Project ALTius
Part V: Practical experiments in FDM 3D-printing.
For those who have not already done so, you will likely want to read or at least familiarise yourself with the first four parts of this series. — Ed.
Last time I concluded with “next month, it’s back to the nuts-and-bolts”: I hope you didn’t take that literally. Instead of “nuts-and-bolts” you will have to settle for some “nozzle-and-plastic”. That’s right: time to do some old-fashioned FDM (fused deposition modelling) 3D-printing! Don’t worry, we will have several parts in the future with hardware laser cutters and 3D-printers to help prepare you for the building season which will be upon us by then.
Five Reasons for FDM
The first reason is quite simple: in our journey between the start point of ‘what’ — that is, the 3D models prepared with SAD/CAD — and the destination point of ‘where’ — the magical 3D resin printer — we definitely need to do a ‘how’ pit stop. In other words we need to get you familiar with a very important tool: the slicer — used for 3D-printing — and its relative the ‘g-code generator’ used with a laser cutter. The purpose of these software tools is very simple: to translate the 3D model or the drawing to the G-code language accepted by 3D-printers and laser cutters. Both are in fact some variants of CNCs with specialised tools.
The second reason is that we need to re-evaluate somehow the FDM printing with the 3D-printers, software and materials available today.
The third reason is more practical: previously we have considered two types of building:
- Balsa geodetic ribs covered in composite — ‘low-tech ALTius’
- 3D-printed core covered in composite — ‘hi-tech ALTius’
In fact there are three additional variations:
- Fully 3D-printed with plastic
- Resin 3D-printed core covered with composite
- Fully 3D-printed with resin
That’s right, for rapid prototyping we need to consider also a quick and cheap build without composites at all. Of the five variations we can consider FDM 3D-printing in at least three.
And the fourth and perhaps most important reason: we need to validate the use of theoretical recipes derived from quick W1 and W2 weight estimations with practical results. In other words, weight and printing time estimations from the slicer software as compared to the actual, measured weight and time for printing. We also need to evaluate the structural integrity of the printed wing and to extrapolate the findings from FDM printing to resin printing.
And maybe a fifth reason: you may already have an FDM printer and you want to put it to good use before starting to build and/or saving up to buy a resin 3D-printer.
In the previous parts of this series we had a “plastic bad / resin good” vibe. Scratch that, we will start with a clean sleight and draw conclusions at the end of the day. One additional remark: I will do the tests on my best printer (Bambu Lab P1P) because it’s a modern, fast printer and I’ve upgraded it and I can print all sort of materials with 0.2mm, 0.3mm, 0.4mm, 0.6mm and 0.8mm nozzles. We will consider three different materials (ABS, PLA and PETG) and see in which part of the project we can use and what kind of printer we need. Finally we will analyse if you need a state-of-the-art printer like the P1P or you can just do your job on a more common, budget-minded printer like Creality Ender 3.
About those plastic ‘sausages’ of which I spoke in previous instalments: there is a German saying along the lines of “if you like sausages it’s better not to know how they are made”.† However, we definitely need to know how to make good ‘sausages’ so it’s time to do some printing.
†Don’t laugh, sausages are big business in Germany and these words may be attributed to Bismarck comparing laws and sausages.
3D-Printing Plugs and Molds
I’ll start with a 0.4mm nozzle and PLA on the standard P1P configuration — just the 0.4 mm nozzle and no enclosure. I will print a mold of the tip — not the full segment, of course, but just the last quarter. This first test is a failure.
In fact, despite appearances, it’s actually quite good. It’s a failure because I used the rest of an several-years-old PLA roll and the filament was not enough to finish the part — the tip is not closed. Then again, you can see the cool-looking gyroid infill. The mold has some elements not suitable for 3D-printed molds: the wall is a very small structure designed to protect the leading and trailing edges when you polish the mold / plug. It’s almost of no use in 3D-printed molds and plugs. I left it there just to see how my new printer is coping with very small details and fast direction changes. All we can see in the surface are some vertical lines — 2mm apart — that are correlated with the GT2 belts. I probably need to adjust the belts tension — after all, this is a brand new printer. Definitely usable if between this printed mold and the carbon/resin we use a FEP release film to ‘smooth’ the surface. All we need to do is to use an offset in the design of the model to compensate for this FEP film and probably use ABS for printing in order to do some thermal curing of the resin . This way we can use also some prepregs. OK, we can use an FDM 3D-printer for molds and plugs — what about wings?
Wing 3D-Printing Test Methodology
We will test two ‘quarters’ of a wing: the first quarter at the root and the last quarter at the tip. These are the easiest and hardest quarters to print, respectively. Also, this is just a ‘plain’ wing — no other additional elements like cutouts for control surfaces, servo pockets, joiners or spars. Of course when you 3D-print an actual wing you intend to fly, you will need to consider those and prepare them either in xflrwing and/or OpenSCAD (see Resources, below). No big deal, you will do a difference() between the wing segment and a cylinder or cone or a square-section tube used for spar — remember to do it a little bit larger and not exactly the nominal size of the spar.
We will compare slicer estimated weight versus printed weight, and then we will estimate the printed weight for the full wing by doubling the slicer estimate for a print of all quarters in a half wing. Note: it is not the ‘virtual’ wing of 75dm² and 6000cm³ we used for weight estimation. It’s a ‘real’ wing with 3.84m span, 24cm chord, 76.12dm² projected area and 6487.80cm³ volume. We will observe and measure three variables: weight, printing time and quality. There is an old saying something like “pick two of them because you will never get all three at the same time” — let’s see if it’s true.
OK, let’s start the first round of wing parts printing. Remember, this is a standard 0.4mm nozzle on a standard printer without enclosure.
PLA+ / 0.4mm Nozzle / One Wall / 1% Infill
It’s a failure. Weight: root quarter 58g, tip quarter 22g, total 80g–81g as compared to 86g estimated by the slicer. Likely there is a small difference in plastic density or my filament if actually a mix of PLA and ABS. Full wing estimation: 615g and about 18h print time. Quality: good airfoil profile on printing bed, not so good — actually, really bad — in the rest. Good adhesion between layers. We notice some 5cm diameter ‘islands’ and a visible sagging of the airfoil between these. Also there is a small but visible ‘layer shift’ toward the tip.
PLA+ / 0.4mm Nozzle / One Wall / 2% Gyroid Infill
It’s also a failure. Weight: root quarter 67g, tip quarter 24g, total 90g–91g as compared to 94g slicer estimation. Full wing estimation: 700g / 22h. Quality: the same but ‘islands’ are smaller and also the sagging is not so bad.
PLA+ / 0.4mm Nozzle / One Wall / 4% Gyroid Infill
It’s OK(ish) but importantly not a failure as there are no structural impediments. Weight: root quarter 85g, tip quarter 29g, total 114g as compared to an estimated 119g. Full wing estimation: 1000g / 26h. Quality: printed parts are tough, no sagging visible, still some layer inconsistencies on the tip.
After the first tests I received from Bambu Lab the package with the upgraded parts — new 0.2mm, 0.4mm, 0.6mm and 0.8mm nozzles along with camera and LED lamp as well as a fan for additional part cooling. We can do some additional tests. No point in testing 0.6mm and 0.8mm nozzles but we definitely want to test the 0.2mm nozzle. Another type of PLA, first round was on yellow PLA+, second round with grey PLA — that is, what was left in the roll after printing the enclosure for P1P.
PLA+ / 0.2mm Nozzle / One Wall / 2% Gyroid Infill
Mixed results. Root is OK(ish), tip not so good. Weight: root quarter 44g, tip quarter 16g, total 60g as compared to 63g estimated weight with the slicer software. Full wing estimation: 640g / 3d and 14h. Wow! Quality: OK(ish). Some islands are still visible but similar in size with 4% infill for 0.4mm nozzle and a very small amount of sagging. Definitely can be covered / hidden by the composite surface. If it’s not covered we should not care too much about it — best to think about how dimples in golf balls positively affect their flight, right? Now, how about the quality of the tip? The tip-of-the-tip is almost unusable: the layer shift is significant compared to the nozzle size and there is a visible adhesion problem between layers at the tip.
PLA+ / 0.2mm Nozzle / Two Walls / 2% Gyroid Infill
This is just the root — I used the last part of the PLA roll and I estimated that there was not enough filament for both root and tip quarters. However, I was wrong — not even the root was finished, it printed only 19.6cm out of 23.75cm so we will extrapolate the weight and … drum roll … it’s OK! Not just OK(ish) but rather it’s fully flyable and airworthy. Weight: this partial print was 58g — so 69g–70g for a full print of the root quarer. No big surprises here, the weight of 0.2mm / Two Walls / 2% Infill is similar to 0.4 mm / One wall / 2% Infill. Full wing estimation: 980g and a total print time of 5d and 7h.
PLA+ / 0.1mm Nozzle / Two Walls / 2% Gyroid Infill
Just kidding, I’m not able to print with a 0.1mm nozzle on the P1P, at least for the moment. I have some 0.1mm nozzles and I’ve ordered some aftermarket hotends for the P1P and maybe I’ll have the patience to do some printing with 0.1mm nozzles. But I finally finished the enclosure for the P1P using some PLA printed parts and acrylic panels from China. Let’s try this new enclosure with some black ABS.
ABS / 0.2mm Nozzle / Two Walls / 2% Gyroid Infill
Actually I started this print but I had to stop because my 0.2mm nozzle clogged due to the old and dusty ABS roll of filament. I ordered some 0.2mm needles used in unclogging the nozzle but what I received are 0.24mm and 0.25mm needles. I probably can unclog the nozzle but in this case it will be a 0.25mm nozzle and the flow of plastic will be increased by 50% (0.2 * 0.2 = 0.04, 0.24 * 0.24 = 0.0576, 0.25 * 0.25 = 0.0625) and the tests will not be relevant any more — we will not be comparing apples with apples. So, I’ll wait for the aftermarket hotend with 0.2mm nozzles. Back to 0.4 mm nozzle and black ABS.
ABS / 0.4mm Nozzle / One Wall / 3% Gyroid Infill
Weight: this print was 58g and a full wing estimation of 800g and 24h print time.
FDM Face-Off
It’s time to pause the FDM printing session and have a contest. So far we had two finalists: Mr. Golf Ball Dimples (PLA+ / 0.2 mm nozzle / 1 wall / 2% infill), Mr. Nice Guy (PLA+ / 0.2 mm / 2 walls / 2%), and right at the wire a new finalist: Mr. Stinky (ABS / 0.4mm / 1 wall / 3%) — ABS has a very bad odour when printing. Let’s order these contestants with regard to four criteria. Three were mentioned previously — weight, strength and printing time — but let’s now add how nice they look.
- Weight: no surprise here, Dimples. But Nice and Stinky are in a tie for second place.
- Strength: Stinky as it was rock solid, two very thin lines of 0.2mm PLA force-cooled can’t beat a single 0.4mm line of 270C-fused ABS without part cooling.
- Printing time: Stinky wins by a large margin.
- Beauty: of course Nice, hands down. Both Dimples and Stinky have dimples — but Stinky subjectively looks better.
Interesting combination: Stinky wing estimation was 800g and a printing time of 24h. Let’s estimate AUW (all up weight): another 100g for carbon spars, 80g–100g for tails, 250g–300g for fuselage. Another 300g–400g for ‘dead weight’. Estimated total AUW: 1500g–1700g. If I will print this with a 0.3 nozzle I can get the wing weight down to 600g and total AUW 1300g–1500g. Pretty good — the wing was designed for AUW 1380g. Another thing: I printed this segment four times with four different speeds which I will call Silent, Standard, Sport and Ludicrous. What I wanted to test was consistency of weight, extrusion and layer adhesion at high speed.
ABS / 0.4mm nozzle / One Wall — Speed and Consistency Test
There is no big difference between these four prints: actually Sport and Ludicrous look better. With this high speed the wing printing time is reduced to 12h –14h. This means I can print the whole glider in one day using 1kg of ABS. The 0.3mm nozzle can print with a layer height of 0.2 mm and the printing time will be the same but with a reduced weight. The problem with this combination: it uses a standard 0.4mm nozzle but not every printer can handle this type of ABS printing. Not your average Ender 3 anyway — ABS is notoriously bad at warping. And with extreme speeds there are definitely some adhesion problems, not visible at printing time but rather later — internal tension leads later to some cracks on the leading edge.
Good Things Come in Small Packages
Meanwhile I received a small package from China with a couple of aftermarket hotends and some nozzles: hardened steel 0.2mm, 0.4mm and 0.6mm along with CHT (core heating technology) 0.4mm and 0.6mm. CHT nozzles are for printing with increased speed. The filament is driven through three different paths inside the hotend in order to obtain a faster, more consistent melting of plastic and hardened steels for abrasive filaments like carbon fiber plastic. The CHT 0.6mm I’ll use for structural parts and CHT 0.4mm for wing parts where I want to obtain the maximum speed. However, the real trick is an old 0.3mm brass nozzle mounted on this aftermarket hotend using standard 6mm nozzle thread.
Time for Round Three
This round is for tuning in printing with this 0.3mm nozzle. We can achieve the same print time with 0.3mm nozzle / 0.2mm layer height and 0.4mm nozzle / 0.2mm layer height — but our goal now is weight reduction. Back to PLA+ yellow.
PLA+ / 0.3mm Nozzle / One Wall / 3% Gyroid Infill
Weight: 59g root and 20g tip, total 79g as compared to slicer estimated 81g.
PLA+ / 0.3mm Nozzle / One Wall / 2.5% Gyroid Infill
Weight: 56g root and 20g tip, total 76g as compared to slicer estimated 81g.
PLA+ / 0.3mm Nozzle / One Wall / 2% Gyroid Infill
Weight: 52g root and 19g tip , total 71g as compared to slicer estimated 76g.
Now let’s see how low we can go with the weight by printing with transparent ABS — the layer adhesion is better without any dye in the plastic using 0.2mm nozzle.
Transparent ABS / 0.2mm Nozzle / One Wall / 2% Gyroid Infill
Weight: 55 g, as compared to slicer estimation for the whole wing of 600g and print time of 91h — ouch! Even if printed using the fastest speed I can get with this printer it will be two days or similar. But the print looks great and the structure is fine. Probably that’s the lowest value for weight at 600g and maybe I can push it to 550g. But — damn! — printing time is huge.
Some Interim Conclusions
Weight and structural strength are linked, which is normal after all: more material equals more strength. However, good looks and low printing times are not very good friends, I’m afraid.
- Weight Estimations — Quite good, for FDM 3D-printing we can trust the estimations of the slicer software. And in some cases what we got from the printing plate is even lighter — around 95% of the weight estimated by the slicer.
- 4m Wing 3D-Printed Under 1kg — This was my original goal many years ago. The answer is yes, definitely, it can be achieved. But if you want some weight savings you need to print with ABS (1.05g/cm³) instead of PLA/PETG (1.25g/cm³). ABS, on the other hand is a lot harder to print. You have to deal with contraction, warping, a heated print chamber and other considerations. Better forget ABS and print PETG instead with a 0.3mm nozzle.
- Print Time — On the very fast P1P printer times are acceptable for 0.4mm/0.3mm nozzles although significantly increased for 0.2mm nozzle. And it’s not simply doubled. At one end of the spectrum — for 0.4mm / one wall / 2% infill — it’s 22h and at the other end of the spectrum — 0.2mm / two walls / 2% infill — it’s a whopping 127h! The same amount of plastic but with 5.77 times increase in printing time! Note that a four times increase is actually what’s expected — double the layer numbers plus double the walls plus double the infill. However, it looks that there are some additional factors such as how the slicer software is handling the extrusion. Also, the printing times above are for Standard speed. With the P1P I mentioned there are four speed settings of which Standard is the second slowest. The faster Sport and Ludicrous settings will certainly speed things up, but at what cost in terms of quality? Finally, these print times were estimations for standard layer height half of the nozzle diameters but in theory you can print with a layer height up to three-quarters and in this case the total print time will be reduced by two-thirds.
- Quality As you will see from the pictures above, it is possible to achieve acceptable quality using FDM 3D-printing with the right settings and some simple precautions.
Next time, I’ll be taking what has been learned with all of this testing and turn it into specific recommendations for optimising FDM 3D-printing for RC sailplanes. If you have any questions feel free to add them in the Responses section below. You get there by clicking the little 💬 below.
Thanks for reading. Until next time, best of luck!
©2023 Tiberiu Atudorei
Resources — 3D-Printers
- Anet A8 V2 — “High cost performance ratio desktop FDM 3D-printer …”
- TwoTrees Sapphire Pro — “CoreXY SP–3 high resolution professional cube 3D-printer …”
- TwoTrees Sapphire Plus — “CoreXY SP–5 fast … large scale 3D-printer …”
- Tevo Tarantual Pro — “3D-printer FDM (235mm x 235mm x 250mm) …”
- Bambu Lab P1P — “Fast 3D printing right out of the box …”
- Creality K1 — “Speedy 3D-printer …”
- Kingroon KLP1 — “a fast-print-speed CoreXY 3D-printer with Klipper firmware and a protective enclosure …”
- VORON2.4 — “We build space shuttles with gardening tools so anyone can have a space shuttle of their own …”
- BLV mgn Cubes — “an open-source 3D-printer project …”
- Anycubic Photon — “colorful touch screen equipped with Photon system and bring new functions …”
- Creality LD–002H — “High precision resin 3D-printer with large printing volume …”
- Anycubic Photo Mono X — “A large build area for your 3D-printed creations: 192mm x 120mm x 245mm …”
Resources — Other
- G-code on Wikipedia. — “the most widely-used computer numerical control (CNC) programming language. It is used mainly in computer-aided manufacturing to control automated machine tools, as well as from a 3D-printing slicer app…”
- OpenSCAD — “software for creating solid 3D CAD models. It is free software…”
- RCGroups thread for Project ALTius. — “altius, citius, fortius — sounds familiar? That’s the Olympic motto where ‘altius’ means ‘higher’. But the spelling (ALTius) is related also to my initials — Atudorei Lucian Tiberiu…”
- xflrwing — “STL generator for an XFLR project wing…”
All images by the author. Read the next article in this issue, return to the previous article in this issue or go to the table of contents. A PDF version of this article, or the entire issue, is available upon request.
