Cubex Trio Revival Part 1

I saved a Cubify Cubex trio from the trash. This tri-extruder printer, and it's cheaper dual extruder but otherwise identical cousin "duo", were made by Cubify (a 3Dsystems brand) in 2013. The retail was ~$4000, which is squarely in the premium category of printers. It has a X-Y gantry with a Z bed that lowers as it prints.



It came with a custom controller, firmware, and software, three proprietary filament cartridges (1.75mm PLA), proprietary extruders and hotends, etc. They sponsored a lot of ads and fuss over this pre-launch. For $4000, people were probably expecting ultimaker level quality. But it turned out to have several major design flaws that made them basically useless as 3D printers. The motherboard would short, cartridges were unreliable and very expensive, jams common, poor bed frame design causing leveling problems, poor quality control, acrylic broke easily, max print speed 25mm/s, etc. I'll show some pics of other questionable design decisions below.

Here are some pics of the printer as I received it.

Missing controller, cartridges, cooling fans (which just blew down onto the build surface), build plate, and a few other things. First major design flaw I noticed is the fact that filament has to be pulled through about 5 feet of tubing that makes four 90 deg bends...even if that's teflon, that's still a lot of friction. I guess they were trying to keep all the filament contained inside the ~24" cube, but the filament routing is just dumb. The cable routing was actually pretty well done. Another pic of the routing:

I'm not sure what the weird acrylic box thing hanging off the two posts was for...maybe filament dribble? I threw it out. Once I figure out how to take the back wall plate off, I'll take those posts off, too.

Here's a pic of the x axis stepper on one side of the y axis gantry.

The little black nub sticking out slightly over the left x axis rail is a reed switch. They used these + little magnetic cylinders for limit switches on all of the axes. They're certainly more expensive than mechanical switches, and are inherently less repeatable than mechanical switches. Not really sure why they decided to use those.

Closer view of reed switch

Z axis reed switch removed

Magnet cylinder next to 12mm linear shaft

Here's a look a closer look at a corner of the base frame.

I guess they wanted to T off of one of the 12mm steel shafts, so they milled a hole in it...yeah, expensive and stupid. Here's a closer look at the bed + arms.

Let's attach a stepper to two linear shafts by milling flats in the
 shafts and drilling holes through them ...freakin' brilliant. *facepalm.
Not sure what the giant standoffs were for.

That lexan piece behind the Z stepper is milled. Looks extraordinarily expensive and not very rigid. 

The bed's aluminum A arm is supported by two linear rails + bearings at the back of the printer, and moves up and down with a single lead screw driven by a big NEMA23 stepper. This means the bed is basically cantilevered, which isn't ideal, though not necessarily a bad thing if it's stiff enough. Many corexy/hypercube printers have cantilevered beds. There's a lot of slop in the linear bearings, though. I can tilt the end of it about +/-1/8" up and down, which is not ok. The bed has a bunch of cutouts for clearing filament cartridges, along with other things. Again, expensive...

I started taking the extruder assembly apart next.

One of the motors. Has a custom bevel gear on it.
Given the pattern on the front, it looks like these
had optical encoders....not sure why. 

Custom hot ends. I think one set of wires is a
thermistor, and the other a heater band. The
ceramic covering looks like it was paste or something. 

Hard to see, but there is a small stainless steel
heat break pressed into the aluminum.

View of the nozzle. 

One of the extruders. I'm guessing these had fans on the fronts.
Two part aluminum pined together. Bearing supported roller
that pressed filament into the extruder gear.

Clamp for the hot end.

X carriage plate. Laser or waterjet steel.

Fold down tabs for fan mounts (I think)

You can probably guess what I'm about to say...expensive. Jeez. I don't have close up shots of them, but even the X axis belt tension-ers were little aluminum cylinders that probably had about 15 lathe/milling operations. There are definitely simpler/better ways to design most of this stuff.

This X bearing was shot...it had ~2mm of lateral play.
Oh, did I mention all of the bearing blocks were custom machined?

You know what? I get why this thing cost $4000. There is a ton of custom fabrication work that had to go into it. But it really didn't have to be that way...there were clearly many stupid design decisions that caused it to be unreliable and way more expensive that it should have been.

There have been other open source conversions of cubex duos/trios. Reprap wiki has a decent guide. My basic plan at the moment:

• Goals: maximize build volume, print quality and reliability at least as good as my wanhao i3 v2.1 printer
• Replace X axis bearing blocks with more standard ones with new bearings
• Design and make a new single extruder X axis carriage plate. These two things should add about 2 inches of X travel. 
• The extruder design seems decent, so I might use one of them. But I might also replace it. 
• Add e3d v6 hot end
• Replace bed plate with a square one (still aluminum). May not need this depending on how I mount the build surface. 
• Add a large AC heated bed + glass build surface. 
• Replace the flimsy lexan Z axis plate thing
• Remove the filament routing tubes
• Replace the reed switches with mechanical switches. 
• Add a standard control board. Put the screen/control interface where the old one was. Will need a relay for the AC bed. May need something fancy to drive the big Z axis stepper. 
• New x axis belt tensioner
• Design and laser cut top cover and side covers. Maybe use magnets to hold them on. 

Possible other changes:

• Replace the NEMA23 with two NEMA 17's, one on each side of the build plate. This would probably help with stability. Common in larger corexy printers. Another common thing to do is use belts instead of two motors. Not really sure why they used a NEMA23 anyways...seems way overkill. 
• Maybe add a third Z linear shaft. Depends if replacing bearings helps eliminate the slop
• Might change the X rails to one on top of the other instead of side by side. This is more versatile for extruders/hot ends and might save about an inch in y travel. Will likely do this if the Y axis bearings are also shot. 
• Design/make a tool changer to change out heads. 

No idea of time frame on this. Might be awhile...