The modification in the last post did not work.
Another session with Google AI yielded another, slightly different, piece of code.
That modification has been made.
An error in the PHP datetime code had been resulting in the return of the DST problem.
The addition of '$' to the existing 'date' statement corrected the problem.
The problem resurfaced when I retrieved an old copy of the PHP script without making the updates.
Whether the latest modification works remains to be seen.
Data gaps is a misnomer. What happens is the PHP script runs but the Triron charge controller does not output data sometimes. The script does run the python NOAA2 script and writes elevcalc and azicalc the the csv file.
I used Google AI to pinpoint the problem and come up with a solution. I gave the agent relevant data and a result was formed from analyzing the github software sources and my data. The agent explained what was happening and pinpointed the PHP library from github as the origin of the problem.
The AI wrote a simple software modification to prevent the script from writing blank lines to the csv files.
The code:
========================================================================
// --- Your existing code above ---
// VALIDATION: Ensure both power data points are present and valid
if (
isset($tracer->realtimeData[2]) && $tracer->realtimeData[2] !== '' && $tracer->realtimeData[2] !== false &&
isset($tracer->realtimeData[5]) && $tracer->realtimeData[5] !== '' && $tracer->realtimeData[5] !== false
) {
$i = 2;
$dat[$i] = $tracer->realtimeData[$i] . ",";
fwrite($myfile2, $dat[$i]);
$i = 5;
// Make sure to append your newline character (\n) here if this is the end of your row!
$dat[$i] = $tracer->realtimeData[$i] . ",\n";
fwrite($myfile2, $dat[$i]);
}
// If the data is missing, it falls through here and writes nothing.
} // <-- This is your existing closing curly brace for the parent 'if' statement
There have been occasions when data gaps occurred in the CSV files collected daily.
The problems are unique to the power file. The azimuth and elevation are printed, but nothing else.
Further, that data is collected outside the schedule. Somhow, the NOAA2.py script is being run outside of schedule, and printed to the CSV files. The EpD3Dy.php script is run by a service/timer. The NOAA2.py script ius called from within the php script.
The problem may disappear if the python script is run in a different service/timer pair.
I'm examining an epevermodbus python script to read and record settings separate from the php script. This will allow me to examine the data for problematic results, such as negative values or other data anomolies.
I reset the timer to a time after the sun has risen. Unfortunately resetting the timer startup very season is not practical. I may be able to set the startup time from within the service or the script.
This problem is relatively recent. Perhaps this is a result of the differences between the Epever Tracer and Triron charge controllers.
Autumn and the shorter day is coming.
I need to raise the panel array up higher since the sun will not be rising as high in the sky until next Spring.
The best solution would be a concrete base for the bottom, 1/2"+3/8" rebar reinforcemant for the steel axis pipe, a more permanent support for the high end.
I need to get short 'samples' of the rebar to insure the fit before buying the ten foot long pieces.
I am making a working set of files and folders available for free.
I've done this before but so much time has passed, and changes made, I need to do this again.
Instead of posting each file on the blog, I'm providing a Dropbox public link.
The battery bank has been wired in parallel with system connections at the closest battery. This creates an unbalanced load on the bank.
The modification: instead of direct wiring to the closest battery, one connection to the closest battery (+) and one connection to the most distant battery (-).
The ideal arrangement would be separate connections to the system for each battery, a radial configuration.
I finished the installation of the replacement lever actuated terminal connector WAGO 2606-1103.
This afternoon, I had occasion to run the inverter/generator. I plugged the unit in while the Xantrex inverter was on. The system was maintained. Notably, the RPPi4 experienced no discontinuities in the data stream.
I count this as a success.
There was a brief lapse in the lightiing, about 1/3 second, but this was partly due to a setting on the iniverter. The transfer is set up for APL, appliances, but could be set up for sensitive electronics. The second option warns against transfers with inductive loads such as electric motors (water pump, fridge) running.
