Printing the unprintable using breakaway support on the Kraken

Printing the unprintable using breakaway support on the Kraken

Now we have finished doing all the pre-launch testing of The Kraken which involved printing rather dull and un-interesting objects to test performance we have finally got round to having some fun! I wanted to share our experiences in the hope that some of our techniques may be of use.

Over the last few days we have been working to print conventionally unprintable objects. In the process we used three combinations of materials:

  1. PLA using _XT as support
  2. _XT using PLA as support
  3. ABS using HIPS as support

We sliced with:

  1. Slic3r
  2. Cura

First, lets talk about the printer setup.

The printer that we have been running the beast on is a self-sourced Mendel90 with a Rumba board. We are looking forward to trying out the Duet and Azteeg X3 PRO soon but were keen to keep the number of variables to a minimum. (At least to start with!)


We were running the latest version of Marlin pulled straight from GitHub, configuration was really easy and took us less than an hour. Just work through Configuration.h and set the parameters - nothing scary.

The Kraken:

The Kraken was configured as sold using the four 0.4mm nozzles. Others sizes are available.

The leveling technique we used was as follows:

  1. Raise all but one of the heat-breaks up into The Kraken to keep them out of the way for now.
  2. Level your bed. It is very important that your bed is very level in X and Y. This is more important than usual because there are 3 other nozzles that risk interfering with your print. Note also that adjusting the lead-screws to improve alignment in X later is not an option because that rotates the entire X-Axis and thus Kraken bringing the nozzles out of line.
  3. Once happy that your datum-nozzle is level, bring it into contact with the bed and lower the other nozzles to also touch the bed. Leave any nozzles that you are not using up and out of the way.

We did not use a radiator in the cooling system, and the total volume of water was under 1L. You can see in the photo above that the reservoir is just a tupperware with some holes cut in the lid. High tech stuff.

We used 3 of our all-time favorite extruders - RichRap's Geared Bowden Extruder.

We initially used fairly long (800mm) lengths of bowden tubing but experienced quite a bit of oozing from the nozzles, so early on this was cut down to ~400mm and the extruders mounted to the top of the machine. In general, keeping your bowden tubes as short as reasonably possible allows you to have more control over your extrusion. We were a bit worried by some people buying up to 8m of tubing with their Kraken!

At the beginning we were so keen to get The Kraken up and running we designed an adaptor for the existing Mendel90 carriage. This was so unbelievably ugly I am not prepared to embarrass E3D by sharing it. The moment Tony Lock released his Kraken carriage in this fantastic post about The Kraken we had it printing. If you have a Mendel90, we recommend you doing the same (thanks Tony!).

Our go-to slicer is normally Slic3r (Version 1.0.0R1) because of its configurability and maturity so it is what we started out using. Because path-planning is so important with multi-extruder printing we switched to Cura (14.01) very early on. As it turns out it is much more sensible about a lot of things, most notably;

  1. If it finished a layer printing the support, it would start with support on the next layer which halves the number of material changes.
  2. The prime/wipe pillar (added in Cura 13.12) is great - we actually upped the 15mm2 default area to 30mm2 - not because it wasn't enough to purge and prime the nozzle, but because it was just more stable on the bed. With that said, the prints look rather entertaining when it falls over. For the most part the ooze shield does a good job of wiping your precisely extruded plastic spaghetti off the nozzle and the print stands a good chance of surviving.
  3. Slic3r seemed to generate some spurious support in places for no obvious reason which slowed down the print a lot because it had to switch materials each layer. To be fair, this was in part due to an inactive-nozzle-cooldown of -5ºC.


Filament Spaghetti Monster

Some shortfalls of using Cura over Slic3r:

  1. The ooze shield is a bit ambitious in the overhang angles that it tries to achieve and often collapses and/or delaminates (as seen in the above image). In slicer the ooze shield is vertical all the way up - this makes it easier to print but it is now further away from your object. Ideally we would have a Cura-like shield with a configurable maximum-angle.
  2. The overhang angle threshold is not definable in the UI but is instead hard-coded into the python backend to be 60º. This is probably fine on an Ultimaker printing PLA with a fan and no heated bed, however in our case we fired this right down to 10º which enclosed small underhung parts and stopped them becoming dislodged. Note: This was only a successful strategy because our material combinations detached from each other so well.
  3. Slic3rs interface layers provided a much nicer finish on the underside of parts for less support material used - printing model material over a sparse support structure in Cura resulted in a less-than-perfect surface finish. Good results can be achieved by bumping up the infill density as we did (to 30%) but this uses more material and takes more time than is strictly necessary.

Things are changing very fast with the slicers at the moment and I am sure that things will continue to improve rapidly.

Right, enough about the setup and on to the printing. Our test print for this whole saga has been the human femur by BME_Sundevil from thingiverse.

Printing PLA using _XT as support

PLA and ColorFabb _XT are a promising combination for printing using support material for a number of reasons;

  1. They print at similar temperatures both at the extruder and on the bed
  2. They both stick well to blue-tape (we added some hairspray to make them really stick)
  3. They hold together very firmly when printing, but miraculously detach from one and other very easily once cooled. It takes a little bit of force but you get a satisfying pop when they detach (voila!).

The main drawback was that the _XT oozed whilst the nozzle wasn't printing, despite having a 15mm inactive extruder retraction enabled. It wasn't really a problem because the ooze shield and prime pillar did their job. We would like to do more testing with this combo but initial results are very promising.

Printing ABS using HIPS as support

For aesthetics we used our own E3D Everyday ABS and HIPS filaments which are colour matched by our manufacturer. What this means is that any traces of the support material left behind are invisible. We don't have colour matched HIPS in stock at the moment, but we do sell Natural HIPS for anyone that wants to give this a go.

As with the _XT and PLA combinations a key point here is the similarity in printing parameters. i.e. both stick onto an ABS-Juice coated glass bed running at 125ºC and print within 10ºC of each other.

Importantly ABS and HIPS detach really nicely and this combination got us our best prints.

Our best prints were achieved with the following Cura parameters:

  1. Density of support material at 30%.
  2. Overhang threshold at 10% (set in cura/util/
  3. Prime pillar enabled (30mm2).
  4. Ooze shield enabled.
  5. Inactive Extruder Retraction to 15mm.
  6. XY Spacing 0mm.
  7. Z Spacing 0mm.
  8. Temps were 235ºC for the ABS and 230ºC for the HIPS on a bed set to 120ºC.
  9. Speeds were 50mm/s for printing, 150mm/s for travel.

Back to blog