A Twist in the Flow

This is the true story of the TFN—Twist Flow Nozzle design

It all started with a debate in a Facebook group about the physical principle that allows the CHT nozzle to achieve higher flow. As is often the case, some points were spot on, while others missed the point.

If you’ve seen my earlier work, you know I’ve done extensive thermal simulations and optimization for electronics cooling. The same fundamental principles of heat transfer apply whether you’re cooling electronics or melting filament. It all comes down to thermal conductivity/resistance, the 3D shape of the materials, and the surface contact areas.

Some physical principles

Let’s begin with some behind-the-scenes physics. To simplify understanding, we’ll consider a basic linearized model consisting of four components: a heating element, a metal heat block, a metal nozzle, and a plastic material (filament), as shown in the illustration.

The thermal resistance of the heater element can be neglected since the thermal sensor used in the temperature control loop is usually attached to the heater block; it mainly affects the PID control loop dynamic performance.

The total thermal resistance between the heater element and the bottom side of the plastic can be expressed as a sum of each material's thermal resistance as follows:

· R_total = total thermal resistance, resistance to heat flow (K/W)

· L_block, L_nozzle,L_plastic = material thickness (m)

· k_block, k_nozzle,k_plastic = thermal conductivity of the material (W/m*k)

· A_block, A_nozzle,A_plastic = cross section area perpendicular to the heat flow (m²)

*Thermal interface resistance between the different materials is not considered

The heat transfer rate q:

Our target is the highest possible heat transfer between the heater element and the plastic. This means we must have the lowest possible thermal resistance (highest thermal conductance). Note that to maintain a small temperature difference between the heater (user-set temperature) and the melted plastic (actual plastic temperature), the total thermal resistance must be further reduced to achieve the same heat transfer into the material.

Let's do an exercise:

· L_block = 5mm, K_block =~385 W/(m·K) (C110 copper)

· L_nozzle = 2mm, K_nozzle =~120 W/(m·K) (brass)

· L_plastic = 0.9mm, K_plastic =~0.2 W/(m·K) (ABS plastic)

· A_block = A_nozzle =A_plastic = 1cm² – for simplification

Using the above formulas, we can calculate each material's thermal resistance:

We can conclude that the greatest thermal resistance in the entire heat transfer path is due to the plastic we aim to
melt. The nozzle and heat block materials are not the primary contributors. While using a heat block or nozzle material with better thermal conductivity can improve results slightly, it is not the dominant factor. Even when considering a plastic thickness of just 0.9 mm (about half the diameter of a 1.75 mm filament), the low thermal conductivity of the plastic itself remains the main limiting factor for melting efficiency

For sure, in reality, the components are round or cylindrical rather than simple rectangular prisms. So, for those interested, here’s a formula that more accurately represents the thermal resistance of an actual hotend design:

Now back to the debate: why a CHT nozzle has higher flow.

1 – It divides the filament into three strands, thereby reducing the thickness of the heated plastic. Each of the three holes is approximately 1 mm in diameter, compared to 2 mm for a typical V6 internal hole.

2 – The three-hole design expands the contact surface area between the nozzle and the plastic. As a result, the nozzle has lower thermal resistance towards the core of the plastic, increasing its melting capacity and consequently the hotend maximum volumetric flow. I have to admit, Carl Beck's idea is simple and brilliant.

The Spark of the idea

This debate started me thinking if there would be another way to increase the volumetric flow. Let’s analyze what we can
change in a standard V6 nozzle to increase flow (decrease thermal resistance). The goal is to maximize volumetric flow while minimizing the melt zone length and the amount of melted plastic inside the nozzle.

Reduce the heat-block and nozzle material thermal resistance or make them thinner, an idea already done, and as proved previously, the thermal resistance of the metal parts is not the bottleneck here.

Make the melting zone longer, therefore increasing the contact surface area—like the goliath design, it works, but it’s not in my personal preference because the larger amount of melted material tends to leak out more easily due to gravity

One of my thoughts was if the CHT can slice the semi-melted plastic (my idea should not slice/split plastic to avoid infringing Carl Beck’s patent)—maybe ... I can deform it in a way to reduce the plastic thickness and perhaps increase the contact surface area.

So, I have come up with a very simple design idea: just a simple rectangular slot in the nozzle, which flattens out the plastic.

Obviously, a rectangular slot cannot be drilled directly into the nozzle. So, I started with a standard V6 nozzle, enlarged its inner hole to 3.5 mm in diameter, and machined two identical inserts from a 3.5 mm thick copper rod to create the rectangular slot.

To my surprise the idea worked well, and the flow increased compared to a normal V6 nozzle but was nowhere near the
Bontech CHT performance. But at least I learned something is possible.

Tinkering further, I thought maybe instead of one rectangular slot, I could make two in perpendicular directions, ending up with the cross-shaped section; this would increase the contact surface area.

To my surprise, this actually was performing worse than the first version with only one slot.

Well obviously, this has increased contact surface area, but the distance toward the filament center is increased, thus
overall having a higher thermal resistance and consequently worse melting performance.

After that, I experimented with several variations, adjusting slit positions, lengths, and channel shapes.

I’ve achieved some positive results, but they still fall short of the CHT performance level.

I decided to take a different approach because machining and testing each idea was very time-consuming. Creating small slots between 0.8 and 1 mm proved quite tricky, and achieving precise dimensions was a significant challenge. This made me question whether the results truly reflected the design concept or if they were influenced by the machining
tolerances of my homemade CNC machine, which may have affected the performance outcome.

With support from my good friend George C., who is an expert when it comes to thermal simulations, I turned to simulation.

Basically, I developed a simple thermal simulation model grounded on a simplified model of the Smart Orbiter 3 hotend. Material is introduced at the hotend’s top at a constant speed of 10 mm/s with the heater element set to 250°C.

From the thermal simulation result, I noted down the filament temperature at the nozzle tip center.

The winning shape is the one that, under the given conditions, delivers the hottest plastic to the nozzle tip. The resulting absolute value is solely for performance comparison.

The simulation does not account for the nozzle backpressure and the complex flow dynamics of the plastic material with phase change, but at least to test out and compare different internal shapes, it was good enough.

As first simulation trials, I run the shapes I already tested, a standard, V6, Bondtech CHT, and the two slot profiles. Simulation results confirmed my measurements. The cross-slot profile was worse than a normal single rectangular slot. Better than a standard V6 but worse than Bondtech CHC.

After simulating about fifty-plus different shapes and variants and hundreds of simulation times, I reached a point when I had no more ideas that gave a better result. In simulation I reached way better performance than the CHT design. So, it was time to test some of the most promising variants in real life.

The most promising performance I got was with an internal channel in the shape of a twisting spiral.

The twisting spiral shape increased the contact surface area with the filament and also created a mixing effect between
hot and cold filament, further increasing its melting capacity.

After trying out several versions, I gave the most promising variant to LDO for manufacturing. Here you can see some of
the performance measurement results and some of the test subjects:

The team at LDO introduced the idea of applying an RNC coating to the nozzle, providing enhanced resistance to abrasive filaments.

One puzzling observation from the tests I’ve done when comparing the Twist Flow Nozzle to the Bondtech CHT.

When the flow is high, the plastic exiting the CHT nozzle tends to curl up, whereas with the TFN, that is not the case. I always have seen this phenomenon when I tested the orbiter extruder designs; all of my hotends with CHT nozzles did this phenomenon. Initially I thought this is how plastic behaves under high pressure when it comes out from a very small hole.

Strangely the TFN nozzle has similar performance to CHT but does not exhibit this phenomenon—obviously this means that this is not the plastic behaving weirdly. You can observe this effect in my nozzle comparison video at approximately 7:11.

A brand new super high flow nozzle concept, the Twist Flow Nozzle.

Looking into the details of the CHT design, I think I can explain this phenomenon. The CHT nozzle has three ~1 mm holes at the filament entry side. The area occupied by this surface is bigger than the filament diameter. Also, the tube inside the hotend leading the filament to the nozzle is usually 2 mm in diameter. This means the filament can slightly move sideways, causing the filament to split unevenly.

As a result, in one hole more plastic will be squeezed in compared to the other two. This means that the plastic flow and the pressure of the 3 holes will not be equal, and when they are mixed together near the tip, the hole with higher pressure will create a side force pushing the filament sideways, thus causing the curling-up effect we see.

The picture illustrates the case when the filament is shifted towards the red hole, causing higher pressure (P2) to build up in that hole compared to the blue-marked hole, with a smaller resulting pressure P1 < P2. This causes a force that pushes the filament sideways when exiting the nozzle.

Definitely this means that the layer is not deposited where it should exactly be because the filament is poured not vertically but with a slight angle.

The Legal debate

Given the performance results we’ve achieved, we anticipate strong interest in this design and its future variants. Naturally, this also means a new battle with the cloning industry.

Although I am the author and co-author of several patents, I’m not a great fan of the patenting system; however, it ultimately safeguards the efforts of both inventors and investors in employing a new technology.

Together with LDO, we decided to safeguard our invention against cloning and file for a patent in China for the basics of
this invention - CN120134620A.