X axis step size just changed?

ok this is bizarre. someone reported the x axis was slightly off scale. the machine was configured long ago and shifting is impossible, as it’s set by timing pulley tooth count.

yes, it’s short by about 1%, in x only. i checked with a vector engraving line so no kerf. calipers go to 150mm, at 140mm it’s about 1mm short in x only.
both x and y are 6um in ruida config. setting x step to 5.945um does make it accurate but that’s an absurd way to try to fix something i don’t understand

it doesn’t sound plausible this machine would have a different type of pulley on x vs y- and a difference in tooth count would be larger than this, it has to be built with an integral number of teeth. it would have to be a matter of being a metric on one and inch on the other to be something this small

it doesn’t seem to be losing steps. everything looked accurate just x is 1% too narrow

and this machine has been in heavy use for years. if the x axis was off before, it would have been reported

i had been performance tweaking the machine config last week. but i backed the current config up and loaded a machine config from years ago and wrote it back onto the machine. same 1% too short

so, i looked to see if this was somehow a backlash or loose set screw. i don’t see it, and that would totally mess up bidir rasters but the bidir rasters are fine

i went looking to see if the ruida controller or lb had a backlash comp set wrong… nothing but small comp numbers in LB for rasters

i even had to get a second pair of digital calipers. same numbers.

any idea? this is crazy… i can’t think if anything to investigate in lb, but i have ruled out everything else at this point

A slight digression to set the stage: when I extended the axis of my 3018 CNC machine, I eventually figured out the shiny new leadscrew had a 3.9872 mm thread lead, making the motion 1 mm short over 250 mm. So it’s entirely possible to have the nominal values tell you (well, me) reality can’t possibly be that way. :grin:

Which is obviously not the case with a change in a machine “configured long ago”: something has changed.

However …

The X axis gets most of the action, because it’s heavily used in engraving and right in the fumes.

Perhaps both the belt and drive pulley have worn enough to reduce the effective pitch diameter. The belt isn’t skipping any teeth on the pulley (which would cause abrupt jumps), but the combined wear causes the belt to sit lower in the pulley grooves and not move quite as far as it should on each pulley rotation. Eventually, the belt changes from a toothed timing belt into a smooth band.

Replacing the belt is (relatively) cheap & easy.

While it’s off, take a close look at the drive pulley to make sure its tooth profile hasn’t eroded into a set of smooth gullies. Replacing the pulley may be difficult, as it’s likely a shrink-fit on the motor shaft, but if it’s that bad you may be backed into a corner.

Good hunting …

1 Like

but with a timing pulley, wearing down the surface doesn’t reduce its effective diameter. that’s why the timing belt was created. as long as it doesn’t wear enough to jump a tooth, it’s always going to feed number of teeth * rotations * pitch

As long as the belt has teeth matching the pulley, that’s certainly true.

Imagine wearing both, so the accumulated difference adds up (or subtracts off) over the distances you’ve seen. Nice tight belt with no backlash, but it doesn’t travel quite as far because the pulley is smaller and the belt teeth narrower.

I have poster over my workbench with a list of Debugging Rules, the third one of which is:

Stop thinking and look

It’s served me well over the years, because I tend to rule out impossible things. Like a new leadscrew that must have a 4 mm lead, because that’s what it said in the description. :slightly_frowning_face:

I’m suggesting something that might have the effect you see. I may be wrong (and certainly have been before), but without actually seeing what’s going on in there, neither of us can be sure.

2 Likes

a leadscrew will always have some error. it actually can’t be exactly 4.00000000mm lead

but a 40t timing pulley can’t be 40.4
217t . teeth must be an integral number. and no matter how much it wears, it will feed 40t per rev unless it actually jumps a tooth.

the 5mm belt will be what creates inaccuracy, but the belt hasn’t changed. stretching 1% all down its length all the sudden wouldn’t make sense. the tensioner would detension long before that but the tension seems fine

i had a strong suspicion that the rotary was somehow enabled because that makes it use a different steps per mm parameter. and the u axis config was 6.33mm instead of 6mm which is kinda close to the 1% error we were seeing.

but, 6.33/6 isn’t 1%, the enabling rotary would mess up y but this prob was in x not y, and i checked that the rotary was not enabled, and walked over to a completely different pc with its own install of LB and it still had the 1% error. i did reuse the original ruler project i drafted in the first machine though

i also went into Machine Config and changed the U to 6.000um per step and it didn’t change anything so it’s not using that field for X.

and of course I’d looked at the belts. nothing notable that i could see. and skipping a tooth would be super obvious and create some really dramatic defects in the work, but nothing like that is happening

ok this makes no sense but apparently the htd 5m belt linking the stepper to the x axis was worn enough that it wasn’t acting as a timing belt, more like a tire.

it wasn’t JUST skipping steps, it was freely walking over the teeth and not using them at all. no only was it a small scale error, there was no bumping in its raster where it jumped a tooth, and it was so repeatable that a vector ruler followed by a 1000mm/s raster didn’t screw up or misalign the raster in any way and a second pass at the same job landed the ruler’s vertical ticks in the exact dame place. all while acting like a tire and not a timing belt

and i had that belt off last night. it appeared somewhat worn but not that bad i had ordered a spare awhile ago but found it needed 12mm wide and I had gotten a 15mm. well, sat down with a good pair of scissors and that 15mm belt now identifies as a 12mm. fixed all the probs

crazy!

1 Like

Take a picture of that setup, because on my machine the X axis motor directly drives the long belt to the laser head:

I must remember about little belts, because they will surely wear out before the longer belts running the axes show any problems.

Good find!

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.