Categoria: English

Publicado em in Engenharia, English, Hardware, Impressão 3D, tecnologia, Vox Leone

3D Printing: I’m Making a 500°C Ceramic Hot End

How I intend to steal the Fire of the Gods and give it to the geeks.

Hotend TF-01 Linear comparado ao hotend standard Sethi 3D.
The unfinished prototype of the first model to be tested, the upcoming TF-01 Linear (the white piece), seen here alongside the standard metallic hotend from Sethi 3D, for dimension comparison. Image: Triforma CC-BY-SA-NC

Notice: This is a Portuguese language blog extraordinarily featuring an article in English.

As a 3D printing entrepreneur and enthusiast, I made the decision last year to dedicate a greater share of my time towards this technology, whose best days I believe are still ahead. My goal is to use 3D printing not just for creating prototypes, but for producing final products — like parts and components — for industrial and consumer use.

My aim is to utilize additive manufacturing to work with advanced materials such as nylon and carbon fiber. Achieving high temperatures above 300°C/573°F is crucial for this. Does this ambition make sense? With the increasing popularity and accessibility of new industrial-grade materials like PC, TPI, and PEEK, which are categorized as “engineering superplastics,” the demand for high-temperature capabilities is certainly on the rise. Therefore, I strongly believe that pursuing this direction does make sense.

However, despite the variety of brands, the crucial item for FFF 3D printing, the so-called hotend remains, despite advances, quite limited in its presentation and capabilities. The part consists of a chamber that is heated by a heating element, which is controlled in a closed loop using a thermistor for temperature feedback. The lower part of the set has a removable nozzle that deposits the material on the 3D printer’s printing plate. See Anatomy of a Hot End.

The hotend is typically constructed from steel and relies on Teflon linings or titanium+copper heatbreaks for passive heat control. It is essential for the filament to stay relatively cool until it reaches the heating block, preventing the heat from rising through the hotend body and excessively softening the filament. This phenomenon, known as heat creep, is addressed through the use of heat barriers.

Hot Ends

The standard hotend installed in my Sethi 3D FFF printer– a Brazilian 3D printer manufacturer — is an excellent product. It operates within the lower performance range, constructed from steel and Teflon-coated. This well-crafted and honest product fulfills its intended purposes exceptionally well, especially for printing with PLA and ABS filaments. However, like all other hotends in its category, its essential component is the internal Teflon lining through which the filament runs. Unfortunately, this Teflon core starts to deteriorate above the 240°C/470°F mark.

Impressora Sethi 3D.
Our beautified Sethi S3 on which the project was developed. From factory it has a more conservative look; a closed box. The addition of a transparent panel enhances print control and allows for greater flexibility in media production – Image: CC-BY-SA-NC

The other available choices consist of hotends crafted entirely from metal, also referred to as “all-metal” in the industry. These hotends employ a bi-metallic core, typically composed of titanium and copper, to act as a thermal barrier between the heating block and the heat sink. As a result, they are able to withstand higher temperatures and facilitate the printing of filaments designed for more sophisticated applications. Despite this, the temperatures they can achieve are still comparable to those of standard hotends such as the one manufactured by Sethi, and their primary purpose is to prevent the premature melting of the filament, the previously cited heat creep.

It’s somewhat disappointing to me to see that in 2024, the best the global industry can come up with is using titanium, a metal with a thermal conductivity index of 25 W/mK, as a thermal barrier. Surely, there has to be a better solution.

In practical terms, both hobbyist makers and occasional tech enthusiasts typically encounter limitations around the 350°C/660°F mark. Beyond this point, a shift to a different level of expertise may be necessary. However, is it conceivable to make temperatures exceeding 500°C/930°F, which are currently accessible only to high-level industry players, attainable for the average individual?

Rare Earths Everywhere

Over the past year, I’ve immersed myself in ceramic literature and marveled at the exquisite items crafted from a variety of transition metals and rare earth oxides. As I pondered the constrained selection of hotend options, I started to perceive a potential opportunity worth exploring.

Isn’t a hotend essentially just a glorified tube? If the folks at Sethi managed to create one using their expertise in metals(*), then surely someone like me, with newfound knowledge of advanced ceramics, could fashion a similar device. Why not harness these incredible materials for their inherent purpose, like regulating the heat of an extruder’s hot tip? How has this possibility been overlooked by the industry?

Ideas started pouring in from some hidden corner of my mind. I began visualizing a zirconia-white object. I called upon my expertise in CAD [OpenSCAD, actually] and settled in front of the workstation, accompanied by a monitor displaying the Periodic Table, to sketch my concept of a ceramic hotend suitable for the Sethi S3 3D printer – my own.

A few sleepless nights later, I had the blueprint of my petite yet fearless creation, boasting unconventional shapes for a hotend, a departure from the typical appearance of a conventional heatsink with its stacked metallic discs. With ceramics, heat conduction is significantly reduced, eliminating the need for a large heat dissipation area and offering newfound design flexibility. The material’s texture and color options — perhaps a pink hue for March? — introduce yet another layer of aesthetic appeal. Ultimately, the hotend can emerge as a beautiful piece.

It quickly became clear to me, within just a few days, why there were scarce offerings of ceramic objects in the market, apart from toiletries. This insight also sheds light on the intriguing and highly pertinent design of the initial hotend envisioned by Sethi in 2013. Traditional ceramic shaping techniques simply do not lend themselves to elaborate stylistic innovations or intricate anatomical details.

Until recently, advanced ceramists lacked the technological means to depart entirely from tradition, therefore they were limited to producing basic objects like rings and tubes using methods such as slip casting, pressurized injection, dry pressure molding, and other large-scale industrial techniques. When it comes to constructing a small component for a precision instrument like a 3D printer, ceramic doesn’t appear to be a practical choice.

Sol-gel Comes for the Rescue

Sol-Gel Chemistry, a process that I had studied extensively in the months prior to my insight and that I knew could potentially solve the challenge of working with ceramics in fine mechanics; enabling the creation of ceramic objects with significant structural complexity.

In simple terms, sol-gel chemistry involves the preparation of inorganic polymers or ceramics from a solution by initially converting liquid precursors into a ‘sol’ [derived from SOLution], and subsequently into a network structure referred to as a ‘gel’.

The formation of a sol occurs through the hydrolysis/condensation of particles, and can be broadly defined as a colloidal suspension, covering a wide range of systems. According to the International Union of Pure and Applied Chemistry (IUPAC), a colloidal system is a dispersion of one phase into another where, “the molecules or polymolecular particles dispersed in a medium have, at least in one direction, a dimension between 1 nm and 1 μm”. Well-known examples of colloids include milk and mayonnaise.

Extensive scientific literature has explored the sol-gel process and its applications in advanced ceramics – for further information, please refer to the recommended bibliography. If you recall the ceramic covering bricks of the space shuttle, they are produced using processes similar to these.

Gel Casting

Ceramic shaping techniques are categorized as either dry or wet forming processes. Gel molding, also known as gel casting, falls under the wet category. The wet forming technology we employ, developed by Oak Ridge National Laboratory (ORNL), is capable of producing high-density ceramics with complex shapes, approaching the final product’s shape [‘near net shape’]. This method offers several advantages, including a short molding time, no material restrictions for the mold, high resistance while unsintered (referred to as “green”), and the ability to apply sections of varying thickness.

Molds used for developing the prototypes – Image: Triforma – CC-BY-SA-NC

It’s important to note that the gel molding method allows for the use of virtually any material to shape the ceramic masses, opening up a new and extensive opportunity for the utilization of 3D printers. No longer limited to prototyping, they become active and highly productive assets for manufacturing advanced ceramics. This is due to their capacity to directly and affordably produce complex molds on-site.

This application encompasses FFF printers as well. In gel casting, the distinct layered marks of the molded impression are faithfully transferred to the molded parts. Personally, I plan to incorporate this ‘imperfection’ as the visual identity of the product, revealing its origin. Others may understandably prefer flawlessly precise and uniform molds, whether they are printed in SLA, SLS, or machined in stainless steel.

Overall, leveraging gel molding represents a significant leap in productivity and profitability for any FFF printer, as an ABS mold that it takes 2 hours to deposit can be used to replicate thousands of high-performance ceramic parts in a short timeframe.

This Project

In our ongoing process, we begin by preparing water-based gels. This involves dispersing ceramic powder (including zirconia and yttria) in water. Following this, we introduce gelling agents, such as monomers and initiators, and carefully mix them to create a colloidal suspension, as mentioned earlier. This mixture is then poured into an ABS plastic mold, which has been created using 3D printing, and left to dry, forming a green body. After this stage, the green body is removed from the mold, undergoes additional drying, and is subject to high temperature sintering. Once sintered, the material goes through a final treatment involving machining and partial enameling.

Gelcasting.
The solution is shaped within the plastic molds. The mold in the background is already closed, undergoing the gel curing/drying process. – Image: Triforma – CC-BY-SA-NC

Until recently, acrylamide (AM) was commonly used as a gelling agent. However, due to the neurotoxicity of AM, gel molding was unable to be employed on a large scale in the industry. This hampered the development of this technique and is also, consequently, partly responsible for the relative scarcity of complex ceramic objects on the market. In this project, we use a gel shaping process based on natural polymers like agarose, which removes impediments to taking advantage of the technology.

It’s quite poetic that a technology as ancient as ceramics can still be the most advanced in terms of materials. Although I may make it sound simple, the effort required months of theoretical and practical learning.

In the end, we have an innovative product, a small thermal element weighing just 10 grams. Its seemingly delicate structure is capable of withstanding temperatures above 2300°C/4300°F and transmitting just 1.7 W/mK of the heat it encounters. For comparison, titanium can withstand temperatures of this magnitude but has a much higher thermal conductivity of 25 W/mK, while Teflon is a great insulator, transmitting only 0.2 W/mK of heat, but it cannot withstand temperatures above a mere 240°C/470°F. The low thermal conductivity of the zirconia-yttrium system theoretically can mitigate heat creep and should make the hotend capable of printing any type of filament.

As additional advantages, the process uses less energy, generates no noise at any stage of production, and does not produce significant effluents. And did I mention that the unit weighs just 10 grams?

For the second phase of the project, it’s worth incorporating a zirconia-magnesia heatbreak into the ceramic structure to further reinforce the set. We will be able to reach as far as the hot block can handle. I read about a Japanese group working on a hotend suitable for ~850°C/~1600°F. Extruding a copper wire with my Sethi desktop printer now seems like a completely achievable goal.

Um heatbreak cerâmico junto ao seu molde.
A ceramic heatbreak — and its small mold — designed to withstand temperatures of 1600°F. Image: Triforma – CC-BY-SA-NC

But this is where the problems begin.

Tests

At this moment, I find myself surrounded by recently acquired laboratory equipment such as beakers, Erlenmeyer flasks, scales, a magnetic stirrer, and a variety of substances. What was intended to be the starting point of a small-scale industrial endeavor might, depending on the observer, resemble a room within a pharmaceutical facility or a rapid analysis laboratory at customs. I am on the verge of commencing a pivotal experiment: ensuring that the process outlined above consistently functions for mass production.

In addition to the formal process and the few prototypes obtained with individual molds, it is imperative to conduct a substantial number of tests, including those related to chemical composition/formulation and destructive mechanical, thermal, and performance assessments.

If we succeed in producing a reasonable quantity at a reasonably low cost, which I firmly believe is entirely achievable, I intend to make the product available for direct sale on the Triforma website – launching soon. The goal is to replicate the business model of Slice Engineering, with OpenSource licensing. I am determined to have a minimally viable product by mid-April and will be working tirelessly towards this objective.

I will share the outcomes of this project here. Stay tuned for updates.

(*) While I have made several references to Sethi 3D in this text, I want to clarify that my only association with this esteemed company from Campinas, Brazil, is that of a [satisfied] customer. However, it goes without saying that I would be thrilled to have some connections. I hold great respect for all industrial initiatives, particularly those related to advanced technology and those situated in Brazil, which elicit not only respect but also profound admiration.

Recommended Reading:

Sol-gel

https://en.wikipedia.org/wiki/Sol-gel_process

Sol-Gel Chemistry and Methods

Link to PDF

Gel Casting of Ceramic Bodies

https://onlinelibrary.wiley.com/doi/abs/10.1002/9781118176665.ch6

The evolution of ‘sol–gel’ chemistry as a technique for materials synthesis

https://pubs.rsc.org/en/content/articlehtml/2016/mh/c5mh00260e

Thermal Properties of Ceramics

Link to PDF

A review on aqueous gelcasting: A versatile and low-toxic technique to shape ceramics

https://www.sciencedirect.com/science/article/abs/pii/S0272884218334606

Publicado em in English

CV: Labels are Metadata – let’s treat them as such

TL;DR: This article questions the current practices in image labeling for computer vision and proposes storing image annotations as EXIF metadata, eliminating the sidecar text file.

Zebu cows in a pasture
One of the images in Caprichoso, our zebu cattle dataset (soon to be released) – Image: Cownt CC BY-NC-SA 4.0 Deed

So, I want to get to down to work to understand firsthand what it would be like to deal with annotation data in this paradigm: the image states a truth about itself [EXIF], rather than being described by another entity [JSON]. That seems a pretty reasonable proposition.

The general idea is to experiment with simplifying the dataset file system — by eliminating the sidecar text files, and check if there are any important gains that justify adopting this approach, at least for small and/or proprietary datasets and fine-tuning tasks. Sidecar files by definition store data (often metadata) which is not supported by the format of a source file. That obviously isn’t true with modern day digital image files.

I’m also trying to understand the technical and conceptual – and why not say, ethical – problems related to inserting/writing/reading data in these structures/environments as well as check if there is anything to gain in the training process, least for small datasets and/or fine-tuning tasks.

It goes in the EXIF

Annotation is a fundamental step in Machine Learning. For the casual reader, computer vision annotations are stored in text files connected, or paired, with the correspondent image files within the dataset, e.g. cow0001.jpg → cow0001.json. These files are created and written by the annotation tool in a work session, as the annotator selects the region of the image containing the item to be labeled. They contain the coordinates that allow the AI system to superimpose the “bounding boxes” – those already familiar ‘squares’ that delimit the target detection items on the image – structured in a certain text format.

Some frameworks use json, others xml, and others csv [there will definitely be other formats]. Some require the presence of the annotated text file in the same directory [same level] as the image. Other frameworks require the text file to be located in its own directory within the image directory, which may [must] be called json, xml, label, and a few other requirements. This text file lives in an indissoluble marriage with the image file, and for computer vision purposes they are always referenced together. Is it worth the hassle?

A common argument is that this image/text separation scheme allows greater flexibility in annotations, because they are atomized, etc. But I counter the argument that nothing is very different when both are metadata. In a proper label tag, the text data remains compartmentalized and manipulating them will be no more difficult than manipulating a text file. It is still perfectly possible to keep tag content synchronized with text files kept outside the dataset. The dataset no longer needs a file system.

XMP comes into play

According to Wikipedia “The XMP standard was designed to be extensible, allowing users to add their own custom types of metadata.”

In a perfect world, such user-defined tag would have its own data type. For this exercise we will use the natural inclination that tags have for dealing with strings. We then need to create a tag [or rename an already defined one – there are many] to contain our label; our own EXIF tag[0].

Doing away with the text files

So let’s get rid of the text file and write our labels as an EXIF tag of the image file. There a few modules available in Python for this task, but little diversity. Many are out of date. A search of the Anaconda (conda, conda-forge) and PyPi (pip) channels consistently returns the three modules I plan t test for this project: pyexiv2; piexif and PyExifTool. The latter is a Python wrapper for ExitFool, which is a Pearl application. We will not detail the peculiarities of each one here. I’m also exploring PIL .

With ExifTool it is possible to perform advanced manipulations on tags. Let’s use it to create [or rename an already defined one] a new tag called ‘Label‘:

The process involves editing the exif.config file containing the tags we want to define, as stipulated in the documentation.

# The %Image::ExifTool::UserDefined hash defines new tags to be added
# to existing tables.
%Image::ExifTool::UserDefined = (
    # All EXIF tags are added to the Main table, and WriteGroup is used to
    # specify where the tag is written (default is ExifIFD if not specified):
    'Image::ExifTool::Exif::Main' => {
        # Example 1.  EXIF:NewEXIFTag
        0xd000 => {
            Name => 'Label',
            Writable => 'int16u',
            WriteGroup => 'IFD0',
        },
        # add more user-defined EXIF tags here...
    },

What I want is

def writeToEXIFtag (data)
    #some pseudocode for now
    Image.Exif.Label = data

Instead of

def writeToJSONFile(path, fileName, data):
    fileName = fileName.split(".")[0]
    filePathNameWExt =  path + '/' + fileName + '.json'
    with open(filePathNameWExt, 'w') as fp:
        json.dump(data, fp)

as in what has become [by custom] the canonical process.

In this project, for greater practicality [integration with other modules, etc.], the best way to go seems to be using virtual environments, such as virtualenv and conda. It is hardly possible to bring together the exact same packages on both platforms. At the moment I’m using environments I configured with modules that I put together through the not very clean practice of mixing conda+pip. I still have things to figure out – don’t know much about Pearl and I’m having a hard time having everything [namely exiftool + pyexiftool] working together.

Pros

  • More efficient processing [to be verified].
  • Dataset image files can be renamed and can be used in any other dataset without additional work.
  • No problems with different formats. These Xlabels [EXIF Labels] can coexist with the paired annotation files.
  • Simplicity brings pedagogic gains; a less steep [human] learning curve.
  • Cameras can automatically pre-annotate images – at least universal categories, like COCO’s [is that really a ‘Pro’?].

Cons

  • Increased dataset size [to be verified]
  • Less control over datasets and annotations [to be verified]
  • The usual surveillance stuff [cameras detecting, identifying, classifying…]
  • <Insert your con here>

Epilogue

I would really like to know if there is any proposal similar to this one out there, as I’d also like to know if I’m actually arriving late to an already rejected solution. I’m still in the very early stages and receiving any feedback is an integral part of the process.

I hope the idea is worthy of receiving inputs from the readers. I will be reporting any progress. I have the skeleton of the repository on GitHub[1], and will be polishing and finalizing the initial version in the next few days. It’s a modest project, as the idea is really simple as everyone can probably see. Everyone’s invited to participate.

* * *

[0] The question of creating new tags, or renaming existing ones [e.g. UserComment → Label], or both, or yet another option with another datatype, is open at this moment, as is the question of whether to use single or combined tags.

[1] https://github.com/VoxLeone/XLabel