MAR K NEWTON
Mark Newton shares the story of how he went from 3D printing novice to producing a half-metre-square rendition of Orkney’s Ness of Brodgar dig
Mark shares the story of how he went from 3D printing novice to producing a half-metre-square rendition of Orkney’s Ness of Brodgar dig.
For my first foray into 3D printing, I decided on a printer that was as close to “plug and play” as possible. I was drawn to the British-made Robox printer ( celrobox.com), with its custom reels of filament that tell the printer what settings it needs. I also liked its unique print heads with shut-off valves to stop the flow of plastic, rather than the conventional method of retracting the filament.
3D printing uses FDM (fused deposition modelling), which pushes a thin plastic filament through a heated nozzle at between 160°C and 300°C. This nozzle is usually 0.4mm in diameter but can range between 0.2mm and 0.8mm, creating layers of plastic as thin as 0.01mm.
One of your first tasks is to decide upon the material you’re going to use. The two main types are biodegradable PLA and the rather smelly ABS, but there are dozens of others – including nylon and wood.
Initially, PLA seems the obvious choice since it’s easy to print with, but it deteriorates in UV and has a “glass point” (the temperature at which it becomes flexible) of 60°C, which renders it useless for many applications. While PLA doesn’t need a heated bed to print on, other plastics do. If a plastic is allowed to cool too rapidly whilst the print is going on, serious warping can occur.
This caused me huge grief in the early days, as I experimented with materials and different print-bed surfaces. On one occasion, the combination of print bed and the plastic chosen bonded so well that I had to remove the bed and use a thin knife to release the print job from it – but at least it didn’t warp away from the print bed!
One thing I’ve learnt while experimenting with lots of different materials is that this can bring its own problems. When you change materials, you have to purge the old material out of the nozzles. If you change from one temperature plastic to a different one, carbonisation can result in a blocked nozzle, which isn’t easy to remedy.
Once you’ve chosen your material, you need to set the temperature of the print head, the print bed, the rate of flow of the filament, and a variety of other parameters. With the Robox all this worry is handled by “smart spools”, whose embedded chip will detect the type of plastic and set the above accordingly. All printers use printing software, complete with preset profiles for their materials. You can tweak these to your liking, but for 3D printing to become mainstream, it needs to be much easier to set up. Even valiant efforts by firms such as Robox have some way to go, as a read of the support forums confirms.
If a print job is to go wrong, it will most likely be on the first layer. If this layer doesn’t stick to the print bed then the subsequent layers won’t adhere and the printer will attempt to print the entire job in mid-air, resulting in an effect similar to the output of a spider on acid!
Placement of the printer is also an issue. The garage isn’t ideal, because of dust and moisture. The plastic filament itself can absorb moisture that will ruin a print run, so if you aren’t using the printer for a while then you should unload the material and store it in dry containers or de-humidify the print room. Then there’s the noise whilst printing: the Robox makes such a racket that it has to be operated at the other end of the house, behind closed doors, and even then I’m aware when the printing stops!
To remedy this, increase print volumes and boost speeds, I decided to buy a beast of a printer in the form of a Raise3D Pro2 Plus ( raise3d.eu). Thankfully, it’s almost silent.
So what do you print? There are lots of 3D models available on the web for you to download. Some are simple, but others require removable supports to be added to the print job by the software. After printing some of these downloaded models, it won’t be long before you’ll want to either alter them or create your own. There are plenty of programs available, but after much trial and error, I’ve found that Autodesk’s Fusion 360 ( pcpro. link/290fusion) is rather excellent, and this normally very expensive software has a free hobbyist licence.
It’s a full 3D design tool with much more functionality than you’ll ever need, and it will export your design as an STL file that the printer software requires. Meanwhile, Autodesk’s versatile Meshmixer ( meshmixer. com) is also free. This enables you to edit and repair STL files ready for printing. The final piece of software I’ve found invaluable is Simplify3D ( simplify3d.com). This emulates the printer by converting your model into the “slices” – one per layer – that the printer requires. You can then “single step” through each layer and see what’s going on, and hopefully detect any trouble spots. It costs $149, but can save you plenty of time, frustration and materials if trying to print badly formed STL files.
All this done, load your model into your printer’s supplied software. This produces a series of instructions that the printer understands known as “G-codes”. The file containing these G-code instructions can be transferred to the printer via an SD card, USB drive or USB lead, but since the printer will often be hidden away due to its noise – and print times are measured in hours and sometimes days – that isn’t always practical. I prefer to use a network connection, either cable or Wi-Fi. My original Robox didn’t have this capacity, but I added it via a
Raspberry Pi box connected to the printer via USB.
Ness of Brodgar
Archaeologist Jim Bright ( sketchfab. com/jimbright) and I originally planned to 3D print some of the 5,000-year-old buildings that make up the world-important Ness of Brodgar dig in Orkney ( nessof brodgar.co.uk). This dig has been going on through the summer for eight weeks each year since 2005. My initial tests were done on the data model of a single wall of Structure One. We showed it to the site director, Nick Card, who suggested that it would be good to see a 1m x 1m square model of the largest part of the site. Sure, we said, no problem.
This model would have to be made up of several pieces, as the printer I had at the time had a build volume of only 210 x 150 x 100mm. The data we had access to was generated in 2016 and generously provided by Dr Hugo Anderson-Whymark ( sketchfab.com/hugoandersonwhymark).
Before you attempt to print this type of data, I recommend examining the model without its overlay of colour from the photos. This colouration convinces the eye that there’s more detail than is actually there. Examining it without gives you a better idea of what your finished 3D print might look like.
You can’t just take this data and throw it at your printer, you’ll get a mess – as my first tests demonstrated. One of the issues is that the data will often consist of just a top “skin” of zero or very small thickness. The printer will see this and print nothing, since it’s too thin for it to reproduce. Our first job was to thicken this first layer using the Solidify function in Blender. We then extruded the model down about 100mm using Meshmixer, which was then sliced at a suitable depth to make a flat base to the model.
Any model you attempt to print should have a complete manifold. This means that the mesh of triangles it consists of is complete, with no holes or broken triangles. If it isn’t complete, the results can be unpredictable, and if you start slicing up then things can go a little crazy. Many programs have options to “fix” a broken manifold, but often they’ll make a dreadful attempt and you’ll end up with large areas that are lacking in detail.
I wish I could offer a definitive way of avoiding this, but I can’t. Experimentation with different methods is the only way forward, since each model is different – but, again, Meshmixer is a handy tool. When we were running our tests, Jim and I almost gave up after about three weeks; we started to think that this particular data wouldn’t convert for 3D printing. But our perseverance paid off, with some tests successfully printed.
We were sensibly just testing with a small segment of the model. Now was the time to put what we had learnt and apply it to the full set of data. This data was generated using photogrammetry, combining some 600 photos to produce a 3D model made up of over 4 million triangles. The complexity of the information nearly brought our machines to their knees!
The next task was to slice the full model into the pieces we intended to print. At this point, I thought it would be sensible to check how much plastic material it would use – and my heart sank when I calculated a price of £600. I should point out that all this was unfunded! But all was not lost: if we halved the length of the sides to a still reasonable 500cm, the volume would reduce by a factor of eight, bringing the material cost to a more reasonable £75.
The total print time ended up being 14 days and nights, but in reality it took longer because the print was done in 11 sections with removal and setup times added. Although 3D printers don’t use a lot of power (about 200W), the extended period of printing can make a significant impression on your electricity bill!
The segments were finally glued together in time for the final public open day at the Ness ( see left). This might be the first time such a thing has been attempted on an active archaeology dig.
We now intend to use these techniques to produce more models of such data, but using the larger print volume of the Raise3D printer. The ability to visualise an object on the computer and then produce a physical version is a fascinating process. But it isn’t easy – far from it.