Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: jajohnsen on December 17, 2016, 01:35:56 PM
-
Hello, New to this forum and the world of CNC and Mach3.
I'm trying to gut a test circle of 49mm but it turns out slightly oval, to eliminate as much as possible i've made the circle in the mach3 wizard.
when i measure the circle it is only 48.5mm on the Y and X axis. but when i meassure on 45 degress to x or y it is the correct Dimension.
What am i missing or where should i look to correct this?
here is the g-code
(Code by Newfangled Wizard, 17.12.2016)
(Version 2.86)
(Program Posted for Wood )
G0 G49 G40.1 G17
G80 G50 G90 G98
G21 (mm)
(***** Circular Groove/Cutout *****)
M6 T7
M03 S24000
M9
G00G43 H7 Z1
G00 X20 Y-2.25
G41 P1.5
G01 X22.25 F1000
G03 X24.5 Y0 R2.25 F1000
G01 Z-0.5 F500
G03 X-24.5 Y0 R24.5
X24.5 Y0 R24.5
G00 Z1
G03 X22.25 Y2.25 R2.25
G40
G00 X20 Y2.25
M5 M9
M30
-
I never use tool radius compensation within Mach3 (it's easier for me to let the CAD/CAM software take care of this) so my knowledge is limited.
First, check that you have your tool table set up correctly then try running your code again but first change the G40.1 to G40
Does this produce better roundness and dimension ?
Tweakie.
-
Thank you for your answer.
Changing from g40.1 g40 doesn't change the Dimensions.
if i try the set steps per unit wizard and ask for 100mm i the lengtht is correct in both x and y.
I tried to make a 50x50mm Pocket With the mach wizard and it turns out square but only 49.8mmx49.8mm
-
initially i tried make the circle in fusion 360 and had the same results. With fusion doing the tool comp.
-
Hi John,
By trying Fusion 360 doing the tool comp. and the same error persisting has, I think, shown that it is not a Mach3 error.
From the information you have given I would suspect the problem is a mechanical issue with your machine, perhaps backlash, a loose set screw, etc.
Tweakie.
-
+1 on checking the backlash.
Mike
-
I suspect its tool deflection especially as your movement check without cuts is correct. Have you tried re-cutting the original machined hole with the same program to see if it takes off more material and corrects the dimensions. When doing a circle the machine constantly changes direction in two axis with no pause hence different pressures on the Z axis spindle. That's why I usually do a finishing cut of say 0.5mm to give me a perfect circle as hobbyist machines are not that ridged as production machines.
-
Ok, thank you.
played with backlash and that definitely helped but in (0.47mm for x and y, measured with dial indicator giving x-10, x0, x8,x0 ), but when i try to make a circular pocket I'm getting a small ridge right when it changes direction on x or y
-
yes i think tool deflection might come in here also, will try as you suggested
-
if your getting a small ridge its got to be tool deflection. If your a hobbyist machinist and time is not a problem always do a finishing cut. as the tool wears over a period of time your deflection will increase hence the finished component size reduction. Some of us if we want to maintain a size on a batch of components have a finishing tool which does very little cutting so can be relied on to keep its size. Just for reference I'm talking about cutting metal but the same applies for wood on a reduced scale
-
ok, so this is my result
the original circle is now 49mm measured on the x and y axis, and 49.4mm in the 45 degree angles whitout backlash comp.
with backlash comp it is 49.4mm on all dimensions, but i get a small ridge just where the compensation is happening. and i can se the router is jumping right when its making the adjustment.
running the same program twice doesn't change anything.
i have adjusted the steps some from when i first started and used a dial indicator for measuring.
yes this is a hobby machine Openbuilds OX, my cutter is brand new. and i only make 1-2mm deep cuts in mdf right now when i try.
I will try to make a circle in fusion with a very light finnishing pass.
-
just to clarify, when i said that i can se the router jump, it's the hole gantry moving. if i hold on to the gantry you can feel it's the steppers making a quick change.
-
Have you seen this?
http://www.openbuilds.com/newtest/index.php?threads/ox-cnc-cant-cut-small-circles.6345/
-
Whatever you do, try to run without backlash comp - it can cause more issues than it fixes sometimes, jerks and steps would be two of them.
I was chasing an off-sized result for some time when it turned out to be the tool slightly worn.
-
Yes, i have seen that. I have checked my grub screws and they are as tight i can get them. I had problems with flat spotting like he mentions and that was because of one loose screw. But that is fixed now.
With backlash correction on, the circle looks round now, but with the noticable jump each time it corrects itself. I'm using wery shallow cuts 1-2mm and feed rate of 1000
-
One last check to confirm if you cut an internal square or circle then an external square or circle just cutting on one edge do you get the same size errors. if the inner one is smaller and the outer on bigger its tool deflection. If they are both the same size its your step settings or backlash.
-
ok thank you for the suggestions,
tried to cut 49mm cirkel
inner one Y48.7 x48.65 45degreo 49mm
outer one y48.5 y48.4 45 degree 48.8
this is whitout backlash comp.
i think there is a compination of backlash and deflection going on here. i just redid my my step settings measuring off the gantry with my dial indicator and it's dead on. but if i touch it slightly i can se neadle deflection and if i push some there is no problem getting the nedle move +- 0.3-4 mm
I
-
Hi jajohnsen,
I remember when I was a boy my uncle teaching me to use the lathe. He set up a dial gauge between the two lathe rails then asked
me to grab the rails and try to spread them apart and lo and behold I got about 1-2 thou of movement. I was gobsmacked, how could
I at nine years old deflect such a beefy casting! The concept of flexure of what appears a rigid component has stayed with me ever since.
I am inclined to forget the lesson until I try to work out why some operation that didn't go as I'd hoped only to realise I was relying on components
being absolutely rigid and unless Hookes law is repealed that's never the case.
Craig