Precision Issue

[Jeffrey Aley]

I removed the belt to check the x-axis rail. It was fine - I was actually pretty impressed with how tight yet free flowing the rail was for a cheap rail system. The belt itself seemed fine as well.

There doesn't appear to be a set screw, the pulley must be somehow permanently attached to the motor.

This is the stepper driver in my machine: https://www.ebay.com/itm/New-Driver-ACT ... .l4275.c10
It is already a "micro-stepper" so there were no toggles to switch. I did verify that the switches were the same between the two steppers. I tried your suggestion of swapping the two drivers and it didn't change anything.

Next Steps I'm thinking:
1) Swap the x-axis stepper. If I do this, I was also thinking I might upgrade to a 3 phase or maybe even a closed loop stepper motor and driver. Cost is not really a driver at this point, getting it accurate is. Any recommendations on brands, where to buy, etc???
2) Replace the belt and idler pulley while I'm at it. It looks fine, but what the hell...

Anything else I should try here? On a positive note, I'm learning a lot about this machine right now!

Cheap rail system? I assume your machine has Hiwin (clone) rails and bearings.

No set screw? Hmm. I'd put a drop of Loctite on it.

Not sure what you meant by "there were no toggles to switch" on the stepper driver. The photos clearly show a set of DIP switches SW1-SW8 and the corresponding microsteps printed on the case. It might not hurt to do the math to see if the stepper resolution is set anywhere close to the 0.4mm error that you're seeing.

You can certainly end up with a more capable machine by upgrading the stepper to a servo. I'm sure some other members have done so. But fundamentally, the machine you have, as designed, should be capable of much better precision. Clearly SOMETHING is wrong / defective, we just haven't figured out which piece

When replacing the belt & pulley, do the mod (twist the belt) that Russ describes in his last "Curtains" video. You'll have to search the video index to see which one it is.

[Jason Chancellor]

Cheap rail system? I assume your machine has Hiwin (clone) rails and bearings.

No set screw? Hmm. I'd put a drop of Loctite on it.

Not sure what you meant by "there were no toggles to switch" on the stepper driver. The photos clearly show a set of DIP switches SW1-SW8 and the corresponding microsteps printed on the case. It might not hurt to do the math to see if the stepper resolution is set anywhere close to the 0.4mm error that you're seeing.

You can certainly end up with a more capable machine by upgrading the stepper to a servo. I'm sure some other members have done so. But fundamentally, the machine you have, as designed, should be capable of much better precision. Clearly SOMETHING is wrong / defective, we just haven't figured out which piece

When replacing the belt & pulley, do the mod (twist the belt) that Russ describes in his last "Curtains" video. You'll have to search the video index to see which one it is.

I just meant the whole machine was cheap, but the rails seem nice - yes they are Hiwin clones based on color.

I'm going to take out the stepper in question today to get a closer look, maybe I'm just missing the set screw. I'll take your advice if not there.

Fair on the toggles - I meant there was no toggle that switched it to microstepping, but clearly I'm in an area I dont understand. Here is what the switches are set to - Are these appropriate?
Y Axis
Peak = 3.31A
RMS = 2.36A
Full Current
Pulse / Rev Table = 2000

Love the Ross videos, have watched many of them but not that one. Will do today!

[Jeffrey Aley]

Here is what the switches are set to - Are these appropriate?
Y Axis
Peak = 3.31A
RMS = 2.36A
Full Current
Pulse / Rev Table = 2000

Basically, you're saying that you're getting 2000 pulses (steps) per revolution.
We need to translate that into linear distance on the axis.
You'll need to measure the diameter of the drive pulley. I'm going to pretend that it's 22mm diameter for this exercise.
The circumference of the pulley pi x D, so in this case it's 3.14 * 22mm = 69.08mm.
So one full revolution (2000 steps) moves the head 69.08mm. A single step would move the head 69.08mm / 2000 = 0.034mm.
Clearly, this number is an order of magnitude smaller than the error you're seeing (0.40). So that tells us that the stepper is CAPABLE of plenty of precision, and the problem lies elsewhere. Note: Even though the stepper is capable, we cannot conclude that it is DELIVERING that precision.

Regards,

-Jeff

[Jason Chancellor]

I'm back! I replaced the x-axis stepper with a 3 phase 1.2degree stepper & driver. I also replaced the belt and the stepper pulley. I confirmed that the rails are tight & smooth. So essentially I replaced all the moving parts on the dang x-axis except maybe the idler pulley.

Tested, and no improvement. Earlier in this thread, there was an idea of using rdworks dot mode to take the Y-axis completely out of the equation. So I tried that - basically put a horizontal line in rdworks and set the dot mode to pulse (dot) every 10mm.

Then I measured every 10 dots which SHOULD equal 100mm. On the left side of the bed, it IS 100.03mm. However, as we move across the X-axis, 10 dots turns into 99.55mm in the center of the bed, and 99.34mm on the right side of bed. Interestingly enough, this behavior is 100% consistent. I can run the job multiple times and it duplicates the above dots exactly over and over.

Based on the consistency, Im guessing its not skipping, but I'm 100% at a loss here.

Help me obi wan

dots.jpg

[Jeffrey Aley]

I'm back! I replaced the x-axis stepper with a 3 phase 1.2degree stepper & driver. I also replaced the belt and the stepper pulley. I confirmed that the rails are tight & smooth. So essentially I replaced all the moving parts on the dang x-axis except maybe the idler pulley.

Tested, and no improvement. Earlier in this thread, there was an idea of using rdworks dot mode to take the Y-axis completely out of the equation. So I tried that - basically put a horizontal line in rdworks and set the dot mode to pulse (dot) every 10mm.

Then I measured every 10 dots which SHOULD equal 100mm. On the left side of the bed, it IS 100.03mm. However, as we move across the X-axis, 10 dots turns into 99.55mm in the center of the bed, and 99.34mm on the right side of bed. Interestingly enough, this behavior is 100% consistent. I can run the job multiple times and it duplicates the above dots exactly over and over.

Based on the consistency, Im guessing its not skipping, but I'm 100% at a loss here.

Help me obi wan

dots.jpg

Wow, this is a difficult problem. The theories below seem very improbable, but you've eliminated all the probable stuff.

Theory #1: the rail is not parallel to the bed. Actually, to explain this behavior, the rail would have to be curved relative to the bed. Please measure both for flatness. It is more likely that the bed is warped than the rail. While you're at it, measure the distance between the laser nozzle and the bed at a few points. If you're handy with CAD, you can probably figure out how much warpage would be needed to cause a 0.5mm error across a 300mm distance.

Theory #2: as the HV power supply (or the tube) heats up, the "turn-on" time for the laser beam changes. I notice that the distance between dots is increasing as we go from right to left. Experiment: draw a second dotted line, immediately below the first one so that the first one cuts right to left and the second left to right [or vice versa], without any appreciable time for the power supply to cool off. TBH, I don't expect this theory to be correct, based on your other data, but it's a simple thing to try, and allows us to rule out thermal changes as the cause of the problem.

Good luck!

[Jason Chancellor]

I found something interesting wanted to share.

Using my dot measurements from before which shot a pulse every 10mm along the x-axis I did some measurements and put it on a chart. I measured every 10 dots which should have equated to 100mm which is what I calibrated the machine at.

This chart should show a flat line at 100mm, but instead shows a Sine curve almost. It made no difference if I used my 3 phase stepper or 2 phase stepper. Also, I spent a lot of time trying to get the belt very well aligned, and that didn't make a difference either.

Clearly I'm not losing steps here. But I'm leaning more and more like it has something to do with the belt or pulleys or something. Any thoughts?

dot-spacing-chart.JPG

[Jason Chancellor]

Wanted to post the eventual solution here - thank god!

Ultimately, I changed a million things with this machine, but I believe the ultimate root cause was belt stretch being caused by the pulleys being out of alignment with the gantry rails. It was out of alignment on both the Y axis and the Z axis.

What really threw me for a loop is when I shimmed and better aligned the pulleys it originally did not seem to fix the problem, but in the end I had managed to throw the machine out of alignment when messing with pulleys. Once I got the machine back in alignment it is cutting much more precise.

The laser precision is still not perfect and the Thunder Laser I also use is much more precise, but my machine is pretty close and good enough for my needs. Ultimately I don't think this Red and Black laser from eBay is designed in a may that it will ever be as precise as the Thunder - but hey, its also half the price - so I guess you get what you pay for,

Wanted to say thank you for everyones help. It is greatly appreciated!

[Gene Uselman]

Thanks for posting the solution, people having similar problems in the future will be helped. As you say, there is a huge range of manufacturing methods
and quality control in China, and [hopefully] spending more money gets you a better product. The Chinese are not good at communication with the English speaking world [and probably the non-Chinese speaking world].

[Josh Bern]

So I'm not really sure if this is the same issue but my lines during scan mode are looking very wavy rather than straight.
I put the interval to .4 in order to adjust the reverse interval.

Here are the pictures of what it is supposed to look like and what it actually looks like on my laser: https://photos.app.goo.gl/tk3Ti2khnizFAK3p9

Could it be a loose laser head or something?
Please let me know. Thank you!

[Doug Fisher]

It could be a number of things but the most common cause is a lens that is not snug in the lens tube. Check to see if it is snug. Snug, not tight because you might crack it.