Why the Frames editor in TrussLab grew out of large printers, spaceframes, and the limits of "print everything".
A few years ago I was trying to build large delta 3D printers, and the frame kept beating me. Not the electronics, not the motion system, not the firmware. The structure. Once a machine grows past a certain size, the frame stops being background hardware and starts deciding whether the printer is a tool or an experiment that needs constant nursing.
By large I do not mean desktop machines made a little taller for marketing reasons. I needed genuinely useful working volume and enough stiffness to keep print quality under control, without turning every prototype into an expensive machining project. At that scale small alignment errors show up directly in the print, vibration becomes hard to ignore, and the mechanical structure quietly sets the ceiling on everything else.
The answer that finally worked was a spaceframe chassis built from stock tubes and 3D printed joints. The tubes supplied length, straightness, and stiffness, so the printer never had to manufacture bulk material. The printed joints handled the awkward three-dimensional geometry: tube angles, mounting features, clearances, and local interfaces. With a basic tube cutter and a 3D printer, I could build a structure far more precise and serious than the simplicity of the tools suggested.
That changed how I saw additive manufacturing. The printer did not need to print the whole machine. It needed to print the parts where geometry was difficult and relationships were dense: nodes, joints, mounts, and interfaces. Everything else could be bought by the meter.
Print the complexity. Buy the length.
The Problem With Printing Everything
For a long time, 3D printing was sold as the future of manufacturing, and the promise was not completely wrong. The technology improved dramatically. Machines became faster, cheaper, more reliable, and capable of better materials. Desktop printers became useful engineering tools, and industrial machines became more accessible than they had been only a few years earlier.
The disappointment came from stretching that promise into the idea that we should print complete objects as often as possible. That works beautifully at some scales and for some classes of parts. It works much less well when the object is large, structural, and mostly made of volume that does not need to be complex.