-
Accuracy problems
Hi
I have a sherline lathe with sherline driver box and i am having the following problems.
After about 40 lines of G code from vector cad cam my lath is about 0.40 mm to 0.60 mm out of tolrence. If i jog say 30mm using mach 2 its perfect no problem but if i run g code i end up geting accuracy problems.
I use vector cad to design my part and turn it in to g code then export it to mach 2
If i used limit switchs to reference zero every few steps would that make my lath more accuret
I would be greatful for any sugestions
C Morton
-
Adding reference switches will not correct for an inaccurate machine. If you run the program without cutting material, do you still lose position?
It sounds like you may be losing steps during the rapid commands. Change all the G00 codes to G01 and see if you still lose position. If not, reduce your max velocity, which is what is in effect during rapids.
This assumes that you have accounted for backlash, part movement in the vise, tool movement from excessive pressure, etc.
Fred Smith - IMService
-
Hi
Thanks for the reply i am going to spend saterday trying to get this mashine accurate.
I still loose position when not Cutting. the driver box is a sherline one with there stepper motors but.
As the box is brand new and if performance is still bad i will have to buy gecko drives but id rather not have to shell out 160 pounds plus shipping.
Thanks chris
-
Hi have slowed the lathe down to a feed rate of 10
Both the x and the y axis are out exactaly the same. 0.04mm
but when single steping there is no error only when i run say 60 lines of G code, the error when i run 132 lines of G code is 0.04mm.
Is the sherline driver box a good controller or have i been riped off.
chris
-
Sherline Controller
Your program probably has rapid moves from the VectorCad-Cam lathe roughing function. In the Vector NC editor, use the Edit-Replace function to temporarily (for testing only) change the G00 commands to G01.
The controller is fine.
Did you change the G00's to G01's like I suggested? Changing the feed rate will not affect the rapids. Only if you change all the moves to feeds will you be able to determine tha maximum traverse rate that the machine and controller will handle.
Sherline has a support forum at
http://www.sherline.com/forums
Most likely someone there will be able to tell you how to adjust your setup to prevent lost steps.
Fred Smith - IMService
http://www.cadcamcadcam.com
-
hi again
I have tried what you have suggested by taking out the G00 comands to no sucsess
After 132 lines of G vode the error is .04mm this doubles when i run the same gcode again.
In mach 2 i can set the step pulse longer than 15us as specified by sherline its set as 14 and the pre direction time wont go any higher than 5us seconds.
The mashine i run it on has only mach2 on no other softwere is running in memory.
I am starting to think this controller was designed for EMC and Not any other softwere.
As there are other people who have had problems with the controller.
chris
-
.04 MM is a far cry from the .4-.6 MM that you reported when this thread started. Is that radial or diametral? On many hobby class machines .04 MM will be within the probable error due to backlash, lead screw variation, heat effects, etc, etc, etc.
However, since you can repeat the error, some more questions may be in order.
Are you operating in Absolute mode(G90)? Incremental mode has some problems that may affect how you measure the error.
Are you returning to the same position when the program ends? In other words, do you program a final motion that returns the tool to the exact starting position?
How are you measuring the error? With an indicator? Is the error accumulative or does it vary + and -.
You didn't say you were trying to rig up Mach2 with the Sherline controller, what other surprises do you have? There is a yahoo group for Mach 2 support.
Fred Smith - IMService
-
The .4 to .6 was for a large g code program
0.04 was jus for 132 lines of code the error doubles when you doble the lenghth of the G code so you can imagine the error is masive when the code is 100 lines long.
When the program ends i hit the go to zero button, by looking at the dials and looking at the DRO on the screen i deduce the error.
When is say there is no error joging, there will be but it is so small and for every move the errors ADD.
The error is accumulative and it is always - . I have sheilded all cables checked the earth, checked analogue and digital noise. but all seem OK. The only point i can think of is this mack 2 wont let me have a pulse time bigger than 14us and sherline specify 15.
I run it on a dedicated PC with nothing running in memory.
Oh both X and Y have the same error exactaly
Thanks
chris
-
Does the error correspond to the number of lines of G-code, or to the distance traveled. You could quickly check this with a program from Vector.
If distance traveled, you may need to adjust the steps per mm(inch).
You didn't respond to my questions about using an indicator to independently verify the error. Your screw pitch may not be set correctly, that would account for the problem that you are having, and the fact that both X and Y are the same. From what you have posted so far, the problem is not clearly software/ hardware, or just adjustment.
Fred Smith - IMService
htp://www.cadcamcadcam.com
.
-
Chris,
I would try a more methodical approach than just "running some Gcode". Make a single axis program, consisting of very short, medium and very long movements. Check the distance moved after each single command is read.
Check the distance moved in jog mode. Does it correspond to the display? Again, try short, medium and long movements.
Check how accurate the machine returns to position after running a simple positioning loop: forward and back to the same two endpoints. Do it once, then several times, to see if you can find a systematic error. This is how you troubleshoot, not by running actual programs.
-
Hi
Both axis are out by the same amount
Y axis
Moving the Y axis in and out 20 mm 2 times there is an error of 0.01
4 times an error of 0.02mm
6 times 0.03mm
8 Times 0.04mm
10 times 0.05mm
And the same for the X
I have tested the lathe dials by cutting a some metal manually and measuring it with a micrometer and its perfect from this I can deduce there is nothing wrong with the mechanics of the lathe.
If i repeat the above with a g code program the error is the same
Chris
-
Check these settings on the controller and CAD/CAM; unfortunately, i am unfamilliar with sherlines.
Does your controller support modal commands?
Are you specifying G21 for metric in the code, and 3 or 4 digits to the left of your decimal?
The circular interpolation tolerances on the controller vs the CAD/CAM.
Circular movement expected by the controller? Full, Half, or Quarter (you can try arcs as lines for testing)
G02 and G03 requirements like Coordinate values (incremental or absolute)
My coordinates have to be absolute, but use incremental I, J, K during circular moves.
The debounce count set too low can cause this as well on some controllers.
Just a few thoughts
-
The G Code is G21
But it should make no difference when i jog and the error is the same.
I dont know if my controller supports modal commands as the specification is poor
chris
-
Yeah, sometimes the documentation is real vague to say the least. You may want to contact the manufacturer for some better detailed information.
I'm sure it will be resolved soon.
Kenny
-
On reflection i think the sherline 8760 driver was a bad buy.
Basicaly if i cant get it working i am screwed as i cant take it back is its implied its ment to run with emc and i am tring to get it run with MACH 2.
Whats even worse is the sherline spec for there stuff it very inaccurate and some of its compleatly wrong. In the mean time i have a lathe that does not work.
Any help is very welcome
Chris
-
So Chris,
The length of the move does not seem to affect the error?
What about several shorter moves in the same direction? Is the error always dependant on a reversal in direction?
What is the screw lead and drive ratio (if any) that you are using on this machine? Would it be possible that you have a fraction of a turn rounding error in your controller setup? What I am suggesting, if there is a fraction of a step required to match one revolution of the mechanical, with one revolution in controller logic? You are talking about a tiny cumulative error and this is why I am thinking this.
-
The length of the move does not seem to affect the error
it must be dependent on direction change.
the screw is 1 turn per mm, its a true metric screw and dials.
The error seems to be change of direction related
-
And are your motors hooked 1:1 ratio? (Its like pulling teeth here, help me out would ya :D)
-
Hi
what is 1 to 1 ratio No Gear box
I am very new to cnc
800 steps per unit
minibea KM23 stepper motors.
Its definitly change of direction related not distance
The problem i have is i dont trust sherline specifications as there is loads of errors in the spec
Motor Connectors:
∙ 5 pin DIN male
∙ Five to 30 volts (better performance at higher voltage)
∙ 2 amp unipolar configuration
∙ 1600 steps per revolution (microstepping)
if i use 1600 i only get half distance
∙ Step pulses are active low.
∙ Step pulses should be at least 15 microseconds long.
It only works with active high steps
i can only get 14us long step pulses
-
The drive ratio would include any kind of drive pulley arrangement which might be used between your motors and your screws. Presumably, this could be a toothed pulley with toothed belt of some sort, or maybe it is direct coupled?
What is the thread pitch of your metric screws?
What I'm trying to determine, is whether whole steps are being lost, and how many.
If you move backwards and forwards by a single step, what do you see happening?