Closed GoogleCodeExporter closed 9 years ago
This issue also affects APM. I have seen it in Servo Roll P and Servo Yaw P.
Original comment by d...@w1ngz.net
on 6 Jun 2011 at 4:09
Original comment by Meee...@gmail.com
on 7 Jun 2011 at 12:20
this is a rounding problem on the apm side, and there is nothing i can do about
it, apart from maybe hide it from you. the saved value on the apm is the one on
the left, the up down box's will always round to 3 sig fig, this is just to
simplify things
Original comment by Meee...@gmail.com
on 7 Jun 2011 at 12:52
Is the saved value really 0.4394531, or is that just a manifestation of it
stored as/sent over mavlink as a float?
Maybe there needs to be rounding on both sides of the link. Both in APM and in
the mission planner.
If you send 0.44, it should be "used" as 0.44. Not 0.4394531 (which is likely
caused because of float precision differences.)
Just my 2 cents on the subject.
chris.
Original comment by ch...@chrisandtennille.com
on 7 Jun 2011 at 2:39
I agree. If we reset PID values to default values and use MP, we see correct
values.
If we save from MP, values get wrong. So it is not how it is stored or
displayed, but how it is saved. Either it gets casted as a float in Arduino, or
it is casted wrong in .NET.
Also if you click on arrows when 0.4394531 is displyed, the value goes up to
4.3 or something really wrong....
Original comment by emile.ca...@gmail.com
on 7 Jun 2011 at 4:17
emile, what os are you using. in the up down box's mine only ever shows 3 sig
fig, apart from if i type something else into it.
Original comment by Meee...@gmail.com
on 8 Jun 2011 at 3:04
ok this error is caused by the conversion from a 32bit float to a 16 bit float.
The planner sends as a 32 bit float, which is required by mavlink. on the apm
the 32 > 16 bit conversion is done and that is where the loss is occuring.
the error is only 0.000469, and not really worth worrying about.
Original comment by Meee...@gmail.com
on 8 Jun 2011 at 3:52
Original comment by Meee...@gmail.com
on 15 Jun 2011 at 3:05
Original issue reported on code.google.com by
emile.ca...@gmail.com
on 6 Jun 2011 at 6:06Attachments: