Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: Ansen on December 05, 2012, 10:25:50 PM
-
I am pecking a fairly demanding pattern in plexiglas and I was hoping that backlash compensation would fix a subtle, but noticable banding pattern in my dots.
Attached is a photo showing (top) backlash OFF and (bottom) backlash ON. There are some big gaps (0.25 in) in the pattern as well as some completely missing holes. I filled the holes with ink to make them show up better.
Can anyone explain this? My backlash settings are X=.009, Y=0.0065, Z=0, shuttle speed=0.005.
Also, the drilling pattern starts on the left and moves to the right in 1 inch vertical strips. Notice how the patttern is about 1/2 inch longer with backlash compensation turned on due to the skipping.
Thanks
-
Hi Ansen,
I am most interested in this but cannot quite see exactly what you mean by 'banding'.
I notice variations in 'shade' in the top example but it appears that not all your dots have been filled with ink.
Presumably you are producing the pattern bi-directionally and that is why you think backlash comp may help but I would like to hear more details of the problem you are seeing without backlash comp being used.
Tweakie.
-
The backlash settings may be wrong or actualy varying some. By varying i am meaning that the force to remove the backlash
varies. More likely to be indicative of play in the thrust bearing.
RICH
-
Here is the real problem that I was trying to address using backlash compensation. The small "rivulets" correspond to the algorithm used in the drilling order, but they do not appear in the original design. They only show up when small holes are relatively close together. I know, pretty minor, but they do show up disturbingly well at a distance.
My theory was that the X axis was doing a lot of back and forth as it travels up the Y to complete a block of holes. Backlash would "short" the distance to the next hole, creating a predictatable gap between the blocks of holes.
-
I understand what you are seeing now.
How is the image created ?? ie. from a horizontal axis travel across the whole picture, one line at a time or some other method.
Tweakie.
Addition;
Just read your first post again and assume that you are doing it horizontally in 1" wide vertical stripes. Although it is a very small amount backlash could well be the issue here.
-
Imagine the whole image broken up into 1 in. squares. The holes are then drilled starting from bottom to top in each square. Hoes are sorted by Y, so the X does a lot of traveling back and forth, but only within that 1 in. box. Then it moves to the next box.
-
OK, I understand how it was done.
If the bottom left hand dot for each box was always approached in the same direction then any backlash would have been taken up in the same direction so would it really be the cause of the problem - I am not sure.
I know it does not solve this issue and I am not sure if it will be of any interest to you but there are alternative methods of producing images from dots or drill holes which produce the image one complete horizontal line at a time which would not be affected by a small amount of backlash which occurs with bi-directional working.
Although I produced this image using a laser it is essentially just a series of small holes or dots which could just as easily have been produced with a drill (although it may take considerably longer to complete due the Z axis movements).
Tweakie.
-
That image looks great.
I'm really more interested in the sculptural quality of the image. I create four different layers in acrylic and paint each layers' holes in a different color--typically cyan, magenta, yellow and black. These are the four colors used in almost all color printing.
Regardless of my original problem, so far the cure has been worse than the disease! The gaps in the original test image are completly software generated. All I did was turn backlash comp on. This pattern of holes is totally repeatable.
I obviously have some kind of hardware issue with 0.009 of backlash, but that doesn't explain why I would get giant 1/4" gaps in the output.
-
Basically, as I see it, you have 2 separate issues here ‘Banding’ and ‘Backlash comp’.
I have never used backlash comp so that issue is outside my comfort zone.
If you would care to Zip and email me a GCode sample file (preferably with approx. equal sized dots) which exhibits the banding issue I would be happy to run it on my machine and post the result. This should (with luck) determine if the banding is machine or software related.
Tweakie.
-
Ansen,
Post the file first so it can be checked. Don't think the 1/4" gaps are due to backlash.
RICH
-
Rich,
I am absolutely sure that the 1/4" gaps are due to backlash compensation. Here's that file. If you want to, run it once with BC on and once with it off. I would be interested to see if you get the same results.
As for the banding, I fixed a farly significant "hardware" issue last night which brought my backlash down considerably. X=0.003, Y=0.001, Z=0. I'm going to try the face file again this morning with BC off.
Thanks for your help, guys!
-
Anset,
Dry running in MACH here with or without BC made no difference.
Backplotting the gcode dosen't show the large gaps nor any missed holes.
So not the gcode in my opinion.
Attach your xml file.
How many runs did you do with BC? Just the one or numerous? If more than one with BC, are the results around the same line number?
ie; Is the skipping location rather consistant?
Steppers or servo's?
Rather hard to say it's a BC setting or your machine at this point.
RICH
-
Ansen,
In your gcode file, the first gap in the bottom row of dots (right side of block 1 to left side of block 2) is 0.1278765 inches wide. This is not caused by backlash. Shows up in my tool path very plainly.
Regards,
John Champlain
www.picengrave.com
-
The spacing varies between the dots........so does the depth.....
RICH
-
OK,
Further exam shows I was not looking at the code properly. My apologies.
This may not be a backlash problem, but may be a screw pitch variation issue or sticky axis bearings . I see similar banding with my laser diode machine when burning images if the oilite bearings and shafts are dirty.
John
-
Pic to portray depth changes......
RICH
-
Rich,
I ran the pattern several times with and without backlash compensation. The results were always the same. Gaps and (even missing holes) in the same locations with BC on.
My machine uses steppers on all axis.
Attached is the XML.
Thanks for your help.
-
One more thing: to drill this pattern, I use a v-groove bit with a 53 degree angle. This make the Z axis plunge depth the same as the hole diameter at the surface of the material.
-
Ansen,
At least you have repeatablity for the problem and that is good in the sense that one can try a few things to see what the cause is.
I would start with trying to find out why the first 1/4" gap is happening ( didn't notice any holes missed in the first section).
Lower your velocities 50%, set backlash speed to 50%, you can play with the shuttle setting or leave as is at .01.
Here is my 50-50-50 rule which you may want to consider:
50% - The max velocity is 50% of where your steppers will start to skip
50% - Set the shuttle Wheel setting in configuration to .0050 to .050 ( .5 too slow )
50% - Backlash speed
----------------------------------------------
I have numbered your gcode file and around line 590 , before/at/ or after is where your skipping.
You don't need to run the whole program so you can edit out a lot of code and see the affect with changes.
See attachements.
RICH
-
Rich, Tweakie and all,
Problem solved. After some testing, all I had to do was move the Shuttle Wheel Acceleration to 0.03 instead of 0.01.
Many thanks for your help. I'm off to the races.
--Ansen
http://ansenseale.com
-
Hi Ansen,
I am pleased that you have got it all sorted.
Now I am looking forward to seeing some pictures of your multi-layer acrylic works. ;)
Tweakie.
-
Good going with a change in the setting. :)
RICH