Delta geometry off, tilting effector plate

[Highcooley]

Hey everybody

In order to get my Kraken extruder working properly, I am currently facing a problem with my printer geometry. After some email exchange with Michael, I think it's time to open up the discussion.

I observe the end effector plate of the printer slightly tilting when I move it along the X axis. Close to the X pillar, the left nozzle gets lifted off the plate by about 0.3mm compared to the right nozzle which gets pressed against the plate. The opposite effect happens at the Y pillar, but it seems not to be as severe as on the X side. At this moment, I am only using these two nozzles, so I cannot comment for any tilting in the Y axis.

Michael and I discussed geometry and angle adjustments and I tried to follow these instructions:
https://www.youtube.com/watch?v=qrrO0elGXJM
https://www.youtube.com/watch?v=E-92PSNINO0
Ironically, they both successfully use my bed levelling aid which I honestly was never able to successfully print (it always gets dragged off the plate, especially the small measuring points which slic3r prints separately).

I still don't get how people are able to precisely measure distances and angle deviations mid-air on this round design without expensive professional measuring equipment. However, I did correct some slight deviations in perpendicularity of the pillars to the printing plate. The angles of the three pillars seem to be right (measured to about 0.5° accuracy), but I could probably still tweak these, as I get some high and low points in positions on the opposite side of the pillars.

All that tried, the main issue still exists and thought over twice can actually only happen if
a) some rods don’t have the same length
b) some rod axles are not exactly mounted in parallel toward each other

According to my measuring and mostly trial and error, I think with about 90% certainty, that the cause of this problem is the Z rods tilting the plate. If I twist the Z pillar or the Z carriage with some force, the tilting gets resolved.

But I still don’t know the root of this problem. The rods seem to have the same length, the axles seem to sit flush in the shafts, and even the distance of the two carriage plates is the same on both sides of the pillar. Also the clearances of the pillar to the different wooden parts at the bottom of the pillar look very even. If somebody has a creative idea how I can measure the position of the axles against to the rest of the geometry, I would be very grateful. So far, I measured the Z position of the carriage axle ends to the printer plate and they both seem to be at the same height.

Have other people been confronted with such issues? What have they done to resolve them? Is there some old mechanics trick to measure such things as axle orientation to circular geometries?

[rpress]

I had the same problem with my v2 rods. I got the carbon rod kit from tridprinting and the effector is now flat.

[Polygonhell]

How are the water cooling lines for the Kraken routed?, when I routed mine to the side of them printer, it was enough force to twist the platform at the edges of the build area.
One of the issues with delta geometry in general is that as the arms get closer to the edge of the build area, the resistance to twisting greatly reduces, and the SeeMECNC arms aren't very stiff which doesn't help the situation.

[Renha]

Polygonhell, am I right that if arms would be longer, problem would be shorter?

[Polygonhell]

Polygonhell, am I right that if arms would be longer, problem would be shorter?

Yes, as the arms get closer to horizontal, they lose the ability to control the twisting, longer arms would provide a steeper angle, but it's something of a trade off.

[Highcooley]

How are the water cooling lines for the Kraken routed?, when I routed mine to the side of them printer, it was enough force to twist the platform at the edges of the build area.
One of the issues with delta geometry in general is that as the arms get closer to the edge of the build area, the resistance to twisting greatly reduces, and the SeeMECNC arms aren't very stiff which doesn't help the situation.

I checked the water cooling lines as well as the main umbilical and in some circumstances the plate gets tilted ever so slightly. However, this is by no means as problematic as the main tilting I experience.

[RegB]

I have been having some thoughts on this lately.
ONE thing that we COULD measure and match during assembly is the degree to which pairs of arms are parallel.
Simple approach; BEFORE attaching the axle carriers to the cheapskates insert the axles and slide the U joints on.
Insert the axles in the effector plate and slide U-joints on.
1) measure across the U-joint arm pins of each cheapskate axle, call these a, b, c.
2) measure across the U-joint arm pins of the effector plate, call these 1, 2, 3.
This gives (center to center plus pin diameter) ignore the pin diameter.
3) swap a. b and c around to get best matches with 1, 2 and 3.
LABEL everything so you assemble what you have matched.
Check that tightening the axle carriers to the cheapskates doesn't change anything.

If you had access to thin precision washers with 1/8 holes.... maybe you could shim out the narrowest ones.
I think there may be advantage to washers inboard of the U-joints anyway, as long as all four within a pair of arms are the same thickness.


EDIT;
Another thing that can be measured quite simply during assembly is how parallel the cheapskate axles are to the plane of the glass plate.
I've just been messing around with this, so what I am about to write probably needs work; criticism is welcome, etc.

1) Start with the hot end, delta arms and effector plate OFF.
2) Clean the glass plate and clip it to the snow flake as usual.
3) Align a carpenter's square to the end of a cheapskate axle and bring the skates down with the knob as when finding Z0 until the axle JUST TOUCHES the square. Record the height.
3a) Repeat for the other end of the same cheapskate axle.
3b) Repeat 3 and 3a for the other two axles.

HOPEFULLY all 6 readings are the same.
I do NOT KNOW what differences would be "significant", I "haven't done the math", but will try to tweak mine to minimize the differences.

Working on this....

Another EDIT:
Part of my reason for thinking about this stuff is that I have been dong a couple of upgrades, veryyyyy slowlyyyyy, but as I re-assemble things
I am looking for opportunities to "tighten up tolerances" - - without over tightening (-:
I haven't been able to dedicate much time to this, but today I tried to check the skate delta arm axles for parallelism to the build plate, as above.
SURPRISE - manually moving the skates up and down with the knob turned out to be very imprecise, i.e. not repeatable.
IOW the skates aren't always precisely where the stepping motor sends them.
The belts are tight, everything is smooth - ESPECIALLY the new trick trucks, which ARE adjusted.
The weight of the effector platform, delta arms and hot end is not there - those would put a bit more load on belt tension, and of course the new trick trucks are significantly lighter than the old cheapskates.
I'll hang some weight on them and try again.
to be continued...

[Highcooley]

Thank you all for your most valuable inputs!

One last thing I have to check (again) is print surface warping. I don't use a metal plate between the heater plate and the glass yet. But I recognised, that after heating for about an hour, the "tilting" seems to be a bit less. What might be is, that I easily compensated the warping by adjusting the printer radius, but the problem recurs with the two parallel nozzles. All right, back to step 1: Get a metal disc. I found aluminium cake-plates for bakeries to be about the right thing to get off the shelf. Now I only need a shop which ships to the Swiss market.

[Generic Default]

I had to deal with this problem until recently. The arms must all be EXACTLY the same length, and they much be very parallel. Rigidity is also important. On one side of the plate, the effector pitches to one direction, and on the other side of the plate it pitches the opposite direction. Two nozzles on this travel axis will end up being level at some point between the two ends of the plate, with one nozzle being higher than the other on the edges of the plate.

It's one of the reasons I designed the Tri hotend to have the nozzles as close together as possible. The further your nozzles are, the more uneven they are when tilted.

I now have magnetic arms on my rostock which are +-0.04mm length consistency and very straight. They have greatly reduced my tilting effector. The backlash is zero too, so it eliminates tilting on direction changes. Tighten your belts too.

I just got in a few very very very flat MIC-6 aluminum plates yesterday for a good price, but since you're in Switzerland shipping would be too expensive. But there are plenty of machine shops and metalworking places in Switzerland so see if you can get some precision tooling plate locally.


EDIT: here is the tread I started a while ago on this same subject.
http://forum.seemecnc.com/viewtopic.php?f=37&t=2696