Circles have flats

[Austire]

Hi, thanks for all the replies, I had another look at the machine too see if I could find anything. Everything seems tight and no play found. Rigidity could be a possibility although I will look into this again.
I ran a simple circle program to see if I could see anything happening, I put the tips of my fingers on the table and I can feel very slight pulses as it’s doing the circle. (Spindle and Z axis where disabled. I’m now thinking I could be tuning if the servo motors which is something I didn’t do. Know to find out how to tune Chinese servo motors.

 

[???]

Cut something at a 45 degree angle in the XY. Tapers very quickly show up if there is an issue with the step settings. You say the ball screw is 10mm, is that the pitch or diameter.

 

[Austire]

Cut something at a 45 degree angle in the XY. Tapers very quickly show up if there is an issue with the step settings. You say the ball screw is 10mm, is that the pitch or diameter.

10mm is the pitch. And I have the servo set to 10000 steps

 

[EmGo]

And I have the servo set to 10000 steps

You have servos "set to steps" ? hmmm ...

 

[Micmac1]

Fusion 360 you need to do two things, I've had this issue on our haas mills before (oddly enough it does not do it on the fanuc machines as bad). First is built in tolerence needs to be changed, its in the edit side of CAM software, change that to like .0001, second is you need to use a contour finish pass to clean it up. Additonally, if the drawing was imported from solid works, if you zoom in on the imported model they always show up as small lines instead of arc's, we usually just redraw the feature if this is the case. Also using a larger diameter endmill may help

 

[eKretz]

Looking at the floor and back wall, my first guess would be excessive FPT.

 

[memphisjed]

You have servos "set to steps" ? hmmm ...

Lots of drives use the step/direction. If op was using steppers it would be a smoother motion. Steppers are better when you can use them. Of course linears are better- just next generation of the stepper.
Second micmak observation jives with what I have noticed. Only I found setting tolerance on adaptive paths high is better results. .004 is good.

 

[BOB-OO]

I'm impressed w the whole PM community - good advice to DIY guy. We are living in the bizzaro world.

 

[eKretz]

I'm impressed w the whole PM community - good advice to DIY guy. We are living in the bizzaro world.

As has always been the case, almost everyone here is usually happy to help with procedural and technique problems. It's just the hobby level machines themselves that are off limits for discussion. I've never seen anyone here balk at helping field questions from even a beginner with a process question.

 

[Austire]

Hi all,
Thanks for all the reply’s and help. I have t had much of a chance this week to get at the machine. I had a quick look at it last night, I did a small manual change on two gains and kept bringing them up slightly until I felt the vibration was gone.
When I did this I was getting cleaner cuts and the flats on the circle did improve. When I did these changes I was getting an oval not a circle but the square cut out dimensions where good. I will have to investigate this further over the weekend and see if I can get software to adjust the gains.
I also noticed that when the compressor kicks in I’m also felling a very slight vibration at on the base of the mill, I will move the compressor away from the mill.
Hopefully I will get at the mill over the weekend and will keep an update on any progress.
Know try and find software to adjust the PID gains that will work with the servo drive

 

[Mtndew]

10mm is the pitch. And I have the servo set to 10000 steps

Can you set it to 1 billion steps?

 

[Austire]

Can you set it to 1 billion steps?

No, from what I have read it’s a17bit encoder, off the top of my head I think it’s 136000

 

[Mtndew]

No, from what I have read it’s a17bit encoder, off the top of my head I think it’s 136000

I wonder if setting it to max would make a big difference.

 

[eKretz]

What is your programmed FPT?

 

[Mtndew]

Hi all thanks for the reply’s. As far as I can tell there is no or very little backlash, I did do a finishing contour finishing cut with .4mm. I have attached the start of the G code I used

Ok so you're at 20k rpm and a feedrate of 533.3 IPM?
Using a 4mm end mill. How many flutes? uncoated or coated? If coated, what kind of coating?

That feedrate is way too fast.

 

[CarbideBob]

I see big flats at the normal 0, 90.180, 270. of the circle. This is one thing.
Also see smaller flats in the arc itself.
Both a concern?
Many may say your machine is not big or rigid enough..... Bullshit. It is a poor machinist that blames his tools.
Tuning a step input to a servo drive can be fussy.

 

[EmGo]

Tuning a step input to a servo drive can be fussy.

That whole thing of talking 'steps' when he said he had servos had me going hunh ?

 

[SteveEx30]

I also noticed that when the compressor kicks in I’m also felling a very slight vibration at on the base of the mill, I will move the compressor away from the mill.

How heavy is your machine? Need a forklift to pick it up or 2 buddies.

 

[Austire]

Hi all, a bit of an update. I hand a go at trying to adjust the gains manually but with out the software it’s very tricky. I also adjusted arc tolerance settings in the UNCNC software that controls the machine.
I have include a pic of how it is cutting now. There is an improvement but still some slight flats. The post is 6mm and the measured accuracy of the post and rectangle cut is good.
I will have to try and get some software so I can monitor the gains as I adjust I think I’m slightly over shooting or undershooting in the servos.
I will try and tune again over the next couple of days and see if I can get any improvement.
The feed rate is in mm and the machine would need two people to lift it.

 

[CarbideBob]

Have your heard of the circle, diamond, square test? Old school but common in the NC days.
I assume you are putting out pluses (or steps) to a servo amp running in position control mode.
(servos amps come in torque, velocity, and position control. Most newer ones can do any of the three by choice of jumper or parameter)
Tuning the PID loop in the amp here can be a bit of a challenge and what works at one speed may not work at another.
No software for monitoring changes here? You are flying blind without some sort of chart of servo error in the system response.