The day before I went on my vacation to Chicago was bright and sunny and I had an opportunity to test the guts of the project.


LiPo test

The night before I ran the unit on the LiPo battery for about 10 hours.  I had fully charged the battery earlier.  With my estimate of 40mA average consumption, I should have consumed about 400mAh or about a third of the battery's capacity.  It took about an hour and 45 minutes to recharge the battery @ 250mA, so I think I'm in the ballpark.  In actual use I'll want to limit the power consumption to much less than this.  Because I'm in test mode, I've maximized my data transmission frequency.  ThingSpeak will allow data updates every 15 seconds and I'm updating at approximately every 17 seconds.  My inverter only provides data at 15 minute intervals, so I'll drop to that rate in normal operation and there also wouldn't be any need to transmit data when it is dark.


Test setup

For my initial test I wanted to check the accuracy of my irradiance reading against the Daystar meter and also compare the data to my inverter readings.

I set up the unit on my deck which is to the west of my house.  This area gets shadowed by my house in the morning, but one of my array strings also gets shadowed by my neighbors tree in the morning so this location is adequate for my window of comparison.  I just realized that I took this picture before I had correctly configured the tilt on the cell.  I attempted to mimic the panel orientation on the roof.

Outdoor test setup


ThingSpeak Plots

Outdoor test TS temperature plot

Outdoor test TS irradiance


SMA Inverter Plot

Outdoor test SMA inverter plot


From the inverter plot the valid time window for comparison is from 11am to 3pm.  This is Oregon, so the good news and bad news is that there are tall trees and I get shadowing both in the morning and afternoon (except June thru August when the sun is directly overhead).  You can see from the ThingSpeak Irradiance plot that my deck gets shadowing before 11am and after 5pm.


Data Comparison in Excel


I normalized the inverter power numbers so that they would match the peak Daystar reading.  I'm somewhat confused by the data.  I can believe that I have an offset that I need to correct for the SPEC, but I'm not sure why the Daystar readings and the solar array power don't track.  I realize that I don't have the temperature component included yet, but it's more probable that I've made a measurement error (the Daystar reading is not automated).


Unfortunately, the weather forecast is for clouds and rain for the rest of the week so I won't have the opportunity to try this again until next week.


One of the lessons that I've learned is that a significant weakness of my design is that I can't update my firmware remotely.  I realize that I can handle a lot by post-processing the data but I can't change parameters like the sample rate.  Need to configure the unit to receive data.