Servo closed loop specifics

[stevespo]

I'd like to better understand how the G320/320X/340 closes the feedback loop. I haven't seen this anywhere in the forum, but please point me in the right direction if it's covered somewhere else.

My primitive (maybe incorrect) understanding is that they are constantly comparing step count with encoder count and when/if there is a mismatch they will apply power to take corrective action. If the difference becomes too great (configurable) then the drive will fault and movement will cease.

So, when accelerating into a cut if the step count starts to outpace the encoder count, additional power is applied to the motor to tap into the torque reserves and keep things synchronized. If this doesn't correct the situation, the drive will eventually fault.

On the other hand, if the encoder count suddenly shot past the step count, would the drive attempt to bring the two back in line via some combination of motor braking or reverse motion?

What I am wondering about is the specific situation when using an upcut mill and having the Z axis pulled down towards the workpiece by the tool. This happens to my steppers from time to time. I'm moving too fast and there isn't adequate torque available to maintain position.

Is this something that the servo drive can catch and correct? Or will this happen so quickly that the drive merely faults and you've probably lost the workpiece anyway?

Obviously, this is something that should not happen. There is probably a problem with the feedrate being too fast, the DOC too large, etc. But I'm wondering on a practical level does a servo drive/motor avoid these types of problems or merely fault when they happen?

Also, how is the controller (Mach, EMC, etc) typically notified when this occurs?

Thanks,

Steve

[Al_The_Man]

Servo loop is based on a PID algorithm. With Mach, the loop is closed back to the drive and not Mach itself.
http://www.jashaw.com/pid/tutorial/
Also the Galil motion site has some very informative instructional video's on Servo PID and tuning and servo's in general that applies across the board, not just to Galil products.
Al.

[stevespo]

Al, thanks for the pointer. I understand the basics of PID and how the loop is closed back to the driver. I was wondering how to handle error conditions that may occur. It looks like the Gecko servo drives have an ERROR pin (output) that can be tied back to one of the port (Mach) inputs to indicate an error has occurred.

From the 320X manual:

"The voltage on this terminal is +5VDC when the G320X is functioning normally. The voltage on this terminal goes to 0VDC whenever the FAULT indicator is lit. This output can be used to signal your controller that an error has occurred. "

I was mainly wondering how people used this. I'll look around for some diagrams. Perhaps those outputs can be tied together and used in conjunction with the eStop and put on a single input to the parallel port. I thinking about a quad input AND gate like the CD74HC21. Any advice here is appreciated.

Steve

[battwell]

from my experience with servos, if you have the right size or bigger to drive your tool or table and they are set up properly or fully auto tuning like the mitsubishi (which i use for everything now) you dont get faults, lost steps or lost work. with ebay you can get good new/used servos for less than some stepper set ups.

[Torsten]

>My primitive (maybe incorrect) understanding is that they are constantly >comparing step count with encoder count and when/if there is a mismatch >they will apply power to take corrective action. If the difference becomes
>too great (configurable) then the drive will fault and movement will cease.

Correct, a Error Register will keep track of the error of position this ER will count the steps received from the controller positive and negative.
If the controller issues 20 steps the ER will go up to positiv 20 if then 5 steps in the opposite direction are commanded the ER will go to positiv 15.
The Drive will constantly monitor this ER and respond by powering the motor to move the ER back to zero in the direction of the error.
The higher the error the more Power is applied to the Drive, once the maximum Power is no longer sufficent to overcome the resistence the ER will if given continueing steps in the same direction eventually exeed the size of its registers and cause the fault signal.
On the other hand errors that never exeed the size of the ER will go unnoticed.


>On the other hand, if the encoder count suddenly shot past the step count, >would the drive attempt to bring the two back in line via some combination >of motor braking or reverse motion?

Yes the motor is driven by the ER count nothing else.

>What I am wondering about is the specific situation when using an upcut >mill and having the Z axis pulled down towards the workpiece by the tool. >This happens to my steppers from time to time. I'm moving too fast and >there isn't adequate torque available to maintain position.
>Is this something that the servo drive can catch and correct? Or will this >happen so quickly that the drive merely faults and you've probably lost the >workpiece anyway?

Exessive Force is just that, you may need more powerful Motors and or Drivers.
This problem may be caused or exagerated by backlash in the system, once the cutter gets pulled into the material the forces due to the increased chipload go up exponentaly and may cause the cutter to break or the Drive Motor to become overpowerd.
This condition may apply to more then one Axes at the same time.

The first to look at is to reduce the backlash at all Axes to the minimum.
Increase the ridgidity of the Machine as far as possible.
If the problem still persists get bigger motors/drivers or slow down the Feedrate.

[stevespo]

battwell and Torsten, thanks very much for the information and advice. The description is helpful. My machine has no measurable backlash (< .001") and is reasonably rigid (8020 and plate aluminum).

When I experienced the slipping/diving, the machine was doing some pretty aggressive 3d cuts in oak with a 1/2" ball nose (upcut) end mill. Even counterbalanced, the Z is moving a heavy, water-cooled spindle up and down. So, it could be that the Z motor was just overwhelmed trying to move and hold position. It's also possible that the motor coupling on the Z axis had slipped, but after slowing down the feedrates I had perfect results on my subsequent cuts.

So, I am currently experimenting with Granite VSD-E drives and 300W brushless DC motors. They are rated at 150 oz-in continuous and 450 oz-in peak, so they will certainly outperform my 425 oz-in steppers under any circumstances I could imagine. Based on the actual specs and estimated cutting forces from my machine, we've calculated that I'll need 59 oz-in or torque to accelerate my gantry fairly briskly. The extra headroom is there as insurance.

It seems like the key to any system (servo or stepper) is sizing the motors appropriately. You want to start with the typical scenarios and then add a "safety margin". If you don't exceed their capabilities, you should not experience faults of failures.

Steve

[battwell]

if you havnt already bought your new motors and drives there are some very cheap 400watt mitsubishi on ebay at the moment
http://cgi.ebay.co.uk/Mitsubishi-MR-...item5ad57add39

these will run 230v single phase. massive resolution in every mode ie torque, speed / step dir or a mixture!
just have to find some hc-mfs/kfs-40 motors which there are always loads of!
i would buy them but i have over 30 here already in that size

[BobWarfield]

Steve, for your situation, I would expect the servo to just fault if it can't be commanded to reduce the following error below the threshold.

People handle the faults in a variety of ways and it can be a little tricky to do something clever with it.

There is also a new Gecko servo drive coming along that has quite a few additional features, though I'm not sure they'd matter for this application.

It does seem like you have a motor sizing/drive ratio issue at work here though.

Cheers,

BW