Based on my experiences with the Multi-Material hotend for the Prusa, I'm trying to build something "better"
Standard Tesselated Geometry - 7.45 MB - 11/27/2017 at 22:52
Standard Tesselated Geometry - 1.09 MB - 11/27/2017 at 06:15
Standard Tesselated Geometry - 223.23 kB - 11/27/2017 at 06:15
So after the failure of the high temp silicone, I tried using some J-B Weld 37901 Extreme Heat High Temperature Resistant Metallic Paste. It failed too, again not bonding well enough to the metal and resin parts to provide any additional physical strength.
So, unless I can find an alternative adhesive or investing in making a metal part,I'm a bit stuck.
New mixer block, but this time glued using a high-temp silicone paste. But ... same story, and not really a surprise given the silicone didn't provide much extra structural support.
Three layers into a simple test print and this happened.
The neck of the mixer, which screws into the heated block, broke away. The material is just too fragile for the pressure exerted on it as the head moves. I'm going rebuild the failed parts and look at ways to better secure the block to the mixer.
Above is a thermal image of the heated hotend while extruding filament. The white is the aluminum heat block. You can see how the heat is being kept in the block and not creeping into the mixer. I was worried that the thermal properties of the high-temp resin would allow that to happen, but all looks good.
The AnyCubic Kossel comes with a single extruder and I need two. Rather than just add another I decided to replace it with two extruders of my own designed (which maximize filament support). I already use a version of this extruder on my Prusa.
The extruder is design to use a pair of Bondtech drivegears which are great for moving the filament effectively but rather pricy at $50 per extruder (there really must be something as good for less money ... anyone?)
The two extruders (plus associated idlers) were printed in Rigid Ink White ABS, assembled with cheap steppers from Amazon, and mounted on the rails in place of the old extruder.
The E-step value for these extruders was 136.2 but obviously your milage may vary.
The TriGorilla board which comes with the Kossel (and unfortunately is no longer produced) supports two extruders, so to get both working was just a matter of reconfirming the AnyCubic version of Marlin to support both.
The hotend was assembled just as in the bench tests, with the heated block screwed into the base of the printed multiplexer. The multiplexer was then bolted into the Anycubic tests frame together with the cooling fan.
The final piece was to glue a small piece of teflon tube into the multiplexer. Gluing teflon is tricky; the very nature of the stuff is that it doesn't stick. To make it happen I first treated the outside surface of the PTFE tubing using Loctite 770 which is a teflon primer. This allows the treated surface to bind to superglue. So, once the surface was treated and the primer had dried, I coated the tube with a little superglue and then stuck it into the multiplexer. I came back an hour later (not entirely sure how long it takes to dry) and the tube was nicely bonded.
Finally, I ran a simple extrusion test using the current extruder to push filament into my new heated hotend. I was rewarded by hot filament dribbling out of the nozzle.
With the AnyCubic's Kossel assembled, I'm replacing the standard hotend assembly with a 3d printed carrier designed to accept the high-temp resin hotend. You can see the frame here:
The hotend will slot into place and be secure with the fan bolts. I modified the hotend wiring to make it modular, using a 5.5mm jack for the heater power, and a smaller 6-pin connector for fans and thermistor. That makes is really easy to pop out the regular Kossel hotend assembly and replace it with my experimental ones.
The new carrier was printed with Rigid Ink ASA and then the 6 rod attachment points were threaded using a standard M3 tap. It's not as rigid as the standard part, but I think will be good enough for these tests.
Just a quick update: to better test this hotend design I decided I would have to build it into a real 3d printer. Unfortunately I don't have one I can cannibalize, so I had to go buy an extra one. Being cheap I bought an Anycubic Kossel from China for about $180. This should arrive this week and I hope to have time to assemble it over Thanksgiving. Once I have it working I'll switch out their hot end for my version.
The PTFE spray experiments suggest that using a better PTFE or other non-stick coating in the heatsink might fix the clogging problem. The way to check this is to insert a small piece of PTFE into the base screw but it has a few problems:
For problem 1 and 3, these are acceptable compromises during testing. For problem 2, in tests the 1mm resin was too thin and would break easily. However, switching to a 3mm OD/2mm ID tube increased the wall to 1.5mm; which was strong enough.
The above shows the new design with a 10mm PFTE insert. In testing, the new PFTE liner did prevent clogging. This design works ... even if it is a compromise.
After ironing out various minor failures, I was able to test an "operational" heatsink with my test rig. To my surprise I was able to push PLA through the heatsink, into the heater block, and out of the nozzle. However, it quickly became harder and harder to push the filament. On a few test pieces, pushing too hard cracked the heatsink at the point it connected to the heater block.
So why was this happening? Well it appears that the high-temperature resin is not particularly slippy and the semi-molten PLA is very happy to adhere to it. As it does this, the path of the molten filament closes until it become impossible to push filament anymore.
One traditional hotend uses a PTFE liner. So a solution to the sticking problem might be to coat the inside of the filament pathways with PTFE (which you can buy as a spray). Experimenting with this did improve the performance, but it only delayed the moment when the pathway became jammed. It's not clear if the PTFE coating didn't stick to the resin effectively or because of some other problem.