One of my many motivations for starting a venture in 3D printing came early on, when I first read about the huge Mammoth printer Materialize had developed and build for their own fleet of 3D printers (http://www.materialise.com/en/blog/materialises-mammoth-stereolithography-3d-printing-on-a-grand-scale).
The specifications of the Mammoth are breathtaking: maximum part dimensions are 2100 x 700 x 800 mm @ 0,1mm layer height. And of course the Mammoth is based on stereolithography, in short SL (or SLA, where the “A” stands for “apparatus”). Just imagine the amount of resin in a big bathtub-like container – so huge you cannot call it a vat anymore! You can still call the Mammoth the mother of all SL apparatuses. Needless to say, that the fairly large-sized 0,8mm focal point UV laser needs a minimum of 3 days to finish its job for a full-sized model.
The Mammoth is of course based on Chuck Hulls famous patent, initially filed in 1984, and, as mentioned, uses a UV laser beam to polymerize a photosensitive resin, layer by layer. I guess, I will do at least one more blog article on photopolymers, but for now let’s just keep in mind the simple but nevertheless mindboggling concept that one can solidify a liquid in a controlled fashion, with just a bit of light.
Since Chuck Hulls’ invention and the birth of the Mammoth, SL technology developed rapidly in different directions. Material-wise, we have to distinguish between the resin, polymer and metal powder processes. I’ll reserve this article only for resin based SL, which branched out into 3 mayor technologies, based on:
- UV-laser
- Texas Instruments (TI) digital light processing (DLP) http://www.ti.com/dlp-technology/markets.html
- LCD masking technology
Now leaving aside the optical quality aspects of those 3 totally different technical approaches, and only looking at the throughput aspect of things, in an approach based on first principles, we can easily tell that the UV-laser-based system is actually the slowest one. There is a tiny 1D laser spot (typically between ca. 50 to 300 microns) being mechanically (!) deflected in order to scan as fast as possible over a comparatively large surface (typically between ca. 100 to 1.000 cm2). The print envelope is something like 4 million times the size of the laser spot surface. Scaling a 3D printer from one laser to 2 or even 4 laser sources does not really change the speed game, here. By the way, the mechanical laser mirror set-up is nowadays as fast as it can possibly get. For the moment I would like to conclude that there is no visible pathway for substantially increasing printing speed for laser based SL systems. And I would go even further in saying that any future speed gains will be offset by the choice of using smaller laser spot sizes in order to increase print resolution.
The SL 3D printing game took a step forward when DLP from TI was introduced into 3D printing. Suddenly, a controlled way of projecting structured light on a plain surface was possible. Early TI-DLP chips had lousy resolutions but just the possibility to scan a given surface in one shot in only ca. 1 to 10 seconds compared to the 100-1.000 and more seconds for laser-based SLAs! The throughput argument was a major driver for heavy development. Some SLA set-ups are even pushing the (chemical) limits down to 0,1 seconds for a layer hardening step.
So we can conclude that, depending on the setup and 3D model in question, DLP-SL is 1 to 4 orders of magnitude faster than laser-SL! This is a huge difference and the main reason for fairly recent and major product developments in that field. Here I refer to Carbon’s CLIP technology (https://www.carbon3d.com/ ) and to 3D Systems Figure4 technology (http://3dprinters.3dsystems.com/figure4), though the latter is a case of history’s irony…
Formlabs’ market success for their laser-based SLA is basically founded on the fact that they had the gumption to be first to fight 3D Systems and Chuck Hull’s intellectual property and come up with a straight forward 3D printer design, reducing the astronomical profit margins to a kind of normal level in offering a 5.000 € desktop printer instead of 100.000 € printers. There is no technical reason why a relatively simple-structured laser-SLA 3D printer should not be “cheap”. I tip my hat to anybody who tries to bring laser-SLAs to the masses!
However, when we talk about industrial use and future-proof concepts, it is clear that DLP has won over lasers. Still, DLP might be dead before having even grown up. A fairly new kid on the block, LCD-SL is fighting for first place in the resin-based 3D printing segment. And the reason why does not lie in either the optical or in the throughput performance comparison. It is still too early to announce a definitive winner on these two fronts, but DLP-SL’s position is jeopardized as LCD-SL clearly wins on pure economics.
To explain this, an approach on first principles will again help us. DLP is based on millions of micromirrors on a silicon semiconductor chip which, by itself, is complicated and expensive enough. Additionally, to function correctly and reliably, this DLP-chip needs high computational power and complicated optics. Moreover, those expensive micromirrors tend to be super-sensitive exactly to the type of UV light needed for photo-polymerization. We are talking about multitudes of € 1,000 for the DLP-chip, for the computer power as well as for the optics, and, inherently, are talking about the same orders of magnitude of Euro if something breaks down. LCD-SL in comparison uses technology based purely on semiconductor technology for LCD screens, which come, let’s say, in the € 10-100 price range. Similarly affordable are the control boards and the optics. That way, even if something breaks, it is quite economical to be fixed and repaired.
So, even if LCD-SL does not necessarily outcompete DLP in terms of printing resolution or speed, it outpaces DLP in the simplest way possible: money, money and again money.
Yannick Bastian
FormWerk Founder & CEO