Problems with station communication: USB locks up completely

[steve]

Thanks, Jim. So we're avoiding reading within 3 seconds of the console writing data to memory, and within 3 seconds of the console moving on to the next location (which just involves changing the data pointer)?

[jim-easterbrook]

Thanks, Jim. So we're avoiding reading within 3 seconds of the console writing data to memory, and within 3 seconds of the console moving on to the next location (which just involves changing the data pointer)?

Yes, although moving to the next location also involves writing data, as the current data is copied to the new location.

[steve]

Yes, although moving to the next location also involves writing data, as the current data is copied to the new location.

Ah, yes, that makes sense, since it happens independently of new data arriving.

[rwilliam99]

Well, as expected my re-arrangement of things didn't make any difference. This morning I lost contact with the console. After much experimentation, I believe I have the EasyWeather.dat arrangement working now. I used the setLogger utility posted here to change the interval to 1 minute and updated the interval in Cumulus to 1 minute to match. We will see how this works. It seems I was getting a pretty regular 5 days or so with the other setup. If I can go a couple of weeks with this, maybe it is OK.

I will keep everyone informed as to what happens with this experiment.

Rob

[rwilliam99]

One odd thing has happened since I made this change - my rain totals are all messed up. We haven't had any rain today, but it keeps increasing the rain for today. Does it have something to do with the logging interval (I currently have it set at 1 minute) and the rainfall display setting in EasyWeather? I'm not sure how that would be, but EasyWeather's rainfall for today is all messed up too. It was not doing this until I changed the way it was logging.

What am I missing?

Rob

[steve]

One odd thing has happened since I made this change - my rain totals are all messed up. We haven't had any rain today, but it keeps increasing the rain for today. Does it have something to do with the logging interval (I currently have it set at 1 minute) and the rainfall display setting in EasyWeather? I'm not sure how that would be, but EasyWeather's rainfall for today is all messed up too.

What should happen is that you possibly get a one-off 'glitch' in Cumulus as it changes from using the rain total from the gauge as it its counter, to the annual total supplied by Easyweather. But it sounds as though EW is confused about the rain figures, so Cumulus will be also. Now that a new day has started, is it all behaving correctly?

[rwilliam99]

It appears to be. Although it doesn't look like we got any rain this morning, it is showing .05 of rain since midnight. Another station nearby shows .01 of precipitation, so maybe it is just a bit of an anomaly. I will keep an eye on it today. It showed over 12" of rain yesterday!!!

[rwilliam99]

Well, it is definitely better today, but the rainfall is inching up as the day progresses (at noon it was .08 inches), and it is sunny today, so I know we aren't getting ANY real rain. Is it due to the 1 min intervals somehow? I'm sure the station is working OK, I haven't touched the station at all since I did the switch.

Suggestions?

[steve]

Is it due to the 1 min intervals somehow?

I suppose it could be, as that interval isn't actually supported in Easyweather. Is the annual rainfall total actually increasing in the easyweather.dat file?

[rwilliam99]

Well, after looking at the file in depth, I think I know what is happening, but I don't know why. Records are being duplicated for some reason. I found 4 instances of one record, and that appears to be consistent. When I look at the size of the file, it is growing way too fast. The data records appear to be about 308 chars per record - 308 chars per record * 1440 records per day should be somewhere in the neighborhood of 443,520 bytes per day. The file currently sits at about 25MB in size. This is after running this way for only about 2 days (I started about 11:00am on Saturday). I would anticipate the file size to be around only 2MB in size.

The other thing I noticed is that the records are not always in order. They seem to appear in chunks together, but for example the file starts on 11-25, then goes back to 11-24, then back to 11-25, etc.

For the records for today, the "rain ticks" and "rain total" (fields 19 and 20) are not changing in the file.

This is the site I got the record layout from:

http://www.jim-easterbrook.me.uk/weather/ew/

Anyone have any idea what is happening? Why? How to correct it? What would happen if I deleted the EasyWeather.dat file and changed the interval back to 5? Would I lose any data? I would think not, since Cumulus is keeping its own tab on the data.

Thanks for the insight.

Rob

[rwilliam99]

Well, that experiment didn't go well.....Data stopped flowing tonight, so that only went two days without resetting. I tried to stop EasyWeather and restart it, but it didn't really seem to work. I ended up resetting the console and re-setting the data interval to 5 mins rather than 1. I'll see if that changes anything. I'm wondering if the shorter interval was messing something up.

If this hangs up in this fashion, I'm just going to go back to regular Cumulus, but that means that this must be some sort of hardware issue, which is a bummer......I need to go back and try and fix up my Cumulus data now.

[steve]

but that means that this must be some sort of hardware issue

I don't think that's in any doubt. Hardware or firmware, anyway. The question is, does Easyweather work in such a way that it is able to avoid the problem, and you appear to have answered that with a "no", which is what I think others have found.

[insx]

I haven't read through all 20 pages(!) but I came across the USB lockup problem for the first time last night. It seems that my rain gauge is constantly giving silly readings so this might have something to do with it. It's a shame I didn't get to record last night's downpour.

The weather station has been up about 2 months and I had to do a full reset of the station to get it talking again. I run Cumulus all day and shut it down at night. It is a WH1081 (from Maplin). I generally get a few 'sensor contact lost' messages each day, usually only 1 or 2 but sometimes a long run of them.

[nikkidev]

I have a Fine Offset (N96GY from Maplin) frequency 433.9MHz purchased this year. I'm not running continuously, probably at most for a couple of hours a day. It locked up for the first time 2 days a go, I never had this problem with my old Fine Offset. I disconnected the USB lead and removed one battery which I replaced after 5 s. I was expecting to lose data since the last upload of data the previous night but I don't appear to have done. I did however need to reset my relative pressure and I got a massive rain rate/rain spike at the time of reset which I think I've mainly managed to remove from the data files now.

The only thing I might have done differently the previous night, was to shut my computer down quickly without exiting from Cumulus properly.

(related but not really to do with this thread: The only record I can't seem to get rid of is the all time high hourly rain which was generated at the time of the reset. I can't see which file I need to edit to get rid of this and what to replace it with. Anyone got any ideas?)

[steve]

(related but not really to do with this thread: The only record I can't seem to get rid of is the all time high hourly rain which was generated at the time of the reset. I can't see which file I need to edit to get rid of this and what to replace it with. Anyone got any ideas?)

Use the built-in all-time record editor. The previous value will be in alltimelog.txt (format in the wiki if it isn't obvious looking at it). Alternatively, find a copy of alltime.ini from the latest backup folder created before the problem, and replace the file in the data folder with Cumulus stopped.