Doing quite a lot of experiments, and learning.

[Preston_Bannister]

Doing quite a lot of experiments, and learning.

First, I built model airplanes, as a kid. Plastic is a kind of lousy material for building flying things. Too heavy, and not strong for weight. Still, possible to make complex shapes, not practical with other materials/fabrication. In particular, single wall FDM prints allow light and stiff structure with little material. Also allow more complex and interesting aerodynamics. Enough to make up the difference?

Also experimenting with larger nozzles. Did the first year of my learning curve with 0.4mm nozzles, as that seemed to be the consensus recommendation.

But … I do mostly functional / structural prints. Have printed exactly one figurine (of Naomi). Interested in function more than pretty detailing.

Swapped in a 0.8mm nozzle, and … interesting.

First, I am discovering the thermal limits of hotends. Put simply, the usual 40-50 watt heaters seem to be a limit. (Very much want to get a Ubis hotend, in time.)

With the 0.4mm nozzle, the physical limits of the printer in terms of moving around the nozzle seemed part of the limit. Was unclear that the hotend was a limit (though was at least in part).

With the 0.8mm nozzle, at the same rate of travel, and the same profile, there is 4x as much plastic extruded. Rate of travel while extruding is limited to about 30mm/s, or the extruder starts skipping. I believe this is strictly a thermal limit in the hotend.

Also, I expected cruder results with the larger nozzle. Got some prints so smooth as to foul up my camera auto-focus. Very much becoming a fan of larger nozzles.

Note that I am using a stronger part cooling fan than stock, and a better duct design to direct the airflow. This was needed for good result with the more complex single-wall prints.

The large rocket is an easy print and 388mm tall. Takes about eight hours at 30mm/s with 0.5mm layer height and 0.8mm line width. Pushes the TronXY X5S printer close to it’s advertised limits (nominally 340mm x 340mm x 400mm). The short version is that print bed design is inadequate. Past 300mm in height, and the bed shifts under the print. (Examine the nose of the rocket.)

Note that I added structure to the underside, to stiffen and flatten the print bed. Seems the two leadscrews, four smooth rods, and sleeve bearings used for Z-motion have a certain amount of “give”. The normal printing force of laying down plastic are enough to make the bed shift. (The print is in effect a 300mm+ lever in this large and stiff object.)

I had doubts about this design for Z-motion, but that was just a guess. Good to put things to empirical test. Also good to know my doubts were justified.

With modifications, the TronXY X5S is a cheap printer that does well enough within a 300mm cube. Note that I am only printing with PLA, at present.

The glider is 600mm from wingtip to wingtip, and from nose to tail. Needs five prints of a larger printer, and is a challenging and much longer exercise. The tail and wings are single-wall “spiralized” prints. The total weight of the final version was 380 grams, of which about 100 grams (or more) could be “payload” in the nose. Not bad compared to other folk’s printed-plastic designs (maybe a notch better?), but still very heavy for a flying model.

What worked very well was using compound curves in single wall prints. This allows light and stiff structure, with interesting aerodynamics. (The appearance is the end result of functional choices, not aesthetics.)

The 3D-printed glider did in fact work, but is not going to replace balsa.

[John_Bump]

I’ve seen people printing wing ribs and then covering with RC wing fabric.

[Preston_Bannister]

@John_Bump Which again limits to simple curves. Also plastic is not good as main structure.

[Ryan_Carlyle]

Large nozzles rock.

Melt zone length (=residence time) versus filament conductivity tends to be a bigger limit than heater power, but you do eventually run out of heat if you extrude fast enough, yeah.

[Cameron_Spiller]

Gliders look great!!
There are a couple of commercially available 3d models of rc aeroplanes that look to fly quite well. Kraga is the commodity making them.

https://www.3dprintedrcplanes.com/planes/

There’s also a spitfire model and a few others doing the rounds.

I have a fiberglass hotliner kit called a Dago Red that I bought years ago . Magnificent looking model. Unfortunately the fuselage is twisted like a cork screw due to a common manufacturing fault. So I never flew it. I have started to model the fuse to print but haven’t decided exactly how to get it strong enough for the power.

I’m thinking spiralize and then glass or carbon on top, or spiralize and carbon braid inflated inside… perhaps just some carbon tubes through it like Kraga. … then fair it and paint…

https://www.3dprintedrcplanes.com/planes/

[Preston_Bannister]

@Cameron_Spiller Note the weights of the models - these are all very heavy, which is expected as plastic is heavy. Saw the videos of the plastic model Spitfire. It flies fast because it must! As too heavy to stay in the air at more normal speeds.

I seem to have done slightly better with total weight in a printed model, but this approach is still too heavy.

With a proper glider, just walking across a field in a slight breeze, the model wants to lift out of your hand. With this plastic glider, it needs a hard shove to get enough speed to fly. This means when the unguided glider comes back to ground, it is travelling much faster, and is likely to break. Not a desirable combination.

Still, an interesting exercise.

[Preston_Bannister]

@Ryan_Carlyle Yeh, my Physics is too rusty. The longer melt zone will help to a point, but in the end we need a minimum amount of power to raise the temperature of the volume of plastic extruded.

In this case extruding 0.5mm x 1.0mm at 30mm/s seems to be pretty much the limit. (The spiralized single wall print allows pretty much continuous motion.) This is the usual “mk8” clone hotend.

How much power is needed to raise the temperature of ~15mm^3 PLA per second from ~20C to ~240C? Any good references on the thermal properties of PLA? (Are the characteristics constant at differing temperature and solid/fluid state?) The minimum power needed should be calculable.

[Ryan_Carlyle]

Let’s see, I made a spreadsheet for this a long time ago, just have to find it.

Polymers don’t have constant heat capacity nor a well-defined latent heat of fusion for melting/freezing, due to some crystallinity and thermodynamic effects that are too complicated to talk about, but roughly speaking PLA’s heat capacity is about 1.8 joules per gram-kelvin (j/g-k) and has 1.25 g/cm^3 density and requires 175 K of heating from room temp to printing temp… so that maths out to around 0.4 watts per 1 mm^/sec melted.

This is probably a bit low due to various crystalline phase changes PLA experiences… but that’ll be blend-specific and heating-rate-dependent, and takes expensive lab equipment to actually measure (differential scanning calorimetry), so there isn’t much point in worrying about it. I would hazard a guess that 0.4 number is probably within +/-30% of reality, but who knows.

(For comparison, ABS is about 1.4 j/g-k and 1.06 g/cm^3 with 195 K of heating, for 0.3 watts per 1 mm^3/sec melted. It’s pretty amorphous so I don’t think there’s quite as much concern about crystallinity changing this number. ABS can definitely be printed faster than PLA with the same hot end so this checks out.)

So, basically, a 30 watt heater can provide more than you’re likely to ever need, assuming your hot block is well-insulated (is it?) and your PID is tuned in a way that lets it actually supply 100% power on demand.

If you want an empirical point of confirmation that conduction is more often the issue than heater power, compare maximum printing melt rates of a PTFE-lined hot end versus an all-metal hot end. The all-metal version can typically print twice as fast in terms of filament flow rate.

[Ryan_Carlyle]

Oh, one more factor. Your motor also adds some heat to the filament through viscous friction. In a screw extruder, shearing provides around half the heating, but it’s probably a lot less for filament extruders. If you run your motor up to stall speed, and multiply its peak torque times its rotational speed with appropriate unit conversions, you can calculate how much mechanical power the motor is delivering to the filament that is turning into heat by viscous friction. It can be a few watts I think.

[Preston_Bannister]

@Ryan_Carlyle For what is on the printer now, the heater block is not well insulated on this (quite basic) PTFE lined extruder. Also the part cooling airflow near to the nozzle is likely taking away some heat.

I have an all-metal hotend (an E3D v6 clone) that is both better insulated and should better control the part cooling airflow - but not yet installed. (Was waiting to complete the prior exercise.)

[Preston_Bannister]

@Ryan_Carlyle Interesting. The same print at 30mm/s works, but skips at 40mm/s. Using your numbers (but assuming a 200C delta), the minimum needed is 6.75 watts, and fails at 9 watts.

So the usual 50W(?) heater cartridge seems entirely sufficient. Maybe.

There is some loss from the heated block through radiation, and from the part cooling airflow. (How much? No clue.)

Phase changes in an organic polymer tend to be complex, so we might be off quite a bit in the energy needed. (How much? No clue. Though one of my friends from college is a professor in organic chemistry, so maybe I should ask …)

Also plastic tends to be a fair thermal insulator, so the temperature of the plastic within the extruder (before the nozzle) is likely not at all uniform.

I have no opinion as to the efficiency of the PID loop in Marlin. (Does anyone?)

Lacking more exact information, and figuring things as a conservative engineer, I would pretty much halve/double on all the unknowns. So twice the energy to melt, half the energy lost from the heat block, and half the energy from the PID loop. That puts the WAG at 54 watts works, and 72 watts fails … which kind of fits.

More exact answers are always welcome.

I suspect the better insulated and all-metal hotend with the longer melt zone will indeed perform better, as you suggest. Want to bet on whether this allows doubled extrusion rates, or not??

[Ryan_Carlyle]

@Preston_Bannister volcanos can do about 30mm^3/sec with PLA and you’re getting 15, so yeah, doubling max speed sounds about right

All the firmwares’ PID code is similar (it’s like 5 lines of code) but a lot of people limit max PWM duty cycle in config or eeprom to less than max, for safety reasons. That might cap heater output to 20 or 30 watts or whatever. 40w is arguably a dangerous amount of power since it can heat the hot block to the auto-ignition point of plastic. 80w can melt the aluminum hot block.

Agreed that these are fuzzy numbers. You can actually measure melt power with just a regular printer if you want to go to some trouble… you first set the heater power to various constant numbers and record the resulting steady-state hot end temp, to chart a “heat loss vs block temp” curve (which will be first order with temp delta to ambient air). Then set the heater power to the constant value required to steady-state reach your desired printing temp (plus 10C maybe). Then you extrude filament at a steady rate and measure temperature sag due to power going into the filament. That temp sag and your heat loss chart lets you calculate power lost to the filament. That and your extrusion rate give you melt power per unit filament.

[Ray_Tice]

Nice write-up! PLA is 20% denser than ABS. Switching filament could save weight.

[Preston_Bannister]

@Ray_Tice The large printer I am using at present (a TronXY X5S) seems a bit dubious from anything other than PLA. The hotend is only safely usuable to ~250C. The bed takes a long time to reach 60C (and that with added insulation on the underside).

For the moment I am content to use this printer for large format PLA prints (which is what I had in mind when bought, and the printer was cheap). Building another large printer, to be more capable, and have an i3 clone (for smaller prints) that does a serviceable job with ABS.

A 20% reduction in density might help with the wings and tail, which are half the weight. So might get a 10% reduction (for a lot of trouble getting this tricky print completed in ABS). Not worth the trouble, as a proper flying model this size should probably be less than 100 grams.

In short, plastic is far too heavy for this use.

[Ray_Tice]

Is “far” a factor of 2? 10? I realized you’ve already reached your conclusion, but did you try fiber filled PLA, such as carbon fiber? (requires nozzle upgrade) The increase in stiffness means you could decrease wall size and therefore weight. Another route to thinner walls is to print thinner and coat with polycrylic or some such. This helps keep vase walls from splitting. Printing widths greater than the nozzle diameter helps with that too.

[Ryan_Carlyle]

Plastic is >50x stronger than balsa in most load orientations (tension across the grain, shear, crush) and is only about 7-8x denser, so I’m sure all-plastic RC planes are perfectly possible… but I think you’d need an unreasonable amount of work to go into creating engineered structures (trusses, spars, structured foams, or whatever) to get the stiffness/weight ratio up. A 1mm thick shell with slicer infill is just too un-optimized for this.

[Preston_Bannister]

@Ray_Tice To be truly comparable, we would want about a factor of four improvement. I am skeptical of carbon fiber filled plastics. To fit through the extruder nozzle, the carbon fiber has to be chopped very small, which is not good for strength. The tests I have seen suggest the tiny lengths of carbon fiber improve stiffness, but not strength - which is what I would expect.

Keep in mind, for light structures you win with less dense materials, even with the same strength per weight. This is why aircraft were built largely from aluminum and not steel.

This is true of bicycles, also. In the 1970s the use of high strength steels reached their peak. I worked at a high-end cycling shop while in college, and got to see a lot of exotic gear. You could buy frames strong and stiff enough for racing, but the steel tubes were so thin you could crush them in your hand. If you got a dent in the tube (very easy!) then you had to throw away the frame. There were some experiments with larger diameter steel tubing (better stiffness per weight), but they were too prone to crumpling failures.

This is when aluminum frames gained traction. The strength per weight of steel and aluminum is about the same, but aluminum is about 1/3 as dense. This allows larger diameter tubes, as for the same weight aluminum is three times as thick. Larger diameter tubes are more efficient structurally, so you can use less weight of material, and still have walls thick enough to avoid crumpling failure.

As @Ryan_Carlyle notes, you can build flying models from plastic. But the denser material puts you at a huge mechanical disadvantage. (Try to fabricate a plastic structure equivalent to a 1/4" Balsa spar. Crumpling is going to be a huge problem.)

My exercise was about pushing the envelope with FDM printed plastics, in particular with compound curves in single-wall prints. Did get a flying glider, and might have done a bit better than others (from published weights). Got what I wanted from the exercise.

[Ray_Tice]

Right, adding chopped carbon fiber in filament improves modulus of elasticity, not yield point. However, in Euler buckling, critical load is proportional to MoE, so the fiber is modifying the right property. (looks like spheres work too, based on the paper below)
It does look like you maxed out what you can do with unmodified PLA through a 1 mm nozzle, and you have some nice prints to show for it.
http://leaders.4spe.org/spe/conferences/ANTEC2017/papers/80.pdf

[Preston_Bannister]

@Ray_Tice First, Google thought your last comment was spam. Unclear why.

I suspect you are right about with improved stiffness comes better resistance to buckling, but probably not enough. Also the fibers seem unlikely to cross layer boundaries. (Bit of a joker.)

Bet someone could get at least a Master’s out of co-optimizing the structure and aerodynamics of a single-wall wing. Pretty sure the computational tools did not exist a few decades back to evaluate the compound curves. Not sure they exist now.

[Preston_Bannister]

@Ryan_Carlyle Forgot to mention … I am a bit skeptical about the numbers for relative strength between balsa and plastic. Do not have any objective data, just old subjective memory.

Keep in mind that balsa is a natural substance with a fairly small market. Balsa for model airplanes was meant to be carefully selected, but even then I would find “soft” (low strength) balsa, and “hard” (higher strength) balsa. (The “soft” stuff was poor for structure.)

There was a period when the airline manufacturers were using balsa in aircraft flooring (do they still?), as a spacer between aluminum skins. Could they find that volume of the higher quality balsa? Did they need the higher quality? The published numbers might represent minimum numbers expected of bulk purchases, not of the select quality wanted for model aircraft.

Then again, the mechanical advantage of larger cross section for the same weight of material might explain my subjective impression. Still doubt that plastic is several times stronger for same weight.