High quality prints

This is for general discussions. Also FAQ can be found here.

Moderator: speedake95

Post Reply
HLA
Posts: 73
Joined: Sat May 05, 2012 11:35 pm
Location: Vienna, Austria

Re: High quality prints

Post by HLA »

After doing the upgrades to the x-axis and extruder described before, I am still cursed with another issue: backlash. I have mentioned this before in this thread, but still cannot resolve it.

The attached figure shows that the backlash presumably comes from the y-axis. Observed magnitude is D-d~0.4mm, which is very noticeable for smaller circles and holes (e.g. 10% for holes with 4mm diameter). I would appreciate any thoughts on what I could do to improve this. So far, the following things have been checked:

- y-belt is well tensioned (gives a nice bass-tone);
- teflon applied to the Y-sliding bearings to reduce friction;
- Kapton tape applied behind Y-sliding bearings to reduce slack (further Kapton tape produces too high friction, so its at the limit now);
- printer upgraded to Revision E;

By the way, this issue was present before any aforementioned modifications to the x-axis and extruder, but I never actually paid much attention to resolving it. Now the results of Guillaume and this thread motivated me to tackle this :)

Thanks for your help!
Attachments
(simplified) Schematic representation of observed backlash in printed circles and holes.
(simplified) Schematic representation of observed backlash in printed circles and holes.
Backlash.JPG (16.14 KiB) Viewed 618 times
Last edited by HLA on Tue Sep 18, 2012 4:59 pm, edited 1 time in total.

danielkschneider
Posts: 49
Joined: Thu Apr 12, 2012 3:29 pm
Location: Geneva
Contact:

Post by danielkschneider »

gfeliksdal wrote:Thanks for the tips!
I would like to dig into the extruder jams more, as I don't seem to be able to reproduce most of them. What PLA are you using? I can print with translucent and solid colored PLA without having extruder jams. I can remember you received 2 types of filament from me the small diameter and large diameter. I still have both and I notice that with the large spool diameter I had to set the temp 10 to 20 deg lower than the small spool diameter in order to have good results.
Here is one use case:
* Using red solid PLA (as in the picture)
* Printing the blue one was ok (translucid PLA at 210)
* To the very right, I printed a floor with red non-translucent PLA at 210, failed in the 2nd layer of the roof (after the filling). I then tried to recover: changing filament, raising temp, but then aborted.
* In The middle (red castle): I printed the whole base plate at 215 and at fairly low speed (40ms). For the rest (castle on top of the base plate) I printed at 210 and 150% manual speed increase. Extruder sometimes stopped working (I wasn't there).
2012-09-18 13.30.58.jpg
2012-09-18 12.42.58.jpg
[Edited/added on sept 19 12:00]

I made a second attempt with the same slicer settings but at 215. The floor printed fine, but then the filament stopped coming out in the upper right corner (picture #3 attached)
the good news:
* my office was still there when I came back to work (so this was an overnight print)
* There is some confirmation about my hypothesis that many pillars + a hex filling pattern may not be easy (with this type of PLA)
Bad news:
* Raising the temperature only helped for the floor
2012-09-19 11.31.44.jpg
I also checked the filament roll (turned fine), but I didn't clean the nozzle or anything else above before printing. IMHO, filament comes out ok enough, but there may be some dirt inside the nozzle walls and maybe bits and pieces above ...

I am going to print this again. this time a different fill pattern, higher speed but the same filament. If this fails too, I then will remove or at least lower the retraction, part of antistringing measures in Dimension I think. I didn't say so before since I forgot, but I am quite sure that without any retraction we would have much less problems, but uglier prints of course. I may sometimes be preferable to have strings (I could not print delicate multiple pillars with the fabbster having retraction on, turning it off helped). It's easy to cut strings off vertical pillars and some quick flame or hot air will remove the rest.

[Edited/added on sept 20 12:00]

I still believe that using 1) denser simpler fills, 2) reduced speed and 3) less retraction can do wonders for printing little towers and/or large roofs over fills and/or large flat 1st layer.
* With little fill, the plastic will bond laterally i.e. a deposited string will bond to the neighbor instead of the fill or the platform underneath. Since the contact to the previous layer is weak, the whole thing then will warp upwards or sideways.
* Too much retraction may lead the nozzle to start printing without filament coming out and then the next layer won't stick.

* However, I finally figured out why I had print aborts where they should not happen, i.e. in some random places, after adopting the strategy above. The red plastic (narrow roll) I got from Guillaume is simply no good :) I didn't notice that before, since mostly the filament is ok (about 1.61-1.68mm). But once in a while and that matters for 10hour prints, huge imperfections are there (like 2.1mm "knots")

Sorry for the bad picture quality:
Bad PLA
Bad PLA
2 above: Bad filament<br />2 below: ok ... abort due to other reasons
2 above: Bad filament
2 below: ok ... abort due to other reasons
A final tip: If you supervise the print and notice that the first layers of a roof are very irregular, turn off the fan for while. This will help the nozzle to slide through the bumps :). Also, always do 3 full layers (top and bottom).
Something like this:
------------------------- top layer <- fan on here
------------------------- 2 <- fan off here (or even below)
------------------------- 1
/ / / / / / / / fill, must be rather dense for a wide area
-------------------------- <- fan on
-------------------------- <- fan off here
--------------------------
|| || || || pillars
Last edited by danielkschneider on Thu Sep 20, 2012 6:38 pm, edited 4 times in total.

stratogavster
Posts: 120
Joined: Tue Jul 17, 2012 9:04 pm
Location: United Kingdom

Post by stratogavster »

Based on HLA's nice diagram related to backlash and printing quality of circles.....I too find this issue....(plus a few others!)

I'm not entirely sure there is much in the way of linear backlash in it's traditional form (as I understand it) but rather I feel it's down to a number of related / similar factors as follows:

1.) Diameter of the pulleys not being as per the original development spec VS the firmware config step/mm (Guillaume suggested this)
2.) Fundamental mechanical design
3.) Mechanical friction factors
4.) General low-precison linear slides


These factors combined would result in what we're seeing manifested as poorly defined circles and / or dimensional errors / quality issues....unless any other guys out there have other ideas or theories?

I've found that on my very first print of a cube 25mm x 25mm x 10mm (Height) each axis was printing 24.5mm x 24.2mm respectively...the Z was spot on at 10mm...well 10.1mm but i'll live with that for such a process :)

So to hit the 'high quality print' holy-grail I think a few upgrades / developments need to happen amongst us all :shock:


Based on point 2 & 3 - My additional thoughts / theories are that the fundamental makeup of the mechanical Y axis - the 'pinched belt' is not ideal and creates a lot of tension in the system for such a delicate operational demand.

Also the X-axis has the 'belt pass through' hole which i'm sure has some kind of standing-wave present which rides against the roof of it's pass-through hole, thus creating some kind of resonant drag.

Last but not least the particular type of low precision linear slides in the design are really fighting against the other merits the design has...

Just all theory...

The only way forwards I guess is to firstly tackle the easy stuff....no. 1....start by calibrating the firmware exactly to each users setup as each pulley will have a manufacturing tolerance....

I've never ever messed with firmware or calibrated any kind of stepper / CNC so any guides on how to do this would be great!

After that I feel we can tackle the other elements I feel are at play here! and start printing at 0.05mm layers :D
Last edited by stratogavster on Wed Sep 19, 2012 9:13 pm, edited 1 time in total.

stratogavster
Posts: 120
Joined: Tue Jul 17, 2012 9:04 pm
Location: United Kingdom

Post by stratogavster »

..after the comments i've added I thought it would be useful to show my latest achievements, which I must admit are very 'mixed'...both bad and good...

After printing a pre-made Gcode file of a Felix 1.0 part (added on the forums by Guillaume) I'm having huge problems getting good quality results and i'm suffering huge z-shift and junk circles as per HLAs findings!....

....but I tried to print the high quality Owl gcode (though cut the print short as I wasn't going to leave the machine on for 12hrs!!) and I had what appears to be a nice print....ok had some blobbing and shift but overall it's heading in the right direction!

See images :) (excuse the fingers)
Owl Part Print 1
Owl Part Print 1
Owl Part Print 2
Owl Part Print 2
Owl Part Print 3
Owl Part Print 3
Last edited by stratogavster on Wed Sep 19, 2012 9:13 pm, edited 1 time in total.

HLA
Posts: 73
Joined: Sat May 05, 2012 11:35 pm
Location: Vienna, Austria

Post by HLA »

@stratogavster: Thanks for highlighting a number of issues. Nevertheless, correcting for steps/mm might be too inaccurate compare to a theoretical evaluation using supplier's datasheets for pulleys and belts. Unless you have some high-end measuring equipment that you didn't mention, besides a digital caliper.

Anyway, I opened a new thread for backlash in Hardware>Mechanics. Posted my progress on this there.

Lioric
Posts: 116
Joined: Wed Aug 15, 2012 1:13 am

Post by Lioric »

Bear in mind that probably the inaccuracy in relation with the expected size might be because of the wall thickness, I have found that in my machine, in its current setup I needed to modify the extruder amount of flow to ~60% of what was originally in the default form, to give exact single wall width (my current extruder multiplier value is 0.642)

The _40x10.stl object used to calibrate this (the first object that is printed after assembling the printer) was producing 0.54mm walls, and after the calibration now produces 0.39mm walls (the exact value expected by the produced g-code)

HLA
Posts: 73
Joined: Sat May 05, 2012 11:35 pm
Location: Vienna, Austria

Post by HLA »

Excellent tip, Lioric. Thanks.
If I understand this correctly, you used 0.642 for the value of Feed Rate in Speed. Or is the E-steps corrector?

stratogavster
Posts: 120
Joined: Tue Jul 17, 2012 9:04 pm
Location: United Kingdom

Post by stratogavster »

As i'm so new to this and i've not even got to 'slicing' yet...i'm assuming these calibration and correction values are within one or all of the slicing softwares out there?

Lioric
Posts: 116
Joined: Wed Aug 15, 2012 1:13 am

Post by Lioric »

I'm really new to 3d printing as well. Been printing for 6 days now

@HLA

Im not using SFACT but Slic3r, its the called "extrusion multipler" from filament settings. But yes in SFACT feed rate ot E-steps will do similar effect. Hopefully I will start experimenting with SFACT soon

HLA
Posts: 73
Joined: Sat May 05, 2012 11:35 pm
Location: Vienna, Austria

Post by HLA »

@Lioric: I would prefer using slic3r to SFACT. It is significantly faster and the latest version has a great user interface. However, some models are not sliced correctly by slic3r in my experience thus leaving SFACT as the only choice. I have no idea why this happens, but until slic3r gets as robust as SFACT regarding different input models, there is no choice left than to use the monstrous SFACT.

@stratogavster: slic3r is the "user-friendly" slicing software out there. It has less settings available to the user, and is faster than e.g. SFACT. SFACT is a slicer for advanced users, with a large amount of levers one can tweak. However, the amount of settings is so huge that is out of human comprehension.

Post Reply