Cross Coupled Gantry Tuning

[kmonroe023]

I've just got done setting up our gantry system and successfully configured the two y-axis motors using the cross-coupled gantry mode. The tuning using the Auto Tuner was actually pretty good, but I think we'll need to fine-tune the gantry tuning for our application. I was wondering if you guys have any advice, or maybe some white papers handy describing how best to utilize the tools in motor Tune for cross coupled gantry mode. Questions: -When adjusting gains, should you only adjust gains for the 'master' or both master and follower? Since each motor does its own servo calculations, I assume I'll need to tweak the gains for each motor. -What about the Cross-Coupling gains? Auto-Tune filled those out for me, but I don't quite know if I should leave them be, or adjust them at some point. I know I'm probably asking more than can easily answered in a forum post. I've looked through the Pmac manuals but I haven't found much regarding the gantry mode. Thanks

[bradp]

There is not much written about this yet. We are adding features faster than documentation. Most systems need to adjust both the leader and follower gains. The auto tuner looks at performance and calculates gains independently which accounts for differences between the gantry sides. Most users however consider the gantry sides as having equal dynamics and friction, unless it is obvious that they do not, and like to keep the gains the same for both sides. There is no real "right way" you just need to make sure both parts of the gantry reach the performance goal. The cross couple gains will control how quickly the gantry sides react to gantry errors (differences in FE between the sides) The dynamics of your system and capablities of the gantry will ultimately determine the response you can acheive. You only need to adjust them if the gantry is not reacting well to gantry errors. Well could be either too sensitive or too slow.

[Omron Forums Support]

You have to adjust both the master and follower gains. The autotuner estimates the hardware gain for both the master and follower (for torque mode) and calculates the servo gains based on this and desired perf. specifications. The autotuner sets up these gains to give an approx. equal tracking performance for each motor and gantry error. (i.e. sets up gantry cross-coupled gains as follows: Motor[x].Servo.Kxpg=Motor[x].Servo.Kp Motor[x].Servo.Kxvg=Motor[x].Servo.Kvfb/2 (if there is a filter in the position, and velocity loop, these gains are setup as the steady-state gains) Motor[x].Servo.Kxig=Motor[x].Servo.Ki). Cross-coupled gantry gains can be fine tuned. You can increse these values if the gantry error performance is more important. The fine tuning strategy is similar to regular pid tuning. A high gain for the cross-couple gantry gain will yield a faster reponse but may create an oscillatory response. Higher cross-coupled derivative gain values will yield a more robust (stable) system, while the response will be slow and may yield a noisy response due to quantization noise, etc.

[Omron Forums Support]

[quote='kmonroe023' pid='1177' dateline='1295991835'] I've just got done setting up our gantry system and successfully configured the two y-axis motors using the cross-coupled gantry mode. The tuning using the Auto Tuner was actually pretty good, but I think we'll need to fine-tune the gantry tuning for our application. I was wondering if you guys have any advice, or maybe some white papers handy describing how best to utilize the tools in motor Tune for cross coupled gantry mode. Questions: -When adjusting gains, should you only adjust gains for the 'master' or both master and follower? Since each motor does its own servo calculations, I assume I'll need to tweak the gains for each motor. -What about the Cross-Coupling gains? Auto-Tune filled those out for me, but I don't quite know if I should leave them be, or adjust them at some point. I know I'm probably asking more than can easily answered in a forum post. I've looked through the Pmac manuals but I haven't found much regarding the gantry mode. Thanks [/quote] In addition to what I have previously mentioned, cross-coupled gantry motors do not use their own servo alg. seperately,they both use a special servo algorithm (sys.gantryxctrl) where all the computations are done simultaneously for both motors. Thus, there is no servo cycle delay for the added cross-coupling servo input as in turbo-pmac. The user manuals are being updated to include the details about these. I am also attaching a write-up which explains step by step x-couple gantry tuning via IDE tuning software.