I was trying to cut a hinge at the exact halfway point.

[Joseph Shawa]

The halves were coming out different sizes.

So I scored 1mm acrylic with a different sized rectangles to see how they compared with my 100mm rectangle that I calibrated the "pulse unit" with.

100mm rectangle on Lasercut comes out 100mm


So I thought to try other sizes and drew rectangles of the following sizes in LaserCut and scored them in acrylic at low power .

25mm came out 24.62

50mm came out 49.53mm

100mm is right on at 100mm just like I calibrated it.

150mm came out 150.40mm

There is some kind of non-linear scaling going on but I cannot figure it out. Anyone else have a problem like this?

Am I going crazy? Did I forget something? Another Calibration setting?

I have always calibrated at 100mm and assumed that any differences in pieces were the result of Kerf or bad focus but this is worse or I didn't check properly before.

Anyone got any ideas on this one?

[Joe Pelonio]

So,e difference in size can be caused by the thickness of the cut, but the same shape sent many times should always be identical. The job I finished a couple of weeks ago was 2,000 pieces from their CAD drawing, and all were the same and exactly as specified, but that was 1/16" ABS. I have no idea why your 100mm came out perfect, the others off, unless cut at different places on the bed, and perhaps the focus was not right due to an undeveloped bed, or mirror alignment problem.

[Joseph Shawa]

Hi Joe,

Finally someone to discuss this with.

My squares are the same size anywhere on the table. It is just when I change the size that the % error goes up or down depending on how far it is from 100mm in size. Try it and see if a 10mm square cuts exactly 10mm. Then try 100mm square. Or 400mm.

Beam is in as perfect alignment.
I have switched to a 2 inch focal length from a 4 in. Might be more touchy regarding the centering. I'll look into it. I switched to a 2.5 in lens and the problem is similar.

I made 59 vertical lines about 5mm each 10mm apart on the X axis.

The distance between each on the X axis don't vary much from RIGHT to LEFT. They are all within .1. Which isn't a big deal...sort of not a big deal. I expect more exactness. I couldn't make a measuring rule that would be very accurate.

I guess I'd just not noticed it before. Makes sense that some of my boxes don't fold together well.

100mm squares come out 100.02 x 100.01

10mm Squares come out 9.64x9.62

I found this from Rich Harmon:
Rich Harman
Member

Join DateMar 2011LocationOlalla, WAPosts1,318

In an open loop belt drive system, we will never get 100% accuracy. If further adjustment is necessary then it will only be applicable to that one specific cut line on that location on the table. If you adjust for a small part, and it comes out perfect, then when you cut a large part it will be off by a significant amount.

My best overall accuracy is a result of entering the settings based upon math as described in my previous post. It's not perfect, and it can't be without ball screws or a linear encoder - but it is better than the way it came from the factory.

[Joseph Shawa]

I still cannot wrap my head around this. It doesn't make sense that a 100mm square on the software that cuts a 100mm square on the machine should cut a smaller that 50mm square or larger than 150mm square. I thing there is something in the pulse settings that I am missing.
I plan on getting this one but I'd take any help.

Edit: It is obvious now that 100mm is NOT the size to go with on a calibration and one shouldn't do cut out pieces.
I am going to go with a nearly full table sized rectangle 1st and another one that is 1/2 that. I will calibrate so that the small one is 1/2 the full one. THAT will give me the constant ratio that I need for various sized pieces and allow the intermediate cuts to be in their proper places within the larger piece.

I will report back.

[Joe Pelonio]

It's hard to compare since Inhave a different machine, power, and 2" lens, but I am getting accuracy over all sizes. It's a bit of work, but maybe go back to the 4" and see if that's the difference? Also, try slower, and less power. Maybe you are doing something on the corners with slippage on the belt.

[Kev Williams]

I got in my parameters and I think I see what you're doing... however, I'm wondering if the controller may not be in sync with the stepper motors, as in, the controller may be factoring in a different number of steps per revolution than the stepper motors used...? Not really sure about these things, but if the math between what the steppers are told to do and what they actually do isn't accurate, it seems like the results would be exactly what you're getting: Accurate at or near the size used in the correction factor, but proportionally wrong at smaller and larger sizes...

Could be some dip switches need changing?

[Leo Graywacz]

Don't stepper motors move to a given position? I don't think they are linear, they are prepositioned for each step. I don't think you can go in between steps. So given that, with limited positions, you can only get so much precision accuracy.

[Joseph Shawa]

I found a clue but it's still irregular to where I cannot approach a solution.

I drew 300mm lines of three colors both horizontally and vertically with hash marks at 150mm along the way.

Seems that the settings I have on the lines make a difference as to where they cut!

Direction and order didn't change anything but when I used the OPTIMIZE option the hash marks were offset more. The vertical lines behaved differently but Optimize still created a problem.

I am leaning toward this being a Software issue. Any ideas on how to further investigate?

Note: The top most horizontal was a separate draw and upload to verify repeatability. Speed changes resulted in identical patterns.

[Chris J Anderson]

Kev might be onto it. It could be that the 'steps per rotation' setting is incorrect.

[Rich Harman]

The problem is not steppers, it is the belts. There is a reason why the high end machines use linear encoder strips - they maintain accuracy all along their length. Even the best belts are not perfect, they stretch, and they can stretch unevenly. Use a linear encoder and you don't have to worry about that.

Use math to get the steps per pulse setting correct. It will be the best all around for accuracy. If the occasion comes along where you need absolute precision, then use the software calibration fudge for that one job, then put it back to the value that the math requires. Otherwise forget about that stupid software feature.

For reference my machine uses 6.0000 for the setting. Or maybe it was 0.00006? I have written before how to calculate it but probably you could just round the current setting to a number that makes sense, since the math will involve integers you should not end up with a crazy number like 5.99337.

[Rich Harman]

Don't stepper motors move to a given position? I don't think they are linear, they are prepositioned for each step. I don't think you can go in between steps. So given that, with limited positions, you can only get so much precision accuracy.

Yes, you can go between steps. It is called micro stepping, without it the motors sound like grinders. A typical motor has 200 full steps per revolution. While you can divide each step up into any arbitrary value, the max useful is about ten (though they sometimes use much higher numbers), or 2,000 steps per revolution. On my machine the motor is then geared down further, I think 3:1. so now it is 6,000 steps per revolution of the output pulley. If the output pulley is 20 grooves with a 5mm pitch that means the gantry moves 100mm with each revolution. 100mm of travel divided by 6,000 steps equals .16667 mm per step, or 60 steps per mm.

[Joseph Shawa]

I got in my parameters and I think I see what you're doing... however, I'm wondering if the controller may not be in sync with the stepper motors, as in, the controller may be factoring in a different number of steps per revolution than the stepper motors used...? Not really sure about these things, but if the math between what the steppers are told to do and what they actually do isn't accurate, it seems like the results would be exactly what you're getting: Accurate at or near the size used in the correction factor, but proportionally wrong at smaller and larger sizes...

Could be some dip switches need changing?

Glad you see how it might be a problem.
Can anyone else duplicate it?
Not sure what dip switches or their effect. Elaborate?

I messed with some settings last night and with only occasional success I was able to draw a dividing line exactly down the center of a rectangle. It seemed related to the size of the piece. I only got the exact division at less than 75mm but not always. I tried adjusting acceleration, high speed, laser gap. But then the perfect divisions would disappear. I will try to get to bottom of but dang I am tired.

[Bill George]

Instead of everyone trying to explain to you how steppers and drives work, go here and the Support page lots of self help> http://www.geckodrive.com/

[Joseph Shawa]

Thanks Bill.

Will check it out.

[Joseph Shawa]

Kev Williams and Bill George and really anyone that kept me thinking, THANK YOU!!

You are EXACTLY RIGHT ON! or as near to right on as one can be.

Bill and Kev, you both said it might be a belt. Well, I thought maybe my belts were a bit gummed up so I cleaned them and some little thing changed. Almost imperceptible but the results were somehow better. So I pulled on my belts and tested the cuts when starting from the left or from the right and still almost imperceptible but nonetheless different. I cleaned my gears and even loosened what I thought were pretty tight belts and got more, ever so slightly different results. I tightened them back up. Tugged the top of the belt before cutting, the bottom part. took notes on all the results all along. I adjusted speed and Pulse Settings and settings that I didn't know the effect of was and I adjusted my underwear too. It just bugged me that different ordering of cuts came out different. Makes sense now. "Belt slipping" in a way. ......And then with your suggestions that it was the belts and it not making any sense otherwise......I finally, FINALLY....
I got pissed and tightened the belts WAY MORE than I thought would be ok and VOILLA! I don't know if this has been going on for 2 years but I think something changed when I squared the gantry. Maybe I have a stiff pulley or who knows what. But now EVERYTHING IS RIGHT ON. As it should be. I'll go back and check the pulleys for free movement.

I was almost ready to get a new machine! I really need to be able to cut where I expect,1/2 mark like at the exact Now as far a size scaling goes, I haven't measured various sized cuts. I'm going to do it now and add to this edit.

Edit: 150mm is only .1mm off now. Acceptable.

Thank you for helping me through this!!!! I tend to never give up but, really, I was close on this one. I just cannot stand imperfection and I am sure you all have the same problem.

Bummer that there isn't a spring tensioner on the belts. Hows a guy to know? especially a blooming novice like myself?....I know, I'll learn to read the cryptic manuals.


Have a great weekend!

Is there such a thing as overtightened belts? I imagine it would wear the bearings and stretch the belts prematurely. The "twang it" belt tensioning method is about as worthless as I can imagine.