What's causing this?

[Spk64]

You have a lot of connections that need to be tight to keep backlash at bay.
Belts, rack and pinion and your linear rails. Looks like all of them are controlled by pulling tight at setup and tighten bolts. You can get away with this on some things and not others. Rack and pinion really needs a spring to keep the gears meshed at all times. As others pointed out the linear rails are really meant to be assembled the other way.
When you first built the machine it was probably tight and you had acceptable results.
Now that it has some time on it everything is loosening up.

With the machine powered up grab each side of the gantry up right and see if you can move it back and forth. If things are tight it will feel rigid and will not move. A dial indicator would be even better to measure any movement.

Then do the same thing with the z slide and see if you have any movement on the gantry. Again should be rigid with no "slack".


Steve

[bald9eagle]

I do have some slack in one side Steve. I am going to change everything up and try it again. I should just go ahead and work on my gantry risers now instead of waiting. I am far too impatient and rushed through some parts of this thing. I am learning a lesson now. Thanks.

[wizard]

You have a mechanical issue someplace with something moving. It could be a pulley or gear on a shaft or something else. Errors like this aren't backlash they are a failure in the machine someplace. There is a good chance you have a pulley or gear rotating on a shaft. In an extreme case you may have the rack and pinion jumping a tooth. In any event the gross errors you are seeing are not backlash.

Good stuff wiz. I think that I may have things mixed up on my axes. Facing the machine I call movement from front to back the X and left to right the Y. The Y is on my gantry while the X is the dual drive RP. The issue is definitely in my X movement. There is one thing you guys would see in my machine that I know some would be critical of and that is my lack of tension spring on my RP arms. What I did was use the grooves of the 8020 to slide my motor bracket up and down and lock it in. This would cause an issue (in my mind) if it came loose but it would be obvious. The plan is to change that later on. Anyway, I ruled that out. I did find the larger belt pulley had worked loose and had some wobble. Last night I measured the distance between my bearing blocks and end of my rail and found them to be .187 out. I then made a manual move and they were then -.125 out. We are talking about 5/16" movement! When I get home I'm going to tear down the motor and see if the pulley is loose on the shaft.

[wizard]

There isn't a lot of advanatge diagnostic wise to moving all over the table. It is best to stick with one program that is known to create the defect and to use that to resolve your problem. Further it really helps if that program is short and sweet. It should be manually audited at the very least to make sure it is actually doing what it is suppose to do. Better yet would be to write the program yourself. When doing diagnostic work on a machine it is important to know what a program is doing exactly. You literally want to know what every step in the program is doing.

I did another trial when I got home. I mounted a board and did the circles again. I moved about 24" from where I was yesterday on the X and then repeated the program. I ran the program on either end of the board and in the center. I moved about 36" from end to end. The circles would change each time. They look better but still not right. I loosened my bearings a little on the rails in case I was binding a little there.

Loose is not your friend here! You don't want binding of course but that should not be a problem with the rails you have. If you disconnect the drives you should be able to push each axis from end to end without problem and without looseness.

It would really help though if each time you post these messages you would post pictures with axis labels on the work piece.

The circles on the far right did like was mentioned above...looked like a half circle and a shift and another half circle. I've got some more pics but they are a little dark.

If you are getting the same basic profile as before then you still have something loose. I'm leaning strongly to a pulley or gear spinning on a shaft. It is odd where your machine goes out and so obviously so as you would not expect huge acceleration changes there. You could have the machine binding or twisting at that point which is facilitating turning of a pulley or gear on a shaft. There is also the possibility of lost steps if this is a stepper based machine but that should create some obvious noise. This makes me wonder if the same problem exists when cutting air and writing with a marker, try this so that the router noise isn't a problem.

[bald9eagle]

@wizard

I ran it through again on the same program without the router. I can definitely tell that the A axis is loose. I can hear a rattle when it moves and can even see it. Interesting though is that the loose side moves before the tight side does. I would bump it back and forth and would hear the rattle in A while watching X not move. With the motor energized I can move the A fairly easy. The X doesn't move.