My modular, multi-input, multi-material, mixing hot end – Part 2

Well the first stage of testing went exactly as expected – very little gained in terms of progress, but very much gained in terms of knowledge. Or using my analogy of a journey, I took a few steps up a hill but came tumbling back down. But now I have a better map, so I think I can navigate around this particular bump in the road.

The featured image above shows the test rig I made. It’s nothing fancy, just a wooden stand onto which I mounted the water pump, radiator and hot end. Naturally, the brackets are all printed parts. But I am already starting to hate this cooling system……..

1st problem. The pump came with quite a large diameter outlet spigot which was also the same size as the radiator fittings so I foolishly decided that it would be a good idea to use the same size spigot when I designed and made the cold tank for the hot end. But that tubing is far too big and stiff to be of practical use when the print head is moving. So I’ll have to use step down adaptors and much small tubing. But for test purposes the tubes would be fine for now.

2nd problem. I used silicone sealant the seal the aluminium tubes that run through the plastic cooler block. “Dow Corning 785” to be precise. It is a medium to high modulus acetoxy curing sealant. Here is what it says on the cartridge “Ideal for sealing sanitary ware and non porous surfaces. It offers excellent adhesion to porcelain, glass, ceramics and painted surfaces”. I can vouch that is all true. What it doesn’t do is adhere well to PLA and/or Aluminium it seems, because I had leaks around those tubes.

3rd problem. More leaks which seemed to be coming from where the tubing fitted onto the spigots on the printed block lid. Fitting hose clips didn’t help and further investigation revealed that the lid itself was porous in the area of the spigots.

Here is a picture of the topmost slice of that part

Cold block top layer

The two spigots are the larger circles to the far left and right. I can’t see anything much wrong with that. I used 3 perimeters, and 60% infill with 3 solid top and bottom layers. The lid was 3mm thick so with a layer height of 0.3 mm I had 10 layers. I’ve printed plenty of vases with those settings which hold water just fine, so I’m a bit baffled as to why it was porous, but porous it most certainly was.

So I printed another block and this time I used an “Araldite” type of two part epoxy to seal the tubes and also applied a liberal coating around the base of the spigots. For good measure, I gave the entire part two coats of “Rustins Clear Plastic Coating” which is a 2 part cold cure lacquer. It’s primary use is as an interior wood finish but it can also be used on plastics. I used to use it when I did wood turning and it is really amazing stuff (I once put a turned wooden bowl coated with this stuff through a dishwasher and it came out completely unscathed). Once cured, it’s impervious to just about anything that you could spill on it and is highly resistant to heat. One of these days I’ll do a bit of a write up here about using it on printed parts but for now – here is a link to Rustin’s web site so you can read about it It does the same job as XTC-3D which I’ve seen people use.

After the lacquer had cured, I filled the cooler with water up to the top of the spigots and left it over night on a paper towel. Next day the water was still at the same level and the paper towel was still dry so I thought I had cured all my leaks (which just shows how wrong one can be).

Next I reassembled the hot end/cold block, lowered my printer bed by 700mm (glad I built a big printer), put the test rig on the build platform and wired everything up. Actually that’s not quite true. To save me altering too much of the printer wiring, I just connected 5 of the heaters to the Duex 5 heater and thermistor connectors – the nozzle/lower mixing chamber heater, the heater that is at the top of the mixer/combining stage, and 3 of the 6 individual filament heaters.

Because of the way I have my Duet boards mounted, access to the heater terminals on the Duex5 isn’t easy. So I made up some leads with a ferrule at each end and connected them to an external connector block which is easier to access. It isn’t pretty but it means I can quickly connect or disconnect this experimental hot end without to much disruption to the printer itself. Here is a couple of pictures.

Before I turned on the power, I thought it best to run some checks to make sure there were no shorts and that all was well. That was just as well because when I measured the resistance of each heater, one of the small ones that I chose for the individual filament blocks gave a reading of around 180 Ohms whereas all the others were around 27.7 Ohms, which is what I would expect for a 20Watt heater using 24V. So I changed that heater for another, which was simplicity itself because there were no screws or other retaining mechanism to deal with. So that part of the design worked well.

Then I made up a configuration files with the new heaters and defined some tools. Initially just 3 tools, 0 to 2. All use the common nozzle and mixer heaters but each one uses a different filament heater. Here is a snip of how they look using the web interface. Note that tool 3 on here is just left over from an earlier configuration.

Having turned on the power and checked that all readings were close to ambient, it was time to test the heaters and I decided the best way to do that was to attempt to tune the PID parameters. It was at this point that I started to doubt my sanity (I dare say that many readers have been thinking that for some time).

To cut a long story short, all sorts of strange things happened before I realised my mistake, which was this. On the main Duet board, the two hot end heaters and their associated thermistors are referred to and labelled E0 and E1. On the Duex5 expansion board, the numbering sequence continues with E2 through to E6 and I connected my heaters and thermistors to these 5 connectors (E2 to E6). Here is a picture of the Duet wiring schematic:

Then in the configuration file, I defined these heaters using M305 P2 to P6. BUT, in firmware, heater number 0 is the bed heater. The hot end heaters start from heater number 1. So in firmware, heater number 1 refers to the heater and thermistor that are connected to the E0 connector. In my case, to define the heaters that were connected to E2 to E6, I needed to use M305 P3 to P7 rather than M305 P2 to P6. Similarly the commands M143 and M570 should use H3 to H7 to refer to heaters that are connected to terminals labelled E2 to E6. I knew this but had forgotten. As an aside I am told that generation 3 boards will be labelled differently and version 3 firmware uses a completely different system to reference inputs and outputs.

This is how the configuration looked after I corrected my errors.

M305 P3 S”E2 Nozzle” T100000 B4725 C7.06e-8 R4700;
M305 P4 S”E3 Mixer” T100000 B4725 C7.06e-8 R4700;
M305 P5 S”E4 FrontLeft” T100000 B4725 C7.06e-8 R4700
M305 P6 S”E5 FrontCentre” T100000 B4725 C7.06e-8 R4700
M305 P7 S”E6 FrontRight” T100000 B4725 C7.06e-8 R4700

M143 H3 S280
M143 H4 S280
M143 H5 S280
M143 H6 S280
M143 H7 S280

M570 H3 P30 T50
M570 H4 P30 T50
M570 H5 P30 T50
M570 H6 P30 T50
M570 H7 P30 T50

Finally, I got to start tuning the heaters but neither the “mixer” heater nor the “nozzle” heater would get to much above 140 deg C so the tuning was aborted. Clearly my choice of heater cartridge was too low in terms of power output. Although I noticed that when tuning the “mixer” heater, the nozzle thermistor was showing a temperature that was only about 7 degrees C lower. Similarly when tuning the “nozzle” heater, the “mixer” temperature was only about 8 degrees C lower. I had expected a steeper temperature gradient than that. Clearly the reason why neither of those two cartridges will attain the required temperature is because of heat conduction to the rest of the blocks. I had expected to see a steeper temperature gradient over the length of the hot end than that. If I was clever enough, I’d have done some thermal modelling or analysis, but I’m not, so I didn’t.

I also noted that when the mixer was at 140 deg C, the individual filament blocks were at around 80 to 85 deg C. That’s too high. It’s already above the glass transition temperature for PLA so when the mixer is at melt temperature, then the filament inside the individual blocks will be too hot to act as a cold plug. There is too much heat conduction in the aluminium tubes connecting the individual heater blocks to the first mixer block. I can probably get around that simply by using stainless steel tubing instead. And/or make them longer, and/or build in heat break but I want to avoid that if possible.

Now I have to make a decision. Regarding the lack of temperature gradient across the hot end, how important do I think it is to have the ability to use different temperatures for mixing than at the nozzle? If I decide it is important then I need to build in some thermal insulation between the mixer blocks and the nozzle block. If I decide it isn’t important, then I can either use both heaters and a single thermistor at the nozzle tip, or one longer heater. Alternatively I could use both heaters with their own individual thermistors but with each heater being capable of attaining the required temperature for the block as whole (so that I can tune the PID parameters). I decided to run a couple more tests using both heaters simultaneously but before I did that, I wanted to see how one of the small, individual heater blocks worked.

So then I started tuning the “front left” individual (20 Watt) heater. That started well and the temperature rapidly reached 200 deg C but as it did so, I noticed a puff of steam so I hit the power switch. Sure enough yet another water leak so killing the power was the right thing to do, before something shorted and took out one or more of my control boards (my Duet boards are very early models, possible even pre-production so don’t have the protection fuses that later boards have).

Right now, I’m asking myself the question “is water cooling really a sensible thing to do with 8 heaters and 8 thermistors various fans and other electrical components in very close proximity?”. I know that people use water cooling for computer processors but all it takes is one very small leak before it becomes very costly. Certainly I have decided that a printed tank is not a good idea. So if I continue down this route, it will be with metal tanks/chambers. Also I’ll do very thorough leak testing before ever considering installing it on a printer.

But something else has come up which means I have some serious thinking to do. As part of the design process, I am always conscious of the need to keep the size down. Losing say 100mm of axis travel for example, would be too much of a sacrifice to make. That was the main reason why I wanted to try this composite tubular design for the heat breaks. It would be much easier to buy “off the shelf” heat breaks but they are all screw in fittings. This means that the blocks they screw into have to be bigger to accommodate the screw thread and if they need a spanner to tighten them, then they have to be spaced quite some distance apart too.

As readers will know, I hate using long Bowden tubes which is why I went to the somewhat extreme lengths of mounting the 5 extruders on a separate XY gantry. So I started to think about how best to mount 6 extruders and how the Bowden tubes would run between the extruders and this hot end (which is narrower than the Diamond).

This OpenScad image shows the best that I have been able to come up with. It’s the most compact way that I can think of mounting 6 Bondtech BMGs with associated stepper motors.

For that to work, I’ll have to move the rails closer together because they obscure the extruder outlets. So it will be a complete re-design of the extruder gantry. But even so, there is a large offset in both X and Y between the outlet position of the extruders and the inlets of the hot end. Which means that the Bowden tubes would have sharp bends which isn’t going to work. Alternatively, I could raise the extruder gantry which would reduce the angle of the bends, but that means extending the Bowden tubes which defeats the whole purpose of mounting the extruders this way. I could abandon my beloved Bondtech extruders and use flex shaft driven extruders like the Zesty Nimbles but 6 of them? That would be a very expensive option and they have high gearing that might cause me some problems with retraction speed due to the maximum step pulse frequency that is available.

So bizarrely, having designed the hot end to be a small as possible, I now have to look at ways to increase the spacing between the filament inputs. Ultimately all the inputs to the hot end have to converge on a single point so that they can be mixed and then extruded out of the nozzle. The best way to achieve that whilst keeping the filament path as short as possible means having a straight line between that common point in space, and each of the extruder outlets. Which means that the inlets to the hot end will each have to be at a compound angle. That’s much how they are arranged on the Diamond hot end where the inlets are spaced around circle and angle outwards at about 20 degrees. Fundamentally it’s a neat design in that respect.

One “plus” is that with the filament inlets arranged in such a manner, I might have room to use “off the shelf” screw in type heat breaks rather than my home made composite tubes (which might not work in any case).

So I have some serious thinking to do. One big consideration is that I have to be able to make this thing with the tools available – essentially a manual milling machine and lathe. I have a plan which involves purchasing a rotary table but I might also need to make some jigs and fixtures. So there is that to do, as well as redesigning and making a new extruder gantry, and ultimately completely re-wring the printer when generation 3 of the Duet boards become available.

For those reasons, further updates on the hot end could be some time away. I have some other things to do so I’ll write them up as and when I get them done. Being retarded retired, I have a bit more time available. But I’ll keep plugging away with this hot end – at least until such time as someone else comes up with a hot end that will do what I want – then I’d just buy one.


8 thoughts on “My modular, multi-input, multi-material, mixing hot end – Part 2

  1. hi Ian! great to see you writing again! i finally made a wordpress account so i can comment and thank you for sharing your fantastic experiments with us.

    why aren’t you flipping the extruders around so the they are in the middle and the motors facing outwards, riding on the rails? it would bring the extruder outlets closer together, wouldn’t it?

    i’ve tried to understand your hotend assembly, but sorry if i’ve failed and this is way off.

    the top 6 elements that lead into the mixing chamber look like regular hotend assemblies only ending in a tube rather than a nozzle, and your heatbrake/pipe stack/water block actually looks longer than an off the shelf hotend.

    couldn’t you just use 6 regular hotends and screw another heatbrake in from below that leads into the mixing chamber? perhaps creality style j-head types that have the heatsinks which are flat on two sides, as they would fit next to each other nicely.

    being threaded together, you also wouldn’t need the difficult to machine stainless steel support structures.
    i think the whole thing could end up shorter, and WAY easier to make.

    for joining the layers of the mixing chamber, JB weld should handle the heat and conduct it between layers too, but then again if there aren’t any leaks without adhesive then it would be great to retain the ability to take the layers apart.

    look forward to reading more as it develops! good luck!


    1. Lots of questions. Everything was designed for a reason.
      Firstly, the mounting screws for extruders go through the body into the motor. So once one extruder is fitted, it is impossible to mount another facing it because you can’t get the screws in. I might yet mount them that way but I could only do it by mounting each set of 3 onto a plate, then fixing those plates into place. But if every I need to disassemble 1 extruder for any reason, I’ll have to firstly disconnect and remove


  2. Sorry, I’m away from home and trying to do this on my phone with very patchy internet access. Anyway, what I was about to say is that I’d have to disconnect and remove all 3 extruders to clear say a blockage in one of them if they were mounted on removable plates facing each other.


  3. The distance between the cooler and the individual heaters is the minimum that allows a heater to be changed – even though they are at an angle of 10 degrees and are only 15 long rather than the usual 20 mm long ones


  4. Standard heater blocks are much wider, which means there would be a greater separation between all the incoming filaments which must ultimately end up at the same point in order to be mixed. Anything that increases the separation between the filaments, increases the volume of molten filament, which is undesirable.


  5. Finally, many of my design iterations got scrapped because ultimately, there needs to be a way assemble it, as well as maintain and service it. So there is no point in having screw in parts close together if you can’t get a tool in to tighten them. Heaters and temperature sensors can fail too, so it’s important to be able to replace them without having to disassemble the entire unit. Extruders can get blockages that need clearing. Etc etc


  6. Hi Ian, I love seeing progress on this monster project. 🙂
    Did you ever contact Daren Schwenke for more information on his mixing hotend?
    There is no public info about the mixing impeller or the (mechanical?) sealing of the mixing shaft.
    I think Daren planned on selling his hotend, but didn’t find the time yet.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s