Non round circles, undersized prints

Hi,

I try to set my printer properly, but I still have one big problem. Everything I print is undersized. As you can see in the attached picture, it makes something about 0.2 mm in X-axis on 20 mm (1%) and 0.5 mm (2.5%) in Y-axis! The undersize also causes that all printed circles are not round and some lines don't touch perimeter (in the first picture are gaps between the hexagon and infill and in the other picture next to the left perimeter).




I thought it is typical backslash. I checked pulleys (sharpened a place for bolt), tightened belt (I thing that it is not possible to tighten belts more, I use simple clothespin belt tensioner), calibrated axises (It could be not perpendicular).

Then, I printed a simple test angle (the second picture). I noticed, that the size error is multiplying. I got error 0.65 mm in X on 50 mm (1.3%) and 1.1 mm in Y-axis (2.2%). I have made myself sure that the X and Y axises are really perpendicular.

(Prusa Mendel i2, Sprinter, Slic3r)

Any ideas?

steps need tweeking abit use the josef calculator basically you put

Demanded move length in mm

Actual move length in mm

Actual steps per mm

and the result will give you the new step value

---

Check my rubbish blog for my prusa i3

up and running
[3dimetech.blogspot.co.uk]

I checked steps by unit, but unfortunately did not help.

More ideas?

Are the belts also perpendicular? Are the level with where they contact points? Or is there any triangulation, and therefor changing error based on distance of belt?
Can we get pictures of your setup?

Edited 1 time(s). Last edit at 06/09/2014 12:38PM by MrDoctorDIV.

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus

As you can see in the pictures, there is no triangulation. I thing that belts (axeses) are really perpendicular, but I am going to measure it again tomorrow.

Thank you for response.

With the motors on, can you move the hot end by hand at all? Bed?

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus

I can move the bed by hand easily, but the x-carriage is harder to move, but I now it is caused by tightened belt.

Oh, sorry I did not notic with motor on. I can not move bed nor x-carriage with motors on. Even a part of milimeter.

Edited 1 time(s). Last edit at 06/09/2014 01:13PM by ncq.

I'm at a loss then, I can move mine about a millimeter and I still get good, accurate prints.
Have you measured the distance moved in different areas? Are your rods perfect straight? You know what, that's probably it. If it was a kit, I'm sure they gave you cheap China rails, I've never seen a perfectly straight one.

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus

It was not kit. I bought rods from a local shop and I use LM8UU. You are right, I have also never seen perfectly straight rods

I will try to allow belts and find out if axises are perpendicular.

There is one more strange thing. I use Slic3er and I print with 0.85 Extrusion multiplier, because the solid infill is overfilled. Note that the extruder motor steps per milimeter are certainly accurate. Setting Extrusion multiplier to 1.00 could make prints fitting the size (I am really not sure), but what about the infill? I will test it.

Thank you for your interest!

You can calibrate your ext. multiplier with a single wall thickness test.
Rods I got from Misumi were as perfect as I can tell with my eyes and home tools. Never seen more beautifully made rods. Never again will I purchase ebay rods.

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus

Yes, I printed the .5 thick wall and it was OK, you are right, it was a nonsense.
The rods are not straight, but how could it affect the undersized print? I thing it is not the real cause.

If a rod is bent on the Y axis, that will cause a shift in that direction according to the bed, so you X axis dimensions will be off as much as your bend [average between bends of both rods holding the carriage, etc] where ever it is off a straight line.
Try printing a really long rectangle, prefereably near your full bed width/length and make measurements of the width along the print, see if it varies.

Edited 1 time(s). Last edit at 06/09/2014 03:21PM by MrDoctorDIV.

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus

It's pretty easy to straighten bent rods using a sturdy table, two v blocks and a g-clamp. One of the 5 12mm rods in my printer was bent (about 0.5mm off in the middle), was a 5 min job to get it perfectly straight.

Hi again,

At first, konwiddak, thank you for your tips.

Today, I checked couple of things again - motor steps per millimeter (I think it is perfect), belts, pulleys (also great) and if are axises perpendicular.

And now hot news! I printed the large rectangle as MrDoctorDIV adviced.
The result is really suprising. The rectangle (in the picture) is hollow and its dimensions are [155,155,1.4] and the inner perimeter has dimensions [125,125,1.4].

I confirmed that axises are perpendicular. It makes error of 0.5 mm per 155 mm of lenght. All sides of rectangle are really straight. I marked points (nozzle is homed at UR) in the picture and there are all measured dimensions:

UL - UR - excatly 155.00
BL - BR - 154.90
UL - BL - 154.65
UR - BR - 154.30

Suprising! I thought that prints are undersized and the size error is relative, but actualy it is not.


Distances between outer and inner perimeters:

CL - 15.22
CR - 15.17
UC - 15.15
BC - 15.05

It is quite OK, but unfortunately the infill has almost no conntact with inner perimeter (most 0.5 mm! as you can see in the picture - between BL and BR). The Infill overlaps outer perimeter, so the infill seems to be enlarged. Perimeter is probably right. I checked gcode with analyzer, but there was no error. Slic3r is right.

I will print the rectangle again and will check if it has the same dimensions.

I consider recording a video of printing, would it be useful?

Video might be useful, though I doubt it.
Just looking at the picture I immediately suspect bent rods. If willing, you could upgrade to Misumi chrome-plated rods, you'll never have to replace them again, build after build [realism implied].
I'd offer my left over ones, but I'm sure you don't need 12mm X 500/1000mm rods
If the dimension repeat, maybe move the rods around and see if they change and repeat again.

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus

I will print it definitely again. I do not know why, but I still do not understand how can bent rods cause the gap between perimeter and infill ... both carrage and bed are almost in the same position

Hm, actually that's a good thought. If it was bent rods then the perimeter and infill would shift together..
Can I grab your STL for that? I'll slice it myself a few different ways, don't need to print full, just see if the perimeter and infill change at all.

Edited 1 time(s). Last edit at 06/10/2014 01:59PM by MrDoctorDIV.

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus

Sure, here are stl and my gcode, you can view it with your gcode viewer, but I guess it will be ok.

It doesn't make sense! GCode is just fine! My slices make not difference! I am at complete loss.
How's the repeating dimensions?

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus

How's the friction on your axis? The was a semi-big problem on my Solidoodle that took a while to find because I never thought about certain parts and friction at the same time. Imagine a half circle and two lines on the same direction going from both sides, because of friction, one of these lines would be set forward more than the others.

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus

Can you post your Configuration.h lines with the steps/mm settings?

Hi, I am glad to hear you again!

I printed the rectangle we spoke about and a result was exactly the same ... dimensions, infill, like twins.

I think, I can now eliminate some possible causes. It could not be caused by belts, pulleys, rods, motors, steps per mm.

@MrDoctorDIV, the friction is a good idea. I had a friction on axises and it resulted in overloading motors (maybe a year ago), because of that there is no friction on rods. The only possible friction is between belts and washers. But, how could the friction cause gap between the perimeter and infill - Y axis? I also tested steps per mm and distances were fine ... I think, It could not be a friction, but I will look at it.

I think, the problem must be somewhere between computer and motors - motherboard?

@pokey9000, sure, you have got it!

Thank you for replies

What belts and pulleys are you using (teeth/pulley and belt teeth/mm specifically)? From experience and what I've seen, manufactured pulleys used with quality belts should work with the calculated steps/mm. I'm not sure about printed though, and all bets are off if your belt has stretched for some reason.

I was hoping that the _AXIS_STEPS_PER_MM was doing an inline divide that would accidentally result in loss of precision, but everything seems ok there.

Your prints are of good enough quality and consistency that I wouldn't suspect friction from bad bearings or lost motor steps to be at fault.

It looks like things are moving towards the center of the bed, perimeters more subject to movement.
Just a theory that is directly related to my Solidoodle experience: Try loosening the belts, not so that you get bad backlash, but to relieve pressure on the body/pulleys. See if that makes a difference.

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus

I am using T5 and printed pulleys with 8 teeth. Pre counted values for steps are X80 and Y80. I checked steps per mm twice. I marked points on rods, determined the measures thereof distance and then I moved bed or carriage ... my values are really right.

It is probably not caused by friction.

I also noticed that during printing the rectangle, first line of perimeter is in a position which matches position of infill, but all other lines of perimeter are elsewhere and they are in the right position. So the firts line of perimeter and infill are not in their places. I also remind non-round circles!

I will shot a video tomorrow, hope it will help.

Edited:

I will then try loosening the belts. It also occurred to me ...

Edited 2 time(s). Last edit at 06/11/2014 01:26PM by ncq.

Can we see the circles?

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus

Quote
ncq
had a friction on axises and it resulted in overloading motors (maybe a year ago), because of that there is no friction on rods.

It took about 3/4 of a year for my idler pulleys to regain friction, causing flat sides on circles in the Y-axis.

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus

Here are examples. Circles are more like elipses. Diameter varies and it is really hard to fit something in them.

One more note, I have been printing for 2 years (my current printer was printed by my previous printer ) and my prints were great before something happened recently!

Quote
ncq
I checked steps by unit, but unfortunately did not help.

More ideas?

So you used the calculator and did the alterations sorry if you have its just that you said you checked and its not just checking it will calculate the steps. For instance mine was 71.11 on the y and x i printed a 50 X 50 L and measured it it was 51 on the y and 52.5 on the x and it calculated mine to be 72.686 on the y and 76.456 etc . if your using a lcd also change it on eeprom as i find those some reason on mine overwrites the firmware when running i was getting oval holes and calibrated the steps.

If your happy with the quality of the prints then look into the steps again, also how much you extruding too more plastic more spreads out

---

Check my rubbish blog for my prusa i3

up and running
[3dimetech.blogspot.co.uk]

I'm sorry for bringing it up so much, but my mind is now screaming friction, looking back at all of the prints you posted it makes perfect sense. Your Y axis [bed?] has friction. The reason it seperates the infill and perimeter is the fact that they move against eachother/towards eachother. The friction makes them stop short of their goal and because they come from opposite directions it causes a seperation. Check your bed idler pulleys for friction under load. When I checked for friction originally I had everything loose, and I found none. Once I applied pressure to my belt against my idlers I found significant friction and the cause of friction like sounds. I added grease and slightly loosened my belts, perfect movements. At the very least, humor me by greasing up your idlers on the Y axis and making sure they have flowing movement under load.

Edited 1 time(s). Last edit at 06/11/2014 02:33PM by MrDoctorDIV.

---

Realizer- One who realizes dreams by making them a reality either by possibility or by completion. Also creating or renewing hopes of dreams.
"keep in mind, even the best printer can not print with the best filament if the user is the problem." -Ohmarinus