XY Steps per mm error

Hi All,

I'm struggling to calibrate my steps per mm and can't figure out whats going on...

I spent a long time calibrating my X and Y steps per mm with a particular object and thought I had them spot on. I went to print a different object that was slightly larger and it came out completely wrong.

I did a bit of investigating and worked out that I calibrated in the lower left quadrant (close to 0,0) but printed my larger part in the middle of the bed. So it appears the error in my calibration gets hugely amplified the further from the origin I go. I guess this makes sense but I had not thought it would make that much difference. Has anyone else experienced this?

Thanks

Andy

What type of printer do you have? Delta, Cartesian, etc?

Sorry - completely forgot to mention!

It is a Reprap Mendel (cartesian). I use Repetier firmware and Repetier Host so I can change the steps per mm in the EEPROM.

The calibration should not matter where on the bed you print the part. Think about how it works- the motor steps/rev doesn't change, the belt pitch doesn't change, and the number of teeth on the pulley doesn't change with XY coordinates. If you calibrate with a very small part then print a big part, the error in the small part calibration will multiply in the dimensions of the larger part. It is better to calibrate with a large part so that when you print smaller parts the error in the measurements on the large part will divide when printing smaller parts.

I see your point.

To try and find out more I made a simple part - a box 50 x 50 x 1 mm. I then printed 3 copies of the part arranged like so:

Y
--------------------------------
|             ___               |
|            |1  |              |
|            |___|              |   
|             ___               |
|            |2  |              |
|            |___|              |   
|             ___               |
|            |3  |              |
|            |___|              |        
|____________________________  _|X

The goal of this was to test the Y axis as I have noticed that the problem is far less prevalent on X. The result is very odd. The same part prints with Y dimensions as follows:
1 - 55.07
2 - 57.80
3 - 49.01
The only thing I can think of is it could be due to the Y belt not quite travelling parallel to the bed. Im only hypothesising but if the belt was secured to the bed a bit lower than the bearings at each end of the frame then the most linear movement would be in the middle but as the bed gets closer to either end of the frame the belt angle increases therefore the actual y motion is less than the length of belt moved.

Does that sound possible? Or is it utter rubbish?

Edited 1 time(s). Last edit at 04/09/2015 06:00PM by awmt102.

It looks like you Y axis rods are not parallel to each other and get wider at your number 3 and closer to number 1 or vice-versa. Measure the distance between them at both ends and at the middle. It gets longer in the middle because the rods can bed there and allows the carriage to slide easier. You will have to re-calibrate your axis once you get both rods at same distance.

Edited 1 time(s). Last edit at 04/09/2015 06:54PM by ggherbaz.

You can check for parallel guide rails by taking the belt off and pushing the print bed/extruder carriage back and forth down the length of the rails. If they are parallel there won't be any binding. If they bind they may cause the motor to slip and skip steps and that will cause the sort of weird behavior you're seeing. While you're at it, check the belt tension and the grub screw on the motor pulley to make sure it is tight. Also check motor current and make sure it is adequate.

Thanks for the advice. I will try it out and see but I think the design of the Mendel mitigates misaligned rods - the carriage is only constrained on one rod, it floats on the other, so slightly misaligned rods should make a difference. I would also have thought that binding of the carriage or underpowered steppers would be a bit more random so each layer would be misaligned with the last, but my alignment is spot on so it is a repetitive non-random error I am seeing.

I checked my belt theory but my belts are running pretty much parallel to the carriage/rods, so I don't think that pans out.

Perhaps a worn belt could result in this behaviour? Since then you would get backlash at exactly the same point every time. I will check that out later today.