Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: zygoat on February 11, 2013, 07:23:50 PM

hi, i have been using mach for a while now on a cnc router that i built myself completely from scratch.
here is the situation. i have calibrated all the axis. everything moves accurately. ie if i move the x axis 1" or 30" it goes exactly that far +/ .005"
when i use any of the mach wizards, or g code from a cam program. the parts come out the wrong size.
if i run a hand written g code program they come out the right size.
here is an example of a simple g code program i wrote to cut the inside of a 1" square. i am using a .125" (1/8") cutter
g01
m03
z.125
x.9375
y.9375
x.9375
y.9375
x.9375
y0
z.1
and here is an example of the squares it cuts.
(http://i168.photobucket.com/albums/u166/Zygoat_D/CNC/20130209_143616_zpsac2a2929.jpg)
and here is an example of what happens with the mach 3 pocketing wizard.
(http://i168.photobucket.com/albums/u166/Zygoat_D/CNC/IMG_20130211_170432_zpsc16363c4.jpg)
as you can see they are not the right size. i am at a loss as to what could be causing this to happen.
i have also cut some parts with multiple passes for depth. ie. .100" doc. 5 times for .500" total. all of the steps line up exactly so i dont think i am loosing steps. and the manual code is also right on.
any ideas or help are greatly appreciated.

Cutter diameter?

Ooops. disregard. Didn't notice it was off .050"

Strange one indeed, the pockets actually look more in proportion than the profiles but that is likely just a trick of the camera.
Do you have backlash comp enabled
Are you feeding both at the same rates?
Hood

Backlash comp is off, but I will double check it again.
Both are being cut exactly the same. I have every thing set very low. Feed rate is 6 ipm accel is set equally low. The g force is .009

Servo's? If set up on loose side this may be your issue.

The machine uses stepper motors. Sorry, I should have specified that.

YOU really need to run the drawing test to eliminate lost steps during cutting. The test are not really equal the way you did them as the pocket test runs from inside out and has a LOT more cutting involved before you get to finish size.
In the drawing test run the first set of code and on the secound test cut only run the last set of perimeter code.
Looks like lost steps to me.
(;) TP

where does one find that drawing test?

Put a pen or pencil in your spindle and draw the objects.
(;) TP

ok, i get that. ill try it.
isn't there a test pattern tho, like a bunch of fractals strung together that is supposed to search out lost steps?

when i use any of the mach wizards, or g code from a cam program. the parts come out the wrong size.
if i run a hand written g code program they come out the right size.
Now that IS wierd!
Can you repeat the test cuts with something other than timber? Maybe plastic?
Cheers

Can you post the entire hand written code file and also the equivilant wizard written code file?
If both gcode files give the same instructions to Mach then your problem is with a setting or the machine.
Measuring wood for accuracy of cut is useless.
RICH

ok, i have been spending a little more time with the machine. i think the steps per unit for the y axis were off, and still are a little bit, but i am getting pretty close.
i have been working on getting a good drawing test, had to spend some time rigging up a pen mount, i think i have something workable for now.
here is a good example of the drawing test.
(http://i168.photobucket.com/albums/u166/Zygoat_D/CNC/20130224_103338_zps23d96472.jpg)
the g code is for a 1/8" end mill 3" square pocket, so the magic number should be 2.875" which the x axis is dead on in the left square.
i think several things happened on the right one.
the x axis motor was getting very hot.
the x axis did not move very smoothly at that speed.
the pen did not write as well at that speed.
as you can see in both the y axis is not quite the correct size. now the x axis (the one that is perfect) is a direct drive from the motor, while the y has a belt drive gear reduction. i am trying to determine if there is some backlash in that setup. any tips on how to measure this or work it out are welcome!

A hot stepper motor can start to behave strangely.
Post both gcode files so we can make sure that the code is correct for the intended machining.
RICH

http://www.machsupport.com/forum/index.php/topic,12513.0.html will help you check backlash
http://www.machsupport.com/forum/index.php/topic,12512.0.html Will help with steps per. Rich also has a steps per Calculator in the members doc board.
Brett

ill post the g code when i get a chance, although i am pretty sure now that it is not a g code problem, it is looking like it is mechanical.

this is the code for a 3" square pocket 1/8" (.125") endmill. i don't think the code is right. the final pass should be at 1.375" i still think my y steps are off and possibly a small amount of backlash.
G00 X0 Y0
G0 Z0.199
G01 Z0.001 F6
X0 F6
X0.025 Y0.025
Y0.025
X0.025
Y0.025
X0.025
X0.05 Y0.05
Y0.05
X0.05
Y0.05
X0.05
X0.075 Y0.075
Y0.075
X0.075
Y0.075
X0.075
X0.1 Y0.1
Y0.1
X0.1
Y0.1
X0.1
X0.125 Y0.125
Y0.125
X0.125
Y0.125
X0.125
X0.15 Y0.15
Y0.15
X0.15
Y0.15
X0.15
X0.175 Y0.175
Y0.175
X0.175
Y0.175
X0.175
X0.2 Y0.2
Y0.2
X0.2
Y0.2
X0.2
X0.225 Y0.225
Y0.225
X0.225
Y0.225
X0.225
X0.25 Y0.25
Y0.25
X0.25
Y0.25
X0.25
X0.275 Y0.275
Y0.275
X0.275
Y0.275
X0.275
X0.3 Y0.3
Y0.3
X0.3
Y0.3
X0.3
X0.325 Y0.325
Y0.325
X0.325
Y0.325
X0.325
X0.35 Y0.35
Y0.35
X0.35
Y0.35
X0.35
X0.375 Y0.375
Y0.375
X0.375
Y0.375
X0.375
X0.4 Y0.4
Y0.4
X0.4
Y0.4
X0.4
X0.425 Y0.425
Y0.425
X0.425
Y0.425
X0.425
X0.45 Y0.45
Y0.45
X0.45
Y0.45
X0.45
X0.475 Y0.475
Y0.475
X0.475
Y0.475
X0.475
X0.5 Y0.5
Y0.5
X0.5
Y0.5
X0.5
X0.525 Y0.525
Y0.525
X0.525
Y0.525
X0.525
X0.55 Y0.55
Y0.55
X0.55
Y0.55
X0.55
X0.575 Y0.575
Y0.575
X0.575
Y0.575
X0.575
X0.6 Y0.6
Y0.6
X0.6
Y0.6
X0.6
X0.625 Y0.625
Y0.625
X0.625
Y0.625
X0.625
X0.65 Y0.65
Y0.65
X0.65
Y0.65
X0.65
X0.675 Y0.675
Y0.675
X0.675
Y0.675
X0.675
X0.7 Y0.7
Y0.7
X0.7
Y0.7
X0.7
X0.725 Y0.725
Y0.725
X0.725
Y0.725
X0.725
X0.75 Y0.75
Y0.75
X0.75
Y0.75
X0.75
X0.775 Y0.775
Y0.775
X0.775
Y0.775
X0.775
X0.8 Y0.8
Y0.8
X0.8
Y0.8
X0.8
X0.825 Y0.825
Y0.825
X0.825
Y0.825
X0.825
X0.85 Y0.85
Y0.85
X0.85
Y0.85
X0.85
X0.875 Y0.875
Y0.875
X0.875
Y0.875
X0.875
X0.9 Y0.9
Y0.9
X0.9
Y0.9
X0.9
X0.925 Y0.925
Y0.925
X0.925
Y0.925
X0.925
X0.95 Y0.95
Y0.95
X0.95
Y0.95
X0.95
X0.975 Y0.975
Y0.975
X0.975
Y0.975
X0.975
X1 Y1
Y1
X1
Y1
X1
X1.025 Y1.025
Y1.025
X1.025
Y1.025
X1.025
X1.05 Y1.05
Y1.05
X1.05
Y1.05
X1.05
X1.075 Y1.075
Y1.075
X1.075
Y1.075
X1.075
X1.1 Y1.1
Y1.1
X1.1
Y1.1
X1.1
X1.125 Y1.125
Y1.125
X1.125
Y1.125
X1.125
X1.15 Y1.15
Y1.15
X1.15
Y1.15
X1.15
X1.175 Y1.175
Y1.175
X1.175
Y1.175
X1.175
X1.2 Y1.2
Y1.2
X1.2
Y1.2
X1.2
X1.225 Y1.225
Y1.225
X1.225
Y1.225
X1.225
X1.25 Y1.25
Y1.25
X1.25
Y1.25
X1.25
X1.275 Y1.275
Y1.275
X1.275
Y1.275
X1.275
X1.3 Y1.3
Y1.3
X1.3
Y1.3
X1.3
X1.325 Y1.325
Y1.325
X1.325
Y1.325
X1.325
X1.35 Y1.35
Y1.35
X1.35
Y1.35
X1.35
X1.375 Y1.375
Y1.375
X1.375
Y1.375
X1.375
X1.4 Y1.4
Y1.4
X1.4
Y1.4
X1.4
X1.425 Y1.425
Y1.425
X1.425
Y1.425
X1.425
X1.4375 Y1.4375
Y1.4375
X1.4375
Y1.4375
X1.4375
G00 Z0.1
G00 Z0.1
M5 M9
M30

here is a screen capture of the settings
(http://i168.photobucket.com/albums/u166/Zygoat_D/CNC/3insquarepocket_zps5507ce33.jpg)
the "clearance" setting was set to .2000" when the above g code was generated. so i could see that being part of the problem. but when i set it lower i get the message show at the bottom of the screen. "msg.clearance <= step depth"

Make sure your steps per are correct, then deterine how much backlash you have. The code looks ok.
RICH

status update!
i have gotten it all sorted out. i finally had the time to sit down and properly calculate the steps per for the y axis. the reality of it is still off a small amount from the calculated, but i know have it very accurate and repeatable. there was also a small amount of backlash that i am now compensating for as well. between the 2 the machine is now right on the money. at least in the y axis.
now i need to fix the x axis. this was the better of the 2 but now that the y is so dialed in, x is now the problem child.
i already know the problem here as well. it is direct drive and can not make small enough steps for the accuracy i am trying to achieve. so gear reduction is in order.
Thanks again to everybody who helped along the way. much appreciated!

:) Glad your makeing progress and thanks for updating us!
Brett