Construct Projects are things that I have worked on while employed at The Construct, the makerspace at RIT. I officially work here as a lab manager, but among my many roles, I am also a WebDev, Graphic Designer, Systems Engineer, and even a  Carpenter!  We’re a small space with a small budget, so many of these projects involve developing equipment and infrastructure for low costs. 

I’ve worked on Gigabot  since April 2016

Gigabot

Sounds Impressive, no? Well, it’s definitely an impressive concept. The devil is always in the details though, especially when it comes to 3D printing.

At The Construct, we work  closely with the local chapter of the Enabling The Future . We help them maintain their printers and provide extra assistance when they have a large number of parts to produce. Unbeknownst to us, they had been given a massive printer, a Gigabot GB2 by Re:3D, that had been part of a vision to mass produce prosthetic components. However, due to overhead and staff issues (I don’t have the full story here), the printer was left unused for several years. It sat until last spring, when it suddenly appeared in our lab.

I was overwhelmed at first. Gigabot is pretty imposing, with a 600mm cubed build area. This was more than 6 times larger than the build areas of every other printer in the space,combined! It also had a lot of things wrong with it, and at the time was not printing consistently. (1) The default extruder on the bot was an old, J-Head – based design, which extruded very slowly, making prints on Gigabot take forever.  (2) The build surface was scarred, and required adhesives to get prints to stick to the glass surface. (3) The controller for the printer was located at the back of the machine (whoever decided the cable to the controller only needed to be 12″ long either had incredibly long arms, or had no idea what “ease of use” means), and was tedious to use, especially when diagnosing the (4) chronic print failures that would occur on larger jobs. Prints would shift, either by small amounts or by catastrophically moving an inch to the left. This was incredibly frustrating to clean up, and prevented the printer from being used for anything important.

We needed to get Gigabot into a state where it could be used with the same confidence as the other printers in The Construct. I came up with a list of necessary improvements, as well as some useability improvements that would make Gigabot more accessible to new users of the space:

1. Improve Print Time/Quality. change the extruder to increase max. layer height, and overall performance.
2. Fix build surface. look through different options available and choose one that is cost-effective.
3. Attach a PC to Gigabot. The built-in controls are very hard to work with, and makes it impossible to  rescue failed prints.
4. Solve print inconsistency. Tune firmware, diagnose layer shifting problems that seem to occur sporadically.
5. Improve safety around machine. Shield off open areas, and make sure the ‘bot isn’t able to maim itself or anybody who decides to put their hands inside its capacious build area.
6. Create Documentation. Determine the pricing model, create a way to handle the inevitable long print jobs that Gigabot will have to take on, and make sure it is clear to new users how control the machine.
7. Make GigaBot mobile. Add a robust, mobile base to make the printer work within our confined workspace.

It’s a long list, but over the Summer (and now fall) months I was able to realize all of these, giving The Construct a machine that is now able to produce giant-sized parts consistently.

I started by getting Gigabot on a new set of wheels. The current base was not robust, and made of scrap wood. we needed something that would allow us to showcase the printer, and move it wherever it needed to be. I put my carpentry and CAD skills to work, and came up with a pretty nice set of cabinets to hold up the 100-lb beast.

   

Next was to improve the interface between the user and the printer. as Gigabot is a free-standing printer, a dedicated PC seemed to be the best choice. Repetier Host was chosen as the controller, due to its sophisticated G-Code sender, which allows users to pause the printer mid-operation, perform filament changes/other corrections, and resume without disturbing the print. Repetier also offers integrated slicing, which means that it can control all of the steps in the printing process.  We installed a PC in the new base, and a monitor/keyboard stand for controlling the machine. At this point I also came up with robust Polycarbonate safety shields to cover the sides and top of the machine, in order to prevent accidental human-machine interfaces. The shields also help to enclose the build area, reducing air currents and help prevent prints from curling.

  

A printer as large as Gigabot shouldn’t have a small, 0.4mm diameter nozzle. This makes plus-sized prints take days and days to complete , and the extra resolution of a smaller print nozzle is lost due to the large, imprecise gantries on the machine. Instead, we decided to install the Volcano Hotend from e3D, a large format extruder which produces line widths up to 1.2mm – 3 times wider than before.  This means that, in theory, Gigabot can print up to 3 times faster, making those ridiculously long print times a bit more manageable.

While the volcano works as expected, and did speed up print times, it did bring along some negatives, including messier print quality, and increased part weight. I’ve found that parts need to have 2 perimeters around them in order to produce a consistent surface finish, but due to the larger nozzle diameter this means  that more plastic is used to make the perimeters than was used before.

Throughout the improvement process, I was also working to improve print quality and diagnose the layer shifting problems that still occured occasionally. Here’s what we tried:

• Doubling down on drivers: Added a second stepper motor driver to the Y axis. The skips seemed to usually occur in this direction, and the single low-current stepper driver pushing current through 2 motors seemed a likely culprit. The a4988 stepper motor driver is only rated to 2A of current, or 1A per motor. if any resistance was applied to the gantry – say a snag on the print or excess tension from the filament/cables, the axis would skip steps and produce a catastrophic mess. A 2nd driver was added, and connected in parallel to the first driver (they recieved the same step/direction commands). It was power separately from the first driver by using an extra header on the printer controller PCB. between this addition, and fine tuning current limits on the drivers, motor torque is now noticeably higher.

• Taking a Chill Pill: we printed a fan shroud for the electronics, and replaced the pointless laptop cooling fans with some very capable 50mm fans from an old server. The heatsinks of the motor drivers no longer overheat, which was one source of layer shifting in the machine (overheat = decreased current output = higher chance of skipped steps).
• Tweaking the Settings: I rewrote the printer’s slicer configuration completely to attempt to reduce the chance of the printhead snagging. This included dropping print speed, increasing retraction settings (the amount of filament pulled back into the printer when moving from one area to the next), and adding “z-hop”. This is probably a major reason why the printer works better now, as it now tells the printer to raise the extruder above the part before travelling to a new location. This reduces the changes that the extruder snags itself on a part of the print and slips.

In the end, Gigabot is now a fully- functioning addition to The Construct’s growing collection of additive manufacturing machines. it has been used for a wide variety of cool, large-format prints, and users are just starting to realize  the potentials of such a massive build area. If interested, feel welcome to check out the page about this machine on The Construct’s website, where you can see our policies on using the printer, and (maybe) a live view of what it’s currently printing!

the orange “skullcap” is due to the current spool running out of filament. This happens pretty often now, which is not a bad thing!