The work those folks do is def a reason why many users are able to fix their aircraft. I am not trying to downplay their work, but merely state that most of the time its just them re-iterating what can already be found in the documentation. Imagine how much more bloated the forum would get if we are now helping folks create a physics-based model of their aircraft (when it really isn’t need for most users).
1 Like
I suppose the additional inputs could lead to more questions on the forums. I would think making them optional would prevent forum bloat. the goal would be for the additional inputs to lead to be better tuning estimates.
I agree that as an optional step it could be useful for some niche projects - but there are already software packages that can help you get PID outputs out of a physics model. Those packages are much better maintained and developed than something we could dedicate our time to.
Here is some documentation on it where you can connect a physics model (they refer to it as Flight Dynamics Model) to ArduCopter SITL and conduct tuning and whatnot.
interesting, using sitl might be an easy way to get estimates. i might look into this. @xfacta have you used sitl to get pid value estimates?
I have played around with SITL and reviewing the logs, which I definitely need to do more of, but haven’t been using it seriously in that regard.
With some of the data I’ve collected and trends, it looks like we could suggest some slightly different initial PIDs (compared to defaults) based on prop size. I’m yet to put that theory into any sort of calcs or practice, but I am thinking about it.
There would be a bit of a bell shaped curve, like low PIDs for very small props, increasing PIDs as we move up to mid-sized props, then reduced PIDs for bigger props. Obviously just a starting point for first flights rather than anything fit for continued use.
I have come up with an estimate of HNOTCH frequency in my spreadsheet, based on prop size and it’s tracking fairly close to realistic. Even that one sounds simple, but in reality a few things feed in to the motor RPM and resultant noise frequency. Of course we could start factoring in motor KV, battery voltage, take-off weight and so on, and get a much better figure - but there might be more mileage in updating the doco and build/tuning process.
It’s all food for thought and ongoing…
2 Likes