So I got a Prusa Research Prusa i3 MK3 kit, and it’s great

After getting into 3D printing back in July of 2016, I’m now on my third 3D printer. I started with a basic Monoprice Maker Select v2 (rebranded Wanhao Duplicator i3 v2) which is a great starter printer. After some life changes I couldn’t keep it around due to the size so I downgraded to a Monoprice Select Mini v2, and after using it on and off for a while I just wasn’t happy with where it was. Then I moved, got the Mini setup again, was working ok so I decided to try to fix some stuff again… and it just got worse.

I could have focused on fixing it but with the move I have the room for a bigger printer again and said fuck it, lets get the Prusa. There must be a reason it’s so universally respected as one of the best printers on the market, right?

Historically, Prusa Research printers are on a pretty steep back-order. There were times when you could order it and it wouldn’t ship for a couple months. They’ve gotten their production ramped up significantly, supply chain is solid, and unless you want the fancy powder coated build plate the lead time was down to 3 days on orders when I ordered mine. Instead of waiting 5-6 weeks for the nicer build plate I figured I’d just pick up one of those separately when they’re available and get the printer itself quicker. Ordered on a Sunday afternoon, got a shipping notice Monday morning, received it Wednesday. That three day lead time was a bold faced lie. They hadn’t announced it quite yet but the standard MK3 kit without powder coated build plate was now fully in stock, no back-order. Bloody fantastic.

Getting the kit means you have to build the printer. The instruction manual is incredibly detailed, although mine was a tiny bit out of date. Going to the support website there’s an even better (and newer) version of the manual that even has comments for each step so you can get further tips from others that have built the kit. I could have recorded the build process or something, but frankly it feels like everyone other person that buys this kit does that so if you’re interested in that, just head to Google.

My experience building the kit took about 12 hours, give or take a bit. I had a fair bit of trouble with certain steps because some of my 3D printed parts were troublesome. Quite frequently I had pieces where you put a nut in place required me pre-placing a screw just to get the nut to pull into the plastic and be snug, then removing the screw so I could actually move on to the step in question. When putting the X-Axis smooth rods in place my parts were extremely snug and it took me about an hour just to get them installed because I was afraid of damaging the rods by putting so much force on the parts just to get them in place. There were a few other places that ended up eating more time than they should have, but easily the most tedious part of the whole thing was installing the extruder. As it’s the business end of the whole thing I was really worried about getting it perfect on my first attempt.

Overall, unless you’ve been 3D printing for a while and are comfortable building the kit, possibly getting a bit frustrated in some places and spending the time on it, I have to recommend spending the extra $250 on the pre-built kit. My build was plenty successful and what not, and it comes down to what your time is worth I think. That 12+ hour build time (I think if you get lucky and get good printed parts it could be more like 10 hours), along with the fear of “did I fuck up?” isn’t quite worth saving $250 to me. On the other hand, the sense of accomplishment once it was done and getting the printer faster was nice. Overall I guess it’s kind of a wash, but worth bringing up.

When you’re all done with the build it’s time to run the pre-flight check, which mostly entails running the setup wizard on the printer itself. Since I wrapped up the build at about 3am on a Friday night/Saturday morning (dedication, folks), I put that off until the next day. This pre-flight check is all about seeing if everything is working as expected (things heat up and move correctly) and that the Axes (is that the right plural of Axis?) are correct. I passed the Axes test with flying colors, pretty proud of that, but there was a SINGLE wiring error. I plugged the filament sensor wire in backwards. Easily fixed, moved onto the rest of the testing. The printer’s mesh bed auto leveling is a key feature in my book so I paid a lot of attention when it got to this step, and the back right corner of the bed just wasn’t quite right. I removed the build plate and went through the mounting screws for the bed again and found that this back corner was a bit looser than the rest so I tightened it up and tried again… and it was pretty perfect when running the mesh bed leveling calibration again.

After all this the printer needs a live adjustment of the Z axis. Basically it prints a single line in long zig zags while you move the Z axis in as fine as 0.002 to 0.003mm increments to dial it in. Most printers end up being set to around .8mm from what I saw online so I went there, was mostly happy with it, and moved on. Printed my first part and it looked pretty damn good, but the first layer went down a bit wonky. It looked over smooshed to me so I went back to the live Z axis adjustment and backed it off.

I started printing a lot of stuff right away. I mostly stuck to typical test prints at first to see how things looked so I could change anything as needed. I kept not being terribly happy with my first layer so I re-ran the Z axis adjustment a few more times until it was better.

After I was content with the test prints I went ahead and started my first 24+ hour print. I really love the look of this Thanos bust by David Östman and wanted to give a shot. I was concerned about the overhangs, specifically the chin, and I’ve never played around with Meshmixer so I figured I’d give it shot. Prusa3D posted a nice guide and video about using it so I followed their steps (Note: Installing Meshmixer via Wine on Linux is pretty simple. Just copy one DLL file and it works pretty flawlessly).

The Meshmixer supports failed pretty spectacularly. I think printing them hollow is a bad move. I also didn’t want to print the mess of tree supports it wanted for the part under the chest of the bust but maybe should have because it came out kind of gross. In the end once it finished and I was able to do some basic post processing, which mostly consisted of getting rid of the spaghetti, and it looks pretty great. The drooping in the unsupported area isn’t great, but the eye catching parts are. I mean look at this mean muggin’ normally-purple-but-in-this-case-silver space Grimmace.

I printed some more little stuff, had a few more issues with the Z axis height so had to change it up a few more time, and decided it was time to dig one of my Raspberry Pi’s out of storage so I could setup some timelapses. Now, another big new feature of the Prusa i3 MK3 is that the new Einsy board can have a Raspberry Pi Zero mounted to it directly just for setting up Octoprint, but a LOT of people recommend against it beacuse the Zero just isn’t powerful enough to do a lot of the fun stuff Octoprint does, namely timelapses. Considering I have an extra one unused right now and my goal here is timelapses, I’m skipping the Zero. This one had been used for RetroPie, but I rarely ever use it. I backed up the SD card image just in case, flashed Octopi onto it, and started setting it up. I printed a mounting bracket for the Pi and ordered a Raspberry Pi Camera v2 and matching cables. I hunted around Thingiverse to check out how people were mounting the camera, and for the most part I wasn’t thrilled with most of the offerings. There are two basic plans most people have, each with pros and cons in my book:

  1. Mount the camera to the bed on one of the corners.
    1. Pros: Traditional timelapses are made stable because you’re attached to the axis that moves. Also, typically a bit more removed from the print bed so you see more of the surface (better for models that are large in the X/Y axis).
    2. Cons: If you’re printing tall objects you could miss a lot of it. Also, the bed can and will make big motions and I’m worried about the camera’s cable ribbon being damaged in that process.
  2. Mount the camera to the X Axis motor.
    1. Pros: Camera moves with the print up the Z axis, so you always see the top of it. Camera will move steadily and slowly up the Z Axis the vast majority of the time, so little concern towards the ribbon.
    2. Cons: Due to positioning you are usually missing a decent bit of the build plate because you’re right up on it. Putting extra mass on an axis you usually don’t want extra mass on.

Since my goal here is fancy time lapses, I’m planning on using Octolapse (a plugin for Octoprint), which eliminates the main pro for mounting the camera to the bed. I also really love WildRose Build’s timelapses that show off the gyroid infill being printed, so I want to see the top of the print all the time.

I found a few options to mount the camera to the X axis motor, printed one while I was waiting on the Pi camera’s ribbon to come in, and used a Logitech C615 USB webcam to see how things might look. With that I realized lighting was lacking. I have an overhead moving arm lamp clamped to the table the printer’s on, but the initial layers are still difficult to see without direct light. I picked up a couple of RGB LED strips at Fry’s and after test fitting them around certain parts of the printer thought, “If I could mount the LED strips to the camera arm, the light will move with the camera and provide light to the camera’s field of view…”

So I set out to design my own. I took the mount I printed, loaded it up into Tinkercad, and ended up with this: https://www.thingiverse.com/thing:3069717. All I really kept from the initial design was the ~1 inch that actually mounts to the motor and the camera case.

Here’s the first time lapse filmed with my first attempt:

I’m surprised my first attempt worked so well and the only change I made for the second revision was making adjustments to bring the camera up higher. It’s working much better but I haven’t started another long print to show off what Rev2 is doing just yet. I also had things in Octoprint/Octolapse and with the PiCam setup to defaults so I’ve adjusted those. I still need to dial in Octolapse’s settings but the smaller stuff I’ve printed so far is looking good.

I’ve been so impressed with the Prusa i3 MK3 so far that I decided to also start using up the Maker Box samples I’ve been avoiding using with my other printers. A couple were completely failures but a couple have been nice so far. The failures are possibly due to their age (some are from December 2016 and not packaged for long term storage), but with enough time/patience I could dial things in and get them working… but there also isn’t enough material in these samples to really do that with and have enough to print something worthwhile left over. So I just tossed them and I’m going through the rest.

As is hopefully clear, I’m really happy with this printer. The 3D printing community is right, it’s a phenom. It’s worth every penny you spend, and outclasses a number of much more expensive printers in feature set. By sourcing parts from good vendors, putting them through strict quality control, Prusa Research is ensuring each printer should start out great. The kit is time consuming to build but in no way exceedingly difficult mostly thanks to the incredibly detailed manual, and if you don’t want to faff about with all that you can spend just a bit more for a pre-assembled model. They also regularly release updates for their firmware and slicer, all of which is open source, and… at the end of the day this is exactly where 3D printing should be in 2018. Bravo Josef Prusa, bravo.

Posted by JP Powers