two large ACC spikes

Static, Dynamic, Log File Analysis, Troubleshooting

Re: two large ACC spikes

Postby 13brv3 » Sun Feb 24, 2013 10:25 pm

Things seem to be going OK, but I won't know until it flies, or until one of the steps fails to work. Step 1 had a definite preference for one of the 3 static files, so I used that one only.

For step 2, I again tried using all 3 of the static files, but it wouldn't complete the step. From that point on, I used only the one static file that step 1 preferred.

Step 3 took 4 iterations to stop dropping significantly.

Step 4a is running now with the MAE dropping very slowly, but steadily at 390 loops and counting.

Gotta love that i7 CPU :mrgreen:

Rusty
13brv3
 
Posts: 152
Joined: Wed Oct 24, 2012 1:10 pm

Re: two large ACC spikes

Postby Kisssys » Mon Feb 25, 2013 12:06 am

The MAE will start to drop very slowly but I would let it run until the last column is out beyond 6 digits and preferably farther. The last column represents how much that value changed in the last pass. The MAE might not show much change after a while but as long as those numbers are changing your getting closer to a better solution.

As you can see from these GYO numbers which ran about 50 times the Scaling is out to 6 digits, it will take it much longer to find a solution that gets all of them out beyond 6 digits.

#define IMU_GYO_SCAL_X +0.114818316600 // 0.000000099987 +0.000002177513
#define IMU_GYO_SCAL_Y +0.114794840264 // 0.000000099978 +0.000002375610
#define IMU_GYO_SCAL_Z +0.114713113928 // 0.000000100045 -0.000001676895
#define IMU_GYO_ALGN_XY -0.008468743372 // 0.000106468563 -0.000165143456
#define IMU_GYO_ALGN_XZ -0.009401762453 // 0.000103405967 -0.000248725281
#define IMU_GYO_ALGN_YX +0.005491711608 // 0.000102242106 -0.000191408043
#define IMU_GYO_ALGN_YZ +0.002302748325 // 0.000101710008 -0.000198586691
#define IMU_GYO_ALGN_ZX +0.022557016319 // 0.000100201682 -0.000143889368
#define IMU_GYO_ALGN_ZY -0.011996301887 // 0.000100549670 -0.000252472692

This was after a long run on the MAGS and the worst is out to -8

#define IMU_MAG_SCAL_X +0.187317803117 // 0.000000542757 +0.000000000911
#define IMU_MAG_SCAL_Y +0.196977559400 // 0.000000233019 -0.000000000834
#define IMU_MAG_SCAL_Z +0.185440303827 // 0.000000334762 -0.000000000215
#define IMU_MAG_ALGN_XY -0.015016015444 // 0.000096253583 -0.000000001184
#define IMU_MAG_ALGN_XZ -0.034615224976 // 0.000086688554 -0.000000013672
#define IMU_MAG_ALGN_YX -0.049949212878 // 0.000077322329 -0.000000005929
#define IMU_MAG_ALGN_YZ +0.025972368672 // 0.000076698419 +0.000000003710
#define IMU_MAG_ALGN_ZX -0.053902537844 // 0.000058746766 +0.000000000190
#define IMU_MAG_ALGN_ZY +0.005963703224 // 0.000097821243 +0.000000007377
Steve
Kisssys
Kisssys
 
Posts: 1340
Joined: Sat Jun 23, 2012 9:23 pm

Re: two large ACC spikes

Postby 13brv3 » Tue Feb 26, 2013 2:01 am

Just a final update on this. I got through the cals without any further issues, and uploaded the file to the board. I've only done a quick flight inside due to weather and darkness, but it seems pretty stable. I'll get it outside in a couple days and see how it behaves.

Thanks,
Rusty
13brv3
 
Posts: 152
Joined: Wed Oct 24, 2012 1:10 pm

Re: two large ACC spikes

Postby LPR » Tue Feb 26, 2013 2:27 am

Rusty

I have built three AQ copters with your hardware and to my great surprise they all fly great with default PIDs except Tilt Rate D usually has to be lowered to between 4000 to 5800. The first AQ copter was an octo that weighs 9 lbs and the second is a quad with 15" props and weighs 4 lbs and the third is a small quad with 9" prop and weighs 3 lbs. If your AQ copter can hover without oscillations with default PID then just lower the Tilt Rate D till you get a little oscillations and raise it till the oscillation stop.
Then test RTH. Your copter should stop right at the home point. This method has worked well for me.

Larry
LPR
 
Posts: 1323
Joined: Tue Jun 26, 2012 1:29 pm
Location: MN, USA

Re: two large ACC spikes

Postby 13brv3 » Tue Feb 26, 2013 3:26 am

Larry,

The defaults felt pretty darn good, so I don't imagine much change will be needed. I'm on R56 now, and I'll probably move up to the latest before doing too much testing. This is just a fairly small quad, but I've had the prototype UAP2 Y6 frame just waiting for the second board. Hopefully, I'll have two AQs flying soon.

Cheers,
Rusty
13brv3
 
Posts: 152
Joined: Wed Oct 24, 2012 1:10 pm

Re: two large ACC spikes

Postby LPR » Tue Feb 26, 2013 5:04 am

Rusty

Be aware that if I did not lower the Tilt rate PID my AQ copter would do a oscillation of death when using RTH. To be safe just do some short RTH and then go longer a little at a time. If there is going to be a problem it shows up first as overshooting the Home position. I know your a dealer and props are cheap for you but not free. If your AQ overshoots by very much then it will likely sometimes go into a rapid oscillation while atempting a RTH and crash.
For me learning this with an octo was not a good thing.

Larry
LPR
 
Posts: 1323
Joined: Tue Jun 26, 2012 1:29 pm
Location: MN, USA

Re: two large ACC spikes

Postby 13brv3 » Tue Feb 26, 2013 1:39 pm

I seem to recall tilt rate D being around 4500, but now that I think about it, these may not be defaults since Steve actually flew the board after repairing it. I'd better make note of the settings before updating the software, assuming that wipes the settings. I've got a lot of tinkering to do for sure, but it should be fun.

Rusty
13brv3
 
Posts: 152
Joined: Wed Oct 24, 2012 1:10 pm

Re: two large ACC spikes

Postby LPR » Tue Feb 26, 2013 2:42 pm

Rusty

I use the print screen button to copy all the pages of setting and than compare with new setting and default PIDs that I get with new firmware.

I sure the Steve gave you the PIDs he uses on his quad.

Larry
LPR
 
Posts: 1323
Joined: Tue Jun 26, 2012 1:29 pm
Location: MN, USA

Previous

Return to Calibration

Who is online

Users browsing this forum: No registered users and 6 guests

cron