3D printer V1.1 - Bootstrapping
In a previous article I described how I built an FDM 3D printer using mostly wood and hand-tools. At the end of that process I had a working 3D printer, but just barely. The first things I made with the printer were parts to improve the printer itself, essentially 'pulling myself up by my bootstraps' to a better and more functional printer. This article describes that bootstrapping process.
By the way, click on any photo to see a larger version!
Endstop switch holders
The first parts I printed with my 3D printer were endstop holders. I didn't have any good mount for the z-axis endstop switch, and the x-axis endstop assembly was a combination of a big, ungainly bracket and a microswitch with an absurdly long lever arm that was not steady at all and not really in the right place to start with. I don't recall exactly what holder I used, it was either this design from Thingiverse or something very much like it. The first one I printed didn't quite fit the smooth rod of the printer as well as I'd like, so I adjusted the dimensions in the OpenSCAD file associated with that thing and made a couple more.
The photos at left show the first holder being printed, and the X and Z axis switches mounted on their printed holders. (The Y-axis switch was installed in just the right spot on its bracket underneath the build platform -- purely by luck -- so I left it as-is.) The holder shown still on the build platform was my very first 3D print that wasn't the test cube. I was printing on bare glass at that point, and as you can see at the back of the part, I was having adhesion problems. I think I had coated the platform with sugar water -- I remember doing that for a time before switching to kapton tape (and I don't see any tape in that photo), but I don't remember if I first tried sugar water before making these parts. You can also see that I printed the holder on top of a raft. That was done partly in an attempt to improve adhesion, and partly because I was exploring the various print options in Repetier / Slic3r.
One additional reason for adhesion problems was that I was still using the PCB heatbed from Sainsmart, which had too high a resistance and therefore couldn't get up to the necessary operating temperature of ~100 °C (that's black ABS I was printing with).
This simple endstop switch holder design isn't really ideal for the Z axis. The ones I printed proved to be a little wobbly. That's not a big deal for the X axis, but the Z axis needs to be very consistent so that the gap between the hotend nozzle and build platform is correct for each print. Getting that gap correct in the first place can be a struggle, and an adjustable Z axis endstop holder would make that much easier. I didn't print an adjustable holder because I didn't appreciate how useful it would be at the time, and because I was struggling to get anything printed at all!
Build platform mount and Y axis
Next I tackled the Y axis. My wooden bearing blocks holding the build platform were rather crooked, which warped the wooden platform and put extra force on the Y axis bushings, making its movement rough. So, I printed better bearing blocks for the Y axis.
The endstop holders were designed in OpenSCAD; that's what the original uploader of that thing used, so I stuck with it when making my dimensional adjustments. OpenSCAD reminds me of POVRay, which I played with when I was a teenager ... a long time ago. Since then I have done a lot of design work with traditional 3D modeling CAD systems (Unigraphics, etc.) and that now feels more natural to me; working with OpenSCAD felt a little clunky. There aren't a ton of free 3D modeling CAD systems out there. At the time I was doing this the only option I could find was FreeCAD, so that's what I used to design this part. I have to say, FreeCAD drove me up the wall. Making changes a few steps back in a part's history usually messed up the associativity of all subsequent features, and the program crashed on me pretty often. I was using version 0.14, and I see that a new version (0.15) is now out. It might work better, but I haven't given it a try.
The photos at right show a closeup of the bushing blocks and how they were installed in the printer. (Look at how flat and level that build platform is!) There are recesses in the base of each block sized to capture #2 nuts. The recesses (as well as the screw holes) are elongated across the part to allow some play. Each block has four mounting holes, though I ended up using just two at opposite corners for each block. I had had significant trouble with my wooden bushing blocks being a bit floppy, which torqued the bushings and caused them to jam, so I tried hard to make these parts rigid. It seemed to work; I never had a problem with them. I wasn't sure that the holes in the blocks for the bushings were perfectly straight, so I glued the bushings in place and then put a smooth through each block while the glue dried to ensure that the bushings in each block stayed perfectly in line with each other.
Even after replacing the bushing blocks the movement of the Y axis was still a bit shaky. One problem was the mount for the Y-axis motor: It was held in place with a bracket that attached to the wooden base plate of the printer. That baseplate was not all that rigid, so the motor and its mount could flex up and down when the motor turned, rather than moving the build platform. So, I reworked the rest of the Y axis. I printed a new motor mount that was fastened to both the base plate and the board that held Y-axis smooth rods, which made a solid mount for the motor and stiffened the printer overall. This new mount also lowered the motor by a few millimeters, so I made a new rear pulley mount as well to match. This new mount was mounted on the front side of the rear Y rod holder, rather than sticking out the back of the printer like the first mount did. The change left less room for connecting the belt to the build platform, so instead of attaching one end of the belt to a (rather large) spring, I spring-mounted the rear pulley.
I can't seem to find any photos of the Y motor mount, so at left is a render of it. The motor was attached to the left side of the mount; you can see cutouts in the two vertical walls for the pulley and for the belt to go backwards through and past the mount. The spring mounting of the rear pulley isn't visible; on the back side of the board the two mounting screws each have a helical spring captured between their head and the board. If the mount is pulled forwards it can move slightly by compressing the springs.
These new parts for the Y axis were not entirely successful. The Y belt would often skip unless I kept the Y axis acceleration low, typically below 2 m/s2. This printer never printed very quickly, and this low acceleration was one of the chief causes of its slowness.
X axis assembly
Finally, I reworked the X axis assembly. You might notice a progression in terms of part size and complexity: first I only attempted to print small things, like the endstop switch holders. The build platform bushing blocks and Y axis motor mount were a bit larger and took longer to print. The parts for the new X axis assembly were larger still, and I tackled them only after I had some confidence that the parts would adhere to the build platform reasonably well and the printer would run long enough to print the whole part.
The photos on the left show the assembly; the first one is a front view and the second is the rear view with belt installed; in both cases the stepper motor indicates the left side of the assembly. I secured the belt to the X axis carriage by printing a slot with ridges matching the tooth profile of the belt, which is barely visible in the rear view photo. I made the slots taller than the width of the belt so I could put holes above them; putting a screw and nut through the holes would prevent the belts from working their way up and out of the slots and would also hold the top of the slot together so the two halves couldn't get pried apart when the belt was put in tension. That proved unnecessary and I can't remember if I ever even installed the screws. I gave the X carriage a single 6mm hole in its front face for securing the extruder. To keep tension in the belt I made the idler pulley on the right side spring-loaded, and that worked pretty well (unlike on the Y axis).
For all the parts of this assembly I reused the same bushings that were in the wooden blocks of the original printer,which means I had to make the holes for the bushings a few different sizes and they were generally not interchangeable.
I made little nutplates to clamp the 1/4" nuts that rode on the threaded rods to move this assembly vertically. Every time I used the printer, up until I had this new assembly installed, I worried that the hot glue holding those nuts in the original parts would break loose. It never happened, but hot glue is generally not a good long-term solution for assembling things in my experience!
I wimped out and left the arrangement of the vertical smooth and threaded rods as-is; I kind of wish I had reworked them and their mounts as well. Recall that the smooth rods guide the assembly, but it is the threaded rods that carry its weight. My arrangement of them was with the smooth rods on the outside and threaded rods on the inside on both sides of the printer. The (rather heavy) stepper motor on the left had to be placed far enough out to avoid hitting the smooth rod, which meant it was cantilevered out a fair distance from the threaded rod on that side. When the X carriage with its extruder was in its home position, all the way to the left, that put pretty much all of the weight of the assembly on the left threaded rod. The rod had no problem carrying the weight, but leaving the right threaded rod with no weight on it meant it tended to ride upwards whenever the assembly was moved downwards. I had to remember to keep a finger on top of the right threaded rod whenever starting a print to keep it in place while homing the Z axis. If the rod were allowed to ride upwards too far it would jam and cause other problems. I could have rotated the blocks holding the vertical rods 180°, leaving the threaded rods in place but putting the smooth rods inside them. There would still have been room for the X carriage to have its full range of motion, but I could have brought the X stepper motor inwards and simplified the mounting of the X pulley on the right side. I didn't do it because I was afraid of changing too many things at the same time. Oh, well!
Closing
With the new X axis assembly, I was finished 'bootstrapping' this printer -- I had replaced all the clunky and/or unreliable hand-made wooden parts in the original printer, and the mechanical components of the printer worked fairly reliably. Keeping the build platform leveled was an ongoing challenge for the entire life of this printer, which I think was a consequence of building it out of wood and having temperature and humidity change with the seasons. Getting good adhesion to the build surface and avoiding curl was also a challenge. The whole time I used this printer I don't think I even cleaned the kapton tape with acetone, which was a mistake. The first time I did that with my current printer, the difference was like night and day! (OTOH, I print PLA with my current printer, whereas this first one was always set up for ABS, so perhaps the difference there would not have been so great.)
The biggest problem with this printer was one of speed. Even though I could print at reasonable speeds (~40-50 mm/s), printing anything took an extremely long time. The root of the problem was low acceleration, which I was not able to increase due to skipping of the Y axis belt. The impact of low acceleration was most obvious when printing something with fairly thin walls, say 4 or 5 mm thick, which involved many short stretches of infill -- the pause at the end of each run of infill as the print head slowed and then moved to the next infill strand stood out very much. I was also concerned with the effect of movement of the build platform on any tall pieces I printed. Anything tall could start to sway forward and backward with movement of the Y axis, making the upper portions of the print sloppy and perhaps breaking it loose from the platform altogether. With the low acceleration I was limited to this proved to not be a problem.
Because of this limitation of the printer, after I finished bootstrapping it I did not spend much more time trying to improve its design. I knew that I wanted to make an entirely new printer, and I would make it focused on speed. More on that in a future article!