Also, the previous linear actuator locked the panel tilt in place whereas the clutch on the drive shaft here is slipping due to wind gradient (wind shear?). When I modeled the fluid dynamics in my head, this didn't happen but it appears that the wind speed at the top of the panel can be significantly higher than at the bottom. This is why we test things.
This was so bad on my first ride that I had trouble getting back home but it was largely because the tilt motor controller was stuck on "follow the sun" mode. I enabled "keep panel level" mode on later rides and the handling was just about manageable even with strong wind but I found I hated having to pick between full range of travel and no travel.
I had the same thought but looking at available wind sensors I didn't see anything that looked small enough and robust enough. I picked up an accelerometer for the Arduino but I'm not confident I'll be able to differentiate between acceptable and unacceptable movement in software. Reaction time might be tricky, as you say. With manual control, I can anticipate but it's kind of taxing to have to think about all the time.
Interesting. I hadn't considered that. Unfortunately, the tilt motor / gearbox / clutch assembly is sized to fit inside the boom tube and to handle a perfectly balanced panel so I don't think I could do that without major re-work. If it's windy enough for the trailer to get pushed around this much, I'm resigned to taking the solar performance hit and leveling out the panel.Cowardlyduck wrote: ↑Jun 06 2020 9:59pmI think a simple solution would be to have the panel able to shift sideways so that the pivot point is further forward into the wind direction. That way the wind will help stabilise the panel rather than catch it like above. Kind of like the trailing edge of a arrow.
The solution I'll be ready to road test tomorrow is relatively low tech: use a handlebar-mounted potentiometer to trim the panel on the fly by setting the maximum tilt angle. One end is full travel and the other end is no travel (horizontal only). This way I can let the sun tracking algorithm follow the sun as much as current conditions will allow based on how the bike is handling. If it works with a manual input, I could try to automate it later.
I might also try adding weight to the trailer using stuff I would otherwise be carrying on the bike anyway, like extra water bottles.