Circles not coming out Circles

brnrd Wrote:
-------------------------------------------------------

> I'm just a beginner compared to most people in
> this forum. I started building in December and I
> was able to produce working parts in less than a
> month that I have been using for my research.
> Thanks to the efforts of people like you in paving
> the way and sharing your experience online. My
> software settings are still far from optimal, but
> I haven't been having the problems that you have.
> The fact that others can do the same would seem to
> indicate that there is nothing wrong with the
> design and you have a specific hardware problem
> that needs to be fixed.

Or maybe beginners on 3D printing as you can't "see" this kind of problems. Maybe you don't have a way, another 3D printer to compare.

I can't forget the words of a very well know developer (Skeinforge):

"There's no point in comparing Mendel/skeinforge versus Up because Mendel is a piece of junk. Makerbot with the default DC motor extruder is also, unfortunately, not worth comparing.

You might want to compare the BFB3000/skeinforge versus Up for your own curiosity because the BFB3000 is constructed reasonably well. A Makerbot with a good stepper extruder may also be worth comparing for your own curiosity."

Edited 2 time(s). Last edit at 02/13/2011 05:34PM by casainho.

---

---
New cutting edge RepRap electronics, ARM 32 bits @ 100MHz runs RepRap @ 725mm/s:

[www.3dprinting-r2c2.com]

casainho Wrote:
-------------------------------------------------------
>
> Or maybe beginners on 3D printing as you can't
> "see" this kind of problems. Maybe you don't have
> a way, another 3D printer to compare.

I've been using a collaborator's Dimension printer as well as having parts printed through Stratasys before I got the Mendel, so I'm not a beginner in 3D printing. Here's our publication on the research that we're now using the Mendel for: [rsi.aip.org].

The example you posted shows errors in the X-axis that is nearly 1 mm. I'm confident that I can spot that in my parts.

Casainho I don't see how someone being a beginner or having access to another printer is relevant. If something isn't circular its fairly obvious and easy to measure. Also, many experienced people have said they don't have this issue. I'll stand by what I said on the wiki, If it was a design problem we'd all be getting it. It must be a problem with setup or calibration.

rhysjones Wrote:
-------------------------------------------------------
> Casainho I don't see how someone being a beginner
> or having access to another printer is relevant.
> If something isn't circular its fairly obvious and
> easy to measure. Also, many experienced people
> have said they don't have this issue. I'll stand
> by what I said on the wiki, If it was a design
> problem we'd all be getting it. It must be a
> problem with setup or calibration.

It's a problem that some of us are having and we don't have the solution.

I think you did well on the wiki text. Maybe the problem is some low quality bearing, belt, etc. I can't know unless I go ahead and spend double of the money to buy again this parts from other sources. I think people should be warned about the possibility of this problems and the fact they may not get solution.

---

---
New cutting edge RepRap electronics, ARM 32 bits @ 100MHz runs RepRap @ 725mm/s:

[www.3dprinting-r2c2.com]

brnrd Wrote:
-------------------------------------------------------
> casainho Wrote:
> --------------------------------------------------
> -----
> >
> > Or maybe beginners on 3D printing as you can't
> > "see" this kind of problems. Maybe you don't
> have
> > a way, another 3D printer to compare.
>
> I've been using a collaborator's Dimension printer
> as well as having parts printed through Stratasys
> before I got the Mendel, so I'm not a beginner in
> 3D printing. Here's our publication on the
> research that we're now using the Mendel for:
> [rsi.aip.org]
> 01_s1.
>
> The example you posted shows errors in the X-axis
> that is nearly 1 mm. I'm confident that I can spot
> that in my parts.

Here are two examples of recent parts printed with PLA. I'm sorry that this are the best example that I can find since I wasn't focusing on the circle problem when I took the photos.

The pegs and holes in these parts are 5 mm in diameter.


This is the 20 mm hexagon.stl example in ReplicatorG printed on glass over the Prusa heated bed. I have lowered the temperature to 55 to reduce the curvature in the bottom.

If it were me I would try a few experiments to see if you can isolate the behavior to one system.

Do they come out the same if you reverse the print (mirror)? Turn the object 90deg? Print in a different part of the print bed?

I agree after seeing your final result that there is likely something causing a hysteresis error in the x axis. The way these printers are assembled, it is most likely a mechanical issue however it is still possible that it is in the drive/motor, electronics, or control.

One advantage with the open loop control is that it isn't terribly difficult to troubleshoot issues.

@shek: are you referring to my photo or casainho? I don't think I have a problem with my circles.

My post was directed to casainho or anyone in general having positional problems.

I sure wish my printer was finished so I could do some testing of my own.

I still am not sure exactly what the symptoms of casainho problems are.

Most issues are usually backlash issues, it seems. The other issue noticed was relating to software/electronics, with skeinforge translating circles into lots of little, quick moves, and possibly overloading the controller chip, or for some peoples machines, the motors.

leonardo's problem looked to be, mainly, backlash. The only other thing he identified that might warrant looking into was how the Y-belt is mounted, with the teeth running against the bearings. Prusa's design does this as well, but uses 608 bearings. perhaps a someone wants to try a y axis redesign with 608 bearings? Might even be possible to mount the bearing right on one of the frame rails for testing.

but the key here is what exactly are we looking to improve? I don't want to spend a lot of time trying to redesign something if I'm not going to see any improvement (or even make it worse).

ps. correct me if I'm wrong, but i think that inherent problems with microstepping will manifest themselves more when you try printing smaller and smaller features. If you truly want higher resolution, and thus more circular circles, you need to start with smaller step-angle steppers.

Buback Wrote:
-------------------------------------------------------
> leonardo's problem looked to be, mainly, backlash.

And Leonard, as me, don't have any solution, but a printer which prints bad quality parts. Look at the iterations he did and the best result:

And now after a day of tests I present all my tests results and an acceptable end result (nº 15).



I wish hardware developers can make a machine that people as us can assembly and don't have this problems, rendering useless the investment made.

---

---
New cutting edge RepRap electronics, ARM 32 bits @ 100MHz runs RepRap @ 725mm/s:

[www.3dprinting-r2c2.com]

I think it's a little hyperbolic to call those results useless. most of the people building mendels don't have access to precision cnc machines like you do, and are happy with the results they get. If they want guaranteed results, they can buy a commercial machine.

but, again, that picture tells me little, other than that he drastically improved over his first attempt by eliminating most backlash, and his blog post is light on details of his methodology. What is the scale of those circles?

I spent over a decade as a machine control technician for a jet engine company (no Honeywell). I don't think there was a single machine that used a stepper motor if positional accuracy was the goal. Most were brushed DC with integral tach and either a rotary encoder or linear encoder (glass slide) with dedicated hardware PID controllers.

Just out of curiosity I looked up a Bridgeport vertical milling machine:

ACCURACY
Evaluation Standard: ANSI/ASME B5.54
Positioning: +/- 0.005 mm (+/- 0.0002in.)
Repeatability Full Stroke (x axis): 0.002 mm (0.00008in.)
Repeatability Full Stroke (y and z axis): 0.002 mm (0.00008in.)

shek Wrote:
-------------------------------------------------------
> I spent over a decade as a machine control
> technician for a jet engine company (no
> Honeywell). I don't think there was a single
> machine that used a stepper motor if positional
> accuracy was the goal. Most were brushed DC with
> integral tach and either a rotary encoder or
> linear encoder (glass slide) with dedicated
> hardware PID controllers.

The problem then is needing high quality PID controllers. Otherwise with regular DC motors you have to deal with overshoot, since telling the motor to stop doesn't stop it right where it is if it's going at speed. Proper control would get expensive. For smaller machines stepper motors are plenty accurate, especially for the resolution required for FDM printing. I don't think the drive electronics or motors are an issue for most RepRappers.

---

Help improve the RepRap wiki! Just click "Edit" in the top-right corner of the page and start typing. Anyone can edit the wiki!

NewPerfection Wrote:
-------------------------------------------------------
> shek Wrote:
> --------------------------------------------------
> -----
> > I spent over a decade as a machine control
> > technician for a jet engine company (no
> > Honeywell). I don't think there was a single
> > machine that used a stepper motor if positional
> > accuracy was the goal. Most were brushed DC
> with
> > integral tach and either a rotary encoder or
> > linear encoder (glass slide) with dedicated
> > hardware PID controllers.
>
> The problem then is needing high quality PID
> controllers. Otherwise with regular DC motors you
> have to deal with overshoot, since telling the
> motor to stop doesn't stop it right where it is if
> it's going at speed. Proper control would get
> expensive. For smaller machines stepper motors
> are plenty accurate, especially for the resolution
> required for FDM printing. I don't think the
> drive electronics or motors are an issue for most
> RepRappers.

Hey! My Zenbot CNC machine and my UP printer, both have stepper motors (UP printer have smaller steppers motors than Mendel) and I get round circles (at least I can't see with my eyes imperfections on them.

---

---
New cutting edge RepRap electronics, ARM 32 bits @ 100MHz runs RepRap @ 725mm/s:

[www.3dprinting-r2c2.com]

I'm just guessing here, but i think stepper motors are better suited for a replicating printer, since it might be difficult to print an encoder. I also think good DC gearmotors or linear actuators are much more expensive, and cheap ones cause problems, like in the makerbot extruders.

casainho Wrote:
-------------------------------------------------------
> Hey! My Zenbot CNC machine and my UP printer, both
> have stepper motors (UP printer have smaller
> steppers motors than Mendel) and I get round
> circles (at least I can't see with my eyes
> imperfections on them.

I know, I was saying that I don't think stepper motor driven axes would have any impact on printing accuracy, especially with the accuracy required for what we are doing. Moving to hugely expensive PID controlled DC motors would be practically useless.

Linear encoders would be quite useful, however as long as the steppers are not skipping and you can mostly eliminate backlash, they wouldn't do much to improve print quality.

---

Help improve the RepRap wiki! Just click "Edit" in the top-right corner of the page and start typing. Anyone can edit the wiki!

lol wow did this thread ever end up getting side stepped. at least i can say i learned a few things from it.

like allot of things in life, the more money and the more time you can spare for it the better the results will be, for now i would just be happy to order my parts that i need..lol

---

[mike-mack.blogspot.com]

Wow indeed... Wow..

Even with my original home made MDF and fixit block Mendel I achived round circles... it has since been dismantled to steel its motors for use in my other printers.

The only problem I found that produced errors was I tried to save heat build up on the steppers by turning them off on X and Y this was a big mistake as it produced an acumulative error on both axis.

Don't think I will ever catch up reading all of the post in the forum... they make interesting reading I noticed Nophead said he only get 35mm of height on his mendel.. I was supprised I expected it to be the 140mm I never check my Mendel as I asumed it would be the 140mm.

---

Bodge It [reprap.org]
=======================================

---

BIQ Sanguinololu SD LCD board	BIQ Stepcon	BIQ Opto Endstop
BIQ Heater Block PCB	BIQ Extruder Peek clamp replacement	BIQ Huxley Seedling
BIQ Sanguinololu mounting	BIQ standalone Sanguinololu or Ramps mounting	Print It Stick It Cut it

My rep strap: [repstrapbertha.blogspot.com]

Buy the bits from B&Q pipestrap [diyrepstrap.blogspot.com]
How to Build a Darwin without any Rep Rap Parts [repstrapdarwin.blogspot.com]
Web Site [www.takeaway3dtech.com]

Hi all,

I was a long time with circles not circles, and squares not squares with Mendel, more precisely said: X lenght was always too little. For example a 20mm square was printed 20mm into Y direction and only 19,5mm into X direction. Accordingly circles were not circles...

After many months trying many things as you can imagine I found THE solution!

I was using Alternate X motor bracket. With this design, the belt is inverted and travels over 4 bearings by the teeth side, with right angles. By the teeth side the belt travels less distance than by the plain side... and then X side of the print is a little smaller.

So yesterday I mounted the original Mendel X motor bracket... circles are now circles, squares are all squares!

2 other problems with the alternate X motor bracket:
-little continuous vibrations because of plain (theeth) then empty parts running continuously on the bearings
-less attachment to the x pulley

---

François

I don't follow your logic. If the pulley rotates one revolution the belt will move by exactly 8 teeth, which is 40mm. Since it has wires in it all the belt will move the same distance and the carriage will travel 40mm.

I use that bracket and my holes come out round. With the old bracket the belt snaps after 200 hours of use.

---

[www.hydraraptor.blogspot.com]

In other words, when belt runs by teeth side on the bearings, the tension is continuously changing because of sequences of "full and empty" belt. So the belt travels a very little less X distance. Otherwise I can't explain why I have now a perfectly working Mendel with the original X motor bracket, that I did'nt have with the other one.

Of course I prefer a working reprap with a belt to change sometimes than a non-working one…

If someone else can solve the same problem with the same solution, shall we say it's a strong solution!

PS: the belt on the Y axe also runs around a bearing, but with a 180° angle (something like 3/4 of the bearing circumference) so the "full and empty" belt effect is less, or not, sensitive.

It's very odd. A few people report this problem, but it works fine for me, and presumably for most other people too. I also worked out the theoretical error due to tooth / gap contact and it was very small due to the difference between a 2.5mm arc and a chord on a 13mm circle being only 0.015mm.

I think it must either be due to belt tension, or perhaps the type of belt. Perhaps the teeth are more rigid on some, while others deform. I trust you have a proper timing belt with steel or kevlar wires in it and not simply a toothed rubber belt?

---

[www.hydraraptor.blogspot.com]

Decreasing the tension on the axis solved my problems, but I think it has to do with building up tension when the stepper rotates and not so much the teeth on the belt.

---

---

reprapworld.com
PLA, ABS, PETG multiple colors (3 and 1.75 mm) €16,49 / kg
Megatronics 3.0 €79.99 / Minitronics €37,19

I had a very similar problem with circles coming out 0.5mm off in one direction. In my case the X-Axis was shorter than it was supposed to be. It turned out that I was using XL belts on both axes, but my pulley for the x-axis was for a 5mm belt and not an US-style XL belt. So on each reversal my X-Axis lost roughly 0.5mm
Replaced the pulley with a new XL pulley and no more problems.

ST

@nophead: yes I have a belt with wires in it, used with your pulley design (many tanks for it!).

Belt from Ultimachine and pulleys profi-printed from Mendel-parts.com.

Reading all your posts: perhaps the problem was coming from a too little belt/pulley hanging into the alternative x-bracket,, not from the side of the belt. Perhaps adding 2 bearings, fixed with 2 of the motor screws, would fix that (see attachment). I'll mount int one day and tell you the result!

Edited 1 time(s). Last edit at 08/09/2011 08:05AM by François Delègue.

I mounted the alternate X motor bracket with 2 more bearing as illustrated into my last post (photo): all is ok as with original X motor bracket, squares are squares.

So I suppose now that the problem was coming from the size of the teeth of the pulley, which are only 1.7mm width: it is not very much, and it's not a problem if the belt turns enouth around the pulley.

I have been trying to fix this problem on my mendel. I've tried so many things... Still cannot seem to find the problem. I just replaced the pulleys with better quality ones, I have tightened, loosened, changed steps per mm, and printed one to many test parts. I also am using ultimachine belts, alternate x-axis and ramps 1.2 w/marlin.

This is driving me crazy. Anyone have any insight into this?

korgon Wrote:
-------------------------------------------------------
> I have been trying to fix this problem on my
> mendel. I've tried so many things... Still
> cannot seem to find the problem. I just replaced
> the pulleys with better quality ones, I have
> tightened, loosened, changed steps per mm, and
> printed one to many test parts. I also am using
> ultimachine belts, alternate x-axis and ramps 1.2
> w/marlin.
>
> This is driving me crazy. Anyone have any insight
> into this?

I am being printing a lot and changing my Mendel also, now I have much better firmware, changed the belts, better pulleys but still nothing!

---

---
New cutting edge RepRap electronics, ARM 32 bits @ 100MHz runs RepRap @ 725mm/s:

[www.3dprinting-r2c2.com]

Have you got 5mm or 6mm belts? If you have 6mm belts have you got the extra washer above the bearing?

I had this problem on my Prusa and it was due to too much friction of the belt against the 608 pulleys. A 608 bearing is 7mm wide with a significant chamfer and M8 washers are 1.5mm thick so the belt rides over the side of the pulley and rubs against the penny washer.

I find the Mendel idler pulleys with the smaller bearings and washers work much better but the instructions are for 5mm belts made with the belt splitter. If you have 6mm you need an extra washer. Also you need to put the penny washers the right way up. They generally have a chamfer on one side and a burr on the other due to being pressed. The smooth side must be towards the belt. Similarly the smaller washers should have the smooth side towards the bearing.

Edited 2 time(s). Last edit at 11/03/2011 06:05AM by nophead.

---

[www.hydraraptor.blogspot.com]

Also where did you get the "better pulleys"? There are some metal 8 tooth pulleys on eBay that don't work with T5 belt because the teeth are too thin, giving backlash, and the holes are not accurately centred.

---

[www.hydraraptor.blogspot.com]