Circles aren't circular??

[antiklesys 3]

Hi all.

I have finished assembling my Ultimaker a week ago and I'm pretty satisfied with the results of the prints I've made so far.

There is just one thing which is bugging my head.

I am printing a couple of small pipes, with inner diameter of 6.2 mm and outer diameter of 7.3 . I interrupted the print after 5mm of the pipe were printed and I noticed the circle wasn't a perfect circle. It wasn't an oval either. I will try to post a picture later on to show what I exactly mean...but has any other experienced this?

Is it how it is supposed to work or is there something wrong with the alignment of my axis? I have printed a test cube and that one looked fine, double checked the design of the circle and that one's perfect too.

Any ideas?

Thanks all in advance

[Daid 306]

If the axis are unaligned then you get an oval, but I've never seen anyone post any proof of this. I think the machine will jam before you get the axes unaligned enough for an oval.

However, more likely, is that you have a "rounded square". Which is caused by not having enough tension in the long belts. This causes some backlash effects which causes a circle to go a bit square. You can solve this with some printed belt tensioners (your first or printed object for your own printer!) viewtopic.php?f=19&t=604&p=3242&hilit=tension#p3242

[antiklesys 3]

Hi Daid, I think you got the "rounded square" thing, it's not much like a square but I think that may be the issue.

Is there any way than printing circles to see if it's an issue related to belt tensions??

I have printed 3 sets of tensioners but I need an advice on what to do with my current situation, please see below my experiments and results:

http://www.thingiverse.com/thing:19892

Good tensioner but doesn't work on the Y belts, it blocks between the top wood and the belt block...shame :(

http://www.thingiverse.com/thing:19818

Interesting design but difficult to regulate properly. Easy slides back during machine operation, creating random uneven tensions across the belts.

http://www.thingiverse.com/thing:10082

Basic design but clever, I'm currently using this one on all belts but I see that they have default uneven tensions. Any ideas? Can't put more than one per belt or it will prevent the axes from reaching the end-stops.

[IRobertI 521]

I've posted these before but it can't hurt to do it again.

http://www.thingiverse.com/thing:17058

 

[Daid 306]

Lowering your print speed might also help, not sure how fast you are printing, but it can help till you solve the actual issue.

The design by Misguided is what I'm using. You can make those thicker to get more tension. However, on the Y belts you'll run into the same problem as with the wave tensioner.

Another thing you might want to check is the tension on the small belts. I don't know it it can cause "circles goes square" problems. But it never hurts to check those. If they feel a bit weak, then unscrew the motors a bit, push the motor down, and screw it down very tight.

[glasswalker 0]

The short belts can definitely cause backlash. Backlash is the primary cause of circles being out of round.

Any loose belt on a pulley can cause backlash (ability for slop, allowing the pulley to turn slightly before the belt finally "locks" and engages, this allows for a "deadzone" where when the pulley reverses direction it can turn a few steps before the belt kicks back in and starts moving. It means anytime you reverse direction you loose a small amount of movement. so circles (which have a reverse in direction at each side) will loose some of their size in the axis that is affected most.

Other key symptoms of backlash:

- Infill won't line up with perimiter loops on one or more sides, leaving small gaps

- The smaller the circle, the more Wonky it will be (larger circles won't be affected as much, this is because the backlash is fixed, for example .5mm so no matter the size of circle it's still out of round by .5mm)

- Higher speed will result in more backlash effect (by a small margin)

- Printing an H pattern (a flat object in the shape of an H) will result in the inner vertical lines of the H not being lined up. because the head is traveling one way on the top and the other on the bottom, and the backlash shifts one from the other. This can be measured with a caliper to determine your backlash. The H must be printed twice (rotated 90 degrees) so that you check the X and Y axis for backlash.

Some other notes:

The belts and pulleys on the ultimaker are very nice. And should not suffer from ANY backlash under normal operation.

If you are seeing backlash make sure your belts are securely tight, use tensioners if need be.

The long belts shouldn't touch one another under light-moderate finger pressure (they should be tight enough to not allow this)

The short belts should barely deflect at all under finger pressure

If the belts are good and tight, and you're still getting backlash the next thing to check is pulleys. If a pulley is slipping on the rod, or slightly loose, it can cause backlash too. So check all pulleys have their set-screws tight, and aren't slipping (all pulleys, including the ones on the x/y rods, the short belts, and the motors.

Lastly make sure you don't have too much friction on your X/Y rods. High friction normally causes skipped steps and so on, but high friction will also contribute to backlash to some degree.

Hope some of that helps.

[mr_seeker 0]

Well, everyone is talking about the machine itself, but nobody ever wondered why it couldnt be in the STL file? Some 3d programs "smooth" rounded objects. A ball for example looks round, but is in fact made out of a lot of triangles.

Open the STL file and look at the wireframe. How does it look?

[antiklesys 3]

The STL file looks good.

I made some adjustments to the machine yesterday and I managed to improve a bit more with belt tensioning.

Now the circle is slightly more circular than before...as the previous results were more like a rounded square.

I printed a test cube of 5mm side and I had a 0.5 mm discrepancy on the Y side.

After the tensioning this was reduced but it's still more like a rectangle than a cube.

I lubed the axis again and moved the pulleys around a bit to make them be as much as aligned as possible.

Any other ideas on how to improve even more?

[mm_build 1]

I printed a test cube of 5mm side and I had a 0.5 mm discrepancy on the Y side.

After the tensioning this was reduced but it's still more like a rectangle than a cube.

I lubed the axis again and moved the pulleys around a bit to make them be as much as aligned as possible.

Any other ideas on how to improve even more?

Actually, I had this problem for the longest time and never did figure out what was causing it. It just sort of went away after a while. I suppose it had something to do with the firmware. I still have some of my old "rectangular" test cubes.

I would suggest upgrading your firmware to Marlin.

[antiklesys 3]

I have this problem with Marlin, haven't tested if it's the same with 5D but I think it is. :/

[Daid 306]

I have this problem with Marlin, haven't tested if it's the same with 5D but I think it is. :/

No amount of firmware changes will fix a mechanical problem.

[antiklesys 3]

Do you think it would be possible to print anything to calibrate accurately the tension on each belt to make it even?