Good to see that you are still doing stuff with the TE923con

If you don’t mind, I would like to borrow your idea for the ‘battery-status’ mail, because I am rewriting stuff in python. And it’s actually cool to display the status on the pages of TE923-Frontend too.

You also fixed some stuff I could not figure out for the UV, jumping from 0.9 to 10.0.

And, I will, if I remember, tell the story why I did the rmmod, on the blog-page

Much Regards!

Jacco

well done for all your work.

I have one of these stations, and I modified a number of other peoples code to run a data collector/database storer on OSX, using java/c++/jndi/jna. Yes got all the HID problems as well, but under OSX its easier (no kext required)

I have embedded your UV fix into my code, my other query is wind gusts, they seem a bit low, I read the data in using Hideki’s software under windows to verify it, but still not totally convinced.

A word of caution regarding the UV detector, they are not waterproof!! and water ingress will either give stupid reading or cause it to fail.

That said, on the old server I found that occasionally the usbhid module would rebind to the weather station device, so I took to adding a script to root’s crontab to check for this and unbind if necessary, and in that script the device id is hardcoded. Checking the ‘man udev’ page, maybe $id is not the value to use. I will try to figure it out at some point, but if you do, please post back.

I am running Debian (Lenny).

Nico

Thanks for your info and your PHP scripts. I got my Ventus W831 working properly now.

It is hosted at my house via ADSL. If the usage is going to be high I will have to host it somewhere else.

Nico

First, here’s without the mouse:

[hal:/home/mrbalky] ll /sys/bus/usb/drivers/usbhid
total 0
–w——- 1 root root 4096 2010-08-23 18:51 unbind
–w——- 1 root root 4096 2010-08-25 18:07 uevent
–w——- 1 root root 4096 2010-08-25 18:07 new_id
lrwxrwxrwx 1 root root 0 2010-08-25 18:07 module -> ../../../../module/usbhid/
–w——- 1 root root 4096 2010-08-25 18:07 bind
[hal:/home/mrbalky] sudo bin/te923con -i-
1282781540:23.50:47:27.10:31:-:-:-:-:-:-:-:-:1011.1:3.4:4:0:10:1.1:0.6:32.8:466

And here’s with:
[hal:/home/mrbalky] ll /sys/bus/usb/drivers/usbhid
total 0
–w——- 1 root root 4096 2010-08-23 18:51 unbind
–w——- 1 root root 4096 2010-08-25 18:07 uevent
–w——- 1 root root 4096 2010-08-25 18:07 new_id
lrwxrwxrwx 1 root root 0 2010-08-25 18:07 module -> ../../../../module/usbhid/
–w——- 1 root root 4096 2010-08-25 18:07 bind
lrwxrwxrwx 1 root root 0 2010-08-25 18:14 1-2:1.0 -> ../../../../devices/pci0000:00/0000:00:1d.0/usb1/1-2/1-2:1.0/
[hal:/home/mrbalky] sudo bin/te923con -i-
1282781688:23.50:47:27.00:30:-:-:-:-:-:-:-:-:1011.1:3.4:4:0:9:1.1:1.0:32.7:466

The RC is laid into bytes 30 and 31 little-endian, so I don’t think it makes sense for byte 29 to be part of the counter on either end.

With my RC at 466, I have obviously rolled over the capacity of a single byte, but have quite a way to go to roll over the capacity of two bytes. Here are the current values in the RC data:

[DEBUG] RAINC BUF[29]=e0 BUF[30]=d2 BUF[31]=01

A counter of 466 is a sane value for my station, and adding e0 either as a high-order or low-order byte of the RC value just doesn’t seem right. I seem to recall seeing that byte flip between e0 and 00, but can’t swear to it.

So given my interpretation of the source code and values I see now, I think byte 29 is not important to the RC value, though I can’t guess what it does mean.

I’ve had no issues with the reliability of the rain and temp/humidity sensors. They’re closer and only have one wall between them and the controller. When I was first setting the station up I did run across this thread about modifying the antenna: http://www.weather-watch.com/smf/index.php?topic=29813.0

That’s interesting about incomplete data with kbd/mouse. Mine is a headless server, so I haven’t got any experience with that. I should give it a try sometime (I also am using Debian).

By missing the last byte, do you mean the high-order or low-order byte? The rain data makes sense to me so far, but in Colorado we don’t get all that much rain (my RC is only up to 466).

When setting the station up, and messing around with my PHP code, I counted clicks of the counter while pouring a glass of water through the rain gauge. The counter data matched my counts, and I used the rain display to come up with the counter -> accumulation scale. The scale was very close to the published resolution of the station, so I thought I was OK as far as the RC goes.

In terms of USB on Linux (Debian in my case), I find that I only get incomplete data if there is a HID device attached. Unplug the keyboard/mouse and I get the full data set, though running as root. Not sure why that’s the case.

Does your rain data make sense to you? I’m currently looking at what may be a bug in the 0.4 code that misses the last byte off the values read for rc. Good luck!

I’ve been thinking of upgrading to one of these fancy new weather stations, but have a sentimental attachment to the boxes with my code running inside.

There are many ways to set this up. In my case, I have a server on my home network that polls my TiVos for the now playing list every 1/2 hour (since that’s when programs tend to start).

I am using a Linux server that is always on so updates can always happen. The update is scheduled through cron.

The server runs a shell script to fetch the now playing list with a command like this:
curl --anyauth --insecure --user tivo:**TiVo MAK** --output tivo_nowplaying.xml "https://**TiVo IP address**/TiVoConnect?Command=QueryContainer&Container=%2FNowPlaying&Recurse=Yes&ItemCount=50"

Replace **TiVo MAK** and **TiVo IP address** with the values for your tivo. Both of these should be available through the TiVo menus, but I don’t remember exactly where. You should be able to find out from the TiVo support site or The Google; just search “tivo mak” and “tivo ip address”.

The shell script then runs:
php tweettivo.php tivotwitterid tivotwitterpw
TweetTivo is here: http://www.mrbalky.com/wp-content/uploads/tweettivophp.txt

You will need the curl application (or maybe wget) to poll your TiVo. PHP is required to run the actual code. You will also need class_tivo_xml.php to parse the specific TiVo XML, which I found over on the tivo community site long long ago: http://archive2.tivocommunity.com/tivo-vb/printthread.php?t=218365&pp=50

It is definitely possible to do schedule this on a windows machine. Binaries for curl and php are available, and scheduling would be done via the scheduled tasks section of control panel.

It’s also worth noting that this is an incredibly quick and dirty application, with barely any testing. It grew out of the script I was already using for the “now playing” page on this site, so it really could all be wrapped up into the same php script. If you know how to use curl from within php, it’d probably be better to skip the exec.

Sorry; that probably doesn’t help much. I just threw this together in a few minutes and I’m afraid I don’t have much time for this script anymore.