|
Post by vulbas on Aug 28, 2014 17:54:30 GMT
hello Thanos, think you can do a software to adjust the PID ?. with a graph. that would be great.
|
|
|
Post by tronicgr on Aug 28, 2014 18:03:36 GMT
hello Thanos, think you can do a software to adjust the PID ?. with a graph. that would be great. Maybe will happen soon. As I mention elsewhere, I will add serial commands for the parameters of the AMC1280USB, so I will be giving this feedback ability as well to be used with PID calibration. For now it has to be done manually, with the help of a cheap oscilloscope perhaps:
|
|
|
Post by riton39 on Aug 31, 2014 18:20:22 GMT
I am interested in this oscilloscope, can you give us some information, a tutorial? it would help me see the sound signals, PID ....
|
|
|
Post by fabi on Sept 1, 2014 0:18:47 GMT
After many many testing session, I came to the conclusion that AMC1280USB should measures the oscillation period (Pu) and print it out for the user. This eliminates the need for an oscilloscope which is not really helpful if the platform is under load (to much noise on the analog signals for a small oscillation). Note-1: A platform under load still works fine with "Step-Changes" even if kp is much too high. So what is needed for a proper PID evaluation (and calculation) is the oscillation period (Pu). Note-2: The platform does not oscillate immediately if there are “worm gears”. I've wrote a software sending movement signals to AMC1280USB so it works also with “worm gears”: See bottom of 6dofabi.blogspot.com/2014/07/controller-fine-tuning-with-schneidy.html
|
|
|
Post by riton39 on Sept 1, 2014 7:03:49 GMT
For test AMC64 Acceleration and Deceleration ramp time from 0.0 seconds •Thanos board: ◦P = 3 ◦I = 1 ◦D = 1 ◦Motor Max = 100 % ◦Motor Min = 1 %
I think the acceleration and decceleration is an additional difficulty. it adds delays and problems adjusting PID?
|
|
|
Post by fabi on Sept 1, 2014 14:40:46 GMT
Assumption you mean by "acceleration and deceleration" the acceleration and deceleration ramps which can be configured on the VDF. If yes, set it to 0.5 seconds, that is fine for a setup with 30 RPM.
In addition: 1. If you have "worm gears", you need to evaluate the Kp by doing a slow motion with the board, otherwise it takes much more P until it starts stepping. 2. From the video I have some concerns about the cable shields, have you validated if the "feedback sensors" are delivering a noise-less signal?
|
|
|
Post by riton39 on Sept 1, 2014 15:33:04 GMT
yes I'm talking about the ramp of frequency. increase the ramp gives the delay. more there late and the PID must have a hard time reacting properly? I do not see how PId loop may work fine if you add the delay ramp in the drives. in a system with servo motor, it does not have a ramp. PID loop should do the job. Also with a ramp, the reaction on the small effect is very bad. ramp removes much effect, because the reaction time is too slow. I do a lot of testing before my 2 DOF, and the best results were with 0.0 Ramp I do not think that putting the ramp improves operation. By cons, so the simulator is more stable in the middle! but it is because we degrade reaction time and therefore the simulator does not move much. if you watch my videos you will see that it is very reactive to small effects. (I'm testing for several years !!) but if I put the ramp, no more vibration. I do not think I am wrong, but I'll try something else and see what happens. the ramp, it's time to go from 0 to the max, so it's late, loss of reaction time. PID is also an optimization of the position is reached, thus a sort of adjustable time between reaction velocity and position.
add the two is obviously a big loss.
Sorry for my Google english
French oui je parle de la rampe des variateurs de fréquence. augmenter la rampe donne du retard. plus il y a de retard et plus le PID doit avoir du mal a réagir correctement ? je ne vois pas comment la boucle PId peut fonctionner correctement si on ajoute du retard avec la rampe dans les variateurs. dans un systéme avec Servo moteur, il n 'y a pas de rampe . la boucle PID doit faire le travail. De plus avec une rampe, la réaction sur les petit effet est trés mauvaise. la rampe supprime beaucoup d'effet, car le temps de réaction est bien trop lent . j'ai fais beaucoup d'essais avant avec mon 2 DOf, et les meilleurs résultats étaient avec 0.0 de rampe je ne pense pas que mettre de la rampe améliore le fonctionnement. Par contre, oui le simulateur est plus stable au centre ! mais c'est parce que nous dégradons le temps de réaction, et donc le simulateur ne bouge plus autant. si tu regarde mes vidéos tu verra que c'est trés réactif sur les petits effets.( j'ai fais des essais pendant plusieurs années !! ) mais si je mettais de la rampe, plus aucune vibration. Je ne pense pas me tromper, mais je veux bien tester d'autre chose pour voir ce que ça donne. la rampe, c'est le temps pour passer de 0 à la la fréquence max, c'est donc du retard, une perte en temps de réaction. le PID c'est aussi une optimisation de l'atteinte de la position, donc une sorte de temps de réaction modulable entre position et vitesse. ajouter les deux est forcément une grosse perte.
|
|
|
Post by riton39 on Sept 1, 2014 15:45:35 GMT
I do not have an oscilloscope to check the noise on the potentiometer wiper. But if you watch the video, the values of the potentiometer on the AMC does not move. the potentiometer cable is shielded the shield on the motor cable and around the potentiometer is to test and see if it improves things. it did not improve things, I'm not sure there is a noise problem. but of course I still have a doubt ... I also raise a question about the dead zone in the center. with a dead area in the center, the engine is not "measure" position by the sensor. the charge carries the engine outside the dead zone, the sensor gives the position ..... and here we go again. I am not sure that the dead zone be necessary in the center.
je n'ai pas d'oscilloscope pour vérifier le bruit sur le curseur du potentiomètre. Mais si tu regarde la vidéo, les valeurs du potentiomètre sur l'AMC ne bouge pas. le câble du potentiomètre est blindé le blindage sur le câble moteur et autours des potentiomètre, c'est pour tester et voir si ça améliore les choses. ça n'a pas amélioré les choses, je ne suis pas sur qu'il y est un problème de bruit . mais bien sur j'ai toujours un doute... j'ai soulever aussi une interrogation concernant la zone morte au centre . avec une zone morte au centre , le moteur n'est plus "mesuré" en position par le capteur. la charge emporte le moteur en dehors de la zone morte, le capteur redonne la position..... et ça recommence. je ne suis pas sur que la zone morte soit nécessaire au centre.
|
|
|
Post by riton39 on Sept 1, 2014 15:55:40 GMT
In my 2 DOf, same motor. just motor AC, no BTK
it's always difficult with a video, but look at the track vibrations !! it's AC motor !! vibrations are even better in real. you can watch the movement of the steering wheel mark.
c'est toujours difficile avec une vidéo , mais regarde les vibrations de piste !! c'est des moteur AC !! les vibrations sont encore meilleur en vrais. tu peux regarder le mouvement du volant en repère
|
|
|
Post by vicpopo on Sept 1, 2014 19:19:23 GMT
Hi Riton ,
It's good to launch the discussion here. When you talk about your tests with your other simulator could you give to all which controller did you use? It was with the amc644usb ?
|
|
|
Post by riton39 on Sept 1, 2014 19:30:14 GMT
yes, for the moment i have just this card. thanos work on my AMC 128 !
|
|
|
Post by vicpopo on Sept 1, 2014 20:01:00 GMT
Thanks for the answer ;-)
|
|
|
Post by riton39 on Sept 4, 2014 9:18:00 GMT
|
|
|
Post by tronicgr on Sept 4, 2014 17:52:00 GMT
Yes, that is the one I use on my videos.
Its easier to get one from Ebay, they are faster in delivery. Look for the cheaper DSO201 you can find...
Thanos
|
|
|
Post by riton39 on Sept 4, 2014 18:54:28 GMT
I found! thank you;)
|
|