simonvfr]

Sympathy for you! These intermittent problems are the most frustrating!!

Just one thought - surely this can't be signal strength related, because this would lead to flat lines not spikes? A spike seems to indicate false data being transmitted or received or generated after reception. Weak signal would give no data and the 'lost sensor contact' error.

I've been astonished at the range and penetration of the signal with the Fine Offset. In my case some 25 metres, through two walls (one of which is 90cm thick stone and rubble).

[Charlie]

Hi Simon. Actually, false data is quite common. I believe the transmitted data does not contain any sort of error checking mechanism - not even something as trivial and easy to implement as a checksum. However, as it stands, it's not unusual to, every few weeks, have the station reporting hundred degree temperature swings within a few seconds, feet of rain sometimes even negative, and hurricane force winds. It's a minor annoyance, and fixable by editing the data although inconvenient to do so.

What is strange is that the problem will go away for months, appear for a couple weeks, then go away again. If we could tie it to a root cause, we could likely resolve the issue. Intermittent problems are every geeks nightmare.

[daj]

A timely reminder that there are an entry that can be included in the cumulus.ini file to help ignore huge swings in readings.

EWTempdiff=x.y

http://wiki.sandaysoft.com/a/Cumulus.ini

Just for the FineOffset/Easy Weather stations

I have an enhancement request in for rain and pressure reading too

[simonvfr]

Hi Simon. Actually, false data is quite common.

I understand the possibility of false data - all I was suggesting, in trying to pin down a cause, was that weak signal would seem more likely to produce flat lines than spikes. It's a shame that there's no way of directly testing signal strength!

[jensdk]

Hi everybody

I have been having troubles with the values reported from my console to Cumulus:
I had removed the batteries from the console, since it is connected to the USB-port and gets the needed power from there. Everything seemed ok until I turned my pc off and the console of course went out. I started my pc and the console started too, but the first set of readings were very odd, -22C and about 1060 hpa, when real temp was 2C and pressure was 1010hpa. After a few seconds Cumulus got the right values from the console and everything was normal, except for my today.htm and records.htm

I learned from this that I always need to have well functioning batteries in the console to avoid powering off the console. Maybe it could be a problem with the contacts in the console resulting in the glitches.

Just a thougt....

I have now installed Cumulus as a service, and configured windows to reboot once every 24 hours and everything has been working fine for about 2 months.

[EvilV]

Hi Simon. Actually, false data is quite common.

I understand the possibility of false data - all I was suggesting, in trying to pin down a cause, was that weak signal would seem more likely to produce flat lines than spikes. It's a shame that there's no way of directly testing signal strength!

There is if you have a receiver that will operate on the required frequency. Such things are pretty expensive for the occasional check though.