reducing polymer filament costs

=manufacturing =suggestion

 

 

Filament for 3d printing could be much cheaper if 3d printers use sensors to compensate for diameter variation.

 

 

the goal

The most common type of 3d printing is FFF, and the plastic filament it uses is much more expensive than the plastic it's made from, often ~10x as expensive. That's true for both cheap (eg PLA) and expensive (eg PEEK) plastics.

Extruding plastic is normally very cheap, so why is filament so expensive? It's a very simple product, and really the only thing that can be special about it is dimensional accuracy. 3d printers control deposition rate by moving the filament, so consistency is important, and if you look at 1.75mm filament products on Amazon they say: "Dimensional Accuracy +- 0.02mm".

Plastic products are generally made from pellets called nurdles, and if going from pellets to filament with that accuracy was easy, people would just use pellets directly. 3d printers using pellets do exist, but they have worse print quality than filament printers, so pellets are usually only used by very large 3d printers with looser tolerances.

So, the problem is dimensional accuracy being expensive. That implies that:

1. If filament cross-section could be measured by the 3d printer using it with sufficient accuracy,
2. it could adjust feed rate based on filament diameter,
3. filament dimensional accuracy requirements could be relaxed,
4. and filament could be made at much lower cost.

 

 

sensor approaches

How could filament cross-section be measured accurately at low cost?

 

capacitive

If we assume filament is round and only the diameter varies, that simplifies the problem significantly. Inconsistent extrusion of filament tends to be flow rate variation that still produces round filament, so that might be good enough.

Plastic has a different dielectric constant than air, so passing a filament between electrodes affects capacitance; that's the basis of capacitive displacement sensors. A non-contact dielectric constant sensor with 4 electrodes around the filament could probably work, but here, there's an easier approach. Assuming non-conductive filament is used, we can just pass the filament between conductive rollers, and measure capacitance between those rollers.

 

hall effect

Looking for prior work on filament diameter sensors, I saw this design which runs the filament between rollers and uses roller movement to measure diameter, by driving a lever connected to a magnetic sensor. But that seems more complex and less accurate than measuring capacitance between the rollers.

 

optical

How do filament producers do quality control? My understanding is they do optical measurement from 1 direction with lasers, and such sensors would be expensive for consumer 3d printers. Also, for printers, compensating for filament color could be a problem with optical sensors.

 

liquid metal

Let's suppose we need to measure cross-section directly (instead of measuring diameter) to get good enough accuracy.

It occurred to me, as it occurred to Archimedes, that a short section of filament would displace liquid around it proportional to its volume, and that liquid displacement could be measured. Most liquids would adhere to the filament surface or evaporate, but a liquid metal (such as Ga/In/Sn eutectic) should have negligible losses. That's expensive, ~$1/gram, but we'd only need maybe 10 milligrams.

So, let's have a small ring of metal containing a liquid metal bead of Ga/In/Sn. The filament goes through the ring, and the bead is held in place around the filament by surface tension. Having a capacitive sensor on both sides of the liquid metal bead seems like the best option; assuming the filament is nonconductive, those sensors would mainly measure displacement of liquid metal by the filament.

 

 

conclusion

Current printers don't have diameter sensors because they're not needed with current filaments, and cheap low-accuracy filament isn't sold because printers for it aren't available. That situation makes adoption harder, but diameter sensors are potentially cheap enough to justify adding them just for detection of bad filament. Compensating for filament diameter variation should make good 3d print quality possible with filament that's much cheaper to make.