Faceting/Linearization leading to poor surface finish

Hi All,

I tried searching for this issue in the forum, but I may not be using the correct terminology to find the info I'm looking for....Or it may not be covered.

I have a Tormach 1100MX and I'm seeing some poor surface finishes on arcs and circles. Normally when I would see this with industrial machines, it would be indicative of linearization - the CAM system splitting arcs or circles into splines, or small line segments. Knowing this, I investigated the CAM output and found it to be true circles - G2's and G3's. I have also properly levelled, and squared the machine with a precision level. This leads me to believe this form of faceting is coming from the control in some fashion. It is not a result of cutting forces as at least one, and sometimes several spring passes are used. It is also not backlash, as I have adjusted this quite well, and these lines/facets don't appear in-line with the X and Y axis'

The faceting is relatively faint, but unfortunately it appears to be bad enough that it will hinder the lower end of the tolerance range that will be achievable with the machine as circularity will have some significant deviations.

I have initiated a support ticket with Tormach and the response was basically "Yeah, we are aware of this - we are looking into it". While I'm sure they are looking into it, I've never been a fan of sitting on my hands and waiting. I'm hoping that perhaps someone here might be able to point me towards the guts of the pathpilot controller, and where I might be able to look to attempting to tune or "play with" some background parameters that may help to solve this issue.

I have attached a picture for reference. It may be a little difficult to see, but there are small lines that are parallel to the cylindrical axis of the circle or arc segments.

Thanks!



 

That's a tough one, could maybe be a servo tuning issue?

What were the feed rates on the cut in question? If it is super fast might be running into a resolution issue from servo-thread speed. (but I think that is unlikely.)

Hi Todd,

Finishing passes were with a 3/8” endmill going roughly 75ipm. Max rapid of the machine in X and Y is 300ipm.

Could be something to look into though. Pretty sure it still shows up at slower feeds but I’ll slow my next part down and see if there’s a difference.

Thanks for the suggestion. Anything/everything helps

Even at 300 IPM, if I guess a 2 inch diameter,
the chord error from doing linear interpolation
at PP's 1 KHz servo thread rate is only about
3 microinches Peak to Peak.

I would think servo tuning, mechanical stick/slip
or step resolution are more likely culprits.

Confirmed to not be feed related.

Two pictures of parts attached. One at 75ipm finishing federate, and the other I slowed down to 20 ipm. Looks the same to me.

Apparently there are “many” machines exhibiting this according to tormach. But they aren’t yet sure if it’s their stepper or servo only machines…mine is servo.

Mechanical stick slip is least likely. As I said, I’ve tuned the backlash, and I would have seen the sticking/slipping during that process.

Servo tuning may be a good way to go. These clear paths probably have a procedure for that…hope it doesn’t require a windows machine!

I shall dig into the tuning process.

Thanks for the suggestion.

A side observation.

The smaller the radius, the less apparent it seems to be.

First picture in original post was a 2.375” circle

Last two picture I posted were a small segment of what is actually a 12” circle.

This same part also has 1/2” rads on the contour, the the lines are significantly less apparent on those. Pics attached. In the one picture you can see the transition from the 12” arc to the small 1/2” arc, and the lines calm down drastically.

Some further experimentation. 

it appears to also show up on angled paths. Not just circles. Basically any time both servos have to work in tandem to produce a line or arc this rippling is showing up.

more pictures:

I'm curious about the step/dir resolution of the Cleapath
servos used. ISTR the cheapest ones only have
800 steps/turn resolution which would give ~1/4 mill
steps with a 5mm pitch ballscrew.