Skip to main content
Submitted by ArnouldEric on Fri, 12/19/2008 - 06:33

Good afternoon,

I have a probleme of accuracy with my systeme.
When i go to a position, for exemple 150500, controller told me i'm at 150500 (TP) but the error can be very important, like -500 to +500 (TE). How can i improve that ? I need to place a probe with a precision of ½m (10^-6) !

Thanks in advance.

Arnould.

Comments 4

Galil_AndyH on 12/19/2008 - 08:46

If your error is changing +/-500 counts, your TP should also reflect that. TE is simply the difference between RP and TP.

Reducing your position error is all in tuning and mechanics. I would start with our manual tuning application note to get a good idea of how the tuning functions effect the performance of your system.
http://www.galilmc.com/support/appnotes/optima/note3413.pdf

Then you should use either GalilTools, or WSDK to scope out the response of your system and tune the system for your required specifications.

ArnouldEric on 12/22/2008 - 02:59

Ok thanks.
So the only solution to correct accuracy error are mechanics and controller tuning.
For the moment, my values are KD=64 - KP=6 - KI=0
I'm going to try with tuning.
Thanks.
Arnould.

Galil_AndyH on 12/22/2008 - 08:19

Those tuning parameters are the defaults, which means the system has not been tuned. You will find an improvement in accuracy and performance once you tune the system.

Once you have tuned the system, if your accuracy is still not within specs, there are other options such as changing your accelerations, speeds and decelerations, or implementing filters such as poles or notches.

ArnouldEric on 01/09/2009 - 02:27

Ok i tuned my PID filter correctly and finally got an error < 10�m.

Just notice that the application note told to set all Kx to 0 and then increase KD until vibration/instability. I just noticed that it was very difficult to find it with KP = 0, most of time i think that KP = 1 to start is better because KD must be very very high to move if KP = 0 ^^

Thanks for your help.