Test G Code shows my Y is off (backlash)

[vios]

This is a small weakness that I also found on Shapeoko.
I dont think that the problem can be eliminate at all but it can be reduce by loosen the bottom V wheel.
As long it can be driven easily that should be enough.

i was thinking to put another stepper for the Y axis once I got the new motor driver from AtomSoft.

[aymanme]

I've actually taken a different approach to fix this problem. I won't say its limited to Y, but Y (at least one mine) is very flexible and therefore easy to get "out of square". You can think of it as a mass-spring system where the mass is the spindle stuck into some material. As you move further away from the Y motor, it is going to apply an increasing torque force to the Y axis.

On mine I an shake it with my hands which is probably way to flexible. My approach (unproven at this point, since I didn't get the parts yet), is to use some gussets to square up the Y axis and apply additional contact points between the Y cross member and the motor plates. This means I'll have to take apart the "X" plates to drill some more holes for the new gussets. In addition (and since it is easy), I am going to add another square piece of Al to stiffen the motor mount where it attaches to the cross member and my new gusset. Ideally I'd do it on both sides, of the plate, but since it is asymmetric, there isn't enough room.

A couple of drawbacks to this approach:
1) it will cost about 30mm or so of Y travel (25mm on one side, and 5mm on the other side).
2) There will be an extra moving mass and unfortunately it is high up on the system which could cause it to vibrate more and become less stable as it accels/decels.
3) It still may not be stiff enough but I am limited on space in where I can add gussets.

I can't do anything to address #1, that is just the way it is. On the other hand, if you are thinking of adding a drive on the other side, you are going to burn some travel there as well since the idler will collide with the Y carriage. On the other hand, the extra drive is lighter and lowers the mass-center, but it is probably not very stiff.

For #2, I am having a couple of different flavors coming of varying strengths and weights. This should allow me to find the best one mix of strength/weight/vibration. I am partial to stainless, but high up on the Y axis might not be the best place for that. It will end up being 12 bolts, washers, nuts, and 6 fairly hefty gussets. I estimated the added mass at 180-240g for the hardware and another 90-100g for the stiffening on the motor mount plate.

I have a couple of ideas for squaring up and stiffening the X axis as well, but I haven't figured out the ones I want to try yet. For the moment, the Y axis is probably the one that needs the most stiffening.

ayman

[jhllt67]

I was thinking along similar lines to you aymanme. I thought maybe it could be done with 2 cables in an X pattern across the gantry, maybe with some turnbuckles to tension them. You could drill holes just above the makeslide at the front and back of the plates, though you would need to be careful with the front placement so it doesn't interfere with the spindle. Shouldn't be any loss of x travel with this method.

[aymanme]

Actually for X I'll probably use corners that fit the extrusion and just bolt those on the outside to a plate. With the small adjustment in the connectors, I should be able to square it up. Again, Y is much less stiff than I feel comfortable using, so that is where I spent my effort so far. I may be wrong, but I am skeptical that drives on both sides or even a shaft is going to help the Y racking much.

[AusRobotics]

I think I'm getting the same problem. Is your cut on the test gcode looking anything like this? :

https://www.dropbox.com/s/sskauape3pckw ... .54.11.jpg

[alpha]

I never used a end mill to cut anything... I will test the S shape soon and post a picture here.

OK, I did draw 8 S shapes from the ShapeOko Hello World file on different areas and as more left the S is as more misshaped it looks like... so I guess your issue is related this this too.

[Colecago]

Crap, this seems like a large issue. Looking forward to see how your fixes do.

Are there people out there with ShapeOko's who don't have this issue?

[AusRobotics]

After Backlash Compensation. I made this cut towards the bottom left of the machine (near the origin) so it should highlight the backlash from the gantry's dead side.

shape.png (100.26 KiB) Viewed 317 times

Before the fixes and software backlash compensation

beforeBacklashComp.png (188.54 KiB) Viewed 317 times

Solutions found so far:


1) Software backlash compensation: measure the backlash of the Y Axis especially, and do the X while you are at it. I measured the Y backlash (with head jogged to the left side) to be about .03mm all the way to 1mm (it was erratic and each measurement was different). I averaged it out to be about .05mm which I added to the LinuxCnc INI file. This seemed to help alot! This is documented well here:

http://www.britishideas.com/2008/06/19/ ... pensation/

With this finding I plan to add another motor to the gantry left and have ordered a few more idlers and bits from Inventables.


2) Slow down, work less: reduce the depth of cut and feed rate, increase spindle speed. This is to minimise the force of the cutting tool against the material, this reducing flex and therefore backlash.

3) Stiffen up: mount the cutting tool further up on the Z Axis to reduce it's length. I've noticed the Z axis flexing quite alot during cutting and this is causing alot of backlash and a "flinging" effect where it bounces forward after coming out of a cut.

[Colecago]

shape.png

beforeBacklashComp.png

Solutions found so far:


1) Software backlash compensation: measure the backlash of the Y Axis especially, and do the X while you are at it. I measured the Y backlash (with head jogged to the left side) to be about .03mm all the way to 1mm (it was erratic and each measurement was different). I averaged it out to be about .05mm which I added to the LinuxCnc INI file. This seemed to help alot! This is documented well here:

http://www.britishideas.com/2008/06/19/ ... pensation/

With this finding I plan to add another motor to the gantry left and have ordered a few more idlers and bits from Inventables.


2) Slow down, work less: reduce the depth of cut and feed rate, increase spindle speed. This is to minimise the force of the cutting tool against the material, this reducing flex and therefore backlash.

3) Stiffen up: mount the cutting tool further up on the Z Axis to reduce it's length. I've noticed the Z axis flexing quite alot during cutting and this is causing alot of backlash and a "flinging" effect where it bounces forward after coming out of a cut.

Glad that turned out so well. I wonder if we can ask for a backlash compensation setting for TinyG.

[alpha]

I did add a second stepper to the y axis and this fixed the issue for me, see picture attachment...