I have had good luck with 2.6 - here is a typical mission
I then loaded 2.7, did and erase, and left all params at default (attached both param files - not any huge differences). I flew in stable, ah, and loiter and it seemed good - very stable, loiter seemed actively holding it within a couple meters. I thought it would be good to try my last mission, so I loaded it and was vvery surprised as it flew very erratically, wildly searching for waypoints-
I finally aborted before the end, it was flying in a very scary way. I do not know what happened - maybe the slight pid default changes did this, but it seemed to be more than pid tuning. Before I took off I did some alt hold and it was good. After I tried some loiter and got bad circling.
I am curious what happened - hopfully Dan's visualizer program might help.
Replies
I had similar weird results on APM2 with 2.7.1:
My speed set to 2.1 m/s, and PIDs reduced as per advice here: NAV_LAT_D,0
NAV_LAT_I,0.085
NAV_LAT_IMAX,1800
NAV_LAT_P,1.855
NAV_LON_D,0
NAV_LON_I,0.085
NAV_LON_IMAX,1800
NAV_LON_P,1.855
What else should I try?
Hi John,
I have similar problem and try investigating it. Found this line in Log.pde
DataFlash.WriteInt(0); // 7 (this used to be compass.heading)
So by some reason 2.7 won't log Mag Heading anymore. Wondering why.
But the reason for erratic behaviour is somewhere else. Did you have luck finding reasons?
John,
I too may be having the same issue as you - I just put a brand new APM2 up in the air with an external ublox on a 3DR hexa frame w/880 motors, and (much to my amazement) it hovered flawlessly at 2-3 feet off the ground, not needing any throttle adjustments in stable mode, which has never happened before - likely a combo of getting away from APM1, and the new code..
I reference the above, because it shows that obviously I'm having good luck with all the sensors (I have sonar and optical as well, all work) and yet when it was up, it would randomly drift in yaw, 50/50 or so cw and ccw - not a fast spin, maybe 10-15 seconds for a full rotation if I didn't correct and put it back in its' place. MP logged the tlog mag info properly - so it looks like it's purely a logging problem with the firmware on the copter..
Misc info:
New APM2.0 w/ublox LEA 6 GPS (amazingly better)
All extra sensors except for airspeed
3DR hexa frame
880kv motors - 11/4.7 props
Edit: I do believe it's time for a bug report - I'm submitting one now..
John,
Had a moment to review your logs. I think you need to lower your Nav_Rate_P value. I can see the quad ramp up speed too quickly and then drop back too low. This tells me your rate_p is too high for the latency in your GPS. Don't mind the jaggedness in the GPS so much. Your high acceleration is the main problem.
There is another param called tilt_comp that is set to 54. It may also be a little high for you, but I would prefer you lowered Nav_Rate_P first and let me know how it goes.
Also, your magnetometer may have been reset during the update and you could have lost your offset. Once you fly around a bit you'll get them back.
Best,
Jason
Here is one leg of your run compared to 2.6, 2.7, and the simulator for 2.7. You can see the Lon speed oscillate in 2.6 and much more in 2.7. A patch was applied in 2.7 to give the copter more authority in the air and stop relying on the Nav_Rate_I term to overcome air resistance. This patch has given you too much authority, and you can now turn down Nav_Rate_P.
Try to disable the Autodeclination, then reduce NAV_P to 1.8, NAV_I to 0.08 and the speed 3m/sec.
With 2.7 the speed was increased (now is real), so you always have to drop a few parameters to obtain the same (or better) performance of 2.6.
I can't see magnetic heading values in your 2.7 log. Did you have compass connected, calibrated and working?
Jani
Did you save your PID values from your old version? Might be interesting to try those, see if it's tuning.
Dan
Oops - here are param files
param 2.6 good 7-23-12.param
param 2.7 BAD 7-23-12.param
Try lowering your waypoint speed to something like 2.5 or 3 m/s (actual internal units are in cm/s).
Jason