Any luck getting your system tuned? It would help to have more information on your system, what is working, and what you have tried.
The basic process I use for bringing up a system is to:
#1 verify I can drive the motor (open loop) forward and backward using console DAC commands
#2 Verify the encoder position counts up and down reasonably
#3 Clear all filters(actually set them to flat gain of 1) on the filter screen
#4 Limit the max velocity to ~ %10 by seting the Max Output on the Step Response Screen to 200 DAC counts)
#5 Set a large following error of ~ 1 encoder Rev of counts (~ 4000 counts)
#6 Set I = D = 0
#7 Set low P gain of 0.1
#8 Download the Configuration and Parameters using the "Enable" button on Step Response Screen
#9 Verify System Servos (holds position and resists turning motor)
#10 Set Size of 1000 counts and time to 3 seconds on the Step Response Screen
#11 Select Position, Command, Output Plot mode
#12 Push "Move" to download settings, move, and plot the response
#13 Tune :}
I hope this gets you started.
Regards
TK
| Group: DynoMotion |
Message: 258 |
From: dunncl66 |
Date: 3/25/2010 |
| Subject: Re: Begining PID numbers for a small mill |
|
Thank you for the advice. I am looking forward to trying it this weekend and I have not got everything tuned. How does a fellow know when his servos are properly tuned?
|
|
| Group: DynoMotion |
Message: 259 |
From: Tom Kerekes |
Date: 3/25/2010 |
| Subject: Re: Begining PID numbers for a small mill |
Well if the performance meets your accuracy requirements and there is good stability with a margin then you should be ok.
Make motions at the speeds that are important to you and plot the following errors. Convert the maximum following error in counts to your units by dividing by the number of counts/unit in your system.
Typically as gains are increased errors will be reduced but at some point the system will begin to overcorrect and overshoot or go unstable and oscillate out of control. Typically gains are then reduced by a factor of 2 or more below this point to provide a stability margin.
When changing PID gains use simple numbers and when making changes make significant changes of at least 10 or 30%. For example changing a gain from 0.3 to 0.4 might make an observable difference, where changing from 0.3123 to 0.3125 would not.
I hope this helps
TK
| |
