|
|
Post by tronicgr on Oct 22, 2014 19:25:08 GMT
I saw that the behavior is different between the high, low and center position.
sometimes the PID is correct bottom-center and bad from top to bottom or vice versa ....
That is because you are trying to convert rotational speed to linear speed. The Rotation speed is constant, but when it goes near the edges (up and down) the Linear speed is reduced, as well the load is getting less on the motor. The 6DOF software fixes difference in speed by applying faster speed for the edges and lower speed for the center using some math for rotation speed calculation (COS,SIN etc)... For your 3DOF you use raw position data from the games that do not take that in consideration. I might have to integrate that kind of math in the firmware eventually for accurate distance and speed on such arm-rod actuators. For the AMC1280USB parameters, I made a quick video of the most basic settings that need attention: www.youtube.com/watch?v=qc_SIUJ796UThanks Thanos |
|
|
|
Post by riton39 on Oct 22, 2014 20:00:48 GMT
yes , yes, then the PID is different.
it's another problem again...
|
|
|
|
Post by riton39 on Oct 22, 2014 20:35:09 GMT
all simulators with cam and crank on this principle have this problem.
My 2 DOF simulator has the same problem.
what is different is the length of the cam 24 cm here.
and a speed of 30 rpm.
but my 2 DOF at this speed (100 Hz.)
I did not know you would correct this for 6 DOF.
it's not a big problem in games, but it's more problematic for the PID.
this makes three different thresholds for PID, up, down and center.
it is on these three position that there are large variations.
|
|
|
|
Post by riton39 on Oct 22, 2014 20:55:55 GMT
different tests I'm doing today
|
|
|
|
Post by tronicgr on Oct 22, 2014 21:01:27 GMT
fabi suggested to have variable PID settings according to the position to compensate for this deviation. I'll try to see if I can make the PID adjustable based on the position of the arm on the arc (perhaps using some math) to adapt the PID to the waveform shape of the actuator output and allow same stability level on all positions:  I'll have to experiment a little bit with it... Edit: Found a nice graphical tool to calculate speed and angle relation: www.desmos.com/calculator/eb4arug8ln thanks Thanos
|
|
|
|
Post by tronicgr on Oct 22, 2014 21:13:06 GMT
different tests I'm doing today As you can see, when you increase the "Sensor oversample" to 5, the motors keeps bouncing on the target. That is when you need to decrease P term and decrease the on-line/offline speed of the motor (LCD menu values). Please try with lower values for motor speed... I see you still run the motor speed in 100%... Also it might help decrease the max RPM value on the VFD side...  Thanks Thanos |
|
|
|
Post by riton39 on Oct 23, 2014 17:43:43 GMT
I do not know if it's important, but to have the symmetry of movement, the middle of the cam should be in line with fixing the link.
The motor axis is more in line.
with a cam of 24 cm, the axis is 12 cm
I tried with your advice, I do not see any difference.
moin is fast, but it's the same behavior.
I load the video ....
|
|
|
|
Post by riton39 on Oct 23, 2014 19:29:29 GMT
|
|
|
|
Post by riton39 on Oct 23, 2014 20:40:32 GMT
|
|
|
|
Post by riton39 on Oct 26, 2014 11:13:25 GMT
I have a problem with the drive, even without load, I have a big floating cam.
the time that the engine react.
I do not know from where it comes. |
|
|
|
Post by riton39 on Oct 27, 2014 14:05:08 GMT
after a few days and many hours ...
I have completely taken over the drive settings.
I decrypted every setting, and the result is:
I do not know if this is the best |
|
