My 6 input mixing hot end – part 4

1. Introduction

Further to my last post (part 3), I have designed some test parts to highlight the “stripey toothpaste” effect and printed these using the Diamond hot end. I’ve also managed to re-configure the extruders on my machine, fit the new hot end, load and flush through some filament and attempt my first print.

2. Test Parts and the “Stripey toothpaste effect”.

Over the last few years, I’ve done an awful lot of printing with the Diamond hot end. This has given me a reasonable understanding of how the filament gets “combined” but not mixed. It can manifest itself in two ways. Firstly a tall object will appear to have different colours when viewed from the side, depending on which side it is viewed from. Secondly the colour of any solid top surface will vary depending on the direction that the infill was laid down. What I mean by that is if for example 45 degrees is chosen and the carriage moves from left to right with each “line”, the colour will be different than if the infill is being laid down from right to left. I find this difficult to put into words but it will become apparent in the following pictures. If an object has an unbroken top surface, the infill is always laid down in the same direction. But if the top surface has a feature such as a hole, then the tool path that the slicer produces will mean that infill gets laid down in both left to right and right to left directions. So the test piece I designed is a simple 3mm tall cuboid with a rectangular hole in the middle which will have this changing tool path on the top surface, but also with a taller cylinder at one end so that the difference in colour around the perimeter can be seen.

Using the Diamond hot end, I printed four of these test parts using different colour combinations with solid (i.e not transparent) PLA. The first combination was 50% blue with 50% yellow which should produce green. The second combination was 50% red with 50% blue which should produce purple. The third combination was 50% red with 50% yellow which should produce orange. The final combination is what I call the “killer combination” and it’s 50% white with 50% red which should produce pink. The reason why I call it the “killer combination” will become apparent.

These test parts are my baseline. This is what a “mixing” hot end such as the Diamond produces and is what I want to improve upon with this design.

Here are the test pieces viewed from different angles. The pictures should be self explanatory.

Now you see why I call the combination of red and white the “killer combination”. The lack of mixing shows as a very stark contrast. The difference in colour on the top surface had me puzzled until I discovered it was related to the direction in which the infill is being laid down (not the direction that nozzle is travelling which is irrelevant).

Here is what I think happens

In the above drawing, the filament bead is represented by the circle, and it is made of 50% colour A and 50% colour B which have been “combined” in the hot end. The nozzle itself is either moving away from us or towards us – that direction has no effect on the colour change that we can see.

Each time a line of filament has been laid down, the nozzle moves by one layer width to the left – this direction is what causes the colour to change. In reality the bead of filament isn’t that shape. It is being squashed at the bottom against the build plate or previous layer. It is also being squashed at the top by the nozzle itself. The right hand side (filament B) is being squashed or at least is being supported or constrained by the previous bead of infill. But the left hand side (filament A) is unsupported and unconstrained so can flow freely to the left. I can’t be sure exactly what happens (we’d need some sort of close up video camera to capture it) but the effect is that the filament bead “rotates” or collapses on one side, so that filament B is on top and filament A is underneath. So filament B becomes the prominent colour. When the direction of infill changes to “left to right” – that is to say filament A is up against the previous bead and filament B is free flow to the right, then filament A will be the prominent colour. Note that in both cases, left to right or right to left, the nozzle moves either towards or away from us but this has no effect on the visible colour. It is only the direction that the infill is progressing in, that affects which colour is more dominant than the other.

So this is what I’m trying to fix with my hot end. If I can get the filaments to actually mix rather than be combined into stripes, then I will have achieved what I set out to do.

3. Machine re-configuration

Readers of this blog will know that, in order to keep the Bowden tubes short, I have the extruders mounted on a second gantry which sits above the hot end gantry. This is my UV axis although it only become a UV axis for homing. At all other times, the motors are mapped to the XY gantry, so the extruders exactly follow the hot end.

The Diamond hot end has five heat sinks arranged in a circle and at an angle of about 20 degrees to the vertical. This means that the filament inlets are much further apart than my design, which in turn means that I had to reposition my extruders so that they line up better with the new hot end. The best way to do that is to mount the extruders facing each other, on the inside of the carriage plates. Whereas with the Diamond hot end, 3 of the extruders are on the outside of the mounting plates facing away from each other. The advantage of having them closer together is that I can use even shorter Bowden Tubes. But the downside is that the extruders become less accessible and it is more difficult to load filament. So in order to alleviate this as much as possible, I re-machined the gantry plates and removed any excess material that I could. Primarily, I took about 8mm off each side making them 16mm narrower. To do that, I had to remove the extruder assemblies from my machine which meant doing this to it.

Here are my 6 (3 left hand and 3 right hand) Bondtech BMG extruders (https://www.bondtech.se/en/product/bmg-extruder/) fitted to their modified mounts.

I can’t get them much closer together. There has to be room for the upper Bowden tube to run between the two lower extruders and I have to be able to get at the coupling on the bottom of the upper extruder. I have to be able to open up the sides too. Obviously this means that I have asymmetrical lengths to the tubes but any other way of mounting the extruders means longer tubes and / or bends in the tubes. Here is a picture of the two extruder plates bolted to the gantry.

I’d love to hear from anyone who has any idea how 6 extruders could be packed more closely while still being accessible.

These pictures shows the two Duet expansion boards (https://www.duet3d.com/Duet3Expansion3HC) fitted but hinged for access (they are normally fixed at the top so that they are vertical).

Access to the extruders is possible but not easy. The lower “thumb” screws holding the expansion boards can be removed and the upper ones fitted which allow the boards to be hinged upwards. Or I can take all the “thumb screws” out and remove the boards completely. But to disassemble an extruder, I have to remove one of the mounting plates holding three extruders because the screws go into the motors from the extruder side. Of course, when the expansion boards are in the vertical position, those six Bondtech BMGs are barely visible at all which is a shame but can’t be helped.

4. Mounting the new hot end.

At this point I’d like to digress a little. People have made various comments and suggestions which include statements along the lines “…because you have such a huge mass…” or “….because your hot end is so massive….” and so forth. For sure it is bigger than most hot ends – it has to be because it is effectively 6 hot ends in one. In an earlier post, I compared this new hot end with a “bare bones” Diamond 5 colour hot end without any fans. I also mentioned that the 5 colour Diamond needs a 50mm fan and modified ducting to get the required 27cfm of cooling that it needs to prevent heat creep. So here are some pictures of the full Diamond 5 colour assembly, complete with its cooling solution and fixed to its three point carriage mount vs the full assembly of my new design with its cooling solution and fitted to a similar three point carriage mount.

The only things missing from those pictures are the Bowden tubes for the new hot end but these come out more or less vertically from the central clips so are well within the envelope of the carriage mount (and are shorter). So although this hot end might be big, it’s a lot smaller than some.

I’m not too bothered about mass. I’ve demonstrated in other posts on this blog that I can print at up to 300mm/sec and my default non-print speed for all prints is 350mm/sec. I know this is contentious but it’s not mass that determines how fast one can print, it’s how fast one can melt and extrude filament. So if mass is not the constraining factor, why go to extreme lengths (as some people do) to reduce it? For sure, excess mass can lead to excessive wear and tear so keeping it low within reason is justifiable, but it’s not the demon that many people believe it to be given that speed and acceleration are ultimately constrained by the attainable extrusion rate. I don’t get “ringing” artefacts and maybe that’s because my higher mass has a lower resonant frequency. In any case, my extruder gantry with 6 Bondtech BMGs and their associated motors, carriage plates and X rails, weigh in at around 3Kgs in the Y direction, so reducing the mass of the lower gantry isn’t going to allow me to print any faster (but travel speeds of 350mm/sec and printing speeds up to 300mm/sec are just fine for me).

For those readers who aren’t convinced by my personal opinions, I did compare the mass of the Diamond assembly and the mass of my design as pictured. The Diamond hot end as shown tips my kitchen scales at 420gms and my design tips them at 260gms (plus a few grammes for the weight of the Bowden tubes). By the way, those kitchen scales are the old fashioned balance type that one puts weights on one side and has a pan on the other. Crude and “old fashioned” but reasonably accurate.

Moving on…. The hot end itself fits to a mount which has three dowels the same as my Diamond hot end mount, and which mate with three “Oilite” bronze bushes fitted to the gantry. So no modifications to the carriage were necessary.

Here is a picture of the hot end mounted with the rear three extruders fitted. You can see that the Bowden tubes are much straighter than they were with the Diamond.

Here is another picture showing all 6 extruders mounted with all the Bowden tubes fitted.

You’ll notice that I use transparent PTFE tube. I find it helps to be able to see that one has loaded the correct colour filament into the correct input of the hot end. I have tried the Capricorn tubing in the past but found that the Bowden clips didn’t grip it well and it tended to slip in the fittings, so I stopped using it.

The tube lengths are 150mm for the lower extruders and 210mm for the upper ones. For the diamond, the tube lengths (excluding the part that lines the heat sinks and heat break) are 180 mm for the shorter ones and 256 mm for the longer ones. So I’ve reduced the Bowden tube lengths between the hot end and the extruders by between 30 and 46 mm. That reduction in length is all due to the fact that the tubes are much straighter than with the Diamond hot end. As you can see from the pictures, because I no longer have the top mounted cooling fan, I could potentially drop the extruder gantry down and use even shorter tubes. But there is a downside to all of this which is that because the tubes are much straighter, they act to put downward pressure on the hot end, which is good in that it holds it in place, but bad when the build plate comes up and lifts the nozzle to trigger my precision switch for Z homing. Previously, it required very little vertical force on the nozzle to lift the hot end off its seat by 0.3mm or so and trigger the switch. Now it requires quite considerable force because it’s trying to compress the Bowden tubes. So I might have to lengthen them slightly so that they have a slight curve and will bend or flex when homing Z.

With the hot end installed, from an access point of view, I can get at the Bowden tubes easily enough, and I can get at the nozzle easily enough. I can also change a heater cartridge or thermistor because they simply slide out vertically. But from a visibility point of view, I can’t see anything of the hot end. This is what I mean. This picture is looking directly at the hot end from outside the printer.

The front rails obscure everything apart from the nozzle. If I push the carriage back, with my camera inside the printer frame, the view of the hot end isn’t all that much better.

I can see the nozzle and the front of a fan but nothing of the hot end itself. From the side, the picture isn’t much better either.

I can just about make out the copper nut at the base of one of the heat breaks but nothing much else. So using a pair of parallel rails with upper and lower carriage plates makes for a really secure mounting arrangement but completely obscures any view of the hot end itself. It shouldn’t really matter because the parts that need to be accessible can easily be reached but you’ll understand later why it does matter.

The printer was designed and built around a Diamond hot end – maybe it’s time for a complete re-think. I will probably end up using a single 2040 extrusion (40mm tall, 20mm wide) rather than hanging the hot end between two 2020 extrusions. But I need to get the thing working first, then I’ll think about redesigning the rest of the XY gantry (maybe).

5. Flushing out the hot end.

Sharp eyed viewers may have spotted a difference in the fan wiring. In the earlier pictures, this was nicely braided, in later pictures it’s bare wires. The reason is that I was stupid. When I first tested and tuned the heater, I left the wires overly long and connected them to the Duet main board, first taking care to set the fan jumper to 12V. Then I tidied up the wires and connected everything to one of the expansion boards. Being the idiot that I am, I forgot that the fan on the Diamond hot end was 24V so I failed to change the fan jumper from 24v to 12V on this expansion board. So having blown up two expensive Sunon Maglev fans, I temporarily fitted the noisier high flow ones (after first setting the voltage jumper to 12V) while I wait for the replacement fans to arrive.

As I mentioned in my first post, I was a bit concerned that there might have been some small particles of debris inside the mixing chamber. So I took a perfectly good E3D style brass nozzle and drilled it out to 1.5mm. Then I heated the hot end and loaded it with filament. I ran the extruders until I could see that each filament was just at the top of the Bowden clips. Then I extruded 10mm at a time from each extruder. This was to ensure that the filaments progressively found their way through the mixing chamber. The danger is that with initial loading of a mixing hot end, molten filament can find it’s way back up into an open input and out the top of a heat break assembly, where of course it would cool, solidify and cause a blockage. I continued with these small incremental extruder moves until all 6 extruders pushed filament out of the nozzle. Then I extruded about 200mm of filament from each extruder and finally another 200 mm from all 6 running at the same time. So far, so good.

Then I removed the 1.5mm “flushing out nozzle” and fitted a 0.5mm nozzle. Yes! My first nozzle change which was simplicity itself and accomplished with only one hand!

6. The first test print.

This was the moment of truth- which went as well as I expected. That’s because my expectations were low. I decided to use the “killer combination” of red and white which I knew to be the worse case. At this stage, I have no idea what values to use for pressure advance or (firmware) retraction so I set them low at 0.2 for pressure advance (I used 0.4 with the Diamond) and 1mm of retraction (I used 3mm on the Diamond). I used exactly the same gcode file which was sliced to use 80mm/sec for infill, with perimeters and top layer at 80% of this (so 64mm/sec). It started well but part way through I noticed some molten plastic hanging down from the carriage, at the side of the nozzle. Clearly I had a leak somewhere but because I cannot see anything of the hot end while it’s printing, I had no idea where the filament was leaking from. I let the print run for a bit longer but eventually this plastic formed a large lump which got caught between the nozzle and the rest of the print, causing the build plate to shift a few mm so I had to abort the print.

For what it’s worth, here is picture of the failed print.

I don’t know if any comments would be relevant given that plastic was leaking out, but dare I say that mixing looks a lot better than the Diamond test prints? The colours look a lot more even and I don’t see the sharp contrast of red and white that I get with the Diamond hot end. But each bead is somewhat stripey in appearance although that might be due to gaps caused by leaking filament giving under extrusion.

7. Next steps

Clearly I need to fix the leak (or leaks). At the time of writing, I haven’t yet taken the hot end off the printer to get a good look. To do that, I need to figure out the best way of unloading the filament without filling the heat breaks with molten plastic. During assembly, I have done something that might help but I’ll leave disclosing that for another time. As near as I can tell, by using mirrors and contorting my body, I think I know where the leak(s) is/are and I think I know why.

Here is an OpenScad mock up of the problem.

With a normal hot end, the heat break assembly would butt up against the nozzle but I can’t do that because I have a “combining block” and mixing chamber in between the two. So the heat break assemblies seal against the top of that “combining block” (that’s the big block with rounded corners). I don’t have the ability to make a single block that would have threaded, flat bottomed holes to take the heat breaks, then with smaller holes drilled at compound angles from the bottom of these flat bottomed holes to a common exit point. So I have to make it as two parts. The upper part that the heat breaks screw into I call the “inlet block”. The threaded part of these heat breaks is about 4mm long so the inlet block is just under 4mm thick, to ensure that the bottom of the heat break assemblies make good contact with the combining block.

I’ve shown the inlet block and combining block as being separated. In reality, they are bolted together using four bolts which fit into the countersunk holes.

From what I can tell, filament is leaking from the junction between the inlet block and the combining block at the extreme left and right hand edges. Now aluminium has a low modulus of elasticity so 4mm thick aluminium will bend quite easily – and when heated to around 200 deg C, the modulus of elasticity is even lower. The four retaining bolts are towards the centre of the inlet block and the edges of the copper part of the heat breaks are very close to the edges of the combining block. So I suspect that the pressure of the incoming filament is deforming that plate. The edges would only have to lift a fraction of a millimetre to allow molten filament to escape.

Off hand, I can think of half a dozen ways to fix that but they all involve making a new, wider, combining block that will take extra screws at the edges. This is by far the most complex part to machine. Maybe if I made the inlet block from brass or even stainless steel, it might work with the combining block as it is. But I think what I’ll do is make a new combining block which has a wider top section and a new, wider inlet block. Then I’ll use two extra screws at each end as well as the four in the middle, so eight screws in total. I might also make the inlet block out of brass which has a much higher modulus of elasticity than aluminium, but I need to be careful of using metals with different thermal expansion coefficients. Whatever I end up with, I’m confident that I can find a solution. One thing I am pleased about is that, as near as I can tell, there is no leakage from between the plates which make up the mixing chamber. I’ll know better when I get the hot end out of the machine.

8. Conclusions

So far the design is looking promising. It looks like it might actually be trying to mix. Whether that mixing is good enough or needs further refinement remains to be seen. Compared to a 5 colour Diamond hot end, my design is significantly smaller, lighter and quieter, despite having one more input, and I can change nozzles (using only one hand)!

I have a few things to think about and some machining to do. But for the past few weeks I’ve put the rest of my life on hold while I’ve been working on this project. Putting together these blog posts alone takes up a lot of my time. So I’m going to take a break for a couple of weeks or so and catch up with some of the other things I should have been doing apart from playing with my toys. 🙂

But rest assured that I will continue – I believe that I’ve made good progress and I’ve invested too much time and effort to stop now.

Back in a few weeks or sooner…….

Ian