ws28xx initial installation and some issues
Bruce Petrie
Hi
I have been using a C86234 standalone for a while. Hooked it up to a PC dual bootable between windows and linux. Did not like the Heavy Weather Pro software since 1) required windows, 2) too restrictive functionality, 3) prefer linux.
Weewx really looks
close to what I need and want. Really appreciate the hard work
everyone has put into weewx. Maybe I can help debug the ws28xx.py....
Installed weewx 2.4.0 using the setup.py script, using the default sqlite DB. Pretty uneventful install once I understood to use ws28xx rather than fousb and set DEBUG = 1. Helps to RTFM a little more than I wanted to. <grin> Tailing weewxd output into a disk file for troubleshooting.
Ftp to web site works well most of the time, handles the web site down condition easily.
Observed issues:
1. 'barometer' = None, and 'pressure' = the manually set C86234 console relative pressure setting, 'altimeter' = 36.xxxx inHg which is wrong. Probably the important issue that I want to help fix. See next section for more in depth discussion about this.
2. 'rain' reading 10X actual values observed using a WMR100N/PCR800 rain gauge and also comparing with nearby CWOP and NWS sites. Adding “rain = rain / 10” in weewx.conf [StdCalibrate] fixed that. Not sure why this was an issue....
3. weewx: bogus date entries in /var/log/messages which apparently is known issue and not a big deal.
4. Tried the same pairing method used with Heavy Weather (press C86234 console Down arrow) which did not work, but would pair at the top of the hour reliably. Looked in ws28xx.py a little and discovered pressing set button method and pairing works perfectly now. Suggest putting that in an FAQ for ws28xx unless I missed it?
5. Streaming certain audio or video streams on other devices in house causes ftp timeouts to external weather web site. Have 20 Mbit/sec connection, fast router, can run multiple streams without issues on other devices including wireless devices. The PC running weewx has a gigabit wired connection path to the router so is not wireless related. Appears to be a too quick timeout on each ftped file. Have not done more in depth analysis yet. Any known issues/fix for that?
6. The web site is updated via one URL (upload.xxxx.xxxx), but has to be accessed via another URL (welbyweather.home.comcast.net). This delta messes up the links in the RSS feed. Could hardcode the correct URL in the RSS skin but not the best solution. Weewx.conf would be better. Again not a huge issue, might tackle this in the future myself.
Investigating Pressure issue in depth:
Commenting out barometer statement in [StdQC], 'barometer' and 'altimeter' now are 36.xxx rather than None. Remember that 'pressure' matches the C86234 console rel pressure (like 29.xxxx or 30.xxxx). My altitude is 5197 feet. So one thought was to just set the console rel pressure to the station pressure listed on a nearby NWS web site (like 24.6xx).
Unfortunately the C86234 console only supports setting pressures between 27.17 to 31.90 inHg (920 to 1080 hPa) so using station pressure is not possible. Look for DIA and Mus. Nat. Sci. pressure triads at http://www.crh.noaa.gov/bou/include/webpres.php?product=webpres.txt for typical values.
I have dug into how 'pressure' gets set in ws28xx.py which is coming from get_observation() which calls sp2ap() and sp2bp() to compute the 'altimeter' and 'barometer' pressures. After reading http://www.crh.noaa.gov/bou/awebphp/definitions_pressure.php it appears that the altimeter pressure should come directly from the C86234 console pressure setting rather than setting sp first. I think sp/bp should then be computed from ap.
Am I missing something obvious before I dig into mod'ing the above python def's to fix the pressure problems?
TIA!
Bruce Petrie
Background info about station
Celeron single core 1.8 GHz, RAM: 1533 MB,
32 bit fedora, python 2.7.8, using external web page via ftp
C86234 console FW: 332 which is setup with US/English units
TX59UN-1-IT Therm/Hydro sensor
TX63U-IT wind sensor 9 feet high
TX58UN-IT Rain Sensor
USBTRX-10 USB FOB (model # ERF-USB)
startup from messages file:
Aug 25 14:44:42 e weewx[5392]: wxengine: Initializing weewx version 2.4.0
Aug 25 14:44:42 e weewx[5392]: wxengine: Using Python 2.7 (r27:82500, Sep 16 2010, 18:03:06) #012[GCC 4.5.1 20100907 (Red Hat 4.5.1-3)]
Aug 25 14:44:42 e weewx[5392]: wxengine: Using configuration file /home/weewx/weewx.conf
Aug 25 14:44:42 e weewx[5392]: wxengine: Loading station type WS28xx (weewx.drivers.ws28xx)
Aug 25 14:44:42 e weewx[5392]: ws28xx: MainThread: driver version is 0.2
Aug 25 14:44:42 e weewx[5392]: ws28xx: MainThread: frequency is US
Aug 25 14:44:42 e weewx[5392]: ws28xx: MainThread: altitude is 1584.0456 meters
Aug 25 14:44:42 e weewx[5392]: ws28xx: MainThread: pressure offset is None
Aug 25 14:44:42 e weewx[5392]: wxengine: StdConvert target unit is 0x1
Aug 25 14:44:42 e weewx[5392]: wxengine: Using config file archive interval of 300
Aug 25 14:44:42 e weewx[5392]: wxengine: Record generation will be attempted in 'hardware'
Aug 25 14:44:42 e weewx[5392]: wxengine: Using archive database: archive_sqlite
Aug 25 14:44:42 e weewx[5392]: stats: stats database up to date.
Aug 25 14:44:42 e weewx[5392]: wxengine: Using stats database: stats_sqlite
Aug 25 14:44:42 e weewx[5392]: wxengine: Starting up weewx version 2.4.0
Aug 25 14:44:42 e weewx[5392]: wxengine: Starting main packet loop.
Aug 25 14:44:42 e weewx[5392]: ws28xx: MainThread: manufacturer: LA CROSSE TECHNOLOGY
Aug 25 14:44:42 e weewx[5392]: ws28xx: MainThread: product: Weather Direct Light Wireless
Aug 25 14:44:42 e weewx[5392]: ws28xx: MainThread: interface: 0
Aug 25 14:44:42 e weewx[5392]: ws28xx: MainThread: base frequency: 905000000
Aug 25 14:44:42 e weewx[5392]: ws28xx: MainThread: frequency correction: 97464 (17cb8)
Aug 25 14:44:42 e weewx[5392]: ws28xx: MainThread: adjusted frequency: 949058745 (38917cb9)
Aug 25 14:44:42 e weewx[5392]: ws28xx: MainThread: transceiver ID: 181 (b5)
Aug 25 14:44:42 e weewx[5392]: ws28xx: MainThread: transceiver serial: 010122081200181
Aug 25 14:53:00 e weewx[5392]: ws28xx: RFComm: console is paired to device with ID b5
Typical output from weewxd:
LOOP: 2013-08-25 18:40:40 MDT (1377477640) {'barometer': 36.087779100941745, 'windchill': 82.4, 'dewpoint': 54.21343792841772, 'outTemp': 82.4, 'outHumidity': 38, 'rainRate': 0.0, 'signal': 100, 'heatindex': 82.4, 'rain': 0.0, 'dateTime': 1377477640, 'windDir': 270, 'pressure': 30.150620200826936, 'windSpeed': 1.342161810275146, 'inHumidity': 50, 'inTemp': 67.28, 'windGust': 2.684323620550292, 'usUnits': 1, 'altimeter': 36.243650927765124, 'battery': 0, 'windGustDir': 270}
REC: 2013-08-25 18:40:00 MDT (1377477600) {'interval': 5, 'barometer': 36.086271854913186, 'windchill': 82.52599999999998, 'dewpoint': 53.959813840344204, 'outTemp': 82.52599999999998, 'outHumidity': 37.5, 'rainRate': 0.0, 'signal': 100.0, 'heatindex': 82.52599999999998, 'rain': 0.0, 'dateTime': 1377477600, 'windDir': 290.11439399112334, 'pressure': 30.150620200826943, 'windSpeed': 0.8276664496696734, 'inHumidity': 50.0, 'inTemp': 67.244, 'battery': 0.0, 'windGust': 2.684323620550292, 'altimeter': 36.24365092776513, 'usUnits': 1, 'windGustDir': 270}
LOOP: 2013-08-25 18:41:15 MDT (1377477675) {'barometer': 36.087779100941745, 'windchill': 82.4, 'dewpoint': 54.21343792841772, 'outTemp': 82.4, 'outHumidity': 38, 'rainRate': 0.0, 'signal': 100, 'heatindex': 82.4, 'rain': 0.0, 'dateTime': 1377477675, 'windDir': 135, 'pressure': 30.150620200826936, 'windSpeed': 0.8947745401834307, 'inHumidity': 50, 'inTemp': 67.1, 'windGust': 1.342161810275146, 'usUnits': 1, 'altimeter': 36.243650927765124, 'battery': 0, 'windGustDir': 135}
mwall
thank you for the detailed report. this is very helpful!
On Sunday, August 25, 2013 10:39:04 PM UTC-4, Bruce Petrie wrote:
2. 'rain' reading 10X actual values observed using a WMR100N/PCR800 rain gauge and also comparing with nearby CWOP and NWS sites. Adding “rain = rain / 10” in weewx.conf [StdCalibrate] fixed that. Not sure why this was an issue....
this is fixed in the ws28xx driver as of r1300
4. Tried the same pairing method used with Heavy Weather (press C86234 console Down arrow) which did not work, but would pair at the top of the hour reliably. Looked in ws28xx.py a little and discovered pressing set button method and pairing works perfectly now. Suggest putting that in an FAQ for ws28xx unless I missed it?
i'll add a section to the weewx user guide once the ws28xx driver stabilizes. until then, anyone using ws28xx should read the comments at the beginning of the ws28xx.py file.
Investigating Pressure issue in depth:
Commenting out barometer statement in [StdQC], 'barometer' and 'altimeter' now are 36.xxx rather than None. Remember that 'pressure' matches the C86234 console rel pressure (like 29.xxxx or 30.xxxx). My altitude is 5197 feet. So one thought was to just set the console rel pressure to the station pressure listed on a nearby NWS web site (like 24.6xx).
Unfortunately the C86234 console only supports setting pressures between 27.17 to 31.90 inHg (920 to 1080 hPa) so using station pressure is not possible. Look for DIA and Mus. Nat. Sci. pressure triads at http://www.crh.noaa.gov/bou/include/webpres.php?product=webpres.txt for typical values.
I have dug into how 'pressure' gets set in ws28xx.py which is coming from get_observation() which calls sp2ap() and sp2bp() to compute the 'altimeter' and 'barometer' pressures. After reading http://www.crh.noaa.gov/bou/awebphp/definitions_pressure.php it appears that the altimeter pressure should come directly from the C86234 console pressure setting rather than setting sp first. I think sp/bp should then be computed from ap.
Am I missing something obvious before I dig into mod'ing the above python def's to fix the pressure problems?
just to be sure we are on the same page... there are 3 pressures:
pressure - station pressure, i.e., the pressure reported by the station's sensor
barometer - sea level pressure calculated (approximated) using station pressure and elevation
altimeter - sea level pressure calculated (approximated) using station pressure, elevation, and temperature
the ws28xx driver assumes that the value we get from the c86234 console is simply station pressure.
it sounds like you just need to calibrate the station pressure. you can calibrate by specifying pressure_offset in weewx.conf:
[W28xx]
# pressure calibration offset, in hPa (millibars)
pressure_offset = 2.4
get a valid pressure for your location, x, from a nearby NWS site. the pressure_offset will be the difference between what you see on the station console and x.
eventually the wee_config_ws28xx utility will do this calibration for you...
m
Thomas Keffer
pressure - station pressure, i.e., the pressure reported by the station's sensor
barometer - sea level pressure calculated (approximated) using station pressure, elevation, and temperature
The last two were switched. Or, were you operating under a completely different set of assumptions?
-tk
mwall
you are correct. the brain got ahead of the fingers (or vice versa).
Luc
# pressure calibration offset, in hPa (millibars)
Bruce Petrie
Tom defined the three pressures as:
(SP) pressure - station pressure, i.e., the pressure reported by the station's sensor
(AP) altimeter - sea level pressure calculated (approximated) using station pressure and elevation
(BP) barometer - sea level pressure calculated (approximated) using station pressure, elevation, and temperature
Based on what I have read, these definitions are dead on with one twist. The barometric pressure is referred to MEAN SEA LEVEL PRESSURE by NOAA. http://www.crh.noaa.gov/bou/awebphp/definitions_pressure.php That web page states that the MSL pressure should be avoided since at high altitudes, temperature variation can render this reading invalid. Both NOAA web page (second paragraph) and CWOP_Guide.pdf page 23 recommend using the Altimeter pressure as the console front panel pressure. The NOAA page states the Altimeter pressure (AP) is most commonly used pressure on TV and radio. So AP seems to be what I need to dial into the C86234 LCD display to conveniently compare with TV, etc.
Based on what I have been seeing with weewx running, is that the C86234 LCD pressure reading ends up as 'pressure' in weewx. Reading ws28xx.py seems to confirm that. 'pressure' is being referred to as station pressure (SP) in the comments.
Here is my confusion, since there is no Altitude setting ever entered into the C86234 console, this pressure being read in can not be the station pressure if the LCD is set to AP. This explains why the weewx AP and BP readings are in the 36 inHg range since the altitude is factored in twice. I can't dial in the 'station pressure' quoted on the NOAA/NWS web site since the C86234 will not allow a pressure in the 24.xx range. It only allows 27.17 or greater.
Based on several comments in this thread, I think there is a misconception that the value read in from the C86234 is station pressure. It is in fact the value you dial in on the LCD display which can not ever be below 27.17.
Illustrate with some current data http://www.crh.noaa.gov/bou/include/webpres.php?product=webpres.txt
DIA has the following:
24.692 station pressure
30.150 altimeter pressure
29.884 Mean sea level pressure
My C86234 LCD says 30.11 because I used a nearby approved CWOP station as a reference since DIA is over 20 miles from me. Here is what weewx read:
'pressure': 30.112226816302424
'altimeter': 36.19907579134678
'barometer': 35.967473658858054
So do you see why I am saying if the C86234 LCD is set to AP, the ws28xx driver should be calling the reading AP and then compute SP and MSL/BP? I can not set the LCD to station pressure since it is below 27.17 and MSL/BP should not be used at my altitude....?
mwall
please confirm the following:
1) calibrate your station by entering a verified mean sea level pressure into your station as described on page 13 of the C86234 manual ("relative pressure reference value")
2) calibrate weewx by entering a pressure_offset in weewx.conf
[WS28xx]
pressure_offset = 182.255
3) after doing (2), what are the values for the following:
lcd display
'pressure' in a weewx LOOP or REC packet
'altimeter' in a weewx LOOP or REC packet
'barometer' in a weewx LOOP or REC packet
the c++ api that we get from decompiling the HeavyWeather code refers to the value from the station's pressure sensor as 'PressureRelative', which is the same thing as 'station pressure', i.e., the sensor pressure. if the api is lying to us and 'PressureRelative' is actually a derived pressure such as altimeter or barometer, then we'll have to make some changes.
as far as i know, the C86234 has no idea what altitude is, so i'm not sure how it could calculate a sea level pressure. it probably simply does a linear mapping of the sensor's pressure range to the typical sea level pressure range (27.17 inHg to 31.90 inHg).
m
Bruce Petrie
Aug 26 19:14:14 e kernel: [341121.820772] usb 3-2: usbfs: process 5396 (weewxd) did not claim inte
rface 0 before use
Aug 26 19:14:19 e weewx[6661]: wxengine: Initializing weewx version 2.4.0
Aug 26 19:14:19 e weewx[6661]: wxengine: Using Python 2.7 (r27:82500, Sep 16 2010, 18:03:06) #012[
GCC 4.5.1 20100907 (Red Hat 4.5.1-3)]
Aug 26 19:14:19 e weewx[6661]: wxengine: Loading station type WS28xx (weewx.drivers.ws28xx)
Aug 26 19:14:19 e weewx[6661]: ws28xx: MainThread: driver version is 0.2
Aug 26 19:14:19 e weewx[6661]: ws28xx: MainThread: frequency is US
Aug 26 19:14:19 e weewx[6661]: ws28xx: MainThread: altitude is 1584.0456 meters
Aug 26 19:14:19 e weewx[6661]: ws28xx: MainThread: pressure offset is 182.255
Aug 26 19:14:19 e weewx[6661]: wxengine: StdConvert target unit is 0x1
Aug 26 19:14:19 e weewx[6661]: wxengine: Using config file archive interval of 300
Aug 26 19:14:19 e weewx[6661]: wxengine: Record generation will be attempted in 'hardware'
Aug 26 19:14:19 e weewx[6661]: wxengine: Using archive database: archive_sqlite
Aug 26 19:14:19 e weewx[6661]: stats: stats database up to date.
Aug 26 19:14:19 e weewx[6661]: wxengine: Using stats database: stats_sqlite
Aug 26 19:14:19 e weewx[6661]: wxengine: Starting up weewx version 2.4.0
Aug 26 19:14:19 e weewx[6661]: wxengine: Starting main packet loop.
Aug 26 19:14:19 e weewx[6661]: ws28xx: MainThread: manufacturer: LA CROSSE TECHNOLOGY
Aug 26 19:14:19 e weewx[6661]: ws28xx: MainThread: product: Weather Direct Light Wireless
Aug 26 19:14:19 e weewx[6661]: ws28xx: MainThread: interface: 0
Aug 26 19:14:19 e weewx[6661]: ws28xx: MainThread: base frequency: 905000000
Aug 26 19:14:19 e weewx[6661]: ws28xx: MainThread: frequency correction: 97464 (17cb8)
Aug 26 19:14:19 e weewx[6661]: ws28xx: MainThread: adjusted frequency: 949058745 (38917cb9)
Aug 26 19:14:19 e weewx[6661]: ws28xx: MainThread: transceiver ID: 181 (b5)
Aug 26 19:14:19 e weewx[6661]: ws28xx: MainThread: transceiver serial: 010122081200181
LOOP: 2013-08-26 19:57:56 MDT (1377568676) {'barometer': 42.30126542531797, 'windchill': 79.34, 'dewpoint': 55.50875301039149, 'outTemp': 79.34, 'outHumidity': 44, 'rainRate': 0.0, 'signal': 100, 'heatindex': 79.34, 'rain': 0.0, 'dateTime': 1377568676, 'windDir': 225, 'pressure': 29.923213230950978, 'windSpeed': 0.8947745401834307, 'inHumidity': 49, 'inTemp': 67.28, 'windGust': 1.342161810275146, 'usUnits': 1, 'altimeter': 42.21540177633069, 'battery': 0, 'windGustDir': 225}
Obviously, the pressure_offset just makes the situation worse.
Key observation is every time I have compared the LCD to the 'pressure' value over the past few days, it has been the same value within the capability of the LCD display. I don't know yet if the LCD value is rounding or truncating the 'pressure' value to two decimal places. I agree there is no way that the C86234 would know about altitude since it is never put into the LCD interface during setup.
I will try to tabulate results for a while but I am reasonably certain the reading that the driver calls 'pressure' is rounded or truncated to be displayed on the LCD. No magical transformation occurs.
Here is the 20:13 results which 'pressure' is unchanged:
LCD 29.92
LOOP: 2013-08-26 20:15:59 MDT (1377569759) {'barometer': 42.30381995953686, 'windchill': 79.16, 'dewpoint': 55.34635119757276, 'outTemp': 79.16, 'outHumidity': 44, 'rainRate': 0.0, 'signal': 100, 'heatindex': 79.16, 'rain': 0.0, 'dateTime': 1377569759, 'windDir': 0, 'pressure': 29.923213230950978, 'windSpeed': 0.0, 'inHumidity': 49, 'inTemp': 67.1, 'windGust': 1.342161810275146, 'usUnits': 1, 'altimeter': 42.21540177633069, 'battery': 0, 'windGustDir': 0}
cheers!
mwall
Obviously, the pressure_offset just makes the situation worse.
sorry, i got the sign wrong. use this for pressure_offset:
pressure_offset = -182.255
that will bring down the sensor pressure, which should result in sane values for barometer and altimeter.
Bruce Petrie
LCD 29.93
LOOP: 2013-08-26 20:45:40 MDT (1377571540) {'barometer': 29.422561359260378, 'windchill': 78.44, 'dewpoint': 54.69668224013664, 'outTemp': 78.44, 'outHumidity': 44, 'rainRate': 0.0, 'signal': 100, 'heatindex': 78.44, 'rain': 0.0, 'dateTime': 1377571540, 'windDir': 247, 'pressure': 29.932073242764325, 'windSpeed': 1.1184681752292884, 'inHumidity': 49, 'inTemp': 67.1, 'windGust': 1.342161810275146, 'usUnits': 1, 'altimeter': 29.722490117048586, 'battery': 0, 'windGustDir': 247}
REC: 2013-08-26 20:45:00 MDT (1377571500) {'interval': 5, 'barometer': 29.422561359260378, 'windchill': 78.44000000000001, 'dewpoint': 54.69668224013664, 'outTemp': 78.44000000000001, 'outHumidity': 44.0, 'rainRate': 0.0, 'signal': 99.5, 'heatindex': 78.44000000000001, 'rain': 0.0, 'dateTime': 1377571500, 'windDir': 312.11044936652735, 'pressure': 29.932073242764325, 'windSpeed': 0.4026485430825438, 'inHumidity': 49.0, 'inTemp': 67.01000000000002, 'battery': 0.0, 'windGust': 1.342161810275146, 'altimeter': 29.722490117048586, 'usUnits': 1, 'windGustDir': 315}
I will let this run overnight and try to get more LCD readings tomorrow. We are supposed to have several days of 90 degree weather with some afternoon shower potential.
Bruce Petrie
Here is the weewxd output for the test period and a .pdf comparing my station MSL (aka pressure) readings to DIA MSL. I had to use DIA rather than my nearby preferred CWOP station due to route peering problems with 6 ISP's here in Denver. As far as I can tell the front panel LCD agreed with my station 'pressure' rounded to two decimal positions.
We had a rain front moving through the area starting at 21:57 MDT (9:57 PM) producing 0.16". DIA would have received it about 30 minutes later.
I sent an email to LC support asking what pressure they recommended setting the LCD pressure too. This might take several days for a response from them since they are famously slow.
cheers!
Bruce Petrie
Bruce Petrie
I setup another copy of weewx connected to my WRM100NR minus its wind sensor which met a horrible death at the hands of a six year old a few years ago. The C86234 and WRM100NR Temp and Hum sensors are physically side by side.
I set my altitude into the WRM100NR and let it stabilize. I then set the C86234 to the same value front panel value WRM100NR shows..
It appears that the WRM100NR front panel pressure is an Altimeter pressure by referencing the following and other sites around me 9:45AM MDT:
WRM100NR front panel: 30.18 inHg
NWS DIA altimeter: 30.26 inHg
NWS DIA altimeter: 1024.722 mb
NWS DIA MSL: 1015.800 mb
NWS DIA MSL: 29.997 inHg
CWOP CW9688: 1025.10 mb
Last night, I took the 'pressure' from the WRM100NR DB and subtracted the C86234 "pressure" (aka long version of C86234 front panel reading) then multiplied by 33.6389 which produced -175.983306143. Took that value and put into 'pressure_offset' on the C86234 station and restarted before epoch 1377833400. The WRM100NR instance of weewx has a pressure_offset = 0 statement. As of now, the C86234 front panel is within 0.02 inHg of the WRM100NR front panel.
Have attached from WRM100NR and C86234 DB query "select dateTime,pressure,altimeter,barometer,inTemp,outTemp from archive where dateTime > 1377833400;" Looks like the WRM100NR 'barometer' is not matching the NWS DIA MSL. The altimeter values are tracking pretty close.
Any advice on where to go from here? Thank you!
Bruce Petrie