Can't mill a round circle to save my life

[cncdrive]

Position following error is a possible scenario if you using closed loop drives.
It could cause an ovalness of a circle.
Did you try to run the circle slower and faster also to see if the ovalness changes?
If it is following error then it is very likely that it will become more oval on higher feedrates.

[dhanger]

Position following error is a possible scenario if you using closed loop drives.
It could cause an ovalness of a circle.
Did you try to run the circle slower and faster also to see if the ovalness changes?
If it is following error then it is very likely that it will become more oval on higher feedrates.

It sounds as if you are saying that it could well be a following error, but if so why don't I get an alarm? I have not yet tried accessing the tuning parameters (or ever) since installing the closed loop steppers. Is it possible that the default factory settings of the stepper would allow that much error? Maybe the encoder pulse setting that Dawid92 mentioned? I don't really understand what that setting changes.

I did indeed try changing the feedrate and got different results. At 32 IPM I get the error I first reported--at 4 IPM I get nearly a perfect circle. Of course slowing down that much is unacceptable on a regular basis, so that is not a solution for me. Also, the last thing I did with the project mill before quitting for the weekend was to change the CL stepper on the Y axis to an integrated servo motor that I am experimenting with. Again using the factory settings I got the same result as before, an ellipse instead of a circle. This strongly suggests to me that it's the control software that is allowing a deviation and the stepper and servo motors are just following the commands without complaining about it. Has anyone looked at my CV settings in my profile to see if I have a bad setting there? I would also really like to hear if anyone has run my program and what their results are.

[dhanger]

Incidentally, if someone does try to run my program I just remembered that the spindle speed that is posted is not reasonable since for the moment I do not have a proper spindle drive and I'm just using a place holder speed to activate a relay to start the variac I'm currently using to drive it. So then the spindle speed in the program has to be set properly.

[cncdrive]

The software never makes an oval shape from a cirle.
If the tolerances are high then it will make a smaller circle or a diagonal, but never an oval shape.
So, it is impossible that the software is doing it, no matter what CV settings you using.

If you have some open loop steppers then try those. Those cannot have following error.
Closed loop steppers (the cheap chinese ones) have a factory fix programmed max.128 following error before alarming out.
This means that the motor can have 128 steps error without giving you an alarm.
You know your steps per value, multiply it with 128 and you will get the max deviation this can cause on your axis.

[spumco]

It sounds as if you are saying that it could well be a following error, but if so why don't I get an alarm?

You certainly need to check the following error settings in your drives. This this the only reason I purchased the overly-expensive 'tuning' cable from StepperOnline - to adjust the following error.

My steppers have 1000 line (4k A-quad-B) encoders. The matching drives were set from the factory with a 1000 pulse following error before alarming.

That's 1/4 turn, or 0.05" with my 5mm screws! Unacceptable.

Have a look at your following error, and if it starts alarming all the time you can try to fiddle with the drive gains to tighten it up.

-R

[dhanger]

Okay, sounds like I need to start figuring out how to access and work with the tuning on the motor drives. The CL steppers are the StepperOnline motors, which generally I've been pretty happy with. The integrated servo I mentioned replacing the CL stepper with is also from Stepperonline, so I guess at this point it's safe to assume that anything from them is going to have to be checked out before expecting the kind of accuracy I want? It's funny that after more than two years of using them that I never noticed this issue before, but since I mostly do 3d surfacing it's kinda hard to quantify geometric errors.

[spumco]

Are you using the Y-series (red) CL stepper drives?

If so, you're in for a treat with the software. StepperOnline don't have any sort of user manual or explanation/description of the parameters - although the critical one is pretty easy to figure out (position deviation alarm)

On a hunch, I tried just about every other chinese stepper manufacturer/re-seller's CL tuning software and nothing but the StepperOnline software would connect. So if you get frustrated don't waste your time with JLC, Leadshine, or any of the other vendors' software.

On the other hand, if you got the higher-end TS series I believe those are re-branded Leadshine's, so LS's software works with those. Just having a readable programming manual might be worth the upcharge.

[dhanger]

Are you using the Y-series (red) CL stepper drives?

If so, you're in for a treat with the software. StepperOnline don't have any sort of user manual or explanation/description of the parameters - although the critical one is pretty easy to figure out (position deviation alarm)

On a hunch, I tried just about every other chinese stepper manufacturer/re-seller's CL tuning software and nothing but the StepperOnline software would connect. So if you get frustrated don't waste your time with JLC, Leadshine, or any of the other vendors' software.

On the other hand, if you got the higher-end TS series I believe those are re-branded Leadshine's, so LS's software works with those. Just having a readable programming manual might be worth the upcharge.

For the 'production' mill I installed the TS series motors, but still have never used the tuning software. They worked so well that until I found my round circle problem I haven't had any complaint about them. I've done a number of stepper motor projects over the last 10+ years and these motors are by far the quietest steppers I've run across. In fact, for stepper motors they are actually pleasant to listen to.

For the current project mill (this is the same one I mentioned in the thread you responded to about the spindle motor driver, both mills are identical twins that I bought on sale a few years back) I was going to use the same motors, in fact I ordered a complete set of them and started to install them when I ran across a video by a guy that goes by the handle Clough42 who was reviewing this motor:

https://www.omc-stepperonline.com/nema- ... motor.html

I decided to give these a try and that's what I'm working on now.

So anyway....to update this thread, this morning I removed both X and Y motors and replaced them with the open loop motors, and the result was a perfect circle. Amazingly perfect, in fact, for an old mill. So I'd call that conclusive, it's the motors for sure. I then re-installed the servo motors and ran the program which gave me an oval once again. Not as bad as I was getting with the CL steppers which was about plus/minus 0.005", instead I'm getting plus/minus 0.002", much better but still unacceptable. So the next step will be to dig in and find out how to tune these. I will likely be back with questions about that, but for now a big thanks for the help and pointers from all.

Dan

[cncdrive]

So it is following error as I suspected.
The most important thing with servos is to properly size then to the load.
I mean to make sure that the motor's contrinous rated torque is larger than the torque requirement of the load.
It is also important to have a high enough encoder resolution, because servos always have some following error, but if the following error encoder counts is high gives a large linear error then it is a problem.
The higher the encoder resolution the same amount of following error in encoder counts gives the lower linear error.
For example we now selling the Leadshine ELP series AC servo drives with 17bits encoders which give about 131000 encoder counts per turn resolution.
There is ofcourse a programmable electric gearing setting in the drive with which you can tune down the encoder resolution to any value to make the required step frequency lower for the motion controller, but the drive then still uses the full encoder resolution for commutation and the PID control, so it is still perfectly smooth even if the encoder resolution is geared down to a low value.
And with a 131000 encoder resolution a 10-20 encoder counts following error still gives a very little rotation angle error.