The Oceanic Remote Chemical Analyzer (ORCA) is an autonomous moored profiling system providing real-time data streams of water and atmospheric conditions. It consists of a profiling underwater sensor package with a variety of chemical and optical sensors, and a surface mounted weather station, solar power system, winch, and custom computer and software package equipped with WIFI/cellular communication.
The ORCA buoy controller Daemon (orcaD) is a UNIX server deamon and utilities which autonomously operate the mechanical systems of the buoy.
The design of orcaD grew out of experience with the initial ORCA control program "WinchSurf" designed by John Dunn and Wendi Ruef ( et al ). WinchSurf was designed to run on the TattleTale low-power embedded processor family in a DOS-like environment.
The redesign of the ORCA management software started in earnest in the Spring of 2005 and was initially directed by the following design goals:
1. Develop a data logging and device management
application which will run in a non-real-time
multitasking environment...namely linux.
2. Ease the management burden of the system and
increase it's robustness.
3. Allow for easy reconfiguration of intruments.
Dependencies: Hardware: BitsyX or BitsyXb XScale CPU Keyspan USB 4-port Serial Port Extender Custom I/O circuitry -- See Specs
Installing From A Binary Software Distribution:
1. Extract the contents of the compressed archive distribution
into the program directory.
Make sure the directory doesn't already exist and if
it does move it aside as a backup:
% file /usr/local/orcaD
/usr/local/orcaD: ERROR: cannot open `/usr/local/orcaD'
(No such file or directory)
or if it exists move it aside
% file /usr/local/orcaD
/usr/local/orcaD: directory
% mv /usr/local/orcaD /usr/local/orcaD.old
Extract the software in the /usr/local directory:
% cp orcaD-#.#.tar.gz /usr/local
% cd /usr/local
% tar zxvf orcaD-#.#.tar.gz
2. Check the distribution:
% cd orcaD
% ls
data/ iotest* logs/ orcactrl* orcad* orcad.cfg.tmpl readme
utils/
Installing From A Source Software Distribution:
1. Extract the contents of the compressed archive distribution.
( usually in a source directory seperate from the installation
directory ). Lets say /root/sources/orcaD for this example.
% cp orcaD-src-#.#.tar.gz /root/sources
% cd /root/sources
% tar zxvf orcaD-src-#.#.tar.gz
2. Check the distribution:
% cd orcaD
% ls
general.h serial.h util.c LICENSE.txt
hydro.c orcactrl.c util.h Makefile
hydro.h orcad.c sigtest.c version.c
README iotest.c orcad.cfg term.c
bitsyxio.c log.c orcad.h term.h
weather.c bitsyxio.h log.h parser.c
weather.h crc.c meterwheel.c parser.h
crc.h meterwheel.h profile.c winch.c
ctd.c mytest.log profile.h timer.c
winch.h ctd.h notebook serial.c
timer.h
3. Remove older object files if they exist:
% make distclean
4. Build the distribution:
% make
5. Install the program ( note this will overwrite existing versions )
% make install
You are ready to use the software. See the PROCEDURES section for information on how to start the daemon up automatically at boot time.
The orcactrl program is used to manually operate the buoy systems when the automated program ( orcaD ) is NOT running. It is very important not to run both simultaneously.
Orcactrl may be run as a command driven shell or as a single use command. Here is a list of the current command line options:
usage: orcactrl [-d debuglevel] [-c config file] [command]
The main ORCA control utility. NOTE: This utility should only be used when the orcaD process has been stopped!
General options: -d (debuglevel) - debug level -c (config file) - hardware config file
Commands: view config | voltage | weather pressure | meterwheel | all - Retreive buoy state. set minDepth (meters) | parkingDepth (meters) | maxDepth (meters) | debugLevel (number) - Set various parameters. monitor voltage|pressure|meterwheel|all - Watch buoy state. move up (target meters) - Move the package up. down (target meters) - Move the package down. up-discretely (target meters) (sample1 meters) (sample2 meters) ... - Move the package up discretely. profile (mission name) - Run through a profile. download ctd|weather (filename) - Download data to a file.
See PROCEDURES section for detailed descriptions of its use.
The orcad program is responsible for running the buoy in un-attended mode. It has been written as a unix "daemon". A daemon is a program which is often started when the operating system boots and remains running in the background ( i.e. without user interaction through a terminal ) until the system is shutdown.
Orcad is configured through the use of a config file ( defaults to /usr/local/orcaD/orcad.cfg ) and is started by the operating system process management program "init" or systemd (see installation section for details). Init/systemd has the ability to not only start a program when the system is booted, but to also restart it should it automatically should it crash at any time. It is important that orcad is started in this fashion as it offers some amount of insurance that the winch will be shut off should the program crash during a profile.
Orcad accepts several parameters:
orcad [-v][-d #][-c conf_file][-f]
Where: -c conf_file - configuration file ( default: /usr/local/orcaD/orcad.cfg ) -d # - set debug level and flags ( default: 4 ) -v - show version info -f - stay in the foreground, do not run as a daemon -h - prints this information
For testing purposes you can run orcad in the foreground ( allowing for terminal control of the process ) by using the -f option.
Orcad has the following general program flow:
1. Startup and become a UNIX daemon
2. Initialize bitsyX hardware ( shuts off the winch! )
3. Open up the log file ( defaults to /usr/local/orcaD/logs/orcad.log )
4. Open up & set the PID file ( defaults to /var/run/orcad.pid )
** This makes sure that you don't run more than
one orcad at one time **
5. Read the config file ( defaults to /usr/local/orcaD/orcad.cfg )
6. Go into main loop:
A. Sleep for a minute
B. Check to see if there are profiles to run
C. Run profiles
D. Check to see if we should log weather archive
E. Check to see if we should log weather status
F. Repeat
7. Exit the main loop only if a "kill -QUIT" signal is received
8. Reset the hardware
9. Close all the files
10. exit the program.
Setting up Orcad To Be Run By The Init Daemon:
This procedure will configure the operating system to run
orcaD when the system is booted. It will also restart
the program automatically should it crash or be killed
by a user ( See procedure for stopping orcaD for details
on how to shut it down without the init daemon restarting it ).
1. Edit the init daemon's configuration file /etc/inittab and
add orcad to it's list of managed processes:
% nano /etc/inittab
# Run orcad in runlevel 3
od:3:respawn:/usr/local/orcaD/orcad -d 8 -f
You could also create two copies of the inittab file
in etc...say /etc/inittab_no_orcad and /etc/inittab_with_orcad
( with and without the orcad line above ) and switch between them
by simply using a copy command:
% cp /etc/inittab_with_orcad /etc/inittab
and
% cp /etc/inittab_no_orcad /etc/inittab
2. Tell the init daemon to re-read the /etc/inittab file:
% /sbin/telinit q
Upon re-reading the configuration file the init daemon
detects that it should start orcad.
3. Check that orcad has been started correctly:
% ps -e | grep orcad
#### ? 00:00:00 orcad
where #### is the process identifier.
If orcad didn't startup due to some configuration error the
init daemon will continue trying to restart it. It does this
for up to 10 times before figuring out that something is wrong.
When init figures out that there is a problem it discontinues
attempts to restart the process and writes a line to the
/var/log/messages file indicating the problems as "respawning
orcad too rapidly". You will have to look at the log file
/usr/local/orcaD/logs/orcad.log for more details on why
orcad is not starting up.
Once you have corected the problem go back to step 1 and
start again.
Stopping Orcad:
1. First check that orcaD is running:
% ps -e | grep orcad
#### ? 00:00:00 orcad
Where #### is the process identifier. If orcad is *not*
running this will return to a command prompt and not
return anything.
2. If you are running orcaD using the init daemon you will
have to disable the "respawn" feature first:
% nano /etc/inittab
Find the line which starts orcad. Might look
something like this:
# Run orcad in runlevel 3
od:3:respawn:/usr/local/orcaD/orcad -d 8 -f
And comment out the command using a # character:
# Run orcad in runlevel 3
#od:3:respawn:/usr/local/orcaD/orcad -d 8 -f
3. Tell the init daemon to re-read the /etc/inittab file:
% /sbin/telinit q
4. Check that a profile isn't currently running:
% tail -f /usr/local/orcaD/logs/orcad
This will continuously display the last few lines of
orcad's log file. If you see the last message looks
something like:
Apr 01 2010 12:14:41 orcad: runJobs(): Completed mission: Deep_Cast_Direct
Than the buoy is *not* running a profile and it is
safe to stop the daemon.
% ps -e | grep orcad
10 ? 00:00:00 orcad
You can kill a process by running the kill command. For
example, using the above example orcad is running as
process #10 ( the second column in the output ). The
kill command would be:
% kill -HUP 10
%
It's always a good idea to check that it worked by
looking at the process list again.
% ps -e | grep orcad
%
Weatherd ( if used ) is activated in a similar fashion. For the
convenience of the administrator we have added four scripts to the
util/ directory in the orcaD distribution which automate the a few
of the steps listed above. Once the inittab is setup you can use
the scripts:
util/startOrcad [ Uncomments initab line and restarts init ]
util/startWeatherd
and
util/stopOrcad [ Comments inittab line and kills process ]
util/stopWeatherd
Servicing Procedure:
-
Connect computer cable to the computer
-
Login and shutdown the wifi amplifier ( for amplified WIFI buoys ) % iotest WIFI OFF
-
Check orcad's log file: % more /usr/local/orcaD/logs/orcad
Check that it's not currently profiling...
-
Check that it's not about to profile ( see setting profile schedule ).
- Look at schedule in /usr/local/orcaD/orcad.cfg
- Check current status in the logs "cat /usr/local/orcaD/logs/orcad"
-
Stop orcad ( see procedure above )
-
Run orcactrl if buoy operations are necessary
-
Open up the computer
- Leave the communications cable attached.
-
halt the proccessor ( this will stop orcad automatically ) % halt NOTE: Once you halt the computer you have approx 60 seconds before the hardware ( watchdog timer ) detects the computer is down and restarts the system. I usually wait until I have the computer open to halt it. I then unplug the power immediately after the halt operation completes.
-
Replace CF card if necessary.
-
Reboot ( simply plug power back in and monitor the reboot ).
-
Double check things like:
- You can log in
- You can see orcaD running
- You have checked the startup lines in the log file for errors.
The orcad package does not require any special administration procedures above and beyond what a typical UNIX system does. With that in mind we mention here a few key UNIX administration functions and their impact on the orcad package:
-
Log Rotation Typically UNIX systems store log files in the /var/log of /var/adm directories. Orcad currently stores its logs in "/usr/local/orcaD/logs/" ( this will become a configurable option in future versions). UNIX log files are typically rotated so that sufficient log history is maintained with less risk of filling up the filesystem. Orcad's logs are not currently rotated and need to be removed periodically to avoid filling up the filesystem.
See: UNIX logrotate software, UNIX "df" command
-
RAM & Disk Utilization: It's a good idea to periodically check that the system has adequate RAM and disk space. Use the UNIX "df" command to check disk space and "top" to check RAM utilization.
If RAM is low check the process list for abnormally high RAM utilization by a single process. If disk space is low check that the logs or data directories contain old data which can be removed.
OrcaD creates several types of data files during normal operations. All data files are stored in the programs data subdirectory. This is typically set to /usr/local/orcad/data. Here is a typical example of the files you will find in this directory:
/usr/local/orcad/data/ castIndex.txt weather-status.dat 200512/ ORCA2_CAST0001.HEX ORCA2_CAST0002.HEX .. ORCA2_MET200512021119.MET ORCA2_MET200512021758.MET .. 200601/ ORCA2_CAST0100.HEX ORCA2_CAST0101.HEX .. ORCA2_MET200601051119.MET ORCA2_MET200601051758.MET .. [YYYYMM]/ [Buoy Prefix]_CAST[####].HEX [Buoy Prefix]_MET[YYYYMMDDHHMM].MET
The cast index file ( default data/castIndex.txt ) contains the integer number of the last cast index used by the orcad program. This file is used to save the state of the index between invocations of the program.
The weather status file ( default data/weather-status.dat ) contains the last Davis Weather Station LOOP record. This file is available for download by a weather monitoring package or webservice and is updated periodically ( default 10 minutes ).
MET Files:
Davis Weather Stations:
mm/dd/yy hh:mm:00\tTAB_DELIMITED_DATA
mm = month
dd = day
yy = year
hh = hour
mm = minute
TAB_DELIMITED_DATA = barometer ( davis internal value / 1000 )
outside humidity
average wind speed
high wind speed
prevailing wind direction
rainfall ( davis internal value / 100 )
inside temp ( " " " / 10 )
outside temp ( " " " / 10 )
high outside temp ( " " " / 10 )
low outside temp ( " " " / 10 )
high solar radiation
solar radiation
AIRMAR Stations:
mm/dd/yy hh:mm:ss\tTAB_DELIMITED_DATA
mm = month
dd = day
yy = year
hh = hour
mm = minute
ss = second
TAB_DELIMITED_DATA = barometer [ inches Hg ]
humidity [ percent ]
temperature [ degrees celsius ]
dewpoint [ degrees celsius ]
wind direction - centerline [ degrees True ]
wind speed [ knots ]
max wind direction - centerline [ degrees True ]
max wind speed [ knots ]
solar radiation [ Watts/m^s ]
avg solar radiation [ Watts/m^s ]
latitude [ degrees ]
longitude [degrees ]
DORCA 0.7.0 ( Slow winch )
- bitsyX
- timezone/localtime 7c7836c1a47b82d402e88495c6001abf PST8PDT 1017 Dec 30 2004 HORCA 0.7.0 ( Slow winch -- just slower than the 0.5 thresh )
- bitsyX
- Resolved: comm problems due to bad anteanna dir/height TORCA 0.7.0 ( Slow winch )
- bitsyXb
- timezone/localtime 7c7836c1a47b82d402e88495c6001abf PST8PDT 1017 Dec 30 2004 [ Bad AIRMAR - windspeed always zero ] NORCA 0.7.0 ( Slow winch ) -bitsyXb
- timezone/localtime 7c7836c1a47b82d402e88495c6001abf PST8PDT 1017 Dec 30 2004 [ Winch intermittent - perhaps control box ]
- 10/2013: currently down for servicing CORCA 0.7.0 (Slow winch)
- bitsyXb
- No /usr/share/zoneinfo /etc/localtime c9452f6b9e08d83c6815c38600798964 localtime PORCA 0.7.0
- bitsyX
- timezone/localtime 7c7836c1a47b82d402e88495c6001abf PST8PDT 1017 Dec 30 2004 [ Bad AIRMAR - windspeed 69 knots all the time ]
- 10/2013: currently out of the water for servicing
0.9.11 - aquadopp.c : Added a warning to the download operation when the datafile is >100kb long. A typical file is 14-20kb for our buoys. Also added a "get prompt" step between the two data query commands in the download routine.
0.9.10 - winch.c : Added 3 second delay after WINCH ON in movePackageUp and movePackageDown to allow for slow start up of new winch control cyclinder
0.9.9 - orcad.c : Sync the CTD time with the computer time at the end of every cast. aquadopp.c : Fixed many communication problems that emerged after firmware updates. It appears that the undocumented recorder commands for obtaining file data leave the Aquadopp in a non-command state. Now obtaining the command prompt before trying to send erase command. It also appears that the timing of sending commands is sensitive. It works best if we formulate the command and paramters and send in one transmit call rather than several calls.
0.9.8 - pifilling.c : Fixed a potential deadlock situation in the getADValue()
code. We found that some PIFilling boards (v0.3 batch)
have a bad ADC chip ( ports 0-3 ). The chip reports back
a "9c" status when the configuration register is read. The
high bit is an indicator that ADC data is ready to be read
( "RDY" active low ). On these bad boards the RDY is not
ever being set low and thus the loop waited indefinitely.
I installed a loop counter that permits up to 5k attempts
before quiting and returning with a value of 0. I am not
aware of any condition that would cause these chips to
lock up permanently in this way -- assuming bad chips.
Updated the modbus/ftdi/usb libraries after developing
a new Raspberry PI operating system image. This new
image is compatible with RPI Model 1B, and Model 3B
boards and all PI Filling Boards as well as T-Rex boards.
The modbustest code has been more appropriately named
sunsaver-query.
The standard utility scripts updateIP, wifiCheck,
getStatus etc. have been copied to/updated in the utils/
directory.
start/stopAuxiliaryD.sh have been added to the distribution.
And all the start/stop scripts have been updated to switch
between inittab and systemctl depending on the system they are
installed on.
0.9.7 - hardio.h : added definitions for controlling Odroid and Zoocam power iotest.c : added functions for controlling Odroid and Zoocam power parser.c : added function for ttyUSB serial ports 4-7 ** in "-lowVoltage" version MINEXTVOLT is set to 16.0 volts to account for ** erroneously low external voltage readings.
0.9.6 - weatherd.c : added driver for Soundnine compass and Vaisala anemometer.
0.9.5 - weatherd.c : Wind direction averaging changed from mean to vector average.
0.9.4 - winch.h : Pressure change threshold was lowered in winch.h to accomodate a slower start up of the package at depth using the new winch controller module, changed from 0.3 to 0.25.
0.9.3 - weatherd.c : magnetic declination changed from (-) to (+).
0.9.2 - weatherd.c : PAR conversion equation for UW op-amp board was corrected.
0.9.1 - weatherd.c : PAR is now reported in umol instead of watts; changed variable name to eliminate confusion changed file headers to reflect PAR instead of SOLAR 0.9.0 - weatherd.c : weather-status.dat now reports averaged weatherd.h wind speed/direction readings as well as maximum wind speed/direction. The wind direction is now corrected for local magnetic declination. Support for Licor PAR amplifier is now included.
0.8.9 - weatherd.c : fixed an error in the conversion between raw counts and wind speed for the RM Young anemometer/compass.
0.8.8 - weatherd.c : fixed an error in the baromater conversion between hPa (reported by Gill MetPak) and inhg (calculated for the output file).
0.8.7 - bitsyxio.c : Added a backwards compatible bitsyxio.h : API call for getADVoltage() so that we can compile versions for all computers.
0.8.6 - profile.c : If a profile is started and it bombs profile.h out in the setup ( ie. before CTD orcad.c INITLOGGING is performed ) the system would save the same old CTD data to subsequent CAST files. Now it will only attempt to create a cast file if the error happened after INITLOGGING.
0.8.5 - profile.h : Created a new hardcoded constant for
the minimum external voltage allowed
when running a cast. Typical this is
set to 22.0 Volts. Because of a problem
we are seeing at NORCA with the battery
voltage measurement being off ( low )
I have set it to 19.0 Volts in this
version.
0.8.4 - iotest.c
hardio.h : A confusing API for the ADC led to a
bug in the 0.8.3 version. A new
API function getADVoltage was created
to clear up the confusion and restore
the PAR readings. Also the "iotest status"
function now shows the raw voltages for all
the PI Filling ports.
0.8.3 - profile.c : Contention for the i2c bus appears to
occasionally cause the A/D reading to
fail and return a negative voltage. This
in turn was causing the profile to error
out due to low voltage. Added some checks
for negative values and a second retry
at reading the voltage before erroring out.
0.8.2 - Makefile : First version using Raspberry PI cross
compiler installed on orcabase. Also
added a new external directory to
contain all external static libraries and
header files.
0.8.1 - ctd.c : Fixed a problem with new CTD firmware. The new
version adds a field to the DS line.
Also fixed some old code that was supposed to be
checking for streaming data after initiating the
"STARTNOW" command. It was not checking and allowing
the code to continue. With the newer 19Plus firmware
it appears there is much more of a delay between
commanding STARTNOW and the start of data streaming.
The fixed code now tries for up to 8 seconds ( 2 seconds
at a time ) to see if data is coming from the CTD.
0.8.0 - profile.c : Print out the actual voltage values when we
have made the choice to abort a profile for low
voltage.
pifilling.c : getScaledADValue() should have been returning
-1.0 when it failed to obtain a value.
hardio.h : PAR Sensor ADC port was set to 3. Should have
been 2 ( 0 based ).
0.7.9 - pifilling.c : IO routines opened I2C device but neglected to
close it aftwards. I suspect this ended up
using all the file descriptors and then started
to fail to open the device.
0.7.8 - Improvements for iotest
- stop/start utils now work on both raspberry PI and Bitsy models
0.7.7 - Fixes for RaspPi version.
0.7.6 - Fixed hardware initialization to set the winch direction to down. This keeps the FET off and the solenoids off by default.
0.7.5 - New weatherd to support new weatherstations
0.7.4 - Major update
This includes a build target for the new Raspberry PI
and PI Filling boards. Currently there is not a
cross-compiler on orcabase so this needs to be compiled
on a RPI system directly.
Also:
- This version defaults to the slower winch
setting.
0.7.3 - Two versions of this release were created 10/17/2013:
This version corrects a bug in the previous version where
power was toggled off of the hydrowire only after a successful
profile; power remained on after a profile erred out.
In this version power is toggled in the orcad.c code, instead
of the profile.c code.
The ones marked "SlowWinch" set the Pressure Threshold
( in meter.h ) to 0.2.
orcaD-0.7.3-bitsyXb-SlowWinch.tar.gz ( CORCA, DORCA, TORCA )
orcaD-0.7.3-bitsyX-SlowWinch.tar.gz ( HORCA )
0.7.2 - Two versions of this release were created 10/17/2013:
All buoys have been switched from an underwater
battery pack to a supercapacitor bank. Therefore
there is no need to keep the power on the cable while
we are not profiling. Power is toggled on at the
beginning of the profile, and toggled off at the end
of the profile.
The ones marked "SlowWinch" set the Pressure Threshold
( in meter.h ) to 0.2.
orcaD-0.7.2-bitsyXb-SlowWinch.tar.gz ( CORCA, DORCA, TORCA )
orcaD-0.7.2-bitsyX-SlowWinch.tar.gz ( HORCA )
0.7.1 - NOT A RELEASE VERSION -- for testing
- Orcactrl now has the option of running despite
failure of the hardware initialization routines.
- test ctd_comm ( in orcactrl ) now prints out lots-o-info
to assist with testing the CTD comm.
- Added some logging code to weatherd. It now detects
when GPS acquisition occurs and reports missing data
for the WMDA record ( we have seen this problem before
with bad sensors ).
Added the device number to the "hasSerialDevice()"
debugging line.
0.7.0 - Two versions of this release were created.
I improved the error reporting to indicate
what the pressure threshold is and what the observed pressure
difference is when it logs a EPRST error.
The ones marked "SlowWinch" set the Pressure Threshold
( in meter.h ) to 0.2.
orcaD-0.7.0-bitsyX.tar.gz ( HORCA )
orcaD-0.7.0-bitsyXb-SlowWinch.tar.gz ( TORCA )
orcaD-0.7.0-bitsyX-SlowWinch.tar.gz ( PORCA )
0.6.9 - All buoys
This release comes in three different distributions which
currently map to buoys as:
orcaD-0.6.9-bitsyX.tar.gz (DORCA, HORCA, TORCA)
orcaD-0.6.9-bitsyXb.tar.gz ( CORCA )
orcaD-0.6.9-bitsyXb-WINCH_DIR_SWITCHED.tar.gz ( PORCA )
This release contains modbus support for both architectures.
The TORCA specific changes have been removed.
Nothing else should be different.
Built on Friday, May 20th, 2011
0.6.8 - PORCA specific
New release with winch direction flip still in place.
Also now a bitsyXb buoy
0.6.7 - TORCA specific The new winch on TORCA is very slow. Its only running at .19-.2 pressure change per reading. Had to change the minimum in the winch.h file to make this work.
Also added a -p flag to the modbustest program.
0.6.6 - New release for all buoys - iotest reading of current state fixed. - Meterwheel can be commented out in the cfg file. - New "test ctd_comm" program in orcactrl
Produced a version for bitsyX, bitsyXb, and bitsyX with
the winch direction switched ( for PORCA -- still broken ).
NOTE: Currently modbus support isn't compiled on the bitsyX
versions.
Installed at CORCA, PORCA, HORCA
0.6.5 - New release to coincide with CORCA deployment.
Changes include:
- Support for bitsyXb EABI binaries.
- MODBUS Library support ( for solar charger comm ).
- Meterwheel comm through USB rather than serial.
- Li-Cor parameters are now in the config file.
- USB devices are now configured in the config file.
- New modbustest/ftditest utilities
New Arduino meterwheel counter hardware.
0.6.4 - Lowered max log size. 500kb was a bit large for the cellular buoys -- especially since this is downloaded once an hour.
- Normal winch logic ( NOT PORCA compatible )
- Meterwheel enabled ( NOT PORCA compatible )
Deployed to TORCA on 8/13/10
0.6.3 - Added functions to orcactrl for debugging aquadopp - Changes to aquadopp.c to make communcations with the aquadopp more robust.
NPB BUOY SPECIFIC RELEASE!!!! The NPB computer still has
the reversed winch logic.
Deployed to NPB(PORCA) on July 13th 2010
0.6.2 - Fixed several problems for the NPB buoy.
NPB BUOY SPECIFIC RELEASE:
- For some reason the winch direction was reversed on this
computer. We need to fix this once the computer returns to
the shop. There is a workaround in the bitsyxio.h for this
computer only!
- The meter wheel counter failed on this computer so there is
a workaround in winch.c which sets a constant for the
counter so that the other routines continue to work correcctly.
Also adjusted down the sensitivity of the pressure change
required to keep moving the package.
GENERAL BUOY/EQUIPMENT FIXES:
- Fixed a problem with INITLOGGING not being recognized by
the new 19Plus V2.
- Found a problem with the new Aquadopp data download.
Disabling in the config file for now. Will look into
firmware changes so what we can update our driver.
- Weatherd was not updating the MET files at the right
time. Fixed.
Deployed to NBP on April 9th 2010
0.6.1 - Fixed a problem with static linking of weatherd. Added meterwheel_cfactor parameter to config file.
0.6.0 - Major release for new hardware. This release supports the new Seabird 19+ V2 CTD. The device is still known as a 19Plus to the software but the support for both versions in the driver. Also we wrote a new daemon called weatherd. This is an independent data logging application which currently reads from a AirMar weather station attached to a USB port and a solar sensor attached to an A/D pin. This is a development release -- not installed on the buoys.
0.5.9 - Skipped release. No changes
0.5.8 - Changed the external voltage threshold from 20 to 22.
0.5.7 - Fixed a bug in profile.c. The aquadopp was being started even if the config file didn't have the device setup.
0.5.6 - Created a new config file option. This option is set per mission and controls the auxillary sampling instrument ( on the auxilary serial port ). if "aux_sample_direction" is set to "up" the auxilary instrument will be turned on only for the upcast phase ( note this is on,off,on,off during a normal 1.0 cycle ). If the parameter is set to "down" sampling will only occur during the downcast. A third setting of "both" will sample during the complete cast.
Fixed a bug in the aquadop getAquadoppTime() function.
The function sent the RC command and then looked for a
double "Ack" word to delimit the returned data. Unfortunately
in June the month is set to "6" ( the Ack character ) and so
the returned data is truncated by one byte. NOTE: This could
also have happened when any of the date bytes formed a
consecutive "0x06 0x06" set. Now this reads 8 data bytes
and doesn't look for delmitter.
Fixed a minor debug message in getAquadoppTime().
Fixed a bug in the runJobs() function where the
aquaFD file descriptor wasn't opened.
Deployed to HORCA on Jun 14th 2007
0.5.5 - Changed getAquadoppTime() isdst to -1
Deployed to TORCA on Nov 8th.
0.5.4 - Added code to power down the aquadopp ( PD command ) at the end of the downloadAquadoppFiles() routine.
Back in 0.5.1 we found a problem with the getCTD*Time
routines in which the time was off by an hour. It
was determined that the time_t.isdst flag needed to
be set. It looks like we should have set this to -1
instead of 1. -1 indicates that the status of DST is
unknown and should be inferred from the date/time
and the bitsy's timezone. This was breaking the getWSTime()
( also changed in 0.5.1 ) but oddly not getCTD*Time ( yet ).
I set the isdst flag to -1 in getWSTime() and fixed this
problem for now. NOTE: Of course this happened the first
time we restarted orcad after the Fall DST time change.
Also changed the isdst flag in getCTD*Time to -1.
NOTE: Aquadopp is still 1. Need to consider changing that
as well.
Deployed Monday, October 30th at 10:30am to:
NORCA, DORCA, TORCA, HORCA
0.5.3 - Added code to set NAVG=1 in the CTD 19+ initialization routines.
Added code to work with a Nortek Aquadopp instrument
on the second hydro wire communications line.
Deployed on NORCA with an aquadop on Oct 23rd @ 2pm
0.5.2 - It appears that TORCA is slow at moving the winch up when the battery voltage is low ( ~22v ). This causes the winch routines error out with a low pressure change in 5 counts. It used to be < 1 meter. I changed it to < 0.5.
I also changed the order of the output pressures
in the LOGPRINT statements.
Deployed to TORCA
0.5.1 - Stupid little bug in the getCTD19PlusTime routine. The hour was one off.
Deployed to TORCA, NORCA, and HORCA July 24th 2006 @ 17
0.5.0 - New release numbering as we have finally hit a stable product.
Moved the 19+ time sync'ng code into its own routines.
Added 19 code to get/set time.
Made a general syncHydroTime routine to call either
19 or 19+ code depending on the configuraton.
Only set the date/time of devices if they are off
by alot.
Modified winch.c: Added code to debug the pressure/meter
values inside the critical loop at debug level 6 or higher.
Also changed the way a meter wheel failure is handled.
We now set a warning flag if the meter wheel count is
constant for more than 4 scans. If it's constant for
more scans and the pressure is not increasing
than fail out. Also added code to recognize if the pressure
goes static. Static is hard coded as less than 1 meter
change in 4 scans.
Fixed a problem with parsing the "depths" field in the
config file.
Deployed to HORCA: July 19th 2006 @ 3pm
NORCA: July 19th 2006 later
0.4.26 - Lowered the log rotation limit to 500kb
Added a loging statement to the critical loop
of the movePackageDown() routine. At LVL_VERB
this routine will log every pressure/meter wheel
count.
Modified the logRotate code to rename before it
closes the logfile. This allows for error reporting
on the rename process.
Deployed to horca on 6/21/2006
0.4.25 - Added "outsideHumidity" to the instant weather report.
( On all the buoys as of Mon June 19th -- installed
way before this date...not sure when ).
0.4.24 - Previous attempts at retrying failed commands to the ctd didn't really improve the situation all that much. After sitting down and manually testing commands over the noisy line we determined that the a one second delay after a failed command vastly improved the chances that the retry would succeed. This version makes this change on the CTD19Plus functions.
Deployed to NORCA, HORCA
0.4.23 - Changed ctd19plus INITLOGGING and OUTPUTFORMAT commands to 3 retries ( from 2 ). Hopefully this will get around the noisy comm line problems.
Improved the getCTD19PlusSPrompt routine to
be more robust to CTRL-Z failures.
Deployed on HORCA
0.4.22 - Intermediate version. For testing purposes only.
0.4.21 - Added code to stop CTD logging whenever an error occurs in the profile subroutine. This was leaving the CTD on and draining the batteries.
Added code to sync the CTD19Plus clock with the bitsyX
during initialization. Need to write a similar function
for the CTD19.
Deployed on HORCA Mar 24 2006 17:06:39
0.4.20 - Added code to winch.c to decypher the status codes coming from the move routines. Also added the last pressure and meterwheel counts to the error status output.
Deployed on Hansville Friday Mar 3rd 2006 @ 14:43
0.4.19 - Added function to rotate log files when they reach 1 MB. This enables us to move them off the CF card with a download script without disturbing the running daemon. Also added the year to log messages.
Still had problem with iotest. It had the aliases
for WINCHDIR reversed! Orcad didn't though. Thus
we didn't notice the problem.
Excessive signal loss in a communications line
is causing occasional problems with commands to
the CTD to be received correctly the first time.
Added some retry attempts on the CTD 19+ commands
which are long like "INITLOGGING" and "STARTNOW".
Added the year to the instant weather output. I
know it's extra data but for so many programs
this is really useful to know as you parse the data.
Deployed on Hoodsport Tuesday Feb 21st 2006 @ 2:30pm
Deployed on Hansville Wednesday Mar 1st 2006 @ 11am
0.4.18 - Changed PORTA_DIRMASK to 0x04 from 0x03. The former was wrong and would have allowed writing to one of the A/D lines. This is bad since writing to a an A/D line enables pull up resistors thus messing up the readings. Unfortunately this is not the cause of our current voltage reading problem but worth changing non-the-less.
Found problem with A/D pullup resistors. It was
in setOutputLine(). The bitmask procedure was
only protecting writing over I/O line states. If
PORTA was mixed I/O and A/D *and* the A/D port had
a voltage on it, the subroutine was attempting to
preserve the logic "1" on this line when changing
other lines on the same port. I introduced a
PORTA_ADMASK for this port which always make sure
that a zero is sent to the A/D lines being used.
Added INTVOLT and EXTVOLT commands to iotest so one
may query the system voltage without too much of
a headache.
Deployed on the North Buoy on Jan 30th, Deployed
on Hoodsport on Jan 26th.
0.4.17 - Added a last goodbye message at LVL_ALRT so that we can ensure that all cleanup duties have been run. Somtimes the process does not quit cleanly ( sits there with DW status ). In situations like that this message will inform us if it at least finished all of its tasks.
- Added "pressureDepth" to the logging at the end of the critical
loop in movePackageDown/Up(). This will help us debug why we
are sometimes leaving this loop before reaching the target
depth. Also added code to print out the 4 pressure/meterwheel
values if we are in debug level logging or if something strange
with the pressure occurs.
Deployed on Hoodsport on Jan 23rd.
0.4.16 - Added date/time and 10 minute avg windspeed to the instant weather log.
Use CTRL-Z to stop the logging on the
CTD 19+ per Seabird's suggestion.
Added code to orcad's cleanup routine to
power off all devices and save the weather
archive on shutdown.
Worked around a problem with ADS's non-standard
read behaviour in iotest.c I had already employed
this in bitsyxio.c but hadn't worked it back into
iotest. Perhpaps this will fix the line status
problem.
Fixed the units of rainfall in the archive download.
To obtain in/hr you need to divide by 100.
Fixed readLastCastInfo(). It was not returning the
last cast number and instead was always returning 1.
Deployed at Hoodsport only.
0.4.15 - Documentation changes.
The ability to download the last archive
from the Davis Weather Station is flaky.
So for instant web-based weather data this
version uses the logInstantWeather routines
instead.
Deployed this version at hoodsport on Jan 3rd 2006.
0.4.14 - Changed CELL power i/o port to WIFI Amplifier power in bitsyxio.h and iotest.c
Deployed this version at hoodsport and north buoy
on November 17 and 18th respectively.
0.4.13 - Moved cast index file routines to the util.c file.
Decreased winch resolution to 1m. Hysterisis
is causing a delay up to .5 meters.
Changed how a failure status from stopHydroLogging
is treated in the profile subroutine. Instead of
propogating the failure up ( causing it not to
attempt to download the data ) we log it and
return success.
Fixed a bug with the downloading of MET data. The
daemon was assuming that upon startup it was ready
to download the archive.
Changed .MET file format to be a tab delimited record
format.
This version was never deployed...you know...
superstition and all...
0.4.12 - Deployed Monday Oct 31st - Halloween Day on the Hoodsport buoy.
Fixed a bug in the bitsyxio.c::emergencyPortClear()
routine. The routine was incorrectly identifying an
open system call return status as bad when it was good.
Made config file parse more robust ( ie. name/value
pairs with/without spaces. Added code to print out
a line number for syntax errors.
Added sync() to the logPrint function. We have observed
that the fflush() is not always sufficient to commit
the data to the disk.
Fixed problem with profile loop where attempt is made
to move the package up to a location it is already at.
Changed up/down movement position tolerance from
0.2m resolution to 0.5m. This is due to the faster
speed of this winch and the hysteris of shutting it
off.
Added code to initialize the CTD 19 Plus. Also worked
on getCTD19PlusSPrompt to make it more robust.
Added an extra check in the getCTD19PlusSPrompt() for
the successful halt to logging. The old code appeared
to halting the logging but thinking it didin't.
Documentation changes.
0.4.11 - First version to see the light of day. This release was placed on the Hoodsport buoy for one week on Oct 19th 2005.
Robert Hubley rhubley@gmail.com Wendi Ruef wruef@ocean.washington.edu