I perhaps need to tweak the maths or the sampling interval time in order to linearise the plot. Between 9am and 5pm, the sampling interval is ten minutes. Outside those hours, it's 20 minutes so there is a nonlinearity in the graph. Alternatively, I could interpolate the missing 10 minute intervals or discard alternate readings during the shorter sampling period.
This is what the graph looks like with the 10 minute intervals removed:
I think that does the job. I'm going to compromise and rewrite the logger so it just logs data at identical 15 minute intervals over the whole 24 hour period. It takes a little 'steppiness' out of the graph, too. Until I tweak the hardware, the accuracy of the temperature measurements is +/- 1 step on the digitiser which equates to roughly 1.9 Celcius. Closer tolerance and equal intervals will mean a more readable graph. Even so, as things stand, the performance of the system can be monitored. Even at almost midnight, the tank was holding up at around 42 Celcius. I really don't need hot water then, but it does show that from when the sun comes up, the system doesn't have to lift the water temperature from a very low starting point if there has been good input on a sunny day.
No comments:
Post a Comment