As I mentioned in a previous club meeting if we were to have a natural capacity test of the batteries up on Mt wombat i'd share the results.
To start off a little background we collect for each of the 6 ports on the dc distribution board: power, current, voltage, current shunt voltage.
All the data is stored in a database with a temporal resolution of about 1 second and about 1mA, 1mV and 10 mW.
So if we(I) want to know the capacity of the battery all we need to do is add up all the mA's from the time the mains powered DC was lost to when the battery was disconnected.
While not the right way to do this I'm hopeful that in right ball park is good enough (what we should do is confirm that there is 1 second between each data point, but at worst this method will underestimate the capacity.)
The sql looks something like this (all the times are in unix time for simplicity of constancy)
`
[vk3rgvdcdist]> SELECT SUM(current) AS 'mA Seconds' FROM battery WHERE unix >= 1634619522 AND unix <= 1634748260;
+------------+
| mA Seconds |
+------------+
| 610849116 |
+------------+
1 row in set (0.066 sec)
`
To get to some more standard units a little dividing is needed, first lets get it into amps not miliamps
610849116 mA Seconds / 1000 = 610849.1A Seconds
Now to get to from seconds to hours.
610849.1A Seconds / (60*60) = 169.7 Amp Hours.
We also have the power so we can get the energy as well, this time we will get sql to do the final tidy up of the conversion. (note the power is not stored in the database in watts directly. it needs to be divided by 40.
`
[vk3rgvdcdist]> SELECT SUM(power/40)/3600 AS 'W Hours' FROM battery WHERE unix >= 1634619522 AND unix <= 1634748260;
+---------------+
| W Hours |
+---------------+
| 2032.88363194 |
+---------------+
1 row in set (0.081 sec)
`
2033 Watt Hours.
Final tidy up; lets check for obvious big gaps in the data ans check if our estimate is going to be usable for future comparisons.
Start with how many data points are in the time interval.
`
[vk3rgvdcdist]> SELECT voltage FROM battery WHERE unix >= 1634619522 AND unix <= 1634748260;
....
| 10915 |
+-----------+
128490 rows in set (0.098 sec)
`
128490 rows that should be equal to the the different between the time stamps.
1634748260 - 1634619522 = 128738
Opps, we lost some, but it does not looks like that many, how many?
128738 - 128490 = 248 Seconds of missing data. or about 0.2%
I don't think it will be worth the effort to expend the effort to get a better result. The measurement error and calibration is not that good.
I think the batteries are in fine health, I would feel that it would be good to bump up the capacity though so that we have more that a day of restricted services runtime as its not often that someone with keys can drop what they are doing and resolve a power outage in that time window? topic for next technical report.
