#robotics Logs

Jan 18 2021

03:55 AM flav0r: HIVE-MIND !
09:34 PM rue_mohr: D = (Vp * Kp)/16 - (Vd * Kd)/4 + (Vi / Ki);
09:34 PM rue_mohr: integral is evil
09:34 PM rue_mohr: http://ruemohr.org/~ircjunk/tempimage/foo102.png <--
09:35 PM rue_mohr: the pink line
09:35 PM rue_mohr: it naturally sits, at 0Hz, at +50db
09:43 PM rue_mohr: Their all dead
09:43 PM rue_mohr: their dead dave, their all dead
09:43 PM rue_mohr: they are all dead Dave
09:48 PM Tom_L: there's a colored X in that pic
09:48 PM rue_mohr: one half of it is I the other D
09:49 PM rue_mohr: the dark blue hammoc is the result,
09:49 PM Tom_L: and they appear to be holding up a saucer
09:49 PM rue_mohr: and the yellow line is P
09:49 PM rue_mohr: and thats PID
09:49 PM rue_mohr: the yellow line is D, it uaually tops out at 0db
09:50 PM rue_mohr: the pink is I, it usually bottoms out at 0db
09:50 PM rue_mohr: so, in there, Ki is like 1/512
09:52 PM rue_mohr: ok, well, I did up a PID controller
09:52 PM rue_mohr: used some analog inputs for Kp Ki Kd so I can play with it
09:53 PM rue_mohr: tho, last time my hardware almost did in my motor driver
09:53 PM rue_mohr: well it killed one
09:53 PM Tom_L: so how do you get 2 PID loops to work hand in hand?
09:54 PM rue_mohr: how do you mean?
09:54 PM Tom_L: with quick response and no overshoot
09:54 PM rue_mohr: with each other?
09:54 PM rue_mohr: oh tuning?
09:54 PM rue_mohr: I wrote these instructions once upon a time
09:54 PM rue_mohr: /* these values were tuned for a HP servomotor
09:54 PM rue_mohr: Set Ki and Kd to 0
09:54 PM rue_mohr: Dial up Kp as far as you can with it being stable
09:54 PM rue_mohr: Dial up kd to slow rate of error correction of Kp
09:54 PM rue_mohr: Dial up both to stable limit
09:54 PM rue_mohr: Adjust Ki for correction (low values)
09:54 PM Tom_L: my vfd is using an internal PID and i'm using the encoder back to lcnc for my analog input to it
09:54 PM rue_mohr: IRATE will control speed of integral correction
09:54 PM rue_mohr: */
09:54 PM Tom_L: also on a pid loop
09:55 PM Tom_L: i'm skeptical to mess with any parameters on the vfd
09:55 PM rue_mohr: if you have two pid loops I guess the results multiply
09:55 PM rue_mohr: I generally think it would end up pissed off
09:56 PM Tom_L: they settle pretty quick but they do bounce around a little bit
09:56 PM Tom_L: overall it's way better than what i had
09:56 PM rue_mohr: if you do the motor as proportionate and let lcnc be PID it might be ok
09:58 PM rue_mohr: hmm I feel like, at the moment, my life is a person who's been blindfolded and handed a donkey tail on a pin, spinning around trying for clues on which way the donkey might be
10:10 PM Tom_L: that way
10:10 PM Tom_L: no no... the other way
10:11 PM rue_mohr: "Why is someone screaming? did I miss?"
10:24 PM rue_mohr: hmm, how fast is that stm32 again?
10:25 PM rue_mohr: 72Mhz
10:27 PM rue_mohr: and it a 10 bit adc?
10:28 PM rue_mohr: nope, 12