We have always used a central PID controller and sent analog signals to the servo amps. Does anyone have any experience with closing the each motor's servo loop in it's servo amp. Would like to hear pros and cons.
We have always used a central PID controller and sent analog signals to the servo amps. Does anyone have any experience with closing the each motor's servo loop in it's servo amp. Would like to hear pros and cons.
can you be a little more specific? are your servo amps currently closing something like a current or velocity loop on their own right now, via an analog signal from your central PID controller? and what loop do you expect them to close on their own in this new scenario? position? how would you command them to do whatever higher level parameter you are thinking of having them handle? it may just be another analog signal unless they have a much more sophisticated interface that takes some kind of discrete command like a serial port or logic style input
1) This is a new design
2) Previous designs sent from central PID, an analog signal to servo amps which operated as torque loops. The central controller takes care the position loop.
3) In this new design we are looking at servo amps that have on board PIV (proportional, integral, velocity filter) that close the motor position loop. You send the drives a step and direction signal instead of an analog signal.
ah-hem.... ok i will.... it is actually just the opposite: less robust, less accurate....
so consider....
u have a galil type multiaxis card, or a cnc control, that gets exact position fdbk from all axes... it CAN do true circular interpolation; ie., it can see if a particular axis is not keeping up and thus slow down the other(s) to compensate. Not so with stepdor sent to each drive. ie., X will have no clue Y is falling behind so will continue to run along its proposed circle making a rectangle.
is this bad? no, if you need cheap. it can be a lot less expensive and do the job maybe most of the time. think of steppers.
'snuff said.
I principally use central controllers, analohue controlled Galil Motion etc, and there is distinct advantage, the feature I use the most is electronic gearing, any number of slave axis off of a master, the ratio can be changed on the fly, if needed.
Also electronic cam, although I haven't used that one as much.
Al.
CNC, Mechatronics Integration and Custom Machine Design
“Logic will get you from A to B. Imagination will take you everywhere.”
Albert E.
what about using step and direction with feedback? Pretty much all servo drives have some sort of emulated encoder output signal. I'm used to omron drives and their tuning when using pulse input seems much better then when using +/- 10v.
CNC, Mechatronics Integration and Custom Machine Design
“Logic will get you from A to B. Imagination will take you everywhere.”
Albert E.
Yes but when using dsp type drive that take pulse input and when using pulse input work absolutely great, for Omron I have CX-drive part of their CX-programmer package which will autotune the drive within seconds and work fabulously. Can galil use the simulated pulse output from these types of drives and still slow down axis when needed? I realize that this would really not make much sense as the drive itself is controlling the PID, however if one axis really had to fight a heavy load or what be it and coordinated motion or gearing was needed it could still accomplish it this way yet keeping the simplicity in setup.
My 2 cents. The loop is closed either way, be it in the drive or distributed between a controller and an amplifier. The disadvantages of a distributed controller and amplifier are:
1) If the PID controller is located elsewhere then a digital to analog conversion is required to generate the +/-10V torque command signal and an analog to digital conversion is required in the drive to use that command signal. This expense is unnecessary if the loop is closed inside a drive.
2) Two pieces of hardware is usually more expensive than a single piece. Electronic components themselves are inexpensive but the ancillary support hardware isn't. Ancillary hardware is the printed circuit board, connectors, screws, stand-offs, insulators, enclosure packaging, etc. Adding the electronic components to close a PID loop inside a drive is inexpensive and doesn't impact the cost of the support hardware; it's already there.
3) Compatibility and setup issues can arise, particularly if the PID controller and the amplifier are from different manufacturers. Both the controller and amplifier settings and wiring to connectors must be correct for both to work together. This isn't an issue if the loop is closed inside the drive. Troubleshooting is easier as well because only a single piece of equipment is involved when something doesn't work right.
The advantages of a distributed controller and amplifier are:
1) Greater flexibility. A separate PID controller has more 'bells and whistles' settings and options than one that is integrated into a drive. This is an economics-driven reality because the engineering assumption is an integrated PID drive is used in cost sensitive applications.
2) A multi-axis motion controller can be combined with the PID controllers. This is harder to do with individual drives that have embedded controllers in multi-axis applications where synchronization between drives is required. Ask me how I know; hint: 16-bit MCU and an FPGA.
Mariss
Not sure what you mean by slow the axis down when needed? the drives are basically a transconductance amplifier mode where the motor current is proportional to the analogue input voltage.
There is some excellent instructional videos on most aspects of servo control by Dr Tal on the Galil site, one covers the consequences of including the double PID loop in drive and controller.
They are aimed at theory in general and not all specifically aimed at Galil products.
Al.
CNC, Mechatronics Integration and Custom Machine Design
“Logic will get you from A to B. Imagination will take you everywhere.”
Albert E.
xjdubber,
Any problems with coordinated motion using the smart drives? Circular interpolation? Not sure what you mean by simulated pulse output.
Better tuning in what way? What about coordinated motion moves between x and y axes?