Replies to This Discussion

Permalink Reply by sam d on July 24, 2011 at 12:32am

In the IMU test, it should definitely not change that much when not moving the quad (at least while not moving the quad/APM), but it is normal behavior when moving the quad/APM shield.

At least, thats my case, just after leveling (i did a setup->level in the CLI (command line interface)), i get -5 to 5 for the first and second column, and constant 338xx on the third column WITHOUT moving my quad. Obviously, if i move my quad, those numbers start jumping to different positive/negative values depending how i move my quad.

Have you tried the hand test to feel how the quad reacts to movement and see if it wants to stay stable? i found that test very valuable for me while i was still learning and building my quad.

sam

• ▶ Reply

Permalink Reply by Michael Brown on July 24, 2011 at 5:08am

Thanks for the reply. When I can today, I will do the level command and check the IMU again just to verify the output data. I am not familiar with the hand test. I would think that it is done with a little throttle with no props, correct?

• ▶ Reply

DeveloperPermalink Reply by jasonshort on July 24, 2011 at 10:03am

#4 When you power on quad the IMU test is run. That's when you need to hold it still, not when loading code.

The imu test is pitch, roll, and yaw in degrees * 100. When the quad is on the table, you should see low values that are pretty stable. Yaw could be anything 0 - 36000.

Have you run "level" ? That's the command to tell the copter what is flat and upright. Without level, it won't fly well.

The hand test is with the props, just be careful. give it enough gas to not fly on it's own. It should want to recover immediately as you tip each axis. tip: don't hold the radio in your hand, leave it on a table and you won't bump the throttle.

If the copter wobbles or spins up the wrong motor, you have a config issue and we can solve that.

• ▶ Reply

Permalink Reply by Michael Brown on July 24, 2011 at 3:14pm

Thanks. I will look into that tonight. I use a 6' USB cable so it sits on the floor flat all the time but I will pay attention to what you say. I will let you all know what happens. Thanks again.

• ▶ Reply

Permalink Reply by Michael Brown on July 24, 2011 at 4:35pm

OK, I ran level and got IMU

Gyro offsets: -8.31, -5.00, 14.00

Accel offsets: 8.14, -9.03, -7.30

I ran the IMU test again. The numbers change incrementally by a few digits at a time.

Col 1 goes from -8 to -57 then back down to 0 then to +38 and back down to -33 over and over.

Col 2 goes rom -16 to +12 over and over.

Col 3 goes from 907 to 1007.

I then rolled the quad by hand in all directions.

Col 1 went from -17,983 to +17,798.

Col 2 -2416 to 6790.

Col 3 235 to 33,340.

I put the quad back on the floor and saw the numbers change to:

Col 1 0 to 3

Col 2 -2 to -17

Col 3 520 to 556. I then repeated the same sequence and the numbers did not stabilize as well the second time. It sounds like what is wrong is almost becoming clear, lol.

I have not tried the hand test because seeing these type numbers, I am not sure it will be meaningful but I will try later.

Any insightful thoughts?

• ▶ Reply

Permalink Reply by sam d on July 24, 2011 at 5:25pm

Well, it looks about right, according to jason, those nombers means degrees * 100, so that means that on your first test, the IMU was reading from -.57 degrees to .38 degrees, which in my opinion is fairly acceptable for a non-moving quad. The numbers on your second test mean you pitched the quad about 17 degrees to both directions and you rolled the quad about 67 degrees. When you placed your quad back on the floor (your third experiment), the IMU reported to be at 0 degrees - 0.03 degrees pitch, from -0.02 to -0.17 degrees roll which i see its pretty accurate.

rememeber that those numbers depend on the moving of the APM and how level it is, for example, they might vary a little if they are placed on the floor or in a table, depending how level is the table.

Also remember that those numbers are degrees * 100, so changing from 0 to 100 only means 1 degree change which is not that significant and due to the nature of the sensors, its very rare to have a deadly stable number which does not change at all.

try only moving rolling/pitching/yaw one at a time and see. Only one column should change significantly.

In my opinion, dont worry too much about those numbers, if you see good tendencies on those numbers, go and try the hand test, the error might be in other place.

• ▶ Reply

Permalink Reply by Michael Brown on July 24, 2011 at 5:25pm

OK, I just finished doing the hand test. Whoa, that was exciting. The forward backward pitch, I could hear the motors fighting my change. The yaw CCW and CW I could feel the motors fighting back. Doing the roll left and right I did not feel too much fighting back of the motors. Hmmmm, is that what the response should be? It was hard to tell which motor was responding other than feeling it fighting back, in fact the roll was way different that pitch and yaw which seemed to want to twist back out of my hand. Being as hot as it has been, it was a nice breeze, lol.

• ▶ Reply

Permalink Reply by sam d on July 24, 2011 at 5:45pm

I dont know if thats normal behavior. But, you could try the following.

Does the roll related propellers throw air downwards?

You could also try recalibrating the roll related ESC's manually as described in the second ESC calibration procedure in AC2 manual.

At last, i might suspect of the motor's bullet connectors (if you have them). I had problems with the bullet connectors on 2 of my motors (they would even not start some times). I had to directly solder 2 of my motors to the ESC's and the problem disappeared. I'm not saying they are wrong, and i would not remove em until im completely sure its nothing else and/or im desperate ;p.

let us know how it goes.
sam

• ▶ Reply

Permalink Reply by Michael Brown on July 26, 2011 at 5:05pm

Thanks Sam for the thoughts. I decided to calibrate all of the ESC's manually. One of the roll ESC's would not calibrate. It would not give the correct tones. I did the other three and came back to that one and it calibrated correctly that time. That is telling me that things aren't right with it. There are no bullet connectors so that part does not come into play. I would guess the motor has nothing to do with it either as the ESC should calibrate with or without a motor. Am I headed in the right direction?

And BTW, the props throw air downward as they should.

• ▶ Reply

JDronesPermalink Reply by Jani Hirvinen on July 26, 2011 at 8:22pm

Mikey, have you changed any of your PID settings? You could try following PID settings on your stabilize modes.

 

P: 2.3 - 2.4
I : 0.015
INT: 40

 

Few people has reported that these will make ArduCopter with bigger motors a lot more stable than current default settings. After you upload latest software all earlier settings are wiped out too so your ArduCopter is back on smaller motor/prop combination setting.

 

• ▶ Reply

Permalink Reply by Michael Brown on July 27, 2011 at 4:54am

Jani, thanks. I will try those settings tonight. As I said, I did get the ESC to calibrate but I accidentally turned off the brake setting. What are the settings of the ESC's so I can turn it back on. Also, I came across some 8" props. It wouldn't hurt to use those while I get things straightened out or should I stick with the 12" props.

• ▶ Reply

Permalink Reply by Michael Brown on July 27, 2011 at 12:59pm

Jani, the quad flies pretty good for once. It is finally a good day in West Palm Beach. I am gathering all of my problems stemmed from the one ESC not calibrating correctly, heavy sigh. I was flying it with 8" props and I haven't changed the PID settings. I will do that in a little while the battery is charging. I would still like to replace that ESC and get some spares. I also would like to know the settings for the ESC so I can turn the brake back on.

I was getting ready to change the PID values.

Old: P=4.5, I=0.025 D=0 INT=50 until I get to the YAW values, they were and still are P=4.5 (I can change that, I=0.1 (I will leave it alone) but the INT is 1500, and changing that one to a 40 does not seem like a good idea. Please let me know your thoughts on that one.

 

Thanks for the help and patience everyone. :)

 

Jani, would you e-mail me at mikey_hd@yahoo.com so we get the parts straightened out?

• ▶ Reply