|
Post by lbendlin on Dec 24, 2018 7:12:22 GMT -8
My solar setup suffers from the usual New England winter blues - not enough insolation to last through 24hrs reliably. A pattern has emerged where the board wakes up the Raspberry Pi around 1pm after the battery is deemed sufficiently full, and then runs through maybe 2am or sometimes even 6am before caving in.
Side note: I think I noticed a very clever behaviour where the load would be switched on when the battery didn't quite reach the upper hysteresis limit but the solar input dropped off long enough for the board to (probably) assume that that's it, nothing more coming, might as well switch on the output. BUT - that may just be my imagination
Nevertheless, it got me thinking. Let's assume the sun comes up at 8, and is sufficiently strong to both support the load and put charge back into the battery. What if the board would watch that input for a while and then say "Ah, looks good, let's risk it and jump out of the hysteresis" and would switch on the load with the hope that the good input continues (or maybe even with the knowledge of the weather forecast).
Right now I basically do the same thing manually by cycling the load switch. But it would be nice to have a way to influence the upper limit of the hysteresis, either via some jumpers, via timer, or - ideally - programmatically.
|
|
|
Post by lbendlin on Feb 10, 2019 12:20:23 GMT -8
I also noticed that the lower hysteresis limit might be impacted by the temperature. I see a cutoff at 3.41V at about 34F versus a cutoff at 3.51V at about 27F. Is there some way to compensate for that so the cutoff is at a predictable voltage (ideally the lower threshold of 3.41V)?
|
|
|
Post by SDL on Feb 10, 2019 15:32:14 GMT -8
Lutz,
If you are using the SunControl Board (which I think you are), you can override the hysteresis shutoff and turn on by using the Grove USB control on the SunControl Board.
BP
|
|
|
Post by lbendlin on Feb 10, 2019 19:03:55 GMT -8
But if I would do that then the output client (the Raspberry Pi) would be the one to monitor the battery voltage (reading the INA3221) and then switching itself off via the USB control. Which requires RTC with alerts, switching the USB on again, checking the battery voltage again and making decisions etc.
I know this will be a fun project to think through, but it's not really related to the temperature drift of the hysteresis loop. Now, if both mechanisms (Grove control and hysteresis) could somehow coexist, that would be ideal... like switching off by either hysteresis or grove control, and switching on ditto (within the allowed parameters)
Not to mention that the documentation for the Suncontrol board promises a lower threshold of 3.3V but the actual threshold is quite a bit higher...
|
|
|
Post by lbendlin on Feb 20, 2019 18:45:57 GMT -8
Maybe I can use your approach but the other way round. I noticed that the power boost circuit is good until at least 2.8V battery voltage. I could play it safe and have the Pi switch itself off (via the USB control) at 3.0V. I can also use the USB output voltage to lock the Grove USB control, and override the hysteresis.
That would result in a scenario where the USB output would be enabled when the hysteresis hits the upper limit. At the same time the USB output then switches the Grove USB control to manual mode, disabling the lower hysteresis point, and leaving it up to the Pi to decide when to cut power. I think I can get at least another 5 hours out of the setup this way.
And yes, this would result in an unclean shutdown of the Pi, but frankly that hasn't been a big issue in the past. Worst case I can re-image the SD card with my regular backup. It's just a few yards away, not thousands of miles.
|
|
|
Post by SDL on Feb 20, 2019 18:51:05 GMT -8
Ha! Yes only a few steps away. John has systems located 3500 miles away! One of those, Project Curacao2 is coming back tomorrow on the plane for him to figure out what went wrong with the Solar Power system. Something died. Lucky there is a ton of data in the database on the Pi in the box so he should be able to figure it out. He'll deploy it for a month or so down in Palm Springs and then take it back to Curacao at the end of July. John has a tough life.
BP
|
|