Newbie- CNC conversion typically open loop or closed loop?

[foobar42]

Hi,

I have a very basic question/confusion. From my point of view there are at least two basic types of CN control:

1. The controller/software sends stepper/servo signals according to some code and the hardware controller controls the steppers/servos accordingly (open loop).
or
2. Like 1., but the controller/software reads the DRO data to measure the actual travel (closed loop).

The second alternative allows exact (up to the accuracy of the DRO) absolute positioning regardless of backlash or any other mechanical issues, the first alternative does not.

Now my question: How are hobby conversions typically done: closed, open or something entirely different?

It seems to me that -if the software supports it- a closed loop system has tangible benefits and is not considerably more complicated than an off-the-shelf DRO. Are people doing this?

-Jan.

[Hans_G]

I think the vast majority of benchtop CNCs run w/ Mach3 and EMC use open loop steppers. Whenever the "stepper vs. servo" topic comes up, most people vouch for steppers, citing simplicity (if properly designed) and lower cost. Personally, I use servos, but the encoders are integral to the motor. If you already have the scales for a DRO, it makes good sense to use them. I don't know much about getting the signals back to Mach3, I'm sure someone else will chime in. The glass scales on my lathe dro output single ended quadrature, which any number of servo or data acquisition systems can use.

I agree that it doesn't add much complexity, whether it's a true servo motor or a stepper with position compensation. There are motion controllers that will treat a stepper as a BLDC motor and run it as a servo (Copley Controls for one), but this approach limits motor performance compared to simple micro stepping.

[HimyKabibble]

Originally Posted by foobar42 Hi, I have a very basic question/confusion. From my point of view there are at least two basic types of CN control: 1. The controller/software sends stepper/servo signals according to some code and the hardware controller controls the steppers/servos accordingly (open loop). or 2. Like 1., but the controller/software reads the DRO data to measure the actual travel (closed loop). The second alternative allows exact (up to the accuracy of the DRO) absolute positioning regardless of backlash or any other mechanical issues, the first alternative does not. Now my question: How are hobby conversions typically done: closed, open or something entirely different? It seems to me that -if the software supports it- a closed loop system has tangible benefits and is not considerably more complicated than an off-the-shelf DRO. Are people doing this? -Jan.

First, it is a myth the adding an encoder and "closing the loop" will somehow magically "fix" backlash. It can't and won't. All it can do is reduce the effects under certain circumstances. True precision and accuracy comes ONLY from a well-designed machine, with essentially zero backlash, operated with its limits.

Mach3 is "open loop", regardless of whether you use servos or steppers. Mach3 tells the machine where to do, and it is entirely up to the machine to go where it's told. A properly built, properly operated machine WILL do this 100% of the time, whether using steppers or servos. A badly designed or improperly operated machine may, or may not, do this. "Closing the loop" on a badly designed machine, or one that is operated outside its limits, will NOT magically transform it into a good machine.

EMC, apparently, will do true closed-loop control, at least with appropriate hardware.

Regards,
Ray L.

[Al_The_Man]

You have a couple of descriptions mixed in there.
Steppers generally are never closed loop as there is no feedback device, servo's operate with a feedback device of some kind and close the loop via a PID method.
With a system such as Mach where the control is through the parallel port, the loop is closed back to the servo drive, the loop is not closed back to the CNC controller (Mach).
The controller sends a command and assumes that the position will be reached.
With the addition of a motion card such as Galil, the control can close the loop back to the controller.
This has the advantage of features such as electronic cam and electronic gearing, as well as true synchronized threading and helical milling etc.
If you are also talking of DRO or scales on the final positioning, then dual feedback is required, an encoder on the motor and a scale on the final positioning device, the PID loop is then split between the two feedback devices.
Al.

[foobar42]

Thanks to Ray and Hans for the responses.

Ray, could you elaborate a little more, why e.g. backlash is still an issue with a closed loop system? From my understanding, backlash is a problem in the mechanical coupling of the driveshaft and the table. Assuming I am able to read the position of the table exactly (up to epsilon precision), how would the backlash still affect me?

-Jan.

Originally Posted by HimyKabibble First, it is a myth the adding an encoder and "closing the loop" will somehow magically "fix" backlash. It can't and won't. All it can do is reduce the effects under certain circumstances. True precision and accuracy comes ONLY from a well-designed machine, with essentially zero backlash, operated with its limits. Mach3 is "open loop", regardless of whether you use servos or steppers. Mach3 tells the machine where to do, and it is entirely up to the machine to go where it's told. A properly built, properly operated machine WILL do this 100% of the time, whether using steppers or servos. A badly designed or improperly operated machine may, or may not, do this. "Closing the loop" on a badly designed machine, or one that is operated outside its limits, will NOT magically transform it into a good machine. EMC, apparently, will do true closed-loop control, at least with appropriate hardware. Regards, Ray L.

[foobar42]

Al, you are right, I wasn't particularly precise in my description.
I am referring to a setup independent of the particularities of the drive mechanism.
More abstractly, I am assuming that I have a means to measure the absolute position of the workpiece/table up to some precision, similar to a DRO. If I have this measurement available, can I not use this to control the drive mechanism? E.g.: the computer drives the X-axis until the measurement increased 5mil in value.

-Jan.

Originally Posted by Al_The_Man You have a couple of descriptions mixed in there. Steppers generally are never closed loop as there is no feedback device, servo's operate with a feedback device of some kind and close the loop via a PID method. With a system such as Mach where the control is through the parallel port, the loop is closed back to the servo drive, the loop is not closed back to the CNC controller (Mach). The controller sends a command and assumes that the position will be reached. With the addition of a motion card such as Galil, the control can close the loop back to the controller. This has the advantage of features such as electronic cam and electronic gearing, as well as true synchronized threading and helical milling etc. If you are also talking of DRO or scales on the final positioning, then dual feedback is required, an encoder on the motor and a scale on the final positioning device, the PID loop is then split between the two feedback devices. Al.

[Hans_G]

Yes, if you can measure the table position, you can use this to drive your motion control system, but your motion control system is probably going to be more sophisticated than the typical hobbyist mill.

And you can use DROs as the sole encoder. For steppers you would need a motor driver that accepts encoder feedback and compensates for position error. Again, Copely Stepnet amps do this as an example (I've used them, but I don't like them).

Practically any brushed dc motor servo amp could operate w/ a DRO as the sole feedback. A brushless DC servo amp can work well on just DROs if the BLDC's have hall sensors. If they don't have hall sensors, you'll need an amp that can perform hall-less commutation and a system without much backlash. Most modern digital BLDC amps can operate w/ encoder feedback only (no halls). I have worked with AMC digiflex and Copley Accelnet and Xenus amps in this way.

[Al_The_Man]

Some of the Galil on line instructional video's are worth checking out to get some precise idea of servo mechanisms in particular.
There is one that deals with dual feedback to eliminate backlash as a last resort.
http://www.galilmc.com/learning/tuto...sation-methods
These video's, although geared to Galil, are common to most PID loop systems.
Al.

[HimyKabibble]

Originally Posted by foobar42 Thanks to Ray and Hans for the responses. Ray, could you elaborate a little more, why e.g. backlash is still an issue with a closed loop system? From my understanding, backlash is a problem in the mechanical coupling of the driveshaft and the table. Assuming I am able to read the position of the table exactly (up to epsilon precision), how would the backlash still affect me? -Jan.

When you have backlash, the machining forces can and will push the tool around within the range of the backlash. So, if you have 0.005" of backlash on your X axis, and you command it to 0.000, it may be at 0.000 for part of the cut, then the machining forces can push it, essentially instantly, up to 0.005" one way or the other (depending on which side of the backlash you were on). Servos (and almost *any* closed loop system will be implemented as a servo, using a PID loop for control) do NOT react instantaneously. It takes time for the servo to realize its off position, and more time for it to decide the error is great enough to warrant correction, and more time to decide it needs to do something about it, and more time for the correction to actually take place. So, even in the simplest case, where the servo can quickly correct the position, you'll still end up with a "divot" in your part due to the short time the system was off position. In more complex cases, it is easy to get into a situation where the servo and the machining forces are fighting each other, and the system can start oscillating, leaving visible artifacts in the workpiece. This is precisely why encoders are typically fitted to the motors, not to the table. If there is *any* backlash in the system, there is a risk of instability as a result. The higher the system gain (and gains is what gives a servo "stiffness", the worse this problem. If you're going to be using feedback, you need to make very sure there is zero backlash between the encoder and the motor.

Regards,
Ray L.