Review: My New Prusa I3 MK2S and Home 3d Printing Redux

I chose the path of the kit form of the I3 MK2S. It saved a couple hundred dollars which I suggest you spend if you are in the market... unless you specifically want an object lesson in building and wiring of a pretty precise piece of machinery. Thankfully that is exactly what I was looking for... and I got it :-)

Now I am not going to do an in depth tech review of this printer. The internet is awash in plenty of reviews and even live stream complete builds as I linked above.  What I am going to talk about is what I have learned over the past couple of years of playing around with my 3d Printer, a bit of what it was like for me to build this kit, and why based on that experience I decided to go with this particular one.

Background:

I started 3d printing around Thanksgiving 2015. I purchased a Printrbot Simple Metal based on what I read in reviews by Make Magazine in their outstanding yearly review of printers. It wasn't the best and it wasn't the cheapest. But it was solidly built, produced decent prints comparable with more expensive machines (if you knew what you were doing) and was based on an open design with a good community behind it (also known as help from people who know what they are doing). In short it was their recommended bang for the buck budget printer. Perhaps most notable in the cost was an automatic bed detection Z probe which by all accounts reduced one of the most hair pulling and teeth gnashing pain elements of 3d printing.... calibrating the Z axis for that all important first layer. Thankfully these days that is a fairly standard feature... it wasn't so much when I got the simple. Left to my own devices I probably would have ordered the simple as a kit... but Printrbot was so focused on trying to get folks a good experience they actually charged the same price for a kit or for a fully assembled bot. Their founder also had a refreshingly direct take on 3d printing. "IT IS HARD. EXPECT TO LEARN LEARN LEARN. YOU SHOULD DO IT ".

He did say it was hard.....

According to Octoprint I have run my Printrbot somewhere over a 1000 hours over a little less than two years and I can certainly attest to all three emphasis Brook Drum put on his machine and the whole process of 3d printing. I am glad that I went for an assembled/tested system to start with as if I had been fighting the hardware AND the difficulties of getting successful prints I am not sure I would have stuck it out.

Baby Groot with a pot for a head

Lots of melted plastic there....

What I think I learned most about 3d printing was that people who are into it, or companies trying to sell you one, undersell the toolchain... or process of going from a design through to a print. The other thing I learned was that the accuracy of the machine I had was great for items without tight tolerances unless I wanted to spend inordinate amounts of time calibrating and tweaking designs to its particular issues. As most files of complex assembly items were just not within my reach I settled in to making single pieces. Probably the most ambitious thing I printed of my own design was a full sized poster frame, and a small model of our new house. The poster frame had gross enough tolerances it worked... but it took a surprising amount of fiddling to just make a bunch of pieces create an integrated large rectangle of a specified size. Both models were ones I created myself and though not particularly challenging prints, I found their creation deeply satisfying. Neither of these are the print I am most proud of... but I will talk of that one in a bit.

Toolchain: Big picture of HOW you make something

In order to take something you imagine and print it out on a 3d printer it takes three major steps of which the last is the machine itself doing the printing. You have to be able to create a 3 dimensional model (CAD design, Model design) and you have to use a slicing process that turns the 3d dimensional object in a series of 2d 'slices' that define mathematically how the head moves, how much plastic to melt, how fast to cool it etc...

One of the powerful things about 3d printing is being able to take your own idea for your own specific need and design a solution. There are two main ways to design a model you can print, one is CAD (Computer Aided Design) which is aimed more at gnarly engineering/mechanical type design work and the other is much more free form Model making software like Blendr or Maya which you could use to make a 3d animated movie, or CGI special effects etc...

As for which you should use? really depends on what you are comfortable with and what you are trying to do. CAD excels if you are trying to make something mechanical and it is tied a bit more tightly into 3d printing at the moment. However, model making software is very attractive if you are after more artistic results. But really, you can design any 3d shape from either platform. The biggest fundamental difference is that CAD is grounded in real world measurements. Model software is typically based around graphics coordinate systems as the models are meant for computer generated graphics rather than physical production. It is a subtle difference, but it is kind of the tip of the iceberg. While both sets of software exist to ostensibly solve the same problem of creating 3d models understood by a computer they accomplish it in VERY different ways.

That said, design is the easiest of these steps to shortcut. The internet is awash in a vast array of designs intended to be 3d printed. Thingiverse, YouMagine, MyMiniFactory and dozens of other sites have more models than you could ever browse through much less try and print. And if you want to focus on the puzzle that is slicing up files and getting them to print there are challenges a plenty to focus on. So if design is something that is stopping you, don't let it.

Once you have your model you have to slice it. Quite literally. Imagine slicing your model into wafer thin cross sections no thicker than a sheet of paper. That is effectively what is happening when the object is printed. You generate machine tool path information that provides the outline and interior movement information for a layer of a given cross section. Of these perhaps the most important is layer number 1. That is the one where the head must get enough plastic on the bed to make it stick, but not enough it drags or squishes the plastic to much. This is tricky, as you are talking about needing to have a bed flat enough and a machine accurate enough to keep the end of the nozzle the width of your 'slice' from the print surface. The tolerance of this is measured in microns, or 1/1000 of a millimeter (mm) if things are going well... 1/100 of a mm for things going ok and things go bad if variations start approaching 1/10 of a mm.

Slicing software abounds from both proprietary solutions for specific machines to open source like Slic3r and Cura or paid software like Simplify3d (S3D). Similarly they range in complexity for trying to make it as easy as possible to opening you up to all the nitty gritty decisions that comes with trying to successfully fuse plastic together in the shape you want. Some CAD solutions will also handle the tool path information for their models for 3d printing but it is not something commonly supported at this time... signs are pointing to it becoming much more common. For example the I3 MK2S I got can be used by Win 10 as a 3d printer which enables some compatibility with their creativity suite and with some big name CAD software. It isn't something I am likely to use but it may appeal to many in terms of simplifying the toolchain that gets you from idea to object in your hand.

The devilish part of printing is that design and slicing are both deep enough topics you could specialize in either.... except for the annoying fact that if you don't understand slicing you can't design things that print well and if you don't understand design (mechanical) you can make some horrible slicing choices. There are several intersecting lines of decision with regards to print orientation and overhangs that drives design and slicing choices into a series of compromises. Being good at both helps you have to make fewer compromises. If you skip design... you will soon appreciate greatly designers that understand the limitations of printers. Thingiverse in particular abounds with fantastical designs that will drive you batty trying to figure out a way to actually get printed.

The final element is how to drive the machine itself. Most machines these days can run themselves. It generally requires loading a slice file into a specific directory on an SD card (or similar) and then navigating a menu on the machines screen (assuming it even has one) to kick off the print job. It then gets the G-code (or other machine tool path code) from the file on the card. Another option is to run it directly from your computer. Slic3r most commonly pairs with an open face controller called Pronterface or Repetier, Cura has its own controller as does S3D. But to use your computer as a controller means it has to be on for the duration of the print. And one thing you soon learn in 3d printing is that anything big/cool/fun generally takes a LONG time to print. If your systems is your laptop for instance, you have to leave it connected to the printer. And you generally can't re-start it without loosing the print job.  My personal weapon of choice in this endeavor is Octoprint. It allows you to run the printer with a Raspberry pi, and stream the print job via either a usb webcam or a raspberry pi camera. The Debian Linux branch that Raspian is based on is very stable and I routinely have my pi up for months at a time... generally only restarting it when updating the software.

I wound my way through most of the software listed above and a few others besides. My current choices to get from idea to print are as follows.

• Autodesk Fusion 360 CAD for design (Free For Home Use). This is Autodesks vision of the future of CAD software and it is a powerful full featured CAD program. Youtube abounds with tutorials from other printer enthusiasts. The source I latched onto the most are Noel and Pedro Ruiz who headline most of Adafruit's 3d printing tutorial content and projects. 3d Printing aside I can't speak highly enough of Adafruit in general... if you want to tinker with electronics they are probably the best one stop shop for hardware and learning content anywhere. 
• Simplify 3d (S3D) for slicing the model into G-Code. Other than the printer itself this was perhaps the best $150 I spent. It is as simple or as complex as you want it to be and it seems to be well on its way to being the home printing standard for slicing. 
• OctoPrint Raspberry Pi driving the printer. Octoprint is free, a Pi, case (if not printed), camera and camera mount (if not printed) will run about $100. If you print the case and stand you can do it for about $70. I love this software so much I pitched in to sponsor the creator via Patreon when she lost her support to develop the software through other means. 

So what is it good for? Why did you get it?:

Why? There are a lot of reasons for me personally. But I can trace my specific interest in additive manufacturing technology to the opening sequence of a movie released in1998 called Small Soldiers. If you do not recall it do not worry. It is largely senseless incoherent gibberish about toys accidentally getting military AI chips and acting out their pretend roles for real... basically imagine Toy Story where Buzz Lightyear actually had a laser and could really fly. What I recall the most was the opening sequence where they showed the action figures emerging from a vat of goo. I thought it was just a special effect. I came to understand later it was just a sped up shot of a real technology called Stereo Lithography. I supposed you could say at the moment I realized that programming a computer to build something was possible I wanted in. It took a while but once FDM style plastic melting systems got to the point you could make some useful stuff with it I was in.

Why not subtractive manufacturing (machining|milling, carving, routing, lathes etc..) you might wonder? Well, while those technologies are on the whole inherently more practical and useful they are much more of the past. And on a practical note.... MUCH more expensive. You can get machines in roughly the same cost range but the materials cost vs rolls of plastic is a lot higher... and it is a well trodden area. Additive technology is relatively young and as such the pace of improvement in it is far greater than in the more traditional subtractive technologies. Or in short... I find additive tech overall more interesting right now. 

So what is it good for? Well.... I could be corny and say anything you can imagine. However, that is definitely overselling the tech. It's limitations are legion. Yet plastic is a fundamental part of our world and the ability to form objects with it more or less at will is a pretty powerful thing once you come to terms with how it works. 

This picture of a couple of odd looking funnels are I think my best example from my experience of the power of 3d printing. You can read the whole (somewhat tongue in cheek) saga of them here. But the short version is I had a leak coming from my HVAC unit and if I couldn't find a way to keep it from getting to the floor I was going to have to turn off the AC in the middle of the dog days of summer in Alabama... which likely would have put us in a Hotel. I found the leak after everything was closed. I jury rigged a towel into a bucket to stop the worst of it and went to the computer and designed a flat sided funnel (smaller one on the left). Then after getting it in place and noticing some problems I went back and designed the second one on the right with the indented brim. They took about an hour - 2 hours each to print and before midnight I had the leak directed completly into a bucket. The next day I completed the rig to drain into a 5 gallon bucket outside of the HVAC closet. Service guys were impressed, the insurance inspector was impressed as was the contractor. It was ugly, simple, and effective for what I needed. 

If you search around you can find examples of car parts being made. Toys being fixed and lots of other small simple things, and not so simple that plastic printing has been used to solve, fix, invent or make better. That said... mostly they get used to print Yoda Heads. Perhaps one of the more amazing mixes of the fun and the practical I have seen is for making RC planes. If you look closely at my shelf of printed parts you can see a vaguely plane shaped object and some wings representing an attempt at plane design that showed me how much I had to learn in CAD. I also have made straw construction connectors for my son, and printed out files of scanned dinosaur fossils. And a body for a small micro-controller robot rover. 

I later printed a smaller version of this T-Rex skull completely, but this larger failed print made a much nicer picture. did I mention printing is hard to get right?  

Collision of geek hobbies

Home made robot action figure... just print, add shock chord and watch the smiles roll in.

In the end I would say what it is good for is what you make of it. For some it is for trinkets. For others it is a lively hood. For instance Josef Prusa and his I3 printer design out of the rep rap family of home made FDM systems. All the orange parts you see on pictures of my printer, or the myriad of stock shots online of the Prusa I3 MK2 or I3 MK2S are made by.... Prusa I3 MKx machines.

So Why the Prusa I3 MK2S, and is it any good? 

I stayed away from the hard core rep rap home brew looking builds with my first system and went for the metal build of a very similar design for the simple reason of longevity. Laser cut wooden models had a lot of "street cred" and also a long list of fundamental design flaws in reliability directly related to the wood and high fraction of printed parts. As a philosophy statement of a community in pursuit of the elusive Von Neuman device they were amazing. As someone who wanted to focus on other areas than the machine itself they were a nightmare. Why then did I go with the I3 Mk2S? It has probably the highest content of printed parts of a mainstream home 3d printer. I could pay quite a bit less and get injection molded copies of the exact same design. Or I could pay a little bit more and probably get an enclosed device like the Flash Forge. 

1. Going back to that same source of the Make Magazine home system shootout I decided I could not ignore that they awarded this thing their best printer award. Not best for the money. Best full stop. The results of its prints stood up next to printers costing 3-4 times as much and gave very little if any ground. It didn't look good. It doesn't have all the fancy bells and whistles and It has some niggling upkeep issues. But when it got down to it, it delivered. And LOTs of people echoed the same observations across the home 3d printing community. 
2. Despite the Rep Rap aesthetic there are some serious design advantages over those in the past where they count the most. Firstly the use of metal as the foundation framing strength vs laser cut wood/melamine. Getting all the nuts in just the right spot is a pain in the keister to be sure. But it is sturdy. As sturdy as extrusion or a purpose built case? No. But when you get down to it FDM systems do not exactly crank out stresses like say a milling machine. It doesn't have to be a tank to get the job done. A full metal hot end from E3d. This head seems to be on everything printing the best stuff outside of an Ultimaker and supports pretty much all filament types. And finally, the Prusa PEI mesh leveling PCB bed. The mesh leveling accounts for most of the framing slop. IE as long as you get it close enough to true... the firmware combined with the mesh leveling capability handles the rest. Which means instead of having to be micrometer correct on frame alignment, you literally just have to be eye-crometer level aligned for the software to handle it.



Upon this foundation.... plastic, threaded rods, some 10 and 8 mm washers and nuts. To quote a certain scoundrel " She may not look like much kid... but she's got it where it counts"






Here you can see the central metal frame that joins the X, Y and Z axis. It literally is held in place by 6 nuts. Two jammed up on either side, and one with a second jammed up as a lock. It is a cast iron PITA to get them in just the right place the align the X and Y axis. I really expect a future version to migrate this to extrusion... it is getting cheaper, and the ease of alignment and shorter build time has got to make it worth while. Also... think you have to go there for a much larger build volume.

1. 
   
3. The promise of single extruder multi material. This seems to be the coming rage. Assuming it works (and it is shipping and folks are starting to show of successful projects) you can add 4 filament printing for about 3 hundred bucks. Dual head printing is out there, but the entry price for a reliable system is a lot higher than 1200 assembled, 1000 in kit form. And 4 head systems don't exist. This is kind of the first example as far as I can see of the home market begining to forge its own future out away from the commercial FDM legacy designs they have largely been following to this point.
4. A proven track record of constant design improvements and upgrade paths/kits provided for older designs. Prusa rapidly iterates their designs but do a good job of allowing you to follow along without having to buy the whole new system. This was and remains a similar benefit to the printrbot family as well, though it is not as well paved. 
5. Massive community. In order to have a decent community there has to be a user base. The I3 Mk2 and MK2S have been extremely successful, combined with open source design and manufacturer support of the community to mod/improve/troubleshoot the printer it has created one of the best eco systems out there for entering 3d printing.
6. Finally... price. While it wasn't the deciding factor on which system to get. It was the deciding factor on getting a new system now. 700 for something that prints this good vs the problems I have had with the Printrbot was hard to turn down. 

Is it as good as they say?:

Assembled and ready for action. All tools used to build on the left, all tools used to remove print jobs on the right (hint... just the hammer to tap it off so far). 

Tree Frog and Adalinda pre loaded files. 

Liking this silver PLA the shipped with it. Printed at .2mm layers. 

Tree Frog at .05mm.... 50 microns. Zero cleanup. Very few blemishes... couple strings between the legs on one side... like 1 or two. 

File from scanned Triceratops fossil. LOTS of support... oh no

Supports peeled off.... what??? Ok surely that back side is a horror story of support studs...

Nope. If I just saw this print after the supports had been removed I would have been absolutely stumped as to how the hell it printed given the layer lines showing the orientation. There are almost no marks indicating support structure attachment points. .15mm layer hight. 

The first prints. Printer passed calibration on the first try after I got the LCD cables in the right place. Had one other small snafu I had to fix and about 10 calibration prints getting the Z offset dialed in for first layer hight. Live Z hight manipulation is the bomb!

I think the prints speak for themselves. I have seen them online in many other places but always wondered if that was really the average user experience. I currently have a plate of 16 parts going slated to take ~ 1 day to print. Every time I tried this on the printer bot I got horrible z wobble and normally a ton of stringing between the parts requiring lots of clean up. So far the prints all look as clean as those above. Will see if it holds out. Being able to reliably print fuller build plates is something I am very much looking forward to. 

The build surface itself is something that so far I highly recommend but I just do not have a lot of time with it yet. Getting to build-tak saved the printr bot as I was ready to throw it against a wall dealing with blue tape and glue sticks. But it had the problem of working too well and I couldn't find find the sweet spot between adhesion and cemented to the bed and got to where I dreaded getting prints off the surface. I will be sorting out a PEI surface for the Printrbot, that much I do know. By the time I wear this one out I do really hope you can just buy a new one from Prusa. Right now it seems to the secret sauce defending them from the hoards of clones. I imagine once they catch up with demand they will be available for purchase independently of a kit... and many frankenbot projects (and clones) will benefit from it. This is a slick design if it holds up well.... and so far the general take on it seems to be that it lasts a good while if you treat it right. 

Next up I aim to be tackling some tougher assembly type prints. A full up recorder (the wind instrument), the new squishy turtle design from thingiverse and... Voltron (yes its a toy.... but damn... just look at it, you know you wanted one when you were a kid :-). After that I plan to tackle one of the prosthetic hand projects and turn back to my design ambitions once I have a feel for what level of accuracy I can expect. 

Anyway if you can't tell I am really liking it so far. Unlike the printr bot I think I could recommend the assembled version of this printer to any one interested. Just make sure you read through the above about the various steps involved and make sure you set your expectations accordingly. 2 years on from when I started it is easier than ever to get from printer to part... but the fundamental challenges of design remain the same. It is just that the software is getting easier and material to learn it more readily available. If you are interested and motivated to learn what are you waiting for?