Effector tilt tuning

Jackal · 3 Oct 2016, 12:59

I have built myself a Kossel with E3D v6, magnetic joints, flex3drive and David's mini IR probe.
It prints pretty well but I am trying to level the bed a bit better.
The IR probe result is very repeatable, but I seem to have a bit of end effector tilt along the vertical line from the Z tower.
The bed a about 0.2mm higher than center below the Z tower and 0.2mm lower than center at the opposite end.
I tried moving the bed lower end higher but the IR probe kept giving the same offset result.
I am currently compensating this with the H parameter in bed.g which the height is just calculated from the 0.2mm proportionally along the probe points which gives just ok result.

Is there a better way to do this?

dc42 · 3 Oct 2016, 13:28

Preferably, eliminate the varying effector tilt. Possible causes include:

Rods in a pair not quite the same length
Bearing spacing not quite the same at top and bottom of a pair of rods
Carriage rotated about its face so that the two bearings on it are not the same height (a pair of bearings on the effector at different heights will have the same effect)
Carriage rotated about the Z axis, e.g, due to a warped carriage, or using wheeled carriages with unequal spacers between the carriage and the 3 or 4 wheels
Play in the joints. the sideways force of the Bowden rube or drive cable will typically change direction with effector XY position, causing the play to be taken up in different directions.

I have a small bulls-eye spirit level mounted on the effector so that I can monitor the tilt.

Jackal · 3 Oct 2016, 14:02

Thanks Davaid.
I have bought my rods from Haydn from the google delta forum, they are quite accurate and tight with the magnetic balls so I think length and play in joints shouldn't be a problem.
I printed the carriage and end effector myself so they are the same spacing (at least in the CAD they are).
My carriages are on linear rails so that should be fine.
I actually have bought a small spirit level but currently my end effector is not large enough for me to mount it.
I believe the biggest factor that can cause the tilt is from the flexible shaft from the flex3drive.
Maybe I should print a blank end effector for the spirit level first to eliminate all other factors before going back dealing with the flexidrive.

dc42 · 3 Oct 2016, 15:01

Even though your carriages are on linear rails, they may still be warped, causing rotation about the Z axis. They could also be rotated about their faces, for example if there is too much clearance in the mounting holes in the carriages, or if the tapped holes in the slider are not exactly lined up.

Jackal · 3 Oct 2016, 16:47

Ah, I didn't think about that.
By the way, I have never confirmed that the bed leveling issue is due to end effector tilt, I just speculated that it is the cause.
Is there any other place I should look that may cause such problem with the IR probe leveling?

dc42 · 3 Oct 2016, 17:35

Differences in the bed surface can cause small differences in trigger height, for example if you use glass coated with a layer of glue or hairspray and you don't coat all of the bed because you will only be printing on a small part of it. But effector tilt is very common in deltas.

peridot · 3 Oct 2016, 19:19

Also you may want to check the spacing of the magnetic balls, on both the effector and the carriages, with digital calipers. In spite of being printed on the same printer from the same G-code, my three carriages have slightly different spacings. And the effector spacings are not all the same either.

Of course, I don't know how much difference in spacing adds up to (say) a one-degree tilt of the effector. Someone (me I guess) needs to do some sensitivity analysis on the delta mechanics. This would help designers focus on the alignments that matter most.

piankoc · 3 Oct 2016, 19:37

I have a bulls eye level as well, but was trying to find the best way to attach it to the effector. My floors are not level so my machine is off a little to start as well. I tried using blue tack but it didn't seem to hold the "level" after sitting for a minute would shift back making it hard to determine what is an issue with the level moving and what is an issue with the blue tack releasing a little.

My first delta (custom rostock) has quite a bit of tilt - enough that i could easily notice by just looking at it, but this one (custom kossel) doesn't appear to have any but I would like to be sure and confirm with the bullseye level.

My second thought was hot glue which will hopefully give me enough time to position to initial "level" but harden and not move later on, while also being semi removable if needed.

How did everyone else attach their level?

Jackal · 3 Oct 2016, 19:53

peridot, I agree that there might be differences on the same printed part. I believe that how tight you screw the steel balls on to the carriages and end effector will also cause a slight difference on how high they sit, which I will need to find a way to make sure they are the same.
Unfortunately, unless I go for some CNC parts, I can only 'trust' my printed parts are good enough.
I would love to see the analysis on factors of end effector tilting, too bad I am not that good at maths.

piankoc, my current end effector also does not have a good place to mount the level. Since I am using magnetic joints, I will probably print and swap a blank end effector with a slot for the level to test things out.

dc42 · 3 Oct 2016, 20:36

I attached the spirit level flat on the effector using double sided adhesive tape. Then I levelled the printer by putting pieces of paper under the three feet.

peridot · 3 Oct 2016, 22:01

Jackal, printing is pretty cheap; maybe print extras and pick the parts that are closest to spec. I measured and marked my effector and tried to match carriage spacings to the effector widths. Plus the plan is once the machine is better tuned, re-print the parts with better dimensional accuracy. Note that delta calibration, left to adjust six parameters, may trade away dimensional accuracy for a flatter bed.

Zesty_Lykle · 4 Oct 2016, 07:11

Jackal,
I think you can eliminate the forces of the Flex3drive from tilting your effector.
I have done a lot of testing on that with the Zesty Nimble and was not able to detect any influence. So put your mind at ease there.

As to accuracy of the prints, when I need a really accurate print to test things out or to make use everything is straight, there are 2 things you can do.
1. You can build a jig that you use to mount the parts, so that you know that each part is identical, regardless of the actual geometry of the part.
2. You can print those parts via Shapeways and know that it is accurate enough for the purpose.

That is what I do when I sell an upgrade kit as it needs to be spot on.

DjDemonD · 4 Oct 2016, 08:34

Two bulls eye spirit levels, one on the lower frame, one on the effector. Bluetac works but I rolled it into a ball put it under the spirit level, pressed it down, then removed the spirit level and and trimmed away any bluetac that was at the edge of the spirit level (mine have a little indent in the base), then reattached them so they are held in place but perfectly flat against the surface to which they are attached.

I have adjustable feet, so I levelled the printer first, then moved the effector around to look for tilt.

I am using a metal effector with magnets all at the same level in the effector, rods all custom made to the same length, and to be fair I can't say I've noticed any effector tilt. I have had magnets disconnect so its not all smooth sailing.

Auto calibration comes back sometimes as low as 0.02 deviation. Any issues I've had so far come down to differing IR reflectivity of aluminium with hairspray unevenly applied (soon to be solved with a printbite surface), and getting a good nozzle to bed first layer height which without babystepping (yet) is a bit tedious to set.

Jackal · 4 Oct 2016, 08:59

@Zesty_Lykle:

Jackal,
I think you can eliminate the forces of the Flex3drive from tilting your effector.
I have done a lot of testing on that with the Zesty Nimble and was not able to detect any influence. So put your mind at ease there.

As to accuracy of the prints, when I need a really accurate print to test things out or to make use everything is straight, there are 2 things you can do.
1. You can build a jig that you use to mount the parts, so that you know that each part is identical, regardless of the actual geometry of the part.
2. You can print those parts via Shapeways and know that it is accurate enough for the purpose.

That is what I do when I sell an upgrade kit as it needs to be spot on.

Thanks for the input, I hope that's the case. The flexible shaft is a bit too long for my frame, if I need to mount it as straight as possible then the motor will have to be quite a bit above the frame. The Duet wifi board will be in the way as well, so I currently mount the motor on the side with the shaft in a inverted U shape.
I am looking forward seeing the Zesty Nimble, maybe even try it out myself

Zesty_Lykle · 4 Oct 2016, 14:28

I mounted the Stepper at the bottom of the frame, have a nice loop going up and over to the effector.
My Delta is approx 1 meter tall and I have 90 mm long drive cable.