My 6 input (5+1) mixing hot end – version 2

1. Introduction

As readers will know, back in June 2019 I started to design and make a multi-input, multi-material, mixing hot end. My main problems at that time all revolved around effective heat sinks and keeping the cold part cold. I experimented with water cooling but had problems with leaks. Then along came Duet Generation 3 electronics and firmware, so much of my time since has been devoted to upgrading my printer to take advantage of the facilities that this gives me. Duet Generation 3 gave me the ability to drive 6 extruders plus my XYUVAB and Z axes whereas previously I could only drive either 5 extruders without the load balancing AB gantry, or I could drive 3 extruders with the AB load balancing gantry but not both together. Also, I had a massive cable chain with about 40 plus conductors going to the extruder gantry so being able to mount Can Bus expansion boards directly onto the extruder gantry has vastly improved the wiring. The change over has been much more time consuming that I imagined it would be, mostly because the firmware took longer than I thought it would when I embarked on that particular journey. However, we got there in the end and whilst there are still a few things that are missing from the expansion board firmware, the printer is once again mostly functional. So I am able to get back to designing and making a mixing hot end.

2. Design criteria.

My first attempt used separate heater blocks for each filament before they pass into the combined mixing chamber. The idea was that it would be easier to set different temperatures for different material types. Also, any unused filaments could be kept cold which might have helped with pressure build up and reduced retraction requirements. But this meant that each individual heater block needed to be thermally isolated from both the incoming filament above, and from the combined heater block below. These additional heaters and heat breaks make a very complicated design. So I took a step back and decided that for printing different materials types, which may require significantly different temperatures, then a tool changer type printer would be a better approach than trying to develop a single hot end that could be both a multi-input mixing design and also used for multiple materials. Therefore I have “lowered the bar” somewhat from my initial design criteria and decided to concentrate on a hot end that will mix, but not necessarily be the best choice for printing multiple materials which require significantly different temperatures.

Other than that, my design criteria is the same. That is to say make a hot end that will truly mix filaments together, rather than simply combine them into “stripey toothpaste”. In addition, I want the ability to easily change nozzles. And if it can be smaller and/or lighter then they are added bonuses.

One other thing that I’ve thought about is variants. As I have previously stated, this version has 5+1 inputs. But it wouldn’t be too difficult to make variants that use less inputs. Also, the section which combines the inputs is separate from the section which mixes them together. So it would be simplicity itself to have a “straight through” module which fitted in place of the mixing chamber. This would allow very high speed printing because it retains the combining section which means multiple, parallel melt chambers, each fed from a separate extruder. I know from experience with the Diamond hot end, that this arrangement gives a very high melt rate and have demonstrated printing at up to 300m/sec with a 0.5mm nozzle and 0.3mm layer height elsewhere in the blog.

3. Practicalities.

It is clear to me that thermal management is the key to a successful hot end design. It is also clear that making an effective heat break is beyond the capabilities that I have available in my garage at home. So I decided that I needed to use heat breaks that I could buy ready made. With all but one exception, heat breaks tend to be structural elements and made from a single material. This in itself means that the material wall thickness has to be thicker than is optimal for an effective heat break, in order to support any loads or forces that are applied to it. This additional material thickness means that more heat gets transferred and so larger heat sinks and fans are required to conduct that heat away, in order to keep the incoming filament below it’s glass transition temperature. Being made in one piece, from a single material also compromises the efficiency. The heat break itself needs to be a poor conductor of heat, but the heat sink on the cold side of the heat beak needs to be a good conductor to take away any heat that finds it’s way past the break..

Additionally, mixing hot ends suffer from a unique problem, which is that all of the filaments have to be loaded into every input, but some of those filaments may not be used for every layer of a particular print, or even not used at all. This means that these filaments which do not move forward effectively bridge the heat break and heat can get conducted through the filament itself. PLA in particular is prone to swelling and can jam in the heat break under these conditions. So very efficient heat sinks are required.

I mentioned above that there was one exception to the rule that heat breaks are single material, structural elements and therefore compromise on efficiency. At least there is only exception that I know of, and that is the heat breaks designed and made by Slice Engineering https://www.sliceengineering.com/ in their Mosquito hot ends. These use ultra thin wall stainless steel tubing for the heat break with copper heat sinks. But this lead me to a dilemma in that although these heat break/heat sink assemblies are of the highest quality, they don’t come cheap – at least when one considers that I need 6 of them- and I have no idea if this design will work until it’s built and tested.

So with that in mind, I contacted Slice Engineering, outlining my dilemma. To my eternal delight and gratitude, I had a long video “Hangouts” meeting with Chris Montgomery, one of the Co-founders of Slice Engineering. The upshot of that was that he very kindly sent me 6 of their heat sinks plus mounting hardware and other accessories at no cost to me. (Although DHL hammered me for Import Duty and VAT that UK customs and excise levied based on the list price rather than the discounted price, plus “dispersements” – but that’s another story). Chris has also been very forthcoming with knowledge, none of which I am prepared, or at liberty, to repeat here.

4. Design overview.

The following is mostly a description of my thought processes and the reasons why I have arrived at where I am now. Some of what follows may not be entirely clear but I’ve added pictures and my step by step assembly details later in this post, which should clarify any points that my words alone cannot adequately describe.

4.1. Mixing

For true colour mixing, I believe that we need Cyan, Magenta, Yellow and Black. This is how most desktop printers work. But they all print on White paper which we don’t have so we need to use White filament as well. Which means 5 inputs. There is possibly a case to say that those 5 colours will take care of hue, but for saturation we may need to add transparent, but based on the fact that many filaments, in the non-dyed state are transparent, I’ll ignore that for now.

As I have mentioned before, none of the available “mixing” hot ends that I have encountered actually mix the filaments together. They tend to just combine them and what comes out of the nozzle is akin to stripey toothpaste. In order to mix the filaments together I’ve retained the idea of passing them through a complex, 3 dimensional labyrinth. So firstly the incoming filaments are combined in a manner to any other “mixing” hot end. Then the combined filament passes through a mixing chamber. My theory being that by twisting and turning the combined filament in many different directions as it passes through the chamber, will cause the combined filaments to become more homogeneously mixed. In order to make this 3 dimensional labyrinth, I used a number of flat plates with slots milled in both the upper and lower faces, and circular holes joining those slots in the vertical direction. The slots are rectangular in section but calculated such that the cross sectional area is the same as the circular holes. This means that as well as being turned and twisted, the filament will also be squeezed into varying cross sectional shapes, which should further help with mixing.

The question is, how much twisting and turning is needed? The dilemma is that if the labyrinth is too complex it may need too much force to push the filament through. Also, the larger the mixing chamber, the greater the volume of molten plastic which will need to be purged on colour change and which may be undesirable for other reasons. So this initial design has what I consider to be the minimum amount of twists and turns. If more are needed, I can extend the labyrinth by adding more plates.

4.2. Pressure and retraction

One other thing that I’m concerned about is pressure. Or more to the point, relieving pressure to prevent filament oozing during non-print moves. Anyone reading this is likely to know that the normal way of preventing filament from continuing to ooze out of a nozzle during non-print moves is to partially retract the filament. This works well in all hot ends but I have a suspicion that retracting filament as it enters a complex labyrinth shaped mixing chamber, may not have much effect on the pressure of the filament at the nozzle tip at the other end of that chamber. For that reason, I designed the hot end so that it uses a 6^th filament and this one has a direct path straight to the nozzle tip. That’s why, as I indicated in the title of this post, this is a 5+1 mixing hot end. That is to say, 5 filaments will be combined and then pass through the mixing chamber but the 6th filament has a straight path directly to the nozzle tip. When using firmware retraction as one must when using a mixing hot end, all filaments get retracted concurrently regardless of the mixing ratio. So the intention is that this 6^th filament will act as a plunger. In practice I’ll likely use clear filament and set the mixing ratios so that a small percentage of the filament is always added. This will keep it moving forward slightly during a print and negate any adverse effects of retracting and un-retracting the same piece of filament repeatedly. The other use for this 6^th input is to be able to print a single colour object, possibly at high speed (due to the extensive melt chamber volume). In that case, I would simply swap out the clear filament for whichever one is needed.

4.3. Mounting.

I gave a great deal of thought about how to mount the thing. Essentially I wanted it to be a drop in replacement on my machine whereby it sits between two parallel rails. But at the same time, I gave some thought to how it might be mounted on other machines. So there are two M3 threaded holes on each side which could used to take bolts and fix it to any flat surface.

Because the heat breaks themselves are not structural, there needs to be a way to mount the hot part but at the same time, minimise any heat transfer between the hot block and the cold block. On the Slice Engineering Mosquito, this is done by using 4 very thin wall tubes and two tiny screws. I have borrowed this idea (and used the thin tubes that Chris Montgomery kindly sent me), but because I use two side plates, I have been able to take it a step further and do away with those tiny screws. (Not that they would in themselves transfer much heat because they really are tiny). This will become more apparent in the following pictures but essentially the hot block is clamped between the two side plates by the same four thin wall tubes (two each side) that a Mosquito uses, but without the two tiny screws. The tubes fit into blind holes such that when the side plates are fitted, the ends of the tubes press hard against the hot block and clamp it firmly in place. I was going to use two 2mm diameter threaded rods with lock nuts to clamp the lower ends of the side plates but once I had assembled it using the upper screws, it became clear that these rods are not necessary. Oh and just like the Mosquito, it is possible to fit or change a nozzle with just one hand.

4.4. Heater and thermistor

Because I have to first combine the filaments, then pass them through a mixing chamber, the entire hot end is quite long compared to a conventional single filament hot end. This lends itself to needing a longish heater and to distribute the heat evenly, it needs to be mounted vertically, rather than horizontally. This is actually rather good because it means that the heater and thermistor can be inserted into “blind” vertical holes and need no clips or screws to hold them in place. It is impossible for either of them to fall out. I designed the top plate and side plates with cut-outs which allow the heater or thermistor cartridges to simply be dropped in from above. And of course they can also be pulled out the same way. So should it be necessary to replace a heater or temperature sensor, it can be done quickly, easily, and in a tool free manner. The heater hole is 6mm diameter to take any readily available cartridge. Likewise the temperature sensor hole is 3mm diameter and will take any sensor of this diameter (e.g. thermistor, thermocouple or PT100/1000) as well as Slice Engineering’s own high temperature thermistor.

At the time of writing, apart from my experiences with the 5 colour Diamond hot end, I have no real idea of what capacity heater cartridge will ultimately be needed. On the basis that I was able to print at very high flow rates using a 40 Watt heater, that is what I have decided to use as a starting point. But I need one that is 30mm long and so far, I have been unable to source one here in the UK. I’ve even contacted heater cartridge manufacturers to get one made but without success so far. I’ll continue my search but meanwhile I have ordered one from China which is on a slow boat somewhere and should be with me in a few more weeks time.

4.5. Fans

Because I have two rows each of three heat sinks, it occurred to me that blowing air from one side or another might favour the central heat sinks but leave the outer ones starved of air flow. Or one row of heat sinks would be masked by the other, so the row closer to the fan would get more air than the row further away. I briefly thought about using some sort of fancy ducting arrangement before it occurred to me that the simplest solution would be to use two fans, both blowing inwards towards each other. My theory being that the airflows from each fan will meet in the middle and get forced out sideways, thus ensuring an even flow of air over all 6 heat sinks. 30mm fans fit nicely within the overall envelope and I’ve chosen a couple of Sunon Maglev fans based on their low noise level, rather than air flow characteristics. Only testing will tell if these are of sufficient capacity or if I need to fit higher flow rate ones.

4.6. Nozzle.

The hole for the nozzle is M6 x 1 so it will take pretty well any “off the shelf” nozzle. Chris kindly sent me one of Slice Engineering’s Vanadium nozzles but as the internals of this prototype mixing chamber are made from aluminium, I probably won’t be trying to print with any abrasive materials.

5. The hot end itself

I’ve made these first prototype parts from aluminium because it’s relatively cheap and easy to work with. If the design proves to be successful, I’ll probably make another from Copper and plate it myself or get it plated to prevent it from oxidising.

Here is a picture of the parts I made that make up the cold block. That is to say, the top plate which takes Bowden clips, the heat sink plate into which the Mosquito heat break assemblies screw, and the two side plates which will also mount the fans.

This is a close up of the top plate,

Here it is with the Bowden clips pressed in. I left two of the plastic parts out so that readers can see the brass, press-fit, inserts. These were “off the shelf” parts that I purchased.

Here is the underside into which the top of the heat break tubes fit. I’ve milled out a lot of material in order to increase the surface area to dissipate any heat that does find it’s way past the heat breaks and heat sinks. Maybe that was unnecessary. Astute readers will also notice that this isn’t the underside of the block in the above picture. That’s because in the first version, I didn’t make the holes deep enough to take the clips so I had to re-make this block with less material removed from the underside, and I forgot to take a picture of the underside of the finished block. But you get the idea I hope.

These are the parts that make up the hot block. Sorry but I’m not prepared to share details of the mixing chamber design here so I’ve blanked out parts of the images. I don’t know where this journey will end so I want to protect any potential commercial interests. If one considers that I’ve bought a milling machine and tooling, I’ve invested many hundreds of £s so far, to say nothing of my time. So if there is the potential to recoup some of this expense, then I think it is wise to protect my intellectual property – at least for now.

The part top left is the combining block. You can see 6 holes evenly spaced which are the inputs. The 4 larger holes are threaded M3 to take bolts which hold the heat sink plate. Of the 6 inputs, the centre hole at the front goes straight down to the nozzle. The central hole at the back is at an angle of “xx” degrees (“xx” is for me to know and you to guess) from back to front. All the other holes are at compound angles, about “xx” degrees back to front and about “yy” degrees left to right. The net result is that those 5 holes exit at the same point at the bottom of the block (but at a different point to the single, non-mixed, filament input).

The plates were all made from flat stock but then hand lapped to make them as near perfectly flat as I could. Initial assembly was “dry” but if there is any leakage, I could try using Boron Nitride paste between them.

Here are a couple of pictures of my milling machine set up to make those compound angled holes

Here are some pictures of the entire assembled hot block (but without nozzle).

Here is a picture showing the inlet block bolted to the combining block with the 6 heat break assemblies in place. On a “normal” hot end, the heat break would butt up against the nozzle but obviously, I can’t do that. So the heat break assemblies butt up against the top of the combining block.

I may need to make some changes here. When I designed it, I thought that I could fit and tighten the inner two heat breaks, then fit and tighten the outer 4. But I forgot about the M3 bolt heads which make it difficult to get any sort of tool onto the hexagonal part of the heat breaks. I think I’ve managed to get them tight enough. If not, I may need to tweak the design a bit. Countersunk screws would likely work well.

This picture shows the side plates fitted. If you look very closely you can see a slight gap between the side plates and the upper inlet block. This is because the depth of the blind holes into which the thin wall tubes fit is just slightly less than the length of the tubes. This ensures that the hot block is clamped firmly in place and cannot slide back and forth along the tubes.

Here is a view from another angle and you can see the 4 thin wall tubes which make the only contact between the hot block and the cold part.

Here is the hot end with the mixing chamber bolted to the combining block and a nozzle in place.

Here are a couple of pictures of the full assembly with the fans fitted.

To get a sense of scale this is me holding it

As a comparison, this is the 6 input,mixing (hopefully) hot end assembly with fan, next to a bare bones Diamond 5 colour without shrouds or fan. The one on the left tips my kitchen scales at around 180gms, the one on the right is 235gms.

As another comparison, this the (6 input) hot end loosely fitted to it’s mount next to a full Diamond 3 colour assembly with mount.

Not bad considering mine has twice as many inlets and a mixing chamber.

6. Next steps

The next thing to do will be to fit the heater once it arrives, then run a PID tuning cycle. That will give me some idea if 40 Watts is going to be enough. If not, I’ll have to source a higher wattage, 30mm long cartridge which could take some time. Then ideally, I’d like to take some thermal images to see what the heat distribution looks like and especially to see if those fans will supply enough air to the places needed to keep all 6 heat sinks cool. I don’t have a thermal imaging camera and my pension won’t stretch enough for me to buy one, so I’m trying to beg, borrow, or steal one. If the heat sinks look like they are running too hot, I’ll have to look at using higher flow rate fans and/or design some sort of ducting. After that, I’ll have to pluck up courage and actually try running some filament through it. Hopefully it will come out of the nozzle and nowhere else. I’ve bought a couple of brass nozzles and I have the Slice Engineering Vanadium one to try as well. I’ve cleaned all the parts as well as I am able but it’s possible that there might be some debris still inside. So my intention is to drill out one of the nozzles to around 1.5mm or so, then use this just to flush filament through before I fit a 0.5mm nozzle. Then I’ll see if it actually mixes. If not, I’ll have to see what I can about extending or modifying the mixture chamber. I’ll also evaluate retraction settings and see if my idea of using a straight path for the 6^th filament is helpful or not. Once I get that lot sorted, then it will be a case of tuning, tweaking and exploring flow rates. In theory, because there are multiple melt chambers, each fed by a separate extruder, it should be capable of printing at very high speed (depending on the mixing ratio) but in practice, the convoluted filament path through the mixing chamber might restrict the flow rate.

So lots to do still but I’ll do my best to post further updates about this journey as and when I have something interesting to report.

As ever, I hope readers have fund something useful or interesting in all of the above.

Ian