ONLINE DATA ACQUISITION & DISSEMINATION
The following "Lessons Learned" have been submitted by NWSH Engineering Branch and NWSREP's (DAPM/OPL/HMT/MIT). These situations
are not uncommon and the solutions presented may help you in the field. If you have similar experiences (Lessons Learned) that
you would like to share, please send an email to: LESSONS
LEARNED
Contact David Desrosiers, Engineering Focal Point at 301 713-1845 x115 or by
email for questions or clarification on any of these issues.
NWSH LESSONS LEARNED PAGE (.pdf)
DOWNLOAD DATA GMA - LAPTOP
Downloading to a laptop is fairly straight forward, but takes longer than loading a key so make sure your laptop battery up to snuff.
Downloading
Connect up
Open Hyperterminal
Open the Zeno user interface, "U"
Go to the data menu, "D"
Go to the transfer menu of Hyperterminal
- select Capture Text.... it opens a dialog box to specify where and what to call the captured text file,
- enter a name and select an easy-to-find location, click OK,
- then everything that shows up on the screen will be written to that file.
Back on the Zeno Data menu, ask for the amount of data you want...
- a month is about 2900 records..... the command " L6000" will get two+ months
- or just use the " * " command to get all data, and stop it when you get enough.
When done, go back to the Hyperterminal Transfer menu, select capture,
then Stop to stop the recording and close the file. Exit Hyperterminal.
If you need to send the downloaded data to NCDC instead of a key file, it will need to be restructured to be acceptable at NCDC.
Convert file structure
Find and open the file in Word. Edit out all the menu stuff and data labels, leaving just the data.
Save as a text file.
Now need to reverse the order of all records...
Open Excel to a blank worksheet.
- select Data menu, Import external data, Import Data, select the text file.
- this opens the Text Import Wizard, in step 1, select delimited
- in step 2, select Comma delimiters
- in step 3, make date column format > Text (not general or date)
Need to reverse order of records.
- Select all the data.
- go back to the Data menu, select Sort,
- sort by first column and then by the second, both ascending, click OK.
Need to save it as a csv file.
- Go to File,
- Save As,
- give it a name and location,
- change save as type to CSV,
- click OK to exit.
- Say YES to all questions about saving to this format.
- Leave Excel open for now.
Open file in Notpad, see if it looks right.
- Date should still be in YYMMDD format. If not, go back and correct.
Open any of your other files for this site in notepad, select the first 4 lines and copy to clipboard.
(the header in these files never change, just copy one over to the new file.)
Reopen your new file and paste the 4 lines at the top.
Note the starting date/time and the ending date time.
Select file save as, and name the file according to the DataKey file name convention:
.csv See any other file as example.
- be sure to select save as TYPE... All Files. Otherwise Notepad will append .txt to the file name.
Done. Save yourself a copy, Send to NCDC.
DISPLAY STUCK ON "COASTAL ENVIRONMENTAL" SCREEN
DISPLAY NOT UPDATING
DISPLAY SHOWING GIBBERISH OR ODD CHARACTERS
HOW TO RESET THE SYSTEM
The FPU display operates independent of the Zeno datalogger inside the GMA, normally showing everything the Zeno sends it. But there are
a few instances where the display can get out-of-sync with the datalogger and 'lock up" resulting in the actions above. Most of the time
when this happens, the Zeno data logger is happily working away collecting data as it should - it was designed to do this so no data would be lost.
The only way to re-sync the display to the logger is to do a POWER-ON reset... you must remove all power from the GMA and re-apply power.
Follow these steps:
- either throw a light blocking coat over the solar panel, or disconnect the solar panel connector under the enclosure,
- disconnect the white battery connector inside the enclosure,
- count to 10 (the display should be off. If it is not off, the solar panel is still getting light through your coat, disconnect it and count to 10 again)
- plug the battery back in
CAUTION - the battery connector only goes in one way, do not force it. (the display should light, first with the Coastal splash screen,
and then with the data screen) - plug the solar panel back in.
If everything is working, you are done. Data will appear within 30 minutes... read lessons learned "Where's My Data" to see why it takes so long to appear.
WHERE'S MY DATA?
WHY DO I SEE -1000000?
WHY WON'T THE KEY WORK?
When the Zeno first power's up, it has no knowledge of anything that happened before power came on... and it won't until after it has taken it's FIRST SCHEDULED READING.
Additionally, ancillary functions like telemetry, the keys, some displays, etc, ... functions not in the data measurement pathway, may show no response until the SECOND SCHEDULED READING takes place. This seems odd - WHY?
First, a data logger is not like your PC. The data logger is first and foremost a SCHEDULE driven machine. It is not awake and fully functional between the times of a scheduled reading. If you power up at 12:01, and
the logger is programmed to take readings at 15 min intervals starting on the hour (like the FPU), nothing will happen until 14 minutes later, 12:15, when the logger wakes up and runs through it's programmed operations for
the first time. Until 12:15, if you try to look at stored data, the Zeno will tell you there isn't any, even if the memory is full. If you try to look at the scrolling sensor output, S1,1 command, you will only get the
"no data taken yet flag" of -1000000. If you try to use the key to download data, you will get no action, because the program has not run through for the first time yet and the Zeno does not know that to do with the key
inserted interrupt, yet. After the first SCHEDULED reading time, the program has run once and data should be available to the S1,1 command. but data may not be available from memory (and therefore to the key) yet. Some
processes take multiple samples to develop a reading. If you try to access data that requires use of multiple samples, (remember the key download asks for ALL the data) then the logger will ignore your until all the multiple
samples are taken and it has developed a finished reading. In a nutshell, you must wait until the first reading, and sometimes the second, has been taken to expect full access to data.
Remember this applies to powering on the device, and it also applies to the times when menu actions turn off data sampling, such as entering the Zeno Program Menu to enter cal coefficients. When you exit the Zeno Program Menu,
you may have seen the message "Data Collection Starting In XX seconds", This tells you that data collection has been off (because you were modifying the program) and you must wait until the next SCHEDULED reading time to start
to get new data. So the key may not appear to work for a while after a restart. You only have to look at the Zeno clock to see how long you have to wait.
WATCHDOG ERROR?
SENSOR NOT DEFINED?
CAN'T SAVE CALIBRATION PARAMETERS?
I believe your FPU has "lost" it's configuration file.
The clues to it are in your statements below...i.e.:
*Sensor Record #1 type is not specified*
*Sensor Name: *(nothing)
The configuration file is the code that defines what is connected to the Zeno and what the Zeno is supposed to do with the data it collects. Without a configuration file, the Zeno has no knowlegde of the anything connected
to Sensor #1 input, as seen in the error statement "*Sensor Record #1 type is not specified". *Other things like the datakey also will not work, because they are not defined and the Zeno does not know what to do when it
gets that interupt from a key being turned in the key recepticle.
*
*To recover, you simply need to reload the configuration file. Go to section 10 of the FPU assembly procedural to read about the configuration file (so you know how it works) and to see how to reload it. If you need a copy,
go to: NWSH Documentation page and click on FPU Assembly Instructions, near the bottom.
The configuration files are also available on that webpage. You have a choice of two, as the procedure explains.
BE VERY CERTAIN that after you have reloaded the file and verified that it works, that you use the E command to save the new configuration to flash eeprom in the Zeno... or it may disapear again at a less oportune time in the future.
Once saved to flash memory, there are only a few ways that the configuration will get "lost" and most of them involve damage to the unit:
- lightning
- forgot to save it and power cycled
- memory failure
- somebody monkeying with the unit and saved garbage over it.
When you save it to flash memory, it will always be there and will get reloaded everytime the unit powers ON.
After you reload the configuration, you will need to re enter your site data and re calibrate the unit... again using the E command afterward to save those new parameters to flash memory.
BATTERY NOT CHARGING
BUT REGULATOR, WIRING, CONNECTORS, SOLAR PANEL ALL OKAY?
If any part of the solar panel is shaded, the panel may shutdown and not deliver any real power to the application.
The solar panels we use are made up from mulitiple solar cells, connected in series to give the voltage needed and then in parallel to give the power needed. When a solar cell is shaded it becomes a
high resistance to any current impressed upon it. Thus if any individual cell of a series string is shaded, that cell will block the current generated by the other cells in that string, effectively
shutting off the output. Mount the solar panels where they are in full sun, not in the fringe area under trees or behind guy wires or tower supports. Also problems have been reported where construction
or farming resulted in a lot of dust on the panel (charging always got better after a good rain?!), please train the operator to inspect and clean the panel on an "as needed basis". The FPU is supplied
with a 25' connecting cable. The cable can be extended as far as you need, provided you use good outdoor rated wire of the same size or larger, and make waterproof connections above ground. If you need
to trench the cable, use direct burial rated cable.
Additional information added February 26, 2007
Something you might want to consider for a site that does not get sufficient sun during the winter:
- For winter use, instead of running AC out to the gauge and trying to figure out how to connect in and where to safely put a charger, you could leave the AC and the charger back at the house/shed and run low voltage wiring out to the gauge, connecting in either at the SP connector under the enclosure(order a SP cable to get a wired connector), or at the SP junction box (disconnect the solar panel!). If you have a small charger, it should be left ON, letting the regulator board in the GMA throttle back and control charging of the battery. The GMA regulator will self limit 1.5 Amps charging current and only needs to see ~15+ Volts at it's input to work properly. At this low current and voltage, the 22 awg direct burial cable we stock for the MMTS sensor use would work fine, even at runs of up to 100 feet. Even if there is some voltage drop in the cable, as long as you get at least 15V at the GMA, it should charge.
- With this setup, the battery only has to supply power during AC power outages, not every night and cloudy day, and should last a long time.
- The LARC "wall-wart" charger might work. I think it's a 12V charger, on it's high setting puts out almost an amp, it's designed for continuous operation and it has terminals for wire connection. The only restrictions on the selection of a charger are, it should NOT have an open circuit voltage of more than 22Vdc and any current capacity over 2 amps is over kill, but will still work. It should supply at least 600mA minimum. It must be rated for continuous operation! If it gets hot (warm is OK) to touch after 5 hours of operation, it should not be used. Most cheap-cheap automotive battery chargers are not designed for continuous use, buzz, and contain no output filtering/smoothing - if you have any doubts about the safe operation of the one you have, dont leave it at an observer's house. On the other hand, some expensive automotive chargers shut off all charging if the output falls to less than 2 amps as they consider the battery charged... obviously this won't work in our application. So if your charger has an automatic/manual setting, check to see if it shuts down in "auto", else leave it in the manual setting.
Quick method to verify charging:
- On a sunny day, system asleep, pull battery connector, measure at white connector on mother board,
>should be 13.8V... if so regulator/sp/cables/connections working.
- Wake up system, ... did it wake up?
- Still 13.8V? If so sp/regulator has excess capacity to charge battery with even under system load.
Any No requires troubleshooting.
INSTALLATION IS CONFUSING
CAN'T FIND INSTRUCTIONS TO DO.....?
HOW DO I CHECK.....?
WHERE ARE THE CONFIGURATION FILES?
If you are still using Engineering Instruction, or the Beta Site Installation Procedures, or haven't downloaded a
new copy of the manual since July '05, you should go to the Surface Observations Home Page
click on COOP Implementation
click on FPU Assembly Instructions, June 2005
to get an updated copy that includes recent feedback-driven modifications.
The configuration files are also on this webpage:
FPU Initial Configuration File Code, June 2005 and FPU Configuration File with Battery Voltage Tracking Code, June 2005.
MY CONFIGURATION FILE DOES NOT WORK
I TRIED TO DOWNLOAD THE FILE TO THE GMA BUT IT KEEPS ASKING FOR THE FILE
OR IT NEVER FINISHES...OR IT ABORTED REPORTING NO FILE FOUND
The configuration file is a text file and should look exactly like the listing is Section 10, page 37, of the
FPU Assembly Instructions, June 2005. Use Notepad to examine the file, NOT A WORDPROCESSOR!!! There should not be any black boxes or graphical characters located anywhere within the file.
The last characters of the file should be the EOF at the end of the listing... there should not even be a carriage return or linefeed after the EOF.
Depending upon the browser used to get the file from the Surface Observations Home Page (above), some browsers will embed html codes or control codes in the file when the File Menu/SavePage
method is used to save the displayed listing to the computer as a file. This renders the configuration file useless.
Look at the saved file with Notepad. Notepad is a part of MSWindows, located at the menu chain:
Start/Programs/Accessories/NotePad
If you see a listing that looks different than the printed copy in the manual, delete it. Go back to the browser, highlight the whole listing, and use copy/paste to get a copy into a blank
page in NotePad, and save it from NotePad. Then use this clean copy to download to the GMA.
The configuration file is very restrictive about what is in it and where the information is located. A single extraneous character can cause it to not function properly.
Extra characters in the beginning will cause the GMA to not recognize the file as a configuration file and abort the download. Any extra characters on the EOF line will cause the the GMA
to not recognize the end of the file and it will sit there expecting more, apparently stuck.
FPU DISPLAY READS A LARGE POSITIVE OR NEGATIVE NUMBER FOR RAIN AFTER IT WAS READING FINE
If your FPU was working fine, but now is reading +60 to +99 of -60 to -99, suspect a problem with either the load cell or it's cable.
If a lightning surge burns out one of the 4 strain gauges in the load cell, the GMA will measure a large positive or negative signal, resulting in a ± number in the 80 to 100 range for a typically calibrated rain gauge.
Prior to calibration, for the setup conditions of A=0, B=1000, and C=0, the numbers will be in the ± 60 to 70 range. Check the load cell (click for picture) with a volt/ohm meter set to Ohms, on a scale to show 10K ohms (10,000 ohms).
Disconnect the cable, and probe the Diagonal pin pairs, you should see about 10K ohms on both pairs. Probe from corner to corneron each of the flat sides of the connector,
4 pairs... you should see about 7.5K ohms.
The numbers may not be exactly 10K or 7.5K but it is very important that both diagonal pairs read very close to the same, same for the 4 side pairs. Check the cable to verify that there are 4 good wires and that
none of them are shorted together.
Additional information added February 26, 2007
When the load cell dies, one of the four strain gauges on that silver beam, usually has opened up, which lets the logger see 1/2 the excitation voltage, ~2.5V, instead of the normal 10-20 micro volts.
That only 100,000 times more that it's supposed to see.... and the measurement system gets pinned to the upper or lower rail... the 89 # you are seeing. It can be negative also.
It could also be the cable, or the connections. In the early days, several folks checked their load cell cable by inserting ohmmeter probes into the contact of that white connector.
This promptly spread those stamped copper contacts resulting in poor contact pressure from then on. So I would check those contacts also. If they are sprung open, you can tease them back together
with a straight pin. Oh, and if you need to check the cable, probe the BACK of that white connector.
If you have a test plug, that will quickly tell you that the logger is working. If you don't have one, order one.
WILL LEAVING THE OLD CORK DASHPOT GASKET IN
HELP ISOLATE THE LOAD CELL FROM LIGHTNING? OR AFFECT OPERATIONS?
No, it will not isolate the load cell or protect it from surge damange. There is no electrical connection between the 4 electrical
connections and the block of the loadcell anyway. Yes it will affect operation, please remove all trace of the gasket. The load cell block must be hardmounted to the F&P frame. The sponge-iness of the cork
will act as a spring causing non-linear operation of the sensor, upsetting the response expected by the data logger.
RAIN VALUES VARY EACH CYCLE
NOT GETTING A GOOD CONNECTION AT THE LOADCELL
I have received some load cell cables that were thought to be defective, only to find that the sockets in the square white connector
were spread open as if a Volt-Ohm-Meter probe had been inserted. Forcing a probe into these stamped metal contacts bends them out of position and renders them unreliable. If you must check the load cell cable for
contunuity, probe the BACK of the white connector, along-side the wire - it does not matter if you inadvertantly bend the back of the contact. (PICTURE)
If you have a problem cable, look at the face of the white connector and examine the socket contacts within. They should be a perfect circle, with space all around them from the plastic bore of the connector, and
the two expansion slits should be practically closed. If the slits are wide and the two hemispherical halves of the contact are spread to touching the bore, a needle or sharp pin can be used to GENTLY pry between the
bore and socket half to bend it back into position. If needed at all, reposition BOTH halves to the center, don't bend one all the way over to the other.
When you check the load cell cable for continunity, obviously check for a direct connection between two pins, but also check that there are no other connections to each individual wire, and to the shield. The shield is
connected to the backshell of the circular Mil connector, no connection at the white plastic connector. A lightning damaged cable may have one or more wires shorted together or to the shield. If you find shorted wires,
the cable must be replaced.
COULD DISCONNECTING THE BATTERY PLUG HAVE CAUSED THE PROGRAM TO LOSE CONTACT WITH THE DATA KEY MECHANISM?
A power glitch is a power glitch, no matter how it is caused. Remember there are three parts to the GMA: the logger, the display and the key mechanism. The logger is quite tolerant to fluctuations in power and will
run on anything from 21V down to 7V though it starts to have problems writing to flash memory around 11V. The other two are not so tolerant. Even a momentary disruption to the display or the key receptacle, can upset
them, requiring them to be reset - which you got it, only happens with a power ON reset from the logger.
The battery is a reservoir of power and by it's nature will flatten out most outside imposed glitches. The solar panel is another matter, It's output power is LOAD dependent and it's output can vary faster than
you can measure with your VOM. Moving a cable or connection or walking in front (blocking the sun) of the solar panel can cause the SP output to drop like a rock (in a microsecond) - long enough to upset the display
or key receptacle. It's the battery that smooths that out. So yes, pulling the battery, may cause a "system" interruption, even if it does not affect the logger (program).
Contact David Desrosiers, Engineering Focal Point at 301 713-1845 x115 or by
email for questions or clarification on any of these issues.
D.A.D.