Unsolved Slightly off steps/mm sign of wrong belt tension?

[wilriker]

Yesterday I was checking my steps/mm on X, Y and Z. I'll leave out Z here because for my question it doesn't matter.

I am using 16T GT2 pulleys and they should result in 100 steps/mm. After calibrating (using calipers to compare actual move distance with commanded) both axis showed about 100.3 steps/mm. What came to my mind was the idea if this might be due to wrong belt tension. More precisely a slightly too low belt tension.

I think that might be the case because once X or Y changes direction I can see that the belt on the "pushing" side of the pulley slightly bends towards the pulley.

But since I am not an expert on belt mis-tension-artifacts I wanted to ask here if anyone has more profound knowledge on that topic?

EDIT: Of course I am aware that these differences are a) small and b) could result from production tolerances. I am only interested in whether or not wrong belt tension could lead to steps/mm being off.

[Veti]

@wilriker said in Slightly off steps/mm sign of wrong belt tension?:

I think that might be the case because once X or Y changes direction I can see that the belt on the "pushing" side of the pulley slightly bends towards the pulley.

create a file that makes rapid movements back and forth.
then see if the belt changes position during the moves. if it does your pulleys are misaligned.

[gtj0]

Beware of optical illusion. Our brains can erroneously fill in the blanks, especially with moving objects that have patterns in them (like belt and pulley teeth).

To help figure out where that 0.3mm/100mm is coming from, try performing the same test with different travel move length. I.E. 50, 100 and 200. Perform this set by moving BACK to your start position, then measure forward.

Then without making any settings changes, do the same set of measurements without reversing direction. Move +5mm to preload the kinematics, then continue in the same direction for the measurement amount.

This can tell you a couple of things.
For instance:

If the error amount stays proportional to the measurement lengths (50.15, 200.6), then it could indicate that the pulley diameter isn't perfect. That can be corrected with steps/mm.

If the error amount is different for 100mm when measuring with and without direction change, that could indicate slop/backlash, belt tension, or other things.

In any case, proper belt tension is an art form.

[zapta]

@wilriker said in Slightly off steps/mm sign of wrong belt tension?:

I am only interested in whether or not wrong belt tension could lead to steps/mm being off.

I don't think so. Loose belt creates backlash which is fix and independent of the travel distance (e.g. will give the same error for 50mm and 100mm travels). steps/mm is a 'multiplier' and its error is proportional to the travel distance.

Try repeating the calibration for 200mm travel instead of 100mm. Did it doubled your error to 0.6mm?

BTW, thingiverse have this great backlash test https://www.thingiverse.com/thing:10524

[dc42]

@wilriker said in Slightly off steps/mm sign of wrong belt tension?:

I am using 16T GT2 pulleys and they should result in 100 steps/mm. After calibrating (using calipers to compare actual move distance with commanded) both axis showed about 100.3 steps/mm. What came to my mind was the idea if this might be due to wrong belt tension. More precisely a slightly too low belt tension.

I have read that the quoted 2mm tooth pitch for GT2 belts is only valid when the belt is tensioned with far more tension that is usual in a 3D printer. However, backlash is a more likely explanation. Does the size of the error scale with the amount of movement? If it does, then it's likely because of belt tension. If the error remains constant, it's due to backlash.

Some types of geometrical error can also result in the steps/mm being a little out, for example if the section of the belt whose length varies with position is not exactly parallel to the motion direction.

[bret4]

You mention that the belt seems bend as it moves. I had a problem like that and it was the cheap belts had bends in them in one spot that made it move up and down in a way that could be seen near the pulley. Changed belts and the problem went away. Still cheap belts and pulley's are just that. Cheap. I am now going to upgrade my printer to gates belts and pulleys and see if that helps. Cheap belts also can go bad and make lots of black dust around the pulleys.

All that said you most likely only need to adjust your steps per mm to get it to move the right distance if belt tension isn't way off. That is normal printer calibration. Print a 20mm cube and seeing how big it is and adjusting steps per mm until your printer prints 20mm.

[zapta]

@bret4 said in Slightly off steps/mm sign of wrong belt tension?:

You mention that the belt seems bend as it moves. I had a problem like that and it was the cheap belts had bends in them in one spot that made it move up and down in a way that could be seen near the pulley. Changed belts and the problem went away.

This is also my experience when I switched to this allegedly Gates belt https://www.aliexpress.com/item/32853559638.html . When the stepper changes direction, I don't see a change on the loose end of the bend. YMMV.

[wilriker]

Thanks everyone so far! I will try to do some more experiments to find out if this off-value is fixed or scaling and whether it will reproduce if I preload the kinematics or not.

Regarding belts and pulleys: on the X axis I have an arbitrary mix. Pulley, belt and idler are all cheapos and each one from a different source. On Y axis it's a set of one manufacturer (Powge). It's still not Gates but at least I only heard good of them. Anyway both show about the same behavior so I don't think it's a quality issue. But let's wait for the results of the additional tests.