Wednesday, February 3, 2016

The 3DSystems Cube is a horrible machine

I'm talking about this thing.

I don't own one personally, but I've used two at my school fairly extensively and had an opportunity today to see up close how they're designed.

Just from a capabilities standpoint, it's not a very impressive printer. It has a tiny build volume and no heated bed. It's extraordinarily slow but doesn't make up for that lack of speed in quality. The hotend can only handle temperatures high enough from ABS, and takes 10-15 minutes to even heat up enough for PLA. On top of that it retails for $1300, even though there are printers a fourth of the price with stronger capabilities.

It goes deeper than that.

Under normal conditions, the Cube is designed to be "easy to use". The problem is that you're giving up far too much: it only allows its proprietary ABS and PLA, has a far too simplistic slicer with almost no configuration options, and is compatible with literally zero aftermarket hardware or software that any other printer would allow.

Cartridges are prohibitively expensive, at anywhere from $25-$50 per cartridge of it seems 1/4 of a kilogram. By comparison, on my Repraps I'm able to use spools of standard filament coming in at $15 per kilogram. What's more, cartridges are limited to ABS and PLA, and because the slicer only gives you those two options, you're not even able to manually change the temperature to bypass the cartridges and use more exotic filaments. You're at a complete dead end in terms of plastic.

The slicer, as mentioned before, is far too simplistic and gives you close to no printing options. It's limited to a single layer height, which is around .25mm. There are three locked infill settings, which are 0%, a "strong" setting that seems to be around 20-40%, and 100%, as opposed to any other slicer which have sliders for anywhere from 0% to 100%. The only other print options are raft and support, which you have zero control over how they're generated: you can only choose to either enable or disable them. (Other slicers have tons of options for both, such as raft density, thickness, width, etc. and support angle, fill, pattern, etc.) Basically, your part is printed how 3DSystems wants it to be printed, not how you want it to be printed.

Almost every single part of it is fully proprietary. I've examined the whole machine and the only things I could find that are compatible with industry-standard Reprap derivative parts are the NEMA motors. Every other electronic part is proprietary and can't be replaced, and to make matters worse they don't even appear to sell replacement parts. This means if your Cube breaks outside of warranty, it's broken forever, and the only thing left to do is buy a new printer.

It goes even deeper than that.

Storytime: A few weeks ago I attempted to print a medium-sized part for my robotics team on this printer out of PLA. It printed maybe 20% of the part, then something happened and each successive layer got more and more detached from the previous layer. Eventually, it was just printing a blob that hung on the hotend. Happens to the best of us, right?

Well, unlike most "blobs" which are actually just a tangle of plastic strings, this one actually turned into a solid blob, which proceeded to attach itself firmly to the plastic casing of the print head. Sometime over the night, the print head stopped working at all, and when I came back the next day it had a huge chunk of plastic the size of a golf ball caked on the bottom of it. (Sorry, forgot pictures.)

I resorted to using a hot wire foam cutter to slowly cut away chunks of the blob. A few days of this layer, I was able to get enough of the plastic stuck to the print head off that I could heat the PLA and remove the plastic casing.

The PLA had seeped into the print head and now formed a solid chunk around the barrel of the nozzle. What's more, the hotend was no longer heating. I had no way to get the plastic off with the print head still on, so I started disassembling it and trying to free the nozzle.

The disassembly process on this thing is absolutely horrendous.

Here's how the disassembly of a typical RepRap print head goes: Unclip the hotend fan, unscrew two screws to free the hotend, pull the hotend out. Optional: unscrew and pull out the heater and thermistor.

The whole process is not only easy, it's intuitive and even a child could do it with no instruction just by examining the print head.

Here's the disassembly of the Cube's print head: Unscrew two screws holding the casing, unscrew 3 screws holding the fan, pull the fan off, search for 15 minutes finding the tiny spacers you just dropped. Unscrew the 3 screws holding the extruder motor, pause for 2 minutes not knowing what to do because the motor seems to be glued, finally pull the motor off. Turns out, it wasn't glued, but the washers between the motor and the frame almost act like glue. Unplug the connector holding the secondary frame plate from the main frame plate, unscrew two screws holding the secondary frame plate, pull the secondary frame plate off. Unscrew two screws for a plate holding the nozzle, pull the nozzle out, and finally shake your head in despair as you realize the nozzle isn't heated with a heating resistor or a heater cartridge like any sensible hotend is, but is instead wrapped in nichrome wire, making nozzle replacements impossible without replacing all of the heating elements as well. (Of course, that's a moot point, seeing as they don't sell replacement nozzles).

The whole thing is designed to be as unnecessarily complex and unintuitive as possible, in an attempt to discourage the user from doing any sort of repair, modding, or tinkering whatsoever. This would be fine if 3DSystems did it properly, but they didn't.

When I got it down to a somewhat bare configuration that granted me access to the nozzle, I started trying to figure out what was wrong with the heating. My hope was that the chunk of PLA was simply massive enough that it was absorbing the heat, but that wasn't the case. As I got deeper and deeper, closer and closer to the critical heating elements, I found that the wires actually started thinning. It was obvious the electrical system was never designed to last: It's tremendously underbuilt rather than being slightly overbuilt as it should be.

 And sure enough, as can be expected with thin wires, I finally encountered a break in the 26-gauge or so wiring right above the nichrome wrappings. (I later learned I wasn't even the first to encounter such a problem. Some students before me had already disassembled the print head at one point in order to repair the thermistor wiring, which was also a pretty pathetic gauge.)

The tiny stub of a wire left wasn't enough to solder, and even if it were I didn't have wire strippers thin enough to strip the tiny stub of teflon insulation still left. RIP print head.

The entire printer is now worthless because of a single irreparable broken wire. The only options I see with it are replacing the entire print head (which costs more than the printer itself is worth) or hacking it to turn it into a Reprap (not exactly easy).

The printer isn't just a bad printer; it tries to be the wrong printer at the wrong time. It's designed with principles already applied to consumer electronics, i.e. it tries to be proprietary, easy to use, and locked down with no need for any sort of maintenance or modification by the user.

That's not a bad direction to take. After all, that's what made the modern inkjet and laserjet printers ubiquitous. The problem is at the time it was designed, the technology was inherently unreliable and inconsistent, and was nowhere near the ease of use 3DSystems wanted it to have.

Maybe someday in the future 3DSystems will release a true, plug-n-play, "file>print" type 3D printer. I may even recommend it when it's released. As of today, however, the only thing I can say is stay far, far away from the hunk of plastic trying to be a 3D printer that the Cube is.

Some extra thoughts:

1. The extruder is designed poorly. It will often try to force in a new piece of filament alongside an existing piece after a filament switch, obviously causing a jam. In addition, there's a lip on the inside of the nozzle barrel, which prevents the melted plug of filament from the nozzle from coming back out, meaning the "remove cartridge" function doesn't actually work.

2. Magnetic bed is clever for convenience, but prevents it from being heated. Also, bed leveling is an extremely complex process involving small wrenches and tight spaces rather than the thumbscrews on most printers.

I wonder...would it be possible to create a heated bed utilizing magnets for conductors? Maybe 4 metal-coated magnets, 2 for carrying the heater current and 2 for the thermistor wiring.

3. Belt driven Z axis? Really?

4. Cube Glue is <3. I honestly think the best part of the Cube printers are the glue, which is kind of sad. The stuff works exceedingly well at high temperatures, sticks firmly to any plastic I throw at it, and washes away cleanly with water leaving a glassy smooth bottom surface on the prints.

I wrote this mainly to get some animosity towards the printer I was working on out of my system. Worked well. If you've read to the end, I hope this post has been as entertaining to you as it has been for me to write.

No comments:

Post a Comment