Closing the loop with Path Pilot. Possibly dual (global) closed loop?

Ive been toying around with the idea of adding encoders to the steppers on my upgraded Tormach 1100 pcnc. I know the Mesa 5i25 is fully capable of accepting encoder feedback, as is the LinuxCNC, which PathPilot is essentially a modified version of.

I see some untapped potential here. My Tormach already has glass scales so the idea of integrating those into a possible dual closed loop has me really interested.

Closed loop back to the controller seems like it would be fairly simple for the LinuxCNC guys who can delve into the very DNA of the program and bend it to their whim. For guys like me its all abut research and asking (hopefully the right) questions.

Have any of you attempted anything like this? What all would be involved in getting encoder feedback back to the controller?

What do you imagine you would gain by that? "Closed loop steppers" have been around forever, and have proven pretty useless in most applications, having no real advantage over the typical open-loop use. It is not the same thing as a servo. So, you can go to a lot of expense and bother to add closed-loop control, but it will not make your machine any faster, or any more accurate. So, why bother?

Regards,
Ray L.

Quote:

---

Originally Posted by SCzEngrgGroup

What do you imagine you would gain by that? "Closed loop steppers" have been around forever, and have proven pretty useless in most applications, having no real advantage over the typical open-loop use. It is not the same thing as a servo. So, you can go to a lot of expense and bother to add closed-loop control, but it will not make your machine any faster, or any more accurate. So, why bother?

Regards,
Ray L.

---

Its not quite that simple, adding encoders to a standard step drive systems will not turn them into true servo systems.
They still suffer from possibility of stalling from resonance and torque loss, though stalls can be instantly detected
They also still waste power since they run at full current even when no torque is required.

You do however gain some advantages:

If you use linear scales you can gain a fair bit of accuracy (though you need close to 0 backlash to have a stable feedback loop)
This improved accuracy is because the scale feedback can compensate for ballscrew inaccuracy, thermal expansion, etc
Also since the position reference is the encoder, a stepper stall does no require re-homing, and if the scales
have a index output, very accurate initial homing is possible.

Note that a standard step drive + encoder is not a optimum servo but there are drives the use the
step motor as a true servo and have all the advantages of normal servo drives with the low speed torque
advantages of a step motor. The Leadshine Easy-Servo drives are examples of true step motor servo drives

Wow those Easy Servo drives look sweet! I will definitely keep them on a short list for a future build out I have in mind. Thanks for sharing.

Quote:

---

Originally Posted by PCW_MESA

Note that a standard step drive + encoder is not a optimum servo but there are drives the use the
step motor as a true servo and have all the advantages of normal servo drives with the low speed torque
advantages of a step motor. The Leadshine Easy-Servo drives are examples of true step motor servo drives

---

So from looking at the website, it appears that the "easy servo drive" + "easy servo motor" would give you the benefit of a system that could recover from lost steps without needing the closed-loop feedback through the Mesa card to linuxcnc?

There is a guy on the Shopmaster forum who has done a similar conversion using his glass scales for feedback- he goes into a lot of detail
http://www.cnczone.com/forums/shopma...-software.html

Is'nt the only reason why you would go closed loop or servo....if you were having problems losing position?

But this just doesn't happen unless you have a specific machine fault......

Please explain why.......

Keen

Quote:

---

Originally Posted by keen

Is'nt the only reason why you would go closed loop or servo....if you were having problems losing position?

But this just doesn't happen unless you have a specific machine fault......

Please explain why.......

Keen

---

Here are some reasons to use servos vs step motors/drives:

1. Size (step motors only get up to a few hundred watts before they are not cost/performance competitive with normal servos)
2. Higher speeds are possible and you have much less steep torque drop-off with speed
3. Better accuracy ( because of the feedback, good servos can provide more accurate motion with dynamic loads, stiction etc )
newer servos are capable of extremely smooth motion even at very low speeds (the latest Fanuc encoders have 16/32 million counts per turn)
4. If the feedback goes to the controller you can restart from a fault without re-homing and
if you have absolute encoders you may never need to home
5. More accurate homing (if the hardware can home to encoder index)

These may or may not be all that important to Mini mills/lathes

Quote:

---

Originally Posted by SCzEngrgGroup

What do you imagine you would gain by that? "Closed loop steppers" have been around forever, and have proven pretty useless in most applications, having no real advantage over the typical open-loop use. It is not the same thing as a servo. So, you can go to a lot of expense and bother to add closed-loop control, but it will not make your machine any faster, or any more accurate. So, why bother?

Regards,
Ray L.

---

In my case I would want them to avoid lost steps and detect any loss of accuracy during an operation. With glass scales you can step up to the accuracy of the grain of the scale. Being able to make repeatable moves accurate to 1 micron at the very least has me interested. I dont think ive met anyone who wouldnt trade their steppers for servos on any cnc machine.

Quote:

---

Originally Posted by SwampDonkey

In my case I would want them to avoid lost steps and detect any loss of accuracy during an operation. With glass scales you can step up to the accuracy of the grain of the scale. Being able to make repeatable moves accurate to 1 micron at the very least has me interested. I dont think ive met anyone who wouldnt trade their steppers for servos on any cnc machine.

---

Thanks - I rushed my reading of your intro. I am with you now. You are hoping to use the accuracy of glass scales to improve on limited accuracy of the balls screws and steps .

Cool - All the best ...I am keen to see if you can get it working.

Keen

A build thread where it all could be pulled together from a multitude of researchers would be great,

Thanks
Dave

The problem with a closed loop stepper system is that it is mostly useless.
If the motor is overloaded to the point of losing steps, you are dead. Having a closed loop won't rescue the drive. Oh well, it should flag an error.

A closed loop servo drive will handle a lagging output by increasing the drive voltage: that is how they work. Of course, if you overload one of those it too will fault out.

Accuracy of movement is almost totally separate from the basic drive power. It is governed by whatever measurement system you are running: encoders, glass scales, laser interferometers, whatever, plus the basic resolution of the drive motor. If a single step on your stepper or your encoder is 10 microns at the table, you are just not going to get a 1 micron system.

Cheers
Roger

I see the ability for faster rapids without the fear of getting lost. I have never had just plain lost steps issues on either machine. But any way to speed up tool changes and rapids between parts on fixtures would increase productivity and make me more money. So just go for it.

I had quite a few lost steps the other day when trying to drive Tormach's 1-3/8-in face mill about 1/2-in deep into some aluminum. I'm not convinced either servos or closed loop would have helped there, though.

Quote:

---

I had quite a few lost steps the other day when trying to drive Tormach's 1-3/8-in face mill about 1/2-in deep into some aluminum. I'm not convinced either servos or closed loop would have helped there, though.

---

For that you need a D9.

Cheers
Roger

Quote:

---

Originally Posted by MichaelHenry

I had quite a few lost steps the other day when trying to drive Tormach's 1-3/8-in face mill about 1/2-in deep into some aluminum. I'm not convinced either servos or closed loop would have helped there, though.

---

Ammonium nitrate + Diesel Fuel and a starter charge might do that job, BUT you would surely have to take a light finishing pass after. You may also have trouble if it needed to be anodized later due to unavoidable embedded contaminants.

Quote:

---

Originally Posted by RCaffin

For that you need a D9.

---

And if you REALLY want to be sure, go for a D11R instead.
https://www.youtube.com/watch?v=F-x5MmqTs0s

Quote:

---

Originally Posted by R.DesJardin

I see the ability for faster rapids without the fear of getting lost. I have never had just plain lost steps issues on either machine. But any way to speed up tool changes and rapids between parts on fixtures would increase productivity and make me more money. So just go for it.

---

Not likely.... People seem to believe closed-loop control can magically overcome the limitations of the hardware - it can't. With a stepper, once you've lost a single step, you're pretty much screwed. This can happen either because you try to accelerate too fast, or you put too much load on the motor. That goes double for when the motor is spinning at high speed. It's not like a servo where the driver can increase current, and make the motor catch-up. With a stepper, there is no catching up once a step is lost, as stepper motors are always running at maximum torque - there is no reserve like there is on a servo. Closed loop control can improve accuracy, IF the rest of the system is capable (which is a very big IF....). But it cannot significantly extend the "safe operating area" of the system. If you want faster rapids, use better motors. My Novakon mill originally came with steppers (since changed to AC servos), and with the steppers it would run all day long at 350IPM with no lost steps. Ever. The same motors on a Tormach should do the same.

Regards,
Ray L.

I just purchased a Leadsine 4 NM easy servo motors, but before that I had Nema 34 open loop steppers, and I can say that with open loop steppers my lose of step was major.

Will do a comparison with my new machine, and put a few words about it.

From what I have read, the closed loop has superior RPM's, with much more improved accuracy of an each step. So, if you have a heavy duty gantry, your acc will make your normal stepper to skip a few steps, while closed loop will not allow to over shoot the position.

Servo motors are far worse in high precision machines, as they dont have the resolution of the stepper, as their encoder calculates the overshooting and executes a retraction, which is not good when you have a mold making business. Gets sloppy and not quite precise enough.

Having full confidence in this closed loop easy servo, but we shall give it a test in a few weeks.

Quote:

---

Originally Posted by lopata

Servo motors are far worse in high precision machines, as they dont have the resolution of the stepper, as their encoder calculates the overshooting and executes a retraction, which is not good when you have a mold making business. Gets sloppy and not quite precise enough.

---

Yet the overwhelming majority of high-speed, high-precision machines use servos. Go figure....

Servos are capable of FAR higher precision than steppers, and FAR more power than steppers. Most steppers have a basic resolution of 200 steps/rev. Micro-stepping can reduce the step size, but the size of the micro-steps varies considerably from one to the next, and under load. It is quite common for servo motors to have a basic resolution of 20,000 steps/rev = 100X more than a stepper.

ANY stepper-based system that loses steps is, by definition, either poorly designed, and/or being operated outside of its capabilities. A properly designed stepper system, operated within its capabilities, will NEVER lose steps. The myth that steppers always lose steps is just that - a myth. It is an excuse used by those who don't understand how they work, and how to properly design a robust stepper-based system.

Regards,
Ray L.