search

rdmenezes / qmast

Automatically exported from code.google.com/p/qmast
0 stars 0 forks source link

Troubleshoot sailStraight #17

Closed GoogleCodeExporter closed 9 years ago

GoogleCodeExporter commented 9 years ago 
Presently theres a problem with sailStraight function in alpha5; stops 
responding after a bit - we dont know if this is a pololu or compass data 
problem.

Original issue reported on code.google.com by cdmblair@gmail.com on 1 Mar 2011 at 2:49

GoogleCodeExporter commented 9 years ago

Original comment by cdmblair@gmail.com on 1 Mar 2011 at 2:49

GoogleCodeExporter commented 9 years ago 
Possible causes:
- pololu isnt working properly. This was tested with a basic "turn left, delay, 
turn right, delay" program for pololu, and it seems to time out after a few 
minutes (red light on, orange light flashing... or the opposite, I forget which 
one flashes). This reduces the probability that Pololu is acting up
- compass serial data is being corrupted and no data is being received after a 
few runs. This is likely as presently the success of getting data from the 
compass is highly reliant on delays. Switching to sample mode may fix this, 
also I believe I've figured out why serial data roll-over was failing (this 
should also greatly improve data success, as slow data can still be processed).

Original comment by cdmblair@gmail.com on 3 Mar 2011 at 5:09

GoogleCodeExporter commented 9 years ago 
Could also be how the rudder direction is set:
- calculates how far to turn, and makes it positive or negative, maximum 180 
degrees either way
- has a dead band of 10 degrees either side of 0 change, have to be outside 
this dead band to actually turn the rudder
- when the rudder is turned, it will only turn to 45 degrees either side of 
maximum. This is only 1/4 of the 180 degree input range; anything larger than a 
45 degree turn will appear as a 45 degree turn
- with the 10 degree deadband, this only leaves 35 degrees either side of 0 in 
which the rudder will respond and turn. Hence, this could be why we dont see 
the motor responding.
- to fix this, put a proportional scaling constant in the call to turning the 
rudder, such that 180 degrees maps to 45 and -180 maps to -45. This constant 
will then be later tuned for proper proportionality for proper adjusting

Original comment by cdmblair@gmail.com on 3 Mar 2011 at 8:33

GoogleCodeExporter commented 9 years ago 
Sorted out what the directions should be to turn the rudder, and fixed the 
'rudder' wheel linkage again; this is written on the roller-boat and commented 
in code.

BUT the compass seems to be broken :( There's a note taped to the whiteboard 
that has the angles it is reading in sample mode in the workroom. So, move onto 
sail control testing for now.

Original comment by cdmblair@gmail.com on 8 Mar 2011 at 2:07

GoogleCodeExporter commented 9 years ago 
Compass isnt broken; the issue was that there is a huge magnetic distubance in 
the bay workroom.

Switching to sample mode was a huge improvement.

Sailstraight is working fine now. Not sure what caused it to be not working 
before. Deadband could be adjusted if necessary after water-testing.

Original comment by cdmblair@gmail.com on 10 Apr 2011 at 9:23