Step/Direction servos, a few questions about closing the loop back to Kflop
Hi Tom,
I've been reading up a bit on Kflop, looks interesting if challenging but a solid contender to put on my CNC router, currently controlled by WINCNC.
I've not been too happy with Win CNC / my router recently, although the problems that make me unhappy may only be marginally related to wincnc itself, other areas could be the machine's aging mechanics and possibly poorly tuning of servos.
The machine is a 2001-2002 CNT motion systems and uses a win-cnc control and technic servo drives and motors, AC I believe, with step and direction drivers. It started out as a dos machine and has always been a little less than desirable in it's accuracy and smoothness of cut, though a 14 year workhorse despite all that. I have a smaller techno-isel that I always used for more precise parts.
The problem I had is that the drives are mounted on the machine gantry back, and if a power wire has a intermittently broken connection, as my spindle fan did, it sent spurious step signals to the drives before the loop was closed. Thus, my "closed loop" system was loosing position and squareness until it torqued out and the drive faulted. Very hard to troubleshoot, but once I did so, I realized ANY setup similar could have the same weakness. Drives in the cabinet and proper wired and shielded is a different matter but to get there, that's pretty much a total rebuild anyways, new cabinet, new cables and everything moved over. And, even then, I don't think I could close the loop to wincnc anyways.
So, looking at my options, all kinds of options, I came across Kflop. Closing the loop back to the control sounds great, but I'm unsure of how it would work with the step and direction drives that are so common now. I assume those drives *need* the feedback loop internal or else they will fault (or runaway maybe too). So, I am wondering how exactly a setup would look like if I was closing the loop to both the drive and the kflop board (or daughter board if needed for the type of encoder I have). Would I simply put wires from the drive back to kflop, so the signal went to the drive but continued on to kflop......if kflop detected an movement error it could send a few extra step commands to get the motor back on track? Or, would I have to run two sets of wires from the encoder?
Would I use Kmotion to tune the servo driver, or set the driver to basically allow kflop / kmotion to handle the tuning aspects?
I have not searched thoroughly other threads for this issue, save for some on Gecko 320s that are a similar closed-loop in the driver setup, so my apologies if this has been covered elsewhere. Having had this uncommon problem, I'd like to be able to solve it 100%
Thanks, Greg
Re: Step/Direction servos, a few questions about closing the loop back to Kflop
Hi Greg,
I don't think it would be a good design to expect Step Pulses to be lost and then compensate for them with an additional closed loop. In that case it is probably as likely to expect encoder count errors that would throw the position off anyways.
But to answer your question, yes any drive that closes the position loop will require the encoder feedback to go back to it. The more expensive drives (ie Yaskawa) typically have extra differential encoder outputs that can go back to the controller. You might try to daisy chain the encoders to the drive and then to the controller. But that can be messy for shielding, grounding, and termination. Our Kanalog differential inputs have only moderate termination (470 ohms) to help facilitate this.
With encoder feedback going back to our KFLOP Controller it can be used to detect following errors where the encoder position doesn't match the commanded step position within a specified tolerance. Or it can be used to insert additional steps to drive the encoder to match the desired position. This assumes the encoder position is correct and any error is due to mechanics or lost steps.
With feedback going back to KFLOP you can plot the performance and response in KMotion easily. But the servo tuning adjustments would still want to be made within the drive (ie turning pots) to get the drives response optimized first. If a dual closed loop is used with KFLOP providing another feedback loop you could adjust that and tune it using KMotion.
HTH
Regards
Re: Step/Direction servos, a few questions about closing the loop back to Kflop
Hi Tom,
I just looked at the drives (SST-3000 model) and I see they have encoder outputs on them, unused at the moment. Would using those outputs allow Kflop (w. Kanalog as they look to be differential) compensate for such EMF introduced errors?
Thanks, Greg
Re: Step/Direction servos, a few questions about closing the loop back to Kflop
Hi Greg,
Well as I tried to explain previously as long as the encoder signals are not affected it should. Additionally it would allow you to detect something is wrong rather than simply running blind until the mechanics bind up.
Regards
Re: Step/Direction servos, a few questions about closing the loop back to Kflop
Hi Tom,
Alternatively, it also looks like I could treat the existing drivers as amplifiers, would that be a better approach? Sorry for the piecemeal questions, mostly busy with production work.
I'm also assuming that The Kflop/Kmotioncnc combo can do well at "smoothing out" linear interpolation moves that many CAM programs like to use for fancier things like tabs and ramping entries....I realized I hadn't really seen much on this but is one aspect of motion control I am looking for improvements in. Would that be correct?
Thanks, Greg
Re: Step/Direction servos, a few questions about closing the loop back to Kflop
Hi Greg,
To treat the existing drivers as amplifiers they would need to support +/-10V analog inputs to control velocity or torque. In that case you would need our KFLOP+Kanalog to control them.
I'm not sure what you mean by "tabs and ramping entries" but KMotionCNC uses our Trajectory Planner that handles large numbers of small vectors well. See:
Tool Setup Trajectory Planner
Trajectory Planner
Regards
