Another 3D Printer

[Jorkki]

Joules,

Just waiting for the same Kossel Mini  kit myself (my first), so can't really give any expert advise.
I have used the waiting time to do some reading though and from what I have read, probably  the most common reasons are the retraction settings.
Just thinking about it, if it was a mechanical issue, the printer probably would not recover to properly print the upper layers.

If the Bowden tube would be loose at either end, you would probably also have the issue everywhere.

So it may have something to do either with a too large retraction or a positive value on the 'Extra length on restart'.
Alternatively it might be that the printer is doing short moves and retracts before arriving on the perimeter to continue.
It might make sense to test with the 'Avoid crossing perimeters' and 'Only retract when crossing perimeters' settings.

Also putting 'Random' in 'Seam position' might make a difference.

For delta printers I have also seen a recommendation to put a small value (e.g. 0.1mm) to 'Lift Z' if the nozzle keeps catching the already laid filaments.

Hope you get everything sorted soon!

Jorma

Some links:
http://manual.slic3r.org/expert-mode/fighting-ooze
http://wiki.e3d-online.com/wiki/E3D-v6_Troubleshooting

[Joules]

Hi Jorma and welcome to MadModders, I finally found the problem and it was the hotend being loose.  It has caught on the prints and loosened.  I wasn't keen on the method of bolting into plastic to clamp the hotend, so I will either make an alloy part or some little threaded bushes to reclamp everything, and increase the Z lift   

[Pete49]

Glad you found the problem Joules as I knew the software was good as I had no problem with the print even though I used ABS (black).
Pete

[Joules]

And here is my problem, screwing M3 bolts into ABS was never going to be a good idea, but you can't say I didn't give it a try.



The advice to tighten gently    into plastic.  It failed in use, I tried over tightening the bolts once I had it out and the difference between snug and pulled out is very small, too small to feel if the delta plate your fixing too is a little tight on the bolts.  So, this part is getting a redesign to take an alloy insert that is tapped to take the bolts properly and clamp things firmly.

At this point a cnc mill would be nice.

[mattinker]

The thread is far too fine, something like self tapping screws or wood screws work much better in plastic!

Regards Matthew.

[tom osselton]

Could you not drill through and use some of the threaded inserts used in woodworking that are capped?

[Joules]

Could you not drill through and use some of the threaded inserts used in woodworking that are capped?

I looked at those inserts, but decided on a horseshoe shaped alloy fitting with the five threaded bosses machined in, so none of them can spin and I get good clamping force.  The whole  part above is having a redesign as its a pretty course print, I can route the wiring better and improve the filament cooling fan supports and reduce its footprint with better clamping.   Just juggling time and jobs as a rather nice robot arm got delivered a few days back and could be a major distraction if I'm not careful.  8-)    But that's another story.

[Jorkki]

I received my my kit and have been going through and preparing the parts for assembly.
The hotend mounting really is not something that gives a warm and fuzzy feeling even if I'm not a mechanical engineer.
To complicate that, my effector piece seems to be a not so successful print with deformed rod end mounts so I'll have to trim it down and use some additional washers to make the rod distances equal.

I think the best option would be to order an effector made of aluminium.
At least this one has a 3mm set screw that tightens the hotend directly to the effector plate.
http://www.robotdigg.com/product/396/End+Effector+for+Kossel

Another thing I have noticed  is that many of the Traxxas ball joints have quite a bit of play. I know this is a fairly common issue that people have tried to solve by using rubber bands or springs to pull the rods together to lessen the effect, but annoying nevertheless. We'll see how it will affect the accuracy.

Jorma

[Joules]

Jorma, you can ignore the play in the ball joints, gravity is working in your favour.  The weight of the hotend keeps enough tension to stabilise everything.  The only time you will have a problem is when you ramp the speed up.  I would suggest this is not a fast printer, but when properly setup, a very precise printer.  All my tests indicated very good repetition and quality printing till the hotend worked loose due to crap clamping method.  I am working on a solution now.

Wow, I really like that effector plate solution, you have just changed my design focus.   

[Joules]

Best laid plans of mice and men.  I figured I would get this redesign of the hotend clamp nailed, when work rudely interrupted, since last year was very hit and miss, earning toy money has become the priority.  So the design got shelved and a quick fix sorted.  I modified the original STL file to take a 5mm diameter x 4mm long knurled bush.  These are set into the 3D print 2mm from the clamping surface.  If anyone wants the new STL file, just give me a shout.



I had to work quick in the workshop as it was freezing and I had about 20mins before my hands stopped working, so no pictures of them being machined.  Back inside and warmed up I made use of a cheap table vice I picked up from Aldi, on Friday.  Very handy it is too, for holding small stuff like this.  Each bush tapped 3mm and the faces cleaned of burrs with a needle file.



I used some packing to spread the load on the working face of the print and pulled the bushes in with a 3mm bolt.  The bushes are about 0.15mm oversize with the knurl so they bite well into the 3D print, but hopefully don't stress the walls too much as I wouldn't want them to break loose when clamping this all together.



Here they are all installed and ready to reassemble, then go all the way through calibration again.  This I hope will fix the printer for now.....  Others plans for update have been pushed well back on the back burner.


Something that has started to interest me is the possibility of moving to a cnc lathe.  With my hands being hopeless in these temeratures I am looking at desktop Denford and Boxford slant bed lathes.   Anyone local (South Yorkshire) with similar machines fancy guiding/training me in return for CAD design work.  If anyone also knows of a working machine requiring a good home  !!!  I am on the look out for one.

[awemawson]

I have a Traub that could be re-homed if you are brave and reckless 

[Joules]

Sounds great, so that will go on the kitchen table then.   

[Joules]

Re built the head, and found lots of gap between the clamp and the hotend.  I ended up putting an O ring on the hot end and compressing it against the effector plate.  Everything now calibrated and running some test prints.  Auto bed level worked even with the old setting so I am probably within 1mm of the old Z height.

[Jorkki]

Good progress there Joules 

Mine is printing as well after some initial battles.
All in all a very nice kit with some very good components like the controller and the IR probe. Also everything is very nicely packaged and marked and the instructions are good.

The gripes I have with the kit are:

- some of the printed plastic parts are not of best quality (like my effector plate)
- the use of ordinary 3mm metal screws with ABS plastic, some are even too short (3 for the effector plate have very little grip) and some cause delamination because of the bad print quality.
- some of the Traxxas ball joints are quite loose
- one of the aluminium verticals was a tiny bit bent

The geometry is so complex that it's a little bit hard to pinpoint what exactly caused my difficulties in calibrating the printer.
The good thing is that the IR probe is quite reliably and repeatably triggering at the same height all over the building plate. What is not so good is that the height of the nozzle is varying quite a bit indicating that the effector is not staying exactly level across the build area.
Fortunately the calibration algorithm is quite good and by putting in height offsets in the configuration file I have managed to get decent results.

Joules, you should try to print the spiral test file that is on the SD card. You will very soon see if you get a spiral or just a mess of angel hair like I did on the first try.

[Joules]

Jorma, I was having trouble getting my prints to stick till I realised my Z probe offset was out by 0.2mm.  I also increased the current to the motors by 10%.   I am still having some slippage on my own test prints, but I think that may be down to a problem in my slicer !!!  As everything else prints OK.

I would suggest for any parts you have issue with, reprint them on your machine at something like 0.15mm layer.  You will get a much better part then exchange it on your machine.  Do you have access to a lathe, as making the brass bushes for clamping the hotend to the effector plate made a big difference in print quality.  I just need to see how it holds up long term.

         Joules

[Jorkki]

I had some head banging myself due to too fast printing speed on a small layer so the plastic did not have enough time to cool.
I got better results after I switched from Slic3er to MatterControl. Actually I'm not sure if it was the slicer switch, because I set the Z-lift to 0.15mm and increased the retraction length to 4mm at the same time, but there may be other subtle differences between the slicers.

If you have Z-offsets in the print head height paper test and you can't fix them mechanically, you can do the same as I did and put the offsets in the bed.g file.
The idea is to adjust the Z-offset to 0 at P9 (origin) by adjusting the probe offset in config.g (in the G31 command) and put the offsets for other test points in bed.g.

My values are as follows:
G30 P0 X-73.6 Y-42.5 H-0.10 Z-99999   ; X tower
G30 P1 X0 Y-85 H-0.15 Z-99999      ; between X and Y towers
G30 P2 X73.6 Y-42.5 H-0.10 Z-99999   ; Y tower
G30 P3 X73.6 Y20 H0.10 Z-99999   ; between Y and Z towers
G30 P4 X0 Y67 H0.05 Z-99999      ; Z tower
G30 P5 X-73.6 Y20 Z-99999   ; between Z and X towers
G30 P6 X-36.8 Y-21.25 H-0.05 Z-99999   ; half way to X tower
G30 P7 X36.8 Y-21.25 H-0.05 Z-99999   ; half way to Y tower
G30 P8 X0 Y42.5 H0.05 Z-99999      ; half way to Z tower
G30 P9 X0 Y0 Z-99999 S6      ; centre

The sign is so that e.g at P1 my my bed is high, a 0.10mm paper read 0.25mm on the Z display and at P3 it's low so the 0.1mm paper read 0mm.

I do have access to a lathe at my brothers, but I think I'll gather some experience first and see what else I'll learn.
I just printed Emmett's Gear Bearing from Thingiverse to test. It's a quite nice example of things that you can't build or take apart, you have to print it.

[Joules]

You got really good print quality there Jorma.  That's what's making the niggles worth solving.  This little printer is capable of very high quality work.   My fear about the extruder was justified when a filament jam popped the large gear and pinch roller out 

I hope it was just a one off, but I still have an all metal extruder on my upgrade list.  That said, a second Mini Kossel will probably join this one shortly.

[Jorkki]

Outch!
Can't really beat metal for certain applications.

Of the printed ones, the coolest looking I have seen is Drew's planetary geared extruder:
http://www.thingiverse.com/thing:39138

Also I kinda like the simplicity of the design that Builda3dprinter.eu has in their kits.
The guy is showing the build in this video:
https://www.youtube.com/watch?v=u3npxBuSv0Y
They sell the geared motor and stainless steel spur gear on their web site as well:
http://builda3dprinter.eu/shop/page/2/
The plastic part stl file is in the information/downloads section.
I have no idea of it's durability though, probably would be a good idea to print a spare or two just in case.

[Joules]

Ha I printed Drews planetary extruded when he first released it with the intention of using it on a Rostock Max.  I like the YouTube video and extruded design might exchange the printed part for alloy machined part.  I do like the other Kossel machines on that site, but won't ever go back from a 32 bit controller now.

Setting the printer up for another test print.

[Joules]

First test print with everything dialled in.  These are shell prints, in effect a spiral movement to print a single layer.  The objects you see are nose cones for a small model glider.



The print on the left when I produced it about a year ago was thought to be pretty damned good.  The print on the right is from the Kossel, you see a grain effect where the 8 bit controller interpreted steps on the left.  The right print barely shows the steps, you really have to see these to get a full understanding of the increase in resolution in printing.

My other concern was going to be tolerance from machine to machine as I have a few printers now and will be printing multiple parts and in different colours, so things like snap shut lids could be a problem.



Not a worry...  The translucent white print was from another machine over a year ago, and the black section below is a failed print from last night on the Kossel, they are mating parts printed on different machines and the grain in each print is alined, so you can see the difference.  The failed part has the snap lock printed on it, so it makes a good test part.  They clicked together just fine, I am so pleased with the tolerance I am seeing so far.   Yeah other issues with the printer are getting resolved and worked around, but the print quality is just outstanding and it's still early days.

[Joules]

Oh well, I hate been right....  The extruder mechanism is pants.  The lower bearing is a rattle fit on the end of the 3mm bolt holding this all together, it should have been longer to at least take a lock nut, that would have stopped the extruder wheel grinding a helix and jacking itself out of the printed body.  So another part for redesign and thinking through a better solution.  Now I need to make two, as another Mini Kossel is on it's way.   Despite the hiccups, I really like what this printer can do, and the above mods will be done before I even assemble the next one.

[nrml]

The difference the Duet board and the DC42 firmware have made is amazing. It is nice to see a direct comparison. I never thought it would be such a big improvement.

I appreciate your stand on supporting local businesses whenever possible, but it might be worth a look at these for your next mini kossel. Some of them might save you a bit of hassle and possibly improve the end product.

http://www.aliexpress.com/item/3D-Printer-Accessories-Aluminum-Alloy-Full-Metal-M3-Delta-Kossel-Fisheye-Effector-Hammock-3MM/32437620404.html

http://www.aliexpress.com/item/4Pcs-Lot-3D-Printer-Accessories-Hot-end-Round-Fixed-Aluminum-plate-Sandblasting-Oxidation-Reprap-Kossel-E3D/32314904596.html?spm=2114.10010108.0.124.C1oOhW


http://www.aliexpress.com/item/New-Black-Color-Reprap-Kossel-All-metal-Aluminum-Corner-Piece-Frame-Vertical-Base-Delta-Top-Bottom/32473624719.html?spm=2114.01010208.3.47.P7Uyy2&ws_ab_test=searchweb201556_2,searchweb201644_3_505_506_503_504_502_10001_10002_10017_10010_10005_10011_10006_10003_10004_10009_10008,searchweb201560_1,searchweb1451318400_-1,searchweb1451318411_6449&btsid=0ea26ee5-c1d1-4a60-b03b-063b6a8c0376

http://www.aliexpress.com/item/3D-Printer-Accessories-Full-Metal-M4-Delta-Pulley-Tackle-Moving-Block-Hammock-Effector-20MMx20MM/32434034394.html

http://www.aliexpress.com/item/5Pcs-Lot-3D-Printer-Accessories-Reprap-Kossel-Prusa-Bowden-Step-42-Stepper-Full-Metal-Remote-Extruder/32426938444.html?spm=2114.10010108.100007.18.0UhbRQ

There are a lot of other very interesting products in the 3D printer parts category.

[Jorkki]

I found myself using a head torch constantly to see how a print is coming along.
I decided this is a good opportunity to start learning a little bit of the openSCAD software and decided to design a small clip to put on the hot end fan where a short strip of LEDs could be mounted to shine some light on the work in progress.

This is what I came up with.

I have also changed the hot end heater cartridge to enter from the front side as the leads were pressing against the cooling fan hood and that tended to unscrew the hot end loose.

At the same time I replaced the original 3mm machine screws that hold the hot end carriage to the effector plate with 3.5mm deep threaded universal screws (as I could not easily find the special screws made for thermoplastics). Not as good as Joules's threaded inserts, but a huge improvement anyway.
Might be a good idea to try and get some of these http://www.tappex.co.uk/trisert-inserts or something similar for future projects.

[Jorkki]

I have done some further improvements to my Think3DPrint3D Kossel Mini.

I noticed that to get better retraction performance, it is essential to lock the Bowden tube better in place.
At the extruder end the pneumatic connector seems to be solid, but at the hotend the connector collar will move back and forth with the tube no matter how hard I tried to push the tube in.
The solution is to lock the collar in place with a C-shaped spacer under it.
I printed some in PLA with the ID of 5.1-5.2mm, OD 7.8mm, thickness 1.3-1.6 and 270 degree coverage.
Seemed to fix the issue.

My hotend cooling fan quit, so I used the opportunity to upgrade it with a better one.
I chose the Sunon HA40101V4, which seems to be a very nice one with magnetic bearings.
http://www.tme.eu/en/details/ha40101v4/dc12v-fans/sunon/ha40101v4-000u-999/
Very quiet and practically no vibration.
At the same time I replaced the print cooling fan with a bit beefier Sunon, but I think there is a bit too much back pressure from the small airway opening in the guiding tube to be able the realise the full potential.
http://www.tme.eu/en/details/mb40101v2-a99/dc12v-fans/sunon/mb40101v2-000u-a99/
An axial blower type fan might be better  here.
I have also soldered a 220nF capacitor across the print cooling fan terminals because the fans do not like the PWM signal. The Fan used to start at 100% only, now it starts at around 10-12% and has a quite nice adjustment range.

I have also gone back using springs to pull the carbon fibre rods together to reduce the effect of play in the Traxxas joints. Seems to make the print quality a bit better, but for sure joints with no play would be better.