| View previous topic :: View next topic |
| Author |
Message |
mikewardlow
Joined: 16 Mar 2006
Posts: 42
|
 Posted: Thu Mar 30, 2006 5:33 pm Post subject: How's my thinking |
 |
|
Hows my thinking??
Timer0 [1us] interrupts every 512us using 1:2 prescale...
At each Timer0 int I start an ADC conversion which takes about 26us.....
When the conversion is complete [conversion interrupt] I save the 10 bit value of the feedback pot...,, and allow a pid loop to run.....which will set/clear the cw/ccw and load ccpr1l[pwm] counter.....This looks like it should be ok....
But thinking about this!!!
I have 1/512us = 2khz UPDATE loop running,,,so any changes I make
to cw or ccw will act like 2khz pwm with the length of cw/ccw being the dutycycle as well as direction control.
Therefore making changes to the dutycycle of the hardware CCP is probably only affecting drive current...and I really have a pwm of some period ruuning within a 2khz pwm loop....
Can sombody tell me what this is telling me ??????? |
|
| Back to top |
|
 |
mpthompson
Joined: 02 Jan 2006
Posts: 650
Location: San Carlos, CA
|
| Posted: Thu Mar 30, 2006 6:14 pm Post subject: |
|
|
I followed you on the first half of your post, but I am not sure what the following really means:
| Quote: | I have 1/512us = 2khz UPDATE loop running,,,so any changes I make
to cw or ccw will act like 2khz pwm with the length of cw/ccw being the dutycycle as well as direction control.
Therefore making changes to the dutycycle of the hardware CCP is probably only affecting drive current...and I really have a pwm of some period ruuning within a 2khz pwm loop.... |
Just to recap, you are taking ADC samples of your potentiometer every 512 uS for a 2 KHz sample rate. The ADC position value is fed into your PID function -- which implies it is producing cw/ccw values at a 2 KHz rate as well. The PID function produces your cw/ccw values that are then set in your hardware timer to produce a PWM value of the correct duty cycle. These updates to your timer to alter the PWM duty cycle are occuring at a 2 KHz rate as well.
You will want the timer frequency to be some modest multiple of the 2Hz sampling/PID/update frequency. For example if you use a PWM timer frequency of 10 KHz each ADC sample will result in about 5 PWM cycles of the calculated duty cycle.
Something to remember is that on the AVR (and I assume other architectures as well) the 8 bit timer will actually be set for a 256 multiple of the desired PWM frequency. This is because the 8 bit timer is constantly counting up form 0 to 256. The pwm signal is set when the timer crosses the 0 threshold and reset when timer crosses the dutycyle threshhold giving you the ability to change the dutycyle with an 8 bit resolution.
-Mike |
|
| Back to top |
|
|
mikewardlow
Joined: 16 Mar 2006
Posts: 42
|
| Posted: Sat Apr 08, 2006 9:11 am Post subject: |
|
|
;------------------------------------------------------------------------------
; misc comments:
; currently the CW / CCW signals change at the T2 interrupt rate [100us]
; The pid rate is T0 int [currently 512 US] 1953 hz
; therefore cw/ccw would be same rate??? only chopped up by pwm hdw
; conclude: computing prop err
; set sign to select cw/ccw the set cw/ccw
; if allowed [rather than T2 int]
; this immediately would drive servo
; [ie] prop only control ????
; so How do I get the stopping resolution better???????
;------------------------------------------------------------------------------
I'm having problems utilizing the pwm ccp pic hdw....
I am getting insufficient control/drive...
I must be missing something in Colin's pid8.asm
The big difference I'm trying to use 1 pwm signal anded with direction..
Colin use 2 pwm's 1 cw the other ccw....... |
|
| Back to top |
|
|
mikewardlow
Joined: 16 Mar 2006
Posts: 42
|
| Posted: Sun Apr 16, 2006 2:22 pm Post subject: Test computing PID result for PWM |
|
|
I was trying to get some insite to what pid returns I would get if
the Desired position stayed at 0 and the current position changed
[ie] arm forced to move ....so I tried this Qbasic pgm ...
I can make no sense out of pidFinal........
Can this be done and get meaning full results ???????????
kp = 10 ' 5-10
ki = 5
kd = 20
pidI& = 0
pidIx& = 0
pidLast& = 512 + 32 ''initial pidlast = 1st current
Imax& = 32000
Imin& = -32000
'Syserr = DesiredPos - CurrentPos
DesiredPos& = 0 ' neutral
PRINT ""==========================================================""
FOR CurrentPos& = 512 TO -512 STEP -32 ' 10 bit error value
Syserr& = DesiredPos& - CurrentPos&
pidP& = Syserr& * kp
pidI& = pidI& + Syserr&
IF pidI& > Imax& THEN pidI& = Imax&
IF pidI& < Imin& THEN pidI& = Imin&
pidIx& = pidI& * ki
pidDx& = CurrentPos& - pidLast&
pidD& = pidDx& * kd
pidLast& = CurrentPos&
pidFinal& = pidP& + pidIx& + pidD&
PRINT CurrentPos&; ""/""; Syserr&; TAB(14); pidP&; TAB(22); pidI&; ""/""; pidIx&; TAB(3 ; pidDx&; ""/""; pidD&, pidFinal&; ""/""; INT(pidFinal& / 256)
NEXT CurrentPos&
END |
|
| Back to top |
|
|
andylippitt
Site Admin
Joined: 02 Jan 2006
Posts: 155
Location: Denver, CO
|
| Posted: Sun Apr 16, 2006 3:56 pm Post subject: |
|
|
| The only thing I can point out is a mistake I've made more than once. Your kD term should be negative. |
|
| Back to top |
|
|
|
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
