Diagonal Rod Ends

I am having a horrendous time getting my cheap Kossel clone to level out correctly. I have recently switched over to Repetier from Marlin. Completed all the auto correction tasks. When I do a dry run, the print head digs between the x and y tower and lifts high toward the Z tower. (the Y axis?). I have always been able to level out the center and tower positions, but just can't remove the opposing tower errors (I call them ~x, ~y, and ~z points).

The first time I noticed an {x,y,z} Max Length entry error. (which by the way, isn't even an entry in the repetier.com configuration widget for the 92.8 version). It corrected the big sweeping error, but still have a deviation of 3+ mm.

The one thing I've noticed and need to ask is: While my diagonal rods are fairly even (<.5mm tolerance), I've noticed the fisheye joints vary in their movement. 3 rods are consistently tight, and the others seem to swivel just fine. This I believe is the result of uneven cooling from the manufacture.

Can you tell me how much of a detriment this might be? Could it really have an effect on my leveling issues? (btw, 'levelling' IS spelled with 2 'L's just like 'cancelled' should be and I curse the idiot programmer of auto-correct who is trying to change the face of the english language).

Please help a guy who bit off more than he could chew when he made a delta his first 3d printer.

Some people buy more rod ends than they need and sort out the sloppy ones.
I took another route, I've learned from RC-cars:
Just put a drop of CA glue between the fisheye and the plastic and wait until the drop disappears in the gap.
Then carefully rotate the ball until the CA is cured.
You might mess up one or two, but a little oil film on the ball keeps it from sticking.

Edited 1 time(s). Last edit at 05/02/2017 03:18AM by o_lampe.

0.5 mm difference in the diagonal rod length is a lot. It should be below 0.1 mm, ideally below 0.05 mm. Anyway I doubt it would result in the leveling error of 3 mm by itself. Your tower Z position is probably wrong too. It looks like it is more far away in reality than what is specified in the printer calibration. But it is hard to tell. This all depends on the precise geometry of your printer. Try to get a z-probe a calibrate numerically: dc42's web calibration calculator or this.

Thanks for both replies.

I will try to loosen the joints as well as get some supplies to build new rods (having a surplus to choose from).

I can't recall if I've seen any good methods for checking the build geometry. Knowing if all towers are equidistant from center should require some sort of reference point. The bed itself could be off center, and I don't know if the tolerances on where the cross braces sit are reliable. With that being said, it does look like the z could be .5-1.0 mm further out when compared to the others. I'll have to take the printer apart and file down the corners a bit. The molded parts may seem nice, but I believe their production process has created enough non conformity.

Trying to measure tower angles is even worse. I can print small footprint items, but not anything large enough to use for calibration.

I will try to follow up if anybody is interested.

Start at the bottom and check the distances between each pair of adjacent towers and set them to all be the same(may be worth cutting a piece of wood or such to act as a gauge).

Repeat for the upper triangle and then make sure all the towers are vertical Don't worry about the bed for this you can adjust it later and it doesn't have to be dead centre it will just mean a slight reduction in print volume if it is out.

Make sure all the Diagonal rods are the same length ideally to within 0.05mm. Make sure there is no slop in the joints and the spacing of the joints are equal on all 6 pairs

That should improve things remarkably.

HTH

Doug

Edited 1 time(s). Last edit at 05/02/2017 03:02PM by dougal1957.

There is a subtle bit of tuning that made a LOT of difference to my calibration geometry.
The DELTA_SMOOTH_ROD_OFFSET is important here. You won't get everything to work well unless the distance from the center of your build plate (where the nozzle is placed when you are at X=0 and Y=0) to the leading edge of the V-groove your carriage wheels run in, are all exactly the same. That measurement will be your DELTA_SMOOTH_ROD_OFFSET, or really close to it. This is fine-tuned by adjusting your tower endstops until they all hit together in such a way to give you this. Until this is so, you will have a super hard time "leveling" your bed and getting your prints scaled properly.
Adjusting those endstops and measuring your DSRO is your first step to getting bed "leveling" and part scales correct.

DLC (who found this out the hard way, and had a "head smack" moment when DSRO was finally made obvious.)

A 0.5mm difference in spacing between the top and bottom of the parallel diagonal rods is massive and will easily cause 3-4mm of deviation from the plate. I had a similar situation as you, flying high at two towers and digging in at the other. The closer I get the diagonal rods the flatter it travels. Someone suggested cutting tiny washers out of coke cans if need be.