2021/01/14: Ashtar M (Moving Gantry – Draft) also with IDEX option now
2021/01/13: ooze prevention at rest position added, mechanical conflict resolved
2021/01/12: starting with a first draft, one mechanical conflict to be resolved
Introduction
I have been pondering on a dual independent X axis upgrade or option for a while, but the other designs of the Ashtar Series I wanted to do first (Ashtar D and Ashtar M) those matured by now (2021/01), so I decided to get back to IDEX upgrade for Ashtar K:
Ashtar K
Ashtar K IDEX
For now I like to keep single 2020 V slot alu extrusion for the X beam where the X carriage rides, and route the2nd beltabove for the 2nd X carriage – and this was a quick solution as earlier version of Ashtar K had the belt routed above the alu profile so I just reused the old pieces again.
“Above routed belt” option with its pieces are weaker and possibly need enforcement improved the strength, so it’s a fast start – just took me 2 hours – but needs definitely some fine-tuning. Alternatively the 2nd belt could be routed at the back of the X carriages, but fastening the 2nd X motor would be challenging.
Old “above routed” X motor mount
New stronger “above routed” X motor mount
For now I use the same code base of Ashtar K and introduce IDEX = true flag, and enhance a few existing pieces in parts.scad and optionally add those new pieces when rendering printer-ak.scad.
As I progress with this option or upgrade I update this blog-post.
Draft
X motor #1 routed within
X motor #2 routed above
Ooze prevention in rest position
Ooze prevention in rest position
Both heads in rest position
Issues to Resolve
X carriage #1 belt mount conflicts mechanical with belt 2: redesign xcarriage_beltmount_2020 piece, make it shorted in Z or fasten it inside V module: resolved, shifted 2nd belt a bit Y off, and shorten xcarriage_beltmount_2020(idex=true) by 2mm.
2nd belt slight Y offset so X belt mount #1 doesn’t touch belt #2
xcarriage_short_hmount_motor_2020 which is the base piece which routes the belt within the 2020, with idex=true option provides idler holder on top
X motor #2 is mounted on a x-mirrored version of xcarriage_hmount_motor(20,"left",idex=true) but definitely needs reinforcement, added ooze prevention in case of idex set
Nozzle drip prevention:
using a piece of sheet metal which the nozzle moves over when in rest position left or right, first attempt done (see below)
and/or use purge box with brush to clean nozzle after and before use
make extending “nose” detachable/replaceable as it’s expected to break or overheat otherwise entire X motors mount needs replacement, resolved
xcarriage_nose-idex-left and xcarriage_nose-idex-right with 10mm wide sheet metal insert
how dealing with long resting hot nozzle?
drop temperature by 5-10C in rest position, and heat up when in use again
heat creep possible weakening extending printed nose – heat insulation required attaching sheet metal
Inside and Above 2020 belt routed X motor mount, w/ removable nose with ooze prevention
Nozzle at resting position, a sheet metal preventing oozing
Right-hand side ooze prevention
Gallery
Ashtar M IDEX
And since Ashtar M (Prusa i3 Moving Gantry – Draft) shares much of the Ashtar K design it took me a few mins to add the IDEX upgrade option as well:
A brief history of “Ashtar K“, my first designed 3D printer I actually built – documented also for my own sake:
AluX: Prusa i3 Clone
It started with AluX (abbreviation of ALU-extrusion eXtendable) early June 2018, which used CTC i3 Pro B / Prusa i3 Clone pieces as the X carriage, X motor mount and X idler all in STL format. I coded the frame parametric using 2040 alu extrusions/profiles and using smooth rods as rails:
AluX 220
AluX 400
I realized then quickly I need to design and code my own pieces, every single piece I need to control and make it parametric if it makes sense, and not rely on existing STL files, as editing meshes of the STL seemed a waste of time but rather design the piece in OpenSCAD right away and derive new variants if necessary from the geometry itself.
Ashtar X & W Series: Riding on Smooth Rods
Mid June 2018, AluX became Ashtar X (abbreviated as AX), and Ashtar W were using 2040 alu extrusions but differently oriented at the base, still using smooth rods as rails:
AX 22.22.24
AX 40.40.40
Ashtar W 22.22.24
Ashtar W 40.40.40
At this point I got sufficient experience of the parametric approach and it was obvious to use the frame as rails.
Ashtar T Series: Riding Alu Profiles
Beginning of July 2018, with the Ashtar T series I began to use the frame as rails itself, utilizing 2040 alu extrusions, it also started with the parametric V module (due its shape) composed by 2x V-plates, using 3 wheels which ride on the alu extrusion:
V plate (outer side)
V plate (inner side)
With the parametric V modules the X, Y and Z frame beams became rails as well, simplifying the overall construction compared to earlier designs:
Ashtar T 22.22.24
Ashtar T 40.40.40
The dual Z motors still residing in the front for sake of accessibility, but then I realized I want them in the back and keep the front dedicated to the printhead.
Ashtar K Series: Riding Alu Profiles, Uni-Length Beams
Mid of July 2018 I started the Ashtar K series, I decided to use 2020 alu profiles and focused on the single length of alu profiles, uni-length so I could reuse the beams for other future designs and since all the designs were parametric, it was easy to attain to find an optimum of single length beams and a common build-plate or build-volume:
Mid 2018: 9 beams design
Mid 2018: 9 beams design, X motor mount added
2019: 11 beams design complete, belt routing above 2020 alu profile
Late 2019: belt routing inside 2020 alu profile
2020: multiple extruders, LCD controller
2021: adding parametric enclosure
The 9 beams design turned out too weak when I actually built the printer, so I added two beams back on left and right, and lift up the 9 beam design.
Eventually I decided to use 500mm alu 2020 profiles to achieve ~380x300x360 build volume; Ashtar K #1 used 400×300 build-plate, and Ashtar K #2 300×300 build-plate. Ashtar K #1 was functional in August 2018, and since then became my working horses together with Ashtar K #2, reliably printing.
See more at Ashtar K project page of the current state.
Next Steps
Ashtar Series Genealogy (2018-2020)
After the Ashtar K I did the Ashtar C Core XY cubic frame also with 2020 alu profiles. Late 2020 I started to design Ashtar M, a derivative of Ashtar K but with a moving gantry and static bed, and Ashtar D with Classic XY alike Ashtar C; and also a draft of a parametric enclosure as well to be adaptable to all of my 3D printer designs.
Printing new X motor mount on CTC DIY I3, and replacing it on the new Ashtar K: CTC DIY I3 prints quite reliably – there is nothing to clean up – the piece I attach it right away:
Ashtar K lacked a proper print surface (before I received the black sticker surface), otherwise I would have printed the piece on itself.
Black Sticker as Bed Surface
The 400×400 black sticker arrived, and I cut it into 400x300mm and put it on the mirror – which worked well, and so far I can tell the surface is very very flat, much better than on alu heat bed.
Current bed setup (top to bottom):
400x300mm black sticker (“frosted sticker”), apprx. 0.6mm thick
400x300mm 3mm thick mirror
210x210mm 12V alu heat bed
various cork patches under heat bed
10mm light black foam material
420x320mm 6mm plywood (white painted) as Y carriage
Now that I have a good print surface I finally printed pieces for itself.
Mounting the 400×300 bed (OSB 6mm, white painted) with 200×200 heat bed (which I hardly use, as I started to print on cold bed):
I currently use the white PU steel enhanced GT2 belts, and it produces hard edges, some ghosting, but more precise prints than the black rubber GT2 belts which just stretch too much – I have to research this more closely – about the type of reinforcement and the use with more heavy beds (Y carriage).
Just for the record regarding Y carriage (2018/09):
420×320 carriage:
4mm plywood flexes, but has been quite flat – not recommended
6mm plywood hardly flexes, but has been hard to buy truly flat – and so far my attempt to flatten it did not work well – not recommend unless it’s flat
6mm OSB quite flat, does not flex much (3 or 4 sliders) – recommended
320×320 carriage (for 300×300 bed):
4mm plywood works (3 sliders, 4 sliders recommended)
6mm plywood works (3 sliders, 4 sliders possible if plywood is truly flat <0.2mm difference)
6mm OSB quite flat, doesn’t flex (not yet tested)
Just to explain my thought or decision process for my setup:
the mirror should not be bend (of course)
the support structure should not be the edge mounts, but the foam in between
the carriage can be bent, but not flex
revelation: already bent means the springs with screws might extend the bent further with a flexing carriage, and not counter act – as the mirror should stay flat
so, even though the springs/screws and edge mount can adjust, the carriage should be fairly flat, and not flex at all – this way the edge mounts holding the glass/mirror only stabilize position. Main force to hold the glass/mirror, for my setup, is the foam in between. So, there is no “spring” induced vibration back/forth introduced, but the foam neutralizes such vibrations – and hardly adds weight/inertia.
Sliders & Belt Mount Positions
Top view with see-through (best mark “0,0” on both sides to keep reference).
400×300 vs 300×300 Bed
Originally I focused on 300×300 bed at least, with some tweaking and narrow X carriage I was able to reach 380×300 printable bed, so it was suitable to use 400×300 plate as well.
It takes me about 5min to mount new bed, downgrade from 400×300 to 300×300:
Changes needed:
move Y endstop switch from left to Y carriage extrusion to the right side
Y stopper mounted on the bed needs to placed accordingly
With 300×300 bed the 0,0 is now plenty outside of the bed, with 400×300 the 0,0 is near the printed bed mount.
Setting Offsets for 300×300 bed
With 300×300 bed the 0,0 is now +32mm to right and +25mm deeper, hence the Gcode M206 is set like this:
M206 X-32 Y-25
H Plate/Module as X Carriage
The 3 wheels module riding on the 2020 alu extrusion I named “V plate” due the shape, the 4 wheels module “H plate” providing more stability or rigidity for use as X axis carriage, when the nozzle runs over slightly unclean extrusion and tilts upside. For the X carriage I choose a narrow (48mm wide hole-to-hole) version:
Belt mount and hotend holder using same mounting holes
It’s the first/early version, the adjustment screws (M3x10) are very or too close to the bed for my taste, next version will use M3x8 and give more spacing. I like to keep the hotend close to the X carriage so not to waste Z space.
Additionally I made a new hotend mount so it would use another mounting holes than belt mount:
Belt mount and hotend holder using same mounting holes
Belt mount and hotend holder separate
But now it’s harder to reach the hotend mount holes due the part cooler – oh well.
After few days, I noticed one wheel stopped to turn, no longer touching the alu extrusion – I guess the carriage slowly balanced itself and triangulized, no longer use the 4th wheel. I re-tighten the 4th wheel gently so it would roll again.
Z Couplers: To Wobble or Not To Wobble
As I posted before, I suspect the Z couplers to be the main source for Z wobbles, as the threaded rods may look and are cheap but they are mostly straight – the wobble actually is caused, after close observation, from the misalignment which happens when you screw the metal couplers on, in particular if you attached the lead screw or threaded rod with uneven surface – the thightening screws may or may not attach cleanly – and thereby push the Z rods out of the center of the Z stepper motor – when the Z thread holding the X axis is fixed, the resulting wobble is worse at low Z heights; and if you fasten the Z rods at the top, the wobble gets even worse.
A simple remedy I found is to use printed couplers, two pieces which are screwed together with 4x M3 screws and nuts, a bit of an overkill, and a bit time consuming to fasten: incrementally tighten each screw over and over until all are tight – but I think it’s worth it: the two halves attach evenly and the PLA or ABS or whatever you printed the couplers, is soft enough so the threads of the Z rods carve themselves evenly into the coupler, and self center themselves this way – result is better centric attachment of the Z rods, not perfect but acceptable and better than poorly manufactured metal couplers.
Alu 5mm to 6mm coupler
PLA printed 5mm to 6mm coupler
As mentioned before, I switched from M8 to M6 for the Z axis the M6 provides 1mm movement per full turn, and is more flexible to even out out-of-center wobbles, better than the stiffer M8 threaded rod. If using couplers at all, and likely introduce out-of-center mounting, rather use a more flexible lead-screw or threaded rod than a stiffer one.
It has been a few days (2018/09/04), since Ashtar K happen to be able to print, the heat bed still unfinished, some prints illustrated below are done with no leveling screws, the mirror just taped on the Y carriage – don’t laugh – later prints I had proper carriage and leveling screws included; a proper build surface I still wait for in the mail (400×400 black sticker to be cut in shape) – anyway, here some of the early prints:
40mm XYZ Calibration Cube
The original 20mm XYZ Calibration Cube is printed in 8 mins with 0.5mm nozzle at 0.4mm layer height, and so I thought, let’s print it 2x the size with 0.4mm layer height, merely 40 mins later this:
The quality is . . . impressive, this is just tuning a single day – mostly on the extrusion factor and print temperature – and this is what I hoped for: XYZ positioning almost flawless: there is slight ghosting on X axis (which could be resolved) shown on “Y”, and Y axis shown on the “X” which is fine, given the size of the bed and its weight and inertia this is OK.
I had to increase print temperature +20C from 200C to 220C for 80mm/s infill while printing with the 0.5mm nozzle, I otherwise would hear clicking from the extrusion stepper motor missing steps. I still use the classic E3D V6 (clone) heat block, not the Volcano heat block.
20mm Calibration Cube: Different Layer Heights
Printed with 0.5mm nozzle, left-to-right: 0.1mm, 0.2mm, 0.3mm and 0.4mm layer height, 60mm/s (80mm/s infill), 200C first layer, rest with 210C, pink glowing PLA by Sienoc.
X Carriage: Sliders vs Wheels
While printing with slider carriage on the X axis, I noticed increased stuttering, and regardless if I thighten or loosen the grip, the stuttering remained, and slight horizontal tilt occured when changing direction on the X axis resulting in too narrow prints in X dimension.
X carriage with white nylon wheels (23.mm OD / 7.3mm width)
So, I changed back to wheel-based carriage, first again 23/7.3 white nylon wheels (right photo), but when I printed “L” shape with 200mm length in X and Y and 1mm height in Z, I noticed slight Z sinus form as I saw before – while it rolled nicely, there was a wobble . . . and so I printed a new carriage which holds the black OpenRail Double V (clone) 24.4mm OD / 11mm width, and put it on the X carriage:
X carriage with double V black wheels 24.4mm OD / 11mm width
A brief overview of the carriages riding on 2020 T slot (B-Type) alu extrusion:
Sliders: on the X axis it did not last, the stuttering was not avoidable; the issue is that the X carriage is one of the hardest axis of the Prusa i3 style geometry to handle: it isn’t just X directional rail, but also pressure on the Z with the weight of the print head, and running over overextruded filament – and it’s hard to pull the X carriage perfectly without the carriage have some vertical tilt as well – anyway, I still use the slider option on the Y carriage – and works fine so far.
White nylon 23/7.3mm wheel: rolls nicely, but gives wobble to the Z height when used on X carriage, apprx. 1mm, also doesn’t stay vertical upright, but tilts a bit with pressure – when the print head moves over overextruded print it doesn’t level it, but jumps over it. I currently use white nylon wheels on the Z carriage successfully.
Black double V delrin 24.4/11mm wheel
groove use: rolls very nicely, gives no wobble, and stays vertical. The next days and weeks will tell if the double V wheels do last on the T slot alu profiles – they are meant on proper V slot alu extrusions.
diagonal/edge use: rolls very nicely too, but surprisingly gives less tilt rigidity than groove use – the T slot 6 (B-Type) gives less surface at supposed 90deg edge, but is rather 85deg
Z Axis Linearity
As you may have read in the other post(s), I use M6 threaded rods, it’s flexible and rather aligns with the Z axis itself, whereas M8 is stiffer and misalignment – which by the way doesn’t come from the rod itself, but the mounting with the couplers – won’t impose on the X carriage – this is my own view and it happens to come true again with Ashtar K, after I changed my cheap CTC DIY I3 also to M6.
Now, the 1m long M6 threaded rod, enough for two Z axis each 500mm long, did just cost EUR 0.70, made in China but purchased locally in Germany, and the nylon wheel-based Z carriage happen to work perfectly so far – I expected some slight sinus wobble imposed by the nylon wheels as I encountered on the X carriage, but it seems when there is little force applied on the wheel the carriage works good enough.
Printing 330mm high 10mm diameter cylinder (with slider-based X carriage):
There was some slight extrusion inconsistencies, this is likely due the material, an broken vacuum seal of a newly purchased glowing pink PLA roll, actually, after watching the 2nd print closely, either GCode errors or USB transmission errors, as some segments of the circle (layer of a cylinder) is repeated for some unknown reason and so overextrusion occurs there (needs proper investigation) – but the linearity is very good, and no Z wobble whatsoever.
The “loopy egg” is a good benchmark for retraction settings, and stressing the extruder motor as the short segments making up the loops require a lot of push / pull on the filament. There was still some slight stringing, which I knew will happen, as the retraction is just set to 2mm at 35mm/s giving very good results. More prints will tell if I can stay with these retraction numbers.
Fighting Heat Creep
I currently use E3D V6 clones as hotends, one with 30mm “original” fan, and one with 40mm fan. And with the “original” smaller 30mm fan I experienced frequent clogging up within the hotend: some of the filament melted above the heat break and expanded and blocked any further extrusion – that happened now several times.
I tried to reduce the extrusion temperature but which caused decline of print quality. After trying to determine the root cause of the problem, I concluded that it was heat creep and insufficient cooling above the heat break, hence, the hotend fan, and I switched to 40mm fan – and the clogging disappeared, not quite yet . . . update follows.
30mm Fan (front facing) with 5015 Fan Fang (top)
40mm Fan (front facing) with 5015 Fan Fang (top)
Although both setups look very alike, I had to print out another fan fang which can contain 40mm fan.
Five Platonics
My favorite geometrical forms – aside the sphere – the sacred set:
Mirror as Bed
I’ve got 40cm x 30cm mirror which became my bed base, underneath with some tight springs some 6mm multilayered plywood, which was warped 2-3mm on the edge – but it didn’t matter (much). The mirror was the reference, and the Y carriage had to hold the mirror. That turned out to work very well: the mirror is truly flat, I leveled the bed once for tilting, after a week, I only had to tweak the Z endstop screw slightly, but I didn’t touch the screws mounting the mirrors to re-level the bed anymore.
So, using the mirror as bed worked well so far due the flatness – but the glass didn’t turn out to print good on it, the printed parts often detached before finishing the print, and ruin the print – so I used blue tape sheet as temporary solution until the black sticker arrives which I already use on the other 3d printer.
Reflection
As I designed Ashtar K with larger build volume, I choose 0.5mm nozzle at least, and the max 0.4mm layer really pays off in regards of print speed, while still maintain some details – I’m quite pleased so far.
Well, after merely 3 months (2018/06/06) when I started to code the first lines of OpenSCAD to develop a series of parametric Prusa i3-like designs, and few weeks ago decided to go with the “K” series with 2020 alu profiles: simple 11x 500mm beam T slot (B-Type) alu profiles – the 1st prototype happen to print the 20mm XYZ Calibration Cube as of 2018/08/27:
Ashtar K 38.30.33
The bed is very temporarly fasten with tape, as I haven’t decided on the actual details of the bed mounting yet and leveling details – but I wanted to see how well the mechanics already works – and it performed quite well so far.
1st print came out mediocre, when I realized I had to tighten X and Y belts more, 2nd print came out much better; 0.5mm nozzle with 0.4mm layer height, merely printed in 8mins with 60mm/s print speed and 80mm/s 20% infill:
And just for the fun of it, 0.2mm layer height with 0.5mm nozzle, at 70mm/s:
Incredible quality: X and Y surface very good, some inconsistency at “X”, on the “Y” side some slight ghosting; but most surprising is the edges on the Z axis – I operate with a simple M6 threaded rod and M6 nut – that’s all – moving nylon wheel-based carriage up and down – sure, I require to print more tests, in particular larger prints to really see how well all axis print up to 300mm.
I had to use blue tape on the mirror otherwise PLA would not stick – eventually I will use the black sticker as I used for the CTC DIY printer which worked quite well.
Nylon Wheels vs Sliders
The past 2-3 weeks, while waiting the nylon wheels to arrive, I decided to check alternatives such as sliders with PTFE tubes – and this paid off: the nylon wheels 23.0mm OD with 7.3mm width sit quite nicely into the T slot (B-type) but when used in real life, like with X carriage, I had some sinus wobble in the vertical – apprx. 0.5mm to 1mm – way too much. So, I exchanged the wheel-based X carriage with the slider-based carriage, remounted the hotend with Bowden setup, and after 5mins the exchange was done:
X carriage with white nylon wheels (23.mm OD / 7.3mm width)
Current setup:
X axis: slider-based carriage, holding on top and bottom side with 2 tightening screws
Y axis: simple sliders (just sitting on the groove)
Z axis: nylon wheel (23.0/7.3mm) based carriage
The next days and weeks I will review my options:
slider-based carriage with
1 axis support or
2 axis support
wheel-based carriage with
nylon wheels 23.0/7.3 and 23.0/7.0
double V wheels
both on T slot alu extrusion – I know ideally would be proper V slot alu extrusions, but I like to find out how good it works with the easily available T slot extrusions. Worst case is, I have to use on X and perhaps Z axis proper V slot alu extrusions, on the Y axis it seems the simple sliders (just a block) work fine.
0.5mm Nozzle
Since I deal with nearly 3x the bed surface compared to 200×200 I thought I have to use a bigger nozzle as well, as a bigger build volume would imply larger objects to be printed. The increase from 0.4mm to 0.5mm diameter also implies 1.5x or +50% more material being extruded and I still desire to print with 60mm/s average with 80mm/s infill – this means I have to test well the hotend performs with that speed and higher throughput of material.
Specifications
Current specifications of Ashtar K 3d printer:
380 x 300 x 320 mm build volume (400 x 300 bed)
E3D V6 clone hotend
0.5mm nozzle
Anet 1.0 controller board
210 x 210mm 12V heatbed
TODO
bed mounting & leveling
300×300 or 400×300 220V heatbed
proper print surface (likely black sticker 300×300 or 400×300)
Brief update of motors, belts, threaded rods and end stops mounted:
X Axis
I switched from slider based carriage to the nylon wheel (23.0mm OD, 7.3mm width) based carriage (X and Z axis), apprx. 120cm belt length.
Y Axis Belt
Apprx. 90cm belt length – with some considered cutting X + Y ~ 2m belt length.
This part was tricky – the parts are glued to bottom of the wooden carriage:
mark the positions of the sliders (left: 65mm distance from top and bottom, left: center of top/bottom)
glue 2x sliders on the left (where the Y motor is mounted)
glue 1 slider on the right side
put carriage on the rails, avoid any horizontal movement, push it slightly down (a slight snapping you sense from the sliders)
let it rest (don’t touch or move it) for 30min – glue must dry
move Y carriage gently forward & backward; if there is slight resistance then
loosen all screws of the right 2020 beam so you can move it sideways
then move carriage forward and backward and let the beam slightly find its new position
when the carriage moves gently without resistance
fasten the screws gently
retest and if it’s still resitance, repeat procedure
this is a bit tedious work, but worth it
This part is to do next (once I concluded those PTFE pipe chunk based sliders do their job well):
drill holes and use Zip ties to fasten sliders
glue Y carriage belt mount, let it dry
mount GT2 belt to carriage belt mount
fasten Y carriage belt mount with screws: drill holes from the bottom side
Ashtar K: 3 slider based Y bed
Y bed mount
Detail measurements: 110mm from left to slider, 65mm from bottom; stopper, 23mm further, apprx. 20mm aside of slider.
Slider closeup with trigger for Y endswitch
I will make some short videos of putting the carriage together and mounting it – it’s quite fast to attach and detach with those sliders (no screws to unfasten).
Z Axis
I extended the corner brackets so the Z stepper motors can be inserted, this weakens the part but saves quite a lot of space and hides the threaded rods nicely behind the 2020 Z beams. I might work on those brackets later to increase rigidity again.
Alu 5mm to 6mm coupler
Currently I use M6 threaded rods for the Z axis, one cycle gives 1mm height change.
X, Y, Z Motion & Homing
Flashing the Anet 1.0 board (which I currently use) with latest Marlin, this was required:
using Arduino Uno R3 (clone), installing Arduino ISP on it
cabling Uno R3 with Anet board (Uno powers Anet board with 5V, all stepper motors or power detached)
installing Bootloader (“Burning Bootloader”) with “Arduino as ISP” as writer
downloading latest Marlin, copying Anet Configuration.h and starting to change it
Finally, after hours fiddling around (bad install of Arduino failed to compile and/or upload anything to my Uno R3) the LCD display greeted with “Marlin 1.1.8” 🙂
Moving X, Y and Z axis briefly, 380 x 300 x 320 build volume with the current V carriage with 23mm OD, 7.3mm width nylon wheels – no extruder and no bed heating and leveling yet.
As I finished the frame, I focused on the carriages:
X carriage: moving left to right with the hotend with Bowden setup to keep it light: Wheel-based Carriage
Y carriage: moving bed forward and backward, relatively heavy with 400×300 bed with a mirror to ensure flatness: leaning toward Sliding Carriage
Z carriage: moving up and down with X carriage: Wheel-based Carriage
XZ Frame with X- and Z-Carriages
I made some tests with sliding carriage (composed with PTFE tubes), and finally the Nylon wheels arrived and I began to review two kinds of wheels:
23mm OD, 7.3mm width: even it’s wider it sits better in the T slot 2020 alu extrusion
23mm OD, 7.0mm width: is bit more narrow, but doesn’t sit well on the extrusion
So, I put the 23/7.3 nylon wheel on the V plate to compose a V carriage, and applied to X and Z axis:
Y Carriage
The nylon wheels work very well, given the fine-tuning capability, whereas the sliding approach with 2 axis support (PTFE tubes in sliding direction plus vertical to stay in line) doesn’t give tuning capability.
3D printed slider with PTFE tubes
Currently I lean toward the more simple slider (white PLA) with 2 axis stabilization, as with the wheel-based carriage too much vertical force will be applied to the wheel in a perpendicular manner and wear the wheels rather fast.
One of the challenges is to mount three such sliders on the bed – two can mounted quite freely (with margin of 1-2mm) whereas the 3rd slider needs to be mounted very precise.
Building 3D Printer Ashtar K starting with the frame using 2020 T slot 6 alu extrusions, changing design slightly from 9x 500mm to 11x 500mm as early tests showed the XZ frame wasn’t stable enough toward Y bed – so two additional beams (later photos in the series) to make XZ frame sturdy toward the bed.
and the current state more or less:
While waiting for wheels to arrive I thought to make some tests with sliders composed with PTFE tubes (4mm and 3mm in diameter) – I might actually use them in the X axis and Y axis:
The building volume with the 500mm beams can be stretched to 380 x 300 x 360mm, if the X carriage is short(er) – this means, the bed can be 400 x 300.
2018/08/26: Partial functional, X, Y and Z motors and belts and threaded rods mounted with end stopper, board been flashed with Marlin 1.1.8
2018/08/24: More photos about XZ frame bracket with integrated Z motor mount, Y belt mount and sliders
2018/08/20: Short video testing X and Z axis with nylon wheel based carriages
2018/08/15: Added photos of composing the frame (XZ + Y) and changing design slightly to add 2 more beams so XZ frame is more sturdy, early tests with sliders, as alternative with wheel based carriages
2018/08/01: More details, extruder motor on the right side with belt idler mount, short video showing some details.
2018/07/30: Updated images, more examples of prototyping V modules
2018/07/21: Published with few drawings, short part list.
Introduction
In summer 2018 I pondered on a parametric Prusa i3 3d printer designs, composed with 2020, 2040 aluminium extrusions / profile, hereby I document the development here.
The Ashtar W Series and Ashtar T Series are fully parametric, from 200mm^3 to 500mm^3 build volume, whereas this Ashtar K Series focuses on single beam length construction with 2020 alu extrusions.
Ashtar K #1 (right, white) and #2 (left, pink), both 380x300x330 build volune, but having different build plates
Parametric Designing
Unlike traditional CAD (Computed Aided Design) sketched constructions, a coded parametric design is actually textual coded a design, defining which parts depend on which, and align according some variables, which can be changed. In this case, the input is the building volume X, Y and Z, and all parts are calculated accordingly, using OpenSCAD as programming language.
Following notion has been introduced:
X,Y and Z are the starting point, the printable volume
IX, IY, and IZ are the inner dimension of the construction needed to make X, Y and Z build volume work, hence, IX, IY and IZ are greater than X, Y and Z
all constructions depends on IX, IY and IZ
XE, YE and ZE are the position of hotend ranging between 0..X, 0..Y and 0..Z
XP, YP and ZP is the calculated position of the hotend in physical space
These notions, in retrospect, allowed me to code all the different printer types: Ashtar K (Prusa i3), Ashtar C (CoreXY), Ashtar M (Prusa i3 MG), Ashtar D (Classic XY) coherently.
Ashtar K: 500mm 2020 Extrusions as Rails
This is a single size design optimized: 300(-380) x 300 x 360mm build volume, composed by 11x 2020 500mm B-type or V-slot beams:
Backside:
Bottom view:
The rollers on the Y axis can be likely reduced to minimum of 3 total, instead of 9 (3×3), it really depends how well the rollers have a grip on the extrusions. Majority of the printed parts are custom. I settled with DIY sliders with small PTFE tubes instead, they were simpler and turned out reliable enough for my use case, see below “Y Carriage Slider”.
The Y axis is quite short to match 500mm beam length, and the Y bed fits barely as you can see in this bottom view, but it should work:
Moveable V Modules
The V modules, composed by 2x V plates, which holds the wheels running on the alu extrusion, I document separately at 3D Printing: Wheels on Alu Extrusions and is used:
2x Z axis motion
1x X axis motion
3x Y axis motion (perhaps a dedicated module to reduce amount of wheels) or
3x or 4x Y axis sliders
V Module X Axis
As first I mounted existing direct drive extruder piece to the module, although due the thin 2020 profile I likely have to run it with Bowden setup to make sure the moving extruder is light enough.
40mm fan with 5015 fan fang on top, X carriage with 3 wheels (V module)
40mm fan with 515 fan fan on top, X carriage with 4 wheels (H module)
Small belt mount, 1st version is one sided, 2nd version goes both ways to be more flexible:
V Module Z Axis
Z Axis V module is a bit more complex, it takes the X axis beam and the Z axis leadscrew or threaded rod, and the X motor mount:
Slider-based Carriages
Aside of wheel-based carriage, I thought of trying and playing with some slider-based carriages as well:
for more details see my blog post on 3D Printing: Sliding on Alu Extrusions. It eventually didn’t work that well, with time it became wobbling, and the friction increased – so I switched to wheel-based V modules.
Frame
Some photos of early tests with building the frame. I changed the frame design an add two more beams to stabilize the XZ frame with the Y bed more; using 11x 500mm beams now – and some strong bracket at the bottom:
X Axis Module
Two options are available:
2020 T-Slot 6 (B-Type): using 3x Nylon wheels 23.0 OD / 7.3 wide
2020 V-Slot 6: using 3 or 4x OpenBuild 24.4 OD / 11 wide V wheels
The V-Slot beam is more suitable as the X carriage will be more stable and sturdy when printing – yet, V-Slot 2020 beams aren’t easily available or with high shipment costs.
Four options I tried: the 1st with a slider worked only briefly, then 2nd I switched to white Nylon wheels which wasn’t stable enough but wobbled in Z a bit on T-Slot beam, the 3rd was 3x wheel V module, or the 4x wheel H module on V-Slot which worked best.
Belt mount and hotend holder using same mounting holes
Z Axis Modules
Two V modules (each with 2 plates) assembly for 2020 T slot 6 B-Type beams, per module:
3 x M5 x 30
3 x M5 nuts
4 + 2 + 2 M3 nuts (4 front insets, 2 back insets, and 2 for adjustment screws)
2 x M3 x 14 or x 16 (adjustment screws)
3 x Nylon wheels 23.0 OD / 7.3 wide (do not use 23.0/7.0 wheels, but 23.0/7.3)
X, Y and Z Axis Motors
All motors mounted with belts and threaded rods:
and all 3 axis in motion (without extruder and without bed heating/leveling):
and early tests show with the nylon wheel (23.0mm OD, 7.3mm width) based carriage a build volume of 380 x 300 (+10mm outside of bed) x 320mm.
Other carriage, e.g. the slider based, might result in smaller or bigger build volume.
Controller Board
For now I use an Anet 1.0 controller board (as part of a “CTC DIY Kit”), and it required some preparation:
using Arduino Uno R3 (clone) and upload “Arduino ISP”
attach Anet 1.0 board (detach all other cables) to Uno R3
run “Burning Bootloader” with “Arduino as ISP” as writer
downloading Marlin and edit main Configuration.h to match my specifications
upload new firmware Marlin to Anet 1.0 via USB upload
I mounted the board first in the farther left corner (in the photo), but the Z stepper motor new mount required to move the board in front of the XZ frame on the left side. The position and casing for the LCD display I haven’t decided yet.
Y Carriage
Current bed setup (top to bottom):
400x300mm black sticker (“frosted sticker”), apprx. 0.6mm thick
I currently use the white PU steel enhanced GT2 belts, and it produces hard edges, some ghosting, but more precise prints than the black rubber GT2 belts which just stretch too much – I have to research this more closely – about the type of reinforcement and the use with more heavy beds (Y carriage).
420×320 carriage:
4mm plywood flexes, but has been quite flat – not recommended
6mm plywood hardly flexes, but has been hard to buy truly flat – and so far my attempt to flatten it did not work well – not recommend unless it’s flat
6mm OSB quite flat, does not flex much (3 or 4 sliders) – recommended
320×320 carriage (for 300×300 bed):
4mm plywood works (3 sliders, 4 sliders recommended)
6mm plywood works (3 sliders, 4 sliders possible if plywood is truly flat <0.2mm difference)
6mm OSB quite flat, doesn’t flex (not yet tested)
Just to explain my thought or decision process for my setup:
the mirror should not be bend (of course)
the support structure should not be the edge mounts, but the foam in between
the carriage can be bent, but not flex
revelation: already bent means the springs with screws might extend the bent further with a flexing carriage, and not counter act – as the mirror should stay flat
Sliders & Belt Mount Positions
Top view with see-through (best mark “0,0” on both sides so you keep proper reference) – if your carriage is truly flat, choose 4 sliders, otherwise 3 sliders.
Bowden Extruder
After few weeks I decided to do my own extruder, adapting the design of the “Compact Extruder” which has low complexity and low amount of parts to achieve simple extruder functionality; here my redesign:
and in a functional state:
Assembled Simple Extruder
Assembled Simple Extruder on Ashtar K #1
Assembled Simple Extruder on Ashtar K #2
It’s published at thing:3265864, it’s based on 625ZZ bearing:
625ZZ bearing (16mm OD, 5mm ID)
M5x14: mounting bearing
2x M3x25: one to attach handle, another to hold spring
2x M3x8: mounting to stepper motor
M3 nut: insert into slot
M3 washer: to hold spring
3-8mm OD 20mm long spring
hobbed gear OD 11mm
4mm OD / 2mm ID PTFE for filament guides
PC4-M6 for outgoing Bowden tube
Gallery
Ashtar K with 3 extruders and LCD controller
Ashtar K with parametric clear enclosure
In Action
After 3 months (2018/06 – 2018/08), since I started to code the first OpenSCAD lines, the “K” prototype happen to print the 20mm XYZ Calibration Cube:
And roughly 2 months later Ashtar K #2 (with RAMPS 1.4 board) was printing as well, on a smaller 300x300mm unheated bed:
TODO
proper bed mounting and leveling: done
bed heating: running without heat bed
better cable management (in particular heatbed / Y carriage)
release sources
complete instructions
complete part list (printed / non-printed)
Parts
This is a preliminary part list (no files yet published):
Printed
Most of the printed parts
V plates (2 plates = 1 module) with 3 x or 4 x M5 x 30: Note: for each axis the plates must be printed with the same print settings to be symmetric when assembled, recommended setting: 1.5mm top and bottom thickness and wall thickness, layer height ~60% or less of nozzle diameter
X module (with 3 or 4 x black OpenWheels 24.4/11):
short 3 wheel carriage:
1x v_plate-2020-double-v-244-110-48w-a
1x v_plate-2020-double-v-244-110-48w-b
short 4 wheel carriage:
1x h_plate-2020-double-v-244-110-48w-a
1x h_plate-2020-double-v-244-110-48w-b
2 x Z modules (with 3 x white Nylon 23/7.3 wheels) each
3 wheel carriage:
1x v_plate-2020-delrin-230-73-10-a
1x v_plate-2020-delrin-230-73-10-b
3 or 4 x Y modules (1 module = 1 slider, per slider 4 x 10mm long x 4mm PTFE + 3x 8mm long x 3mm PTFE)
3 or 4 x slider_2020-ptfe=true,td=4,td2=3,axis=2,closed=true,hplus=5,hole=true
X carriage:
1x xcarriage_short_hmount_motor-endstop-left Note: minimum 1.5mm top and bottom thickness, and 1.5mm wall thickness, 30% infill, layer height ~60% or less of nozzle diameter
1x xcarriage_short_hmount_motor-right Note: minimum 1.5mm top and bottom thickness, and 1.5mm wall thickness, 30% infill, layer height ~60% or less of nozzle diameter
2x xcarriage_short_hmount
1x xcarriage_beltmount-y=7,w=25
1x pulley_holder
1x endstop_mount
Printhead/Hotend:
1x e3d_mount
2x M3x12 (mounting to x carriage), 2x M3 nuts for insets for clamp (use M3x12 to draw nuts into inset)
1x e3d_mount-type=clamp
2x M3x16 (clamp E3D v6)
Y carriage:
4x knob_30,8,6 (bed level wheels)
1x ymotor_mount
1x ycarriage_mount-h=15
1x yendstop_bumper
1x yendstop_mount
1x pulley_holder
1x ybelt_mount
Z carriage:
2x zcarriage_short_mount-6,30 (for M6 threaded Z rods)
1x ztop_bracket-left
1x ztop_bracket-right
1x zendstop_mount
Frame
8x c_2020 (simple 2020 corners)
2x l_2020-a (short L bracket)
2x l_2020-b
1x ll_2020-a, 3 perimeters/wall line count
1x ll_2020-b, 3 perimeters/wall line count
1x ll_2020-type=nema17-a (X/Z bracket + Z motor mount) Note: 3 perimeters/wall line count with layer height ~60% of nozzle diameter (e.g. 0.25mm @ 0.4mm nozzle or 0.3mm @0.5mm nozzle)
1x ll_2020-type=nema17-b (same notes as above)
2x c2_2020-a (strong L for bottom frame)
2x c2_2020-b
12x e_2020 (end caps)
Non-Printed (Vitamins)
11x 500mm 2020 alu extrusions (T slot 6 B-type or V-slot 6)
Double or single V slot wheels (OpenWheel 24.4/11) and/or 18x (6 x 3) x Delrin R nylon (23/7.3) wheels (see printed parts above which are needed)
Screws & Nuts:
200x M3 nuts
100x M3 8mm
20x M3 10mm
20x M3 15mm
150x M3 Hammer Nuts for T slot/V slot 6
20x M5 x 30
20x M5 nuts
Y carriage / bed:
420 x 320 or 320 x 320 OSB 6mm as Y carriage
400 x 300 or 300 x 300 3mm thick mirror
400 x 300 or 300 x 300 frosted bed sticker
4 x springs 20mm long, compressed 10mm
Belts:
2x GT2 pulleys (ID 3, OD 16, 6 wide with teeths)
~220cm GT2 6mm belt (200cm might be sufficient but without any cutting margins)
Printhead:
E3D V6 original / clone with 0.4mm or 0.5mm (recommended) nozzle
100cm PTFE 4mm OD / 2mm ID (60cm for Bowden tube, reuse rest for sliders)
1x Pneumatic Connector PC4-01
1x Pneumatic Fittings PC4-M6 Bore 4.3mm for 4mm PTFE
Electronics:
5x stepper motors Nema17 42-45NM (40mm height) with 1m wires
1x control board (with Marlin support), e.g. Anet V1.0 or Makerbase MKS Gen L board
2 x endstops with 1m wires
IDEX Option
In order to run two independent printheads aka IDEX (Independent Dual Extrusion) see this blog-post on Ashtar K IDEX with the details and those new pieces are needed:
Current bed setup (top to bottom):
