Today I did a short run (House batteries basically already charged) after making a small change to the source - trying to fix the Fuzzies during Acceptance phase (See this graph here). The change as to use "<= 0" as opposed to "< 0" when assessing the error calculations for Voltage, Amps and Watts. The prior "< 0" would make sure anyone over their limit had the ball in terms of deciding what to do with the field PWM value, but when that error WAS 0 (meaning we were on-goal for say Vbat), the "< 0" would in effect let the other guys bully their way in and raise the PWM value. And thus the hunting began.
Now, by using "<= 0", if any of the critical values are happy - we will let the Field PWM be.
Notice how the field (and hence Amps/Watts) no longer get the Fuzzies after entering acceptance phase? Timestamp 165,000 on - Fuzzies gone!
Next I will be assessing how the algorithm respond to load changes. Viking Star is an active boat, and during battery charging things like water pumps come on, refrigerator starts, washer/dryers enter drying cycles. Any one of these can dramatically change the Amps draw relative to what a battery is accepting during the Acceptance phase and need to be accounted for. I am wondering if that spike around timestamp 245,000 might be an example.....
I also notice the RPMs drop away during middle of ramping and how the slope of the PWM ramp changes as well. Need to figure out what is happening there.
And it appeas have been sucessuff!