/ engineering

Lessons in Engineering - Assembling the Prusa i3 Mk2s and Multi Material upgrade (Part 2)

Introduction: This was originally going to be a blog about the multi material upgrade build for my Prusa i3Mk2s, but it got a bit lengthy, so I split it into 2 parts. This second part happened a few months after the successful assembly of the Mk2s, when I received the multi material upgrade.

Part one is here

Part the Two.

I'd been anticipating the multi material upgrade (MMU) since it was announced. Quite a few of my designs are multi-coloured and printing seperate parts for later assembly can be a bit of a pain, especially swapping filaments between prints for a rush job and of course the initial modelling with tolerances to allow for easy fitment (RULE 2!).

So I was very keen to put it together once it arrived in my hot little hands. The promo videos looked fantastic and every review made it seem like the best thing to happen to 3D printing. And, eventually, it was pretty OK. It just took a long time to get there.

First up, like the Mk2s kit it was another beautifully packaged parcel. Everything split up into bags for the various steps, all tools provided and a link to the online assembly manual - plus a handbook detailing how to set up MM prints. So far so good. But of course, this wouldn't need to be written if everything was perfect. All of the things I pointed out with the original Mk2s build about tolerances were multiplied here in spades. Housings required way more force than I was comfortable with to fit nuts, every screw hole was too small and required drilling out or the thread would tap itself into place, and the PTFE tubing in the multiplexer required shaving to allow it to fit - despite the measurements provided by the manual I still had to take 1.5mm off the specified 41mm length of tube. It may sound inconsequential, but this was just in the first half day of building the kit!

The stepper motors for the 4 extruders were another tight fit, with the idler screws self tapping into the housing. As I discovered when I picked them up and all the locking nuts fell out of their traps.

Cable management was well documented, if a bit optimistic. The box provided for the new circuit board might have fit perfectly on the drawing board, but trying to fit 300mm of cables into a 200mm box without damaging any pins or breaking the box was a challenge. I ended up leaving an extra ~80mm of cabling hanging out of the top of the box. The single screw to close the case is under a lot of pressure, and I resorted to a long cable tie around the bottom to make sure everything stays where it should. (my initial attempt using 2 short ties joined together failed when the join hit on the extruder as it went up and down the Z axis) The detailed pictures of the new RAMBO layout were very much appreciated, especially when I had to go back into it to find a pin I'd fit the wrong way around (check the 5 into 4 pin cable from the stepper motors is placed properly or the unit will only recognize the first extruder!).

Flashing the new firmware over was quite easy, the online manual gives you a step by step guide to downloading the new drivers, multi material compatible Slic3r and pre-sliced calibration prints, with detailed instructions on how to upgrade the printer using the Firmware Upgrader program supplied.

And then, much like the old joke about the history of Russia, it got worse.

The preflight check of setting the PINDA probe was improved after I replaced the 10mm thread specified with a 12mm one. That allowed me to actually tension the clamp as opposed to the 10mm thread needing to get to the breaking point of the part before reaching the nut. The 'cable tie' technique for setting the probe height was a failure. After a lot of experimenting (and a big gouge in the bed from the nozzle trying to find a sensor point) I set the probe at 0.85mm as opposed to the 1.1mm recommended in the manual.

Do Not Want

06_gouge-1

All of the handling of the unit during assembly threw the frame out of whack as well. The manual suggests you can simply do a Z axis calibration after flashing the new firmware onto the machine, but I erred on the side of caution and ran the full XYZ run, and I'm glad I did.

Now part of this is a mix of ambiguity in the manual and my own bad habit of interpreting instructions in odd ways on occasion (quite frustrating during board games!). But I could not get the PINDA probe to center in the marked circle using the methods in the troubleshooting guide. The fact that this critical bit of information about the probe centering points is hidden at the bottom of a page in the handbook for calibration of the Mk2s, and not during the assembly stage is another issue.

The important bit

15_PINDA-1

For the error message 'XYZ calibration failed. Front calibration points not reachable', the handbook suggests 'move both left/right Y threaded rods in the Z frame away from you'. Which made no sense to me at all, I ended up translating that as moving the Z frame towards the Y stepper motor at the rear of the unit using the M10 nuts.

After repeated failures, a friend at the local makerspace suggested ignoring the handbook, and showed me how to line up the PINDA marking by setting the bed on the rear endstop and adjusting the frame accordingly. As it turned out, after all those adjustments I was nearly 5mm off where I should have been, and in fact should have been moving the frame forwards the whole time. The 100mm measurement for the frame outlined in the Y frame assembly was off by ~1.2mm. After another check for the bed being perpendicular and another XYZ test I received the Calibration OK message and moved on the the next step, nozzle height setting.

Having gone through all this with the initial build, I wasn't overly worried about this part - until the printer simply refused to extrude any plastic for the calibration print. Returning to the internet revealed similar problems with other users, the solution that worked for me was making sure the printer wasn't running on silent mode. Quite why that should matter I still don't understand, unless the stepper motors on the new extruders require a bit more grunt to feed the filament through the PTFE tubes. Either way, it eventually worked, I got the V2 calibration print down and got a Marvin underway.

This next bit is another minor sticking point (foreshadowing!) for me. I have had a lot of trouble with bed adhesion for both PLA and PETG filaments. The heated PEI bed is meant to be the dogs bollocks, duck's nuts and bees knees for 3D printers. And having no other frame of reference to judge from, I really feel for anyone using any other bed surface if this is as good as it gets. With the initial build of the Mk2s, I had filament ice skating around the bed and clinging to the nozzle on the first layer. Tried tape and glue sticks before I got the slicer settings to the appropriate range, and even then I still used glue for smaller parts. Did I mention that the new versions of Slic3r will overwrite all your old settings for filaments? All those hard determined print speeds, support and temp settings, gone...

First level failure

08_layer-adhesion

The first Marvin print fell over about 1/3rd of the way through the initial MMU print. So did the second one. Despite the firm belief of every online source (and a fair few people at the makerspace) that I shouldn't need glue for a PLA print, the third print passed only after glueing the bed. Again, I have no point of reference for how well this bed is meant to adhere, but I now glue any part with a smaller footprint than 45mm diameter. And for MMU prints, I definitely put some glue under the purge block, after having a few lift and get knocked over by the nozzle.

Finally!

00a_cover_MMU

(The DM staff of Natural 20 - available here)

The results are awesome - and will open up a few aspects of modeling that I haven't been able to produce cleanly with single filament prints. I will note that the purge block is quite thirsty on filament, and if you're doing small or one-off prints it pays to overestimate how much material you'll need. As Slic3r doesn't have an estimate for print times I've been using this online gcode viewer, which also gives you the amount of material used in a print. Multiple copies of prints at the same time won't affect the size of the purge block, so maybe keep that in mind if you're on a budget.


So, I was originally going to wind this up here, with a reiteration of how sloppy the build quality is on the parts in this upgrade and finally concluding that - similar to the initial build - it was frustrating but worth it. Except in the 2 weeks it's taken me to write this blog (lots of re-wording after particularly aggravating attempts at calibration) and having had the printer working nicely with both single and multi material prints, I've found another issue with the build.

First spotted when I emailed a photo of some staves in a flyer to The Tech Monkey, he commented that they looked 'a bit wonky'. Putting that down to my less than stellar photoshop skills, I forgot about it until assembling a customer order a few days later and realising that, yes, the staff sections were on a lean. By about ~2 degrees on the Z axis. Hard to spot on an individual piece, but very noticeable over a long assembly. Printing off an XYZ calibration part showed a definite lean to the rear on the Z axis. A bit more google-fu found a few tales of similar woe, with possible causes ranging from bent rods to filament pulling on the frame. After rebuilding my spool holder to give more clearance from the extruder cables, I added some shims under the M10 nuts on the Y frame to try and strengthen the join.

Shims (5mm aluminium flat bar)

11_shims

11a_shims

Another calibration print showed no improvement and a set square on the bed to the X axis found the frame still not perpendicular. My eventual solution? After another few hours online I found these (recommended as a remix of these) and found bundled in with a heap of other modifications here.

The parts were a reasonably quick print (sorry, I didn't have the official Prusa orange!), 40% infill and 3 perimeter wall thickness, and not high enough that the Z lean would affect them. Grabbing some M8 threaded rod, nuts and washers from Bunnings while the parts were printing (it was a Friday arvo, so no sausages), I cut the rod to suit and put it all together that night.

Wow. Such thread. Many spin.

13_brace_02

The final assembly

14_brace03

(Another quick aside – if you do end up using this modification, make sure the 8mm washers you get are small enough to avoid clashing on the front of the Y axis legs)

A quick attempt at setting the adjustment with a spirit level proved again how far off true the frame was, and even from the initial test print showed a drastic improvement.

Square at last!

17_winnar

Now why I would need to do this after several months of good prints on the original MK2s took a bit of thought. The 4 new extruders on top of the Z frame with their stepper motors are a lot of weight at the top of a frame held up by 4 M10 nuts and washers, on very slender rods with an overall contact area of ~20mm each side. The filament pulling theory was given credence during a test print when the filament tangled on the spool and I watched it drag on the extruder before I could free it up. In an ideal world, there would be plenty of space for spool holders and filament would never tangle. But here we are.

Frame warp

10_Prusa-bed-skew-resize

This frame warping will be an inherent problem with the MMU upgrade, as it is basically retrofitting new equipment onto an old chassis that, as pointed out earlier, already had design flaws. I hope the Mk3 addresses these issues, I know that Mr. Prusa takes a lot of interest in online discussions about his printers, so this can't have escaped the attention of his research team. That said - and this is speculation on my part - with the announcement of the Mk3 arriving so soon after the Mk2 and multi material upgrade I can't help but wonder if Mr. Prusa has rushed the production of these kits to concentrate on the development of the Mk3. The build quality of my MMU kit is certainly not up to standard.

3D printing as a technology is well out of its infancy these days but still has a long way to go to become accessible. I'm hoping that the lessons learned from these kits will provide a marked improvement in what is to come.

Ben the Design Monkey

Ben the Design Monkey

World reknowned industrial designer* and self confessed geek, Ben is the design guy for Enstaved. Using his love of comics, gaming and pop culture to create the designs we sell. *(in two countries)

Read More