My 6 input mixing hot end V2 – part 2.

Thermal testing and other stuff…

1. Introduction

Following on from my first post, I acquired a 30 mm long, 40 Watt heater cartridge which I have fitted, along with a cartridge style thermistor. I wired those and the fans temporarily to my Duet generation 3 main board. Eventually, when the unit is installed on the printer it will be wired to an expansion board but there were three reasons for using the main board for testing. Firstly the firmware does not yet support the PID tuning of heaters fitted expansion boards, which I need to do. Secondly, the mainboard is easier to get at than the expansion boards. Thirdly, it means that I can connect everything up without disconnecting any of the existing wiring. So I can switch between my working printer and the “test rig” simply by changing the configuration file. That process only takes a few seconds using Duet firmware.

I mentioned “test rig” (note the inverted commas). This is it –

I simply screwed a couple of lengths of 3mm threaded rods into the mounting holes and clamped these to the front rail of the printer. So the hot end hangs in free air, much as it will when it gets fitted to the carriage.

Astute readers will notice that I decided to fit the 2mm threaded rods which join the side plates together, whereas in my first post, I mentioned that they were not necessary. The main reason for doing so was that originally, I had the heater and thermistor wires coming straight out of the top. But after fitting the heater, I realised that these wires were going to obstruct the air flow from the fans to some extent. So I wanted to have the wires coming out of the sides, well away from the fans and those threaded rods gave me a convenient fixing point for a cable tie. Here is a close up –

I’m a bit concerned that the wires coming out of the heater are bent at too acute an angle which might cause them to fail. So I plan to fit those threaded rods a bit higher up – (see Design Modifications later).

I was also slightly concerned that, when I change nozzles, it’s only those 4 very thin walled tubes which would take the twisting force of the tool tightening or loosening the nozzle. It works well enough on the Mosquito hot end so it’s probably not an issue, but the Mosquito does have screws (albeit very small ones) as well as the tubes. The rods will help to take any torque away from the tubes. I’m not overly concerned about the extra couple of grams or so that they add.

2. Heater tuning

The first thing to do (after verify that everything worked as it should) was to tune the heater PID parameters. This is where I hit my first snag, although it wasn’t unexpected. Essentially the heater would not reach the target temperature that I set to 240 deg C. The tuning cycle aborted at about 170 deg C and I would say, looking at the temperature curve, that if I had left long enough it might have reached about 180 to 185 deg C. Clearly 40 Watts is not enough and I need a higher capacity heater. The surface area of the hot part is a lot higher than a Diamond hot end (which is all I have to go by) so it’s not really surprising. This was without any filament so it may be worse once I have filament loaded in all 6 inputs.

However, there is some good news in that Chris from Slice Engineering thinks that a 20mm long heater will work just fine despite the fact that the hole for it is a bit over 30mm long. So that gives me more scope to source a higher wattage heater because 20mm long ones are much more common than 30mm long ones. I think the ideal will be about 60 watts but so far, finding one is still a challenge. I’ll keep looking but meanwhile, I’ve ordered an 80 Watt one from which is what they supply for the 5 colour Diamond (why oh why, did I throw out the one that I had?).

I might also look at making a “silicone sock” to insulate it. That is more to shield it from the effects of the part cooling fan as anything else. But at the moment, I don’t see an easy way to hold it in place so I’ll leave that for a future date. And of course, if I do end up with something that works, I’ll likely make the hot parts out of copper which might help too.

3. Temperature gradient testing.

Although I could only run the heater at about 160 deg C, I thought this would be adequate to do some sort of evaluation of the temperature gradient throughout the melt and mixing chambers. I’ll run the tests again when I fit a higher capacity cartridge heater, just to be sure.

I mentioned in my first post that I thought some thermal images would be good, and they probably will. But being impatient. I decided to try something else. So I bought a very thin thermocouple (I already have a K Type reader) that will fit inside the filament tubes. Having a straight path from one of the inlets right down to the nozzle tip made this very easy to do. I inserted the thermocouple right down until it was touching the back of the nozzle, making sure that the tip was in contact with the side wall and measured the temperature with the heater set to 160 deg C. Then I withdrew the thermocouple by 5mm, left it to stabilise for a few minutes and repeated until I got to about 5mm from the top of the hot block.

Here are the results –

The first thing of note is the good correlation between the thermocouple and the thermistor which was controlling the temperature at 160 deg C. The second thing is the lack of any temperature gradient apart from the very top of the heater block. This is probably being affected by the fans blowing cold air over the heat sinks, some of which must be blowing over the top of the block. I’m not concerned about a 2 degree temperature drop at this point but it’s something to bear in mind if I increase the air flow (more on this later). So I’m reasonably confident that I have good thermal distribution throughout the hot end.

4. Heat sink fans cooling

There has been some debate on various forums about the potential effectiveness of my plan to use two fans, both blowing in from outside and exhausting out of the sides. I was sceptical too. Also, I had no idea if the fans that I had chosen would be high enough flow rate. My criteria at the time was to use fans with the lowest noise level. When my printer was in my study, right behind where I’m sitting as I write this, noise level was a significant factor. But now that the printer resides in an enclose “booth” which is itself inside my garage, noise level is not really a concern at all.

I thought about various methods of evaluating air flow using some sort of smoke or visible vapour generator but it eventually dawned on me that what I actually need to measure is the temperature inside the heat sink tubes. Any differences between temperature at the same point in the 6 heat sinks will be as a direct consequence of the air flow over those same heat sinks.

So I set the thermocouple up again but this time inserted it into the heat sink assembly about 2 – 3 mm above the heat break. Roughly between the lowest two heat sink fins. I marked this point on the thermocouple using the simple expedient of sticking a piece of tape around the wire.

Here is picture of the set up – I used Kapton tape as it was to hand, so it might not show up well.

Then I ran the heater at 160 deg C, gave it 10 to15 minutes or so to stabilise and recorded the temperature of each heat sink at that point. I did 3 tests. The first was with both fans blowing in towards the heat sinks and exhausting out of the sides. I called this the “Push-Push” test. The second was with one fan reversed. So air blown in towards the heat sinks on one face and being “sucked out” on the opposite face and with the sides open. I called this the “Push-Pull Open Sided” test. The third test was also a “push-pull” but I masked off the sides using tape. I called this the “Push-Pull Closed Sided” test. Obviously, I could not do a “push-push” with closed side test as there would be nowhere for the exhaust air to go.

Here is a picture of the “closed sides” setup –

Here are the results

FL= Front Left, FC = Front Centre, FR = Front Right, RL = Rear Left, RC = Rear Centre, RR = Rear Right.

4.1. Results discussion

Well it’s not really a discussion – this is my blog so I’ll just put down what I think (and I’m the only one with moderator rights ). 🙂

This is my take on it. Firstly, all the temperatures in the Push-Pull Closed sided test were higher that the Push-Pull Open sided test, so I’ll just dismiss closed sided as an option (sorry I.A.S). Looking at the first (Push-Push) test we can see that there is a small difference between left and right with the left hand side of the front and the right hand side at the rear being the lowest temperatures overall. Given that the fans are mirrored, the front left heat sink is the same side of the fans as the rear right heat sink. So front left and rear right correlate well together, as do front right and rear left. Which means that there may be a slight bias in airflow from one side of the fans compared to the other. But more significant is the fact that the centre heat sinks on both front and rear are notably hotter than the outer heat sinks. This may be (as D.C. predicted) that there is a dead zone caused by the two opposing air streams. Or it may just be that there is a “hole” in the air stream caused by the boss that holds the fan blades (as I.A.S predicted). Or a combination of both. Moving on to the push-pull test, we can see that the temperatures are more evenly matched between front, left and centre but the heat sinks at the rear are running hotter than those closest to the air inlet (as I predicted). This is likely because the heat sinks at the rear are masked by those at the front so receive less air.

But overall, the highest temperature was 48.7 deg C at an ambient of about 20 and the hot block running at 160. This is well below the glass transition temperature of about 60 deg C for PLA so I’d be happy with that if I could be sure that it won’t get much hotter. Is it a linear relationship? Can I say that 140 degree increase in hot block temperature (160 – 20 deg ambient) leads to a 28 degree increase in heat sink temperature (48 – 20 deg ambient), in which case at 200 deg C, the heat sink will be at (28/140 x 180) 36 + ambient = 56 deg C? Not with any confidence I can’t. So I’ll have to test it once I get a more capable heater. And the other thing I have to bear in mind is that the filament itself will conduct some heat. Especially those filaments which may be loaded but not in use, so not moving forward.

On balance, I think I might end up needing more airflow. Quite whether to use push-pull or push-push, I’m still undecided. If I use different fans, with higher airflows, I’ll likely get completely different results in terms of variability between heat sinks, because the airflow pattern may change as well as the overall volume of air. I suspect the final design may need some sort of ducting, shroud, baffle or vane arrangement. I have lots of options here.

So with all that in mind, I’ve order a pair of 30mm fans with the highest flow rate that I can find. These ones were rated at 3.5 cfm (and 15 dBA) and the ones I’ve order are rated at 7.7 cfm (but 34 dBA). I can always use PWM to run them at a slower speed if I find the air flow is too high (unlikely). From these tests, I’m fairly confident that I’ll be able to get the required amount of cooling using these new fans. Also of note is that to cool the 5 colour Diamond, I need 27 cfm to cool the heat sinks sufficiently. So even if I double the air flow (to 7.7 x 2 =15.4), I’ll still be using about 55% of what I need to cool the Diamond 5 colour hot end. So I can say with a high degree of confidence that the Slice Engineering heat break/heat sinks are far more efficient that the modified E3D “lite” style as used on the Diamond hot end.

5. Design Modifications

I mentioned in my first blog post that I had to re-make the inlet block which takes the Bowden clips because I didn’t make the holes for the clips deep enough. Which is fine but what I should have done was to make the changes to my OpenScad design first. Instead, I just made a new plate with deeper holes without giving consideration to how this might impact on other aspects of the design (at my age, I should know better). 🙂

Here is what it should look like –

There is an 8mm diameter flat bottomed hole to take the Bowden clip. Then there is a countersink which should guide the filament into the top of the heat break tube as it exits the Bowden tube, to compensate for any slight misalignment (I don’t how good those inserts are at aligning the Bowden tube in the exact centre of the hole).

Here is what it actually looks like –

By making the flat bottom holes deeper, the heat break tubes stick up above the countersinks. So when I practised loading some filament, it had a tendency to catch on the edge of the heat break tube and had to be “jiggled about a bit”. It’s not a huge problem but none the less, it’s not as good as it should be. So I intend to re-make the side plates that this block bolts to but a few mm taller. This will pull the block up relative to the heat sink tubes, so that it looks like the first picture. I think that’s the easiest thing to do.

While I have the hot end partially dis-assembled, I’ll countersink the holes that take the bolts which hold the heat sink block onto the combining block. Then I can use countersunk screws instead of cap head screws. As I mentioned in my first post, this will make it easier to tighten the heat sinks.

Finally, I’ll move the holes for the 2mm threaded rods further up so that when the heater wires get clipped to one of those rods, they won’t be bent at such an acute angle.

6. Next Steps

I’ll do that machining while I’m waiting for fans which I’ve ordered from Digikey and which are coming from the USA, and the 80 watt heater that I’ve ordered from and which is coming from Denmark (nothing gets made in the UK any more – although I suspect that both the fans and the heater originate from China. …………)

I’ll also continue my quest to source a 60 Watt heater.

Breaking news……. As I’m about to publish this, I have received a quotation from a UK supplier for a 60 Watt heater cartridge. They will kindly waived their normal £125 minimum order charge which is good news. Instead, they just want to charge me £45.87 + £14.50 carriage + 20% VAT = £72.44! I’ll keep looking…………..


4 thoughts on “My 6 input mixing hot end V2 – part 2.

  1. I love following your design and thought process. Glad to see someone pushing the envelop instead of just accepting the products pushed out by big manufactures. This isn’t exactly rocket science. There’s no danger in experimenting. So its nice to see you give it a go. I guess my only question is why do you think you prefer a 60 watt heater cartridge? It’s PID controlled, so whats the harm in using an 80 watt or even more? On a second note, the heater cartridges are just resistors. If you can’t find a 30 mm. Maybe find two 15 mm or make room for two 20 mm. Then wire those in series. If running a 24 v system, then you can easily get two 12 V cartridges and run them in series. Or keep it simple and run two 24v ones in parallel. I also am an advocate for a copper material construction for the hot end, as its heat conductivity is twice that of aluminum. If you go copper then I suspect you would be fine with a single 20 mm cartridge as the heat moves so much better. Looking forward to a good youtube video about this some time! That is how I found you originally.


    1. Hiya,
      Thanks for the feedback. It’s always gratifying to find that the hours I spend at the keyboard documenting all this stuff are being appreciated.

      Ref the heater cartridge – the danger in using an overpowered heater is if a Mosfet on the controller board failed. I’m told that although failures are rare, if they do happen they are as likely to fail “closed” as they are “open”. So it’s possible that the heater could get stuck on full power. I know that when I tried an 80 Watt heater on the Diamond, after tuning the PID parameters, the firmware warned that in the event of a failure, the heater had the potential to reach about 660 deg C. That’s the melting point of aluminium! The other thing I found was that despite tuning the PID parameters, it was still difficult to control. It would over shoot a few degrees, then cool a bit then overshoot but not so much. Eventually, the oscillations would settle and the temperature would be stable but it took quite a few minutes. But then I’d have the same problem when the part cooling fan came on. I managed to tame it by limiting the maximum PWM value but then it just took longer to react to the part cooling fan which, even with the best ducting design, can still cause air to get deflected off the bed or off the object being printed, and cool the nozzle. And of course, if a Mosfet failed, I still had the potential fire hazard. So it’s generally best IMO to match the heater to the hot end but I’ve been thinking about incorporating a thermal fuse for a while as added protection. So I might do that in any case. There is always something more ………..


  2. Did you consider water cooling instead, E3D have water cooled hotends for cyclops and chimera.
    Love your work, I also want to replace my A10M 2in1 mixer with a 3in1 or more still deciding wheter mixer or switching, not keen on mixing if it always results in aqua fresh look.


    1. I have thought about water cooling but right now, due to the very low thermal mass of the Slice Engineering heat sinks, I can adequately cool them with nothing more than two very compact and almost silent fans. I could see some benefit if the hot was enclosed in a heated chamber, but right now, it isn’t. So water cooling would just add the complexity and cost of additional tubes, wires, pump, and heat exchanger for no gain.


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 )

Google photo

You are commenting using your Google 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