Abhijit Posted September 17, 2013 Posted September 17, 2013 For monopulse auto tracking in antenna application, I am using cascade loop for auto error to make it zero. However, for sustaining high speeds, finite auto error is necessary. This defeats the requirement of making auto error zero. How to overcome this ?
steve.milici Posted September 18, 2013 Posted September 18, 2013 What is your current implementation of the cascade loop?
Abhijit Posted September 18, 2013 Author Posted September 18, 2013 What is your current implementation of the cascade loop? Currently I am comparing auto error with zero reference and the resultant (negative of auto error) is being fed to cascade loop. Output of the cascade loop is being integrated and fed to the real axis with electronic gearing as described in the Turbo PMAC User Manual. Secant correction is being applied too. This introduces lag during high elevation trajectories which demand high rates in elevation over azimuth mounts.
steve.milici Posted September 19, 2013 Posted September 19, 2013 Does lowering the gearing ratio or loop gains provide any improvement?
Abhijit Posted September 20, 2013 Author Posted September 20, 2013 Does lowering the gearing ratio or loop gains provide any improvement? Electronic gear ratio at the output of cascade is 1. Cascade loop gains lowering is not helping as it results into deteriorating step response.
Richard Naddaf Posted September 20, 2013 Posted September 20, 2013 You want the tracking error to be zero but you don't want it to be zero? Sorry, I don't understand what the problem is or what are you trying to achieve. Can you try to explain better?
Abhijit Posted September 20, 2013 Author Posted September 20, 2013 You want the tracking error to be zero but you don't want it to be zero? Sorry, I don't understand what the problem is or what are you trying to achieve. Can you try to explain better? I want the tracking error to be zero. But I have observed appreciable lag (non zero tracking error) at higher speeds of the load. I am attributing this to the topology of usage of the cascade loop itself. Is there any control method better than cascading ?
Richard Naddaf Posted September 21, 2013 Posted September 21, 2013 What is the resolution of the analog (I suppose) feedback? How do you have the cascaded loop set up? if you have a virtual motor then you can make full use of the PID of that motor to tighten up the tracking. E.g. Quick step move using the PewinPro2 Tuning. It will be as responsive as your input signal is. By the way, inherently, tracking will always have some error during motion. There is no such thing as zero error. You cannot predict where the target is going, otherwise you would just pre-program it.
Abhijit Posted September 23, 2013 Author Posted September 23, 2013 What is the resolution of the analog (I suppose) feedback? How do you have the cascaded loop set up? if you have a virtual motor then you can make full use of the PID of that motor to tighten up the tracking. E.g. Quick step move using the PewinPro2 Tuning. It will be as responsive as your input signal is. By the way, inherently, tracking will always have some error during motion. There is no such thing as zero error. You cannot predict where the target is going, otherwise you would just pre-program it. Resolution of the analog feedback is +/-10 volts maps into 0 to 65535 counts. After applying secant correction (using open servo), I am feeding this signal to a virtual axis through encoder conversion table. I am closing the virtual loop by giving command #1j=32767 which means that I want the error to become zero. I have integrated the output of the virtual loop and then applied to real axis through electronic gearing as described in the Turbo PMAC user manual. I have observed that step response is acceptable when I set the Ix30 of virtual loop correctly. This is a good setting on which the system is running currently. The performance is acceptable even at moderate speeds, meaning that lag is within acceptable limits. It is only at the high speeds of the target that there is a large lag (during zenith pass). I have noted your point about lag being inherently present in the tracking applications.
Recommended Posts