Inaccurate Arcs and Circles at Higher Speeds

Hi - I'm looking for direction in addressing a problem I came across recently. I have a stepper table using #25 chains for both X and Y axis. I ran a plasma cut piece at a higher feed rate and straight lines were OK, but the arcs and circles were somewhat shifted. They were not oblong, but shifted over. If I ran the same piece at half the rate, the paths were proportionally correct. At first I thought I was loosing steps, but am now looking elsewhere because the machine returns to the correct spot when going back to X0Y0 after each cut. The one difference I am aware of is the velocity is a bit different when comparing the X and Y axis portion of the ini file. I am using G64 P0.005 on a metric setup.

Could the velocity difference (250 vs 275) be causing issues at higher feed rates or do I need to look elsewhere? Thank you...

PK
Also: Axis drive slip/slop has been checked and is tight.

Could the velocity difference (250 vs 275) be causing issues at higher feed rates or do I need to look elsewhere? Thank you...

In theory probable not, the fastest speed of a coordinated move is only as fast as the slowest axis can keep up with.

Have you tried adding Q- to your G64 command?

By shifted do you mean the circle arcs are round but not in the correct location?

Sure there is not flex in the system, more so at higher speeds?

Have you tried adjusting your acceleration?

Rick G

Thanks for your time, Rick,

I did forget to update my gcode in adding the Q- part to the G64.

I will attempt to add a picture to help show what I'm experiencing. I don't believe there is flex at high speeds, but maybe sluggish axis as this table was built for a router CNC. I can run rapids over 15,000mm/min and this problem job was specified at 5000mm/min with some arcs running around 2500. When I ran it at a maximum 2500mm/min, all was well.

I have not tried adjusting the acceleration as I set it up to run as fast as possible with as quick acceleration so as to keep the torch moving as consistently as possible. At least that's what I think my goal should be. I am making a list to try each suggestion next time I "get" the garage. Probably next weekend...

Again - Thanks for your help, Rick.

Paul

Thanks John,

I wondered how overboard that .005 was, but threw it in my code without thinking it through considering my machine is setup for metric. I will keep your information on my list, too.

Is it correct for me to think that your 500IPM might not be something I *should* expect due to probable computer latency differences? I believe (without checking my notes) that my maximum jitter was 13,000 using Keling KL-4030 drivers to 495oz motors as my setup.

I appreciate this!
Paul

John's info covers the bases great.

Take a look at your acceleration.

Also keep in mind that as you are not geeting any joint errors and the machine returns to the same postion each time it would seem you are not loosing steps.
I beleive you will find that a plasma gantry is lighter than a router gantry so it is easier to accelerate, and it does not need the strenght of a router table.

Rick G

Thanks guys,

I have a part programmed a couple different ways: one with G61 and then G64. I will check my acceleration and see how things go next time I get a chance. I do have a 36v power supply, but can't seem to find motor specs.

A goal I have is to use the CNC router setup to plasma cut parts to build a plasma gantry/table for a setup better suited for plasma cutting.

Thanks again. Will post results here in case it may help someone in the future.

PK

I ran those parts recently. First ran the one programmed with the G61 and the part was perfect. I next ran the same part using G64 P0.5 Q0.75 and, surprisingly, it too was perfect proportionally. ! I believe I didn't make any other changes, so I'm not sure what caused the problem.

Mostly unrelated, I noticed the post processor I am using was reducing the feed rate on smaller radii (smaller than 10mm). I see more dross on those smaller areas. Should I slow the torch down in these cases?

Thanks for your help....

PK