Design Upgrades, Reliability Testing and Implementation of Engineering Grade Thermoplastics in Prusa MMU2s

(1)

Master Thesis

Design Upgrades, Reliability Testing and

Implementation of Engineering Grade

Thermoplastics in Prusa MMU2s

Ajith Kannoth

Jönköping University School of Engineering

(2)

This thesis has been carried out at the School of Engineering, Jönköping University, in the subject area of Product Development and Materials Engineering. The authors are responsible for the presented opinions, conclusions and results.

Examiner: Supervisor: Scope: Date:

Dr. Mirza Cenanovic Lic. David Samvin

30 hp (second-cycle evaluation) 2020-06-04

(3)

thing that everyone nds, and I am truly thankful for being given this opportunity to work related to a topic of 3D printing.

This thesis work would not have been possible if not for the continuous support, guid-ance and mentor-ship of both my supervisor and examiner Lic. David Samvin and Dr. Mirza Cenanovic respectively. I sincerely thank them for providing me with all the nec-essary resources and tools for conducting the required project work. The exchange of ideas, discussions and constant feedback which I received from them helped me gain a new insight in approaching a problem.

I am also grateful for the support of my friends and family, especially Maria and Manu for constantly urging me to give my best in what I do and thereby help me develop both personally and professionally.

(4)

printers in possession of JTH and how to resolve them; to be able to get a reliable print outputs from engineering grade materials apart from conventional materials like PLA and PETG. The second aspect being the implementation of multi material module 2.0S, hereafter referred to as MMU2s successfully by analyzing and testing the current modi-cations and upgrades currently in the community and suggest any further modimodi-cations, if required, both in terms of hardware and software which is further discussed in the upcoming sections. At present, there are numerous design upgrades and modications over the stock parts in the community which claim to iron out the reliability issues of the multi material unit. But, the success rates of these modications and upgrades vary widely. We tend to look at some of these modications which helps in eliminating the issues associated with the unit while getting it to produce results in a consistent and reli-able manner. The engineering grade thermoplastics which the university plan to use were also taken into account to implement in the printers once the MMU2s setup was tested for reliability. The objective also to create a successful prole sets by tweaking various parameters in the slicing software for the aforementioned engineering grade materials so that a ready-to-print prole is available for the corresponding material.

During the course of project work, the reliability of the multi material unit was in-creased by upgrading few of the components such as idler barrel and selector. Fine tuning of software parameters led to the error free running of the MMU unit by which exten-sive testing was possible. Furthermore, engineering grade thermoplastics was able to be tested and implemented on the current setup by making use of these software and hard-ware changes. Finally, extensive testing of the multi material unit was done coupled with engineering grade thermoplastics which yielded successful results and the conguration settings saved for future use in the university.

(5)

1 Introduction

1.1 Background

Prusa i3 Mk3s is the third generation of rep-rap style printers from Prusa Research based in Czech Republic. The printers are currently the best in the class Fused Deposition Modeling printers, having sold over 200,000 units since their release. Even though these printers work right out of the box for conventional materials like PLA and PETG, the materials in focus here namely Carbon ber infused Nylon (AduraX), PrimaSelect Ny-lonPower Glass ber and Primaselect Polycarbonate requires tweaking in terms of both hardware and software which is not implemented currently and hence makes printing these advanced materials impossible.

The MMU itself has many reliability issues in terms of design, usability and reliability for which there are already few modication in the community which would be imple-mented and tested for their eectiveness. It is important to acknowledge the problems and try to resolve it since the current brand line up of 3D printers in JTH, namely Maker-bots are far inferior in terms of print quality and reliability and hence solving these issues would enable printing of engineering grade materials and combination of multi-materials with ease.

1.1.1 Prusa Research

Established in February 2012, Prusa Research based in the Czech Republic is one of the prominent leaders in 3d printers having shipped over 6000 printers all over the world. The man behind it, Josef Prusa began by building a model called Prusa Mendel.

The popularity of this model led to two mode iterations being released; namely Mendel I2 and I3 respectively. The original Prusa I3 was released in August 2015 and its up-grade MK2 released in May 2016. Traditionally, at that time, the concept of MMU was relatively unknown or not available in the market. In May 2017, Multi Material Upgrade 1.0 was released which enabled users to print up to 4 materials at the same time. The immense popularity of original I3 and Mark 2 led to the unveiling of Prusa I3 MK3 in September-2017, their most successful model yet. The upgraded version of MMU 1.0; namely MMU 2.0 was released and started shipping on 08-2018. Prusa Research remains the fastest growing tech company in Central Europe according to the 2018 reports in Deloitte with a growth rate of 17,118% over the last four years. As stated earlier, the reason of this thesis work being mainly focused on this specic brand are mainly because of the price-to-performance ratio, the quality of prints and sturdy design and electronics. 1.1.2 Multi Material Upgrade 2.0S

(7)

also has an integrated blade to cut the lament o should anything go wrong. The key features that dier from the MMU 1.0 is number of materials that can be printed has increased from 4 to 5. Also, the 4 bowden tubes coming has been reduced to one PTFE tube coming from MMU 2.0 which directly goes to the extruder. The working principle of the unit is detailed further in the upcoming sections.

JTH currently has two Prusa i3MK3s with MMU 2.0S which are prone to jamming thus deeming them unreliable. The printer itself has some rework to be done given its current state of extruder which may require reassembly or even redesign thus making it reliable.

1.2 Problem Description

JTH's focus towards polymer science causes a need for a more reliable and consistent FDM 3D printer that can generate components consisting of several materials and also components which are multi colored.

Currently JTH has numerous Maker-bot Replicator, a Raise 3D and two Prusa i3 MK3s with MMU module which are used for various projects and applications. Comparatively, Prusas are of superior quality and oers a plethora of sensors in aiding to a successful print while maintaining the design to be open source unlike the other brands in possession of the university which makes upgrading and redesign possible. Although these machines are predominantly used for printing PLA and PETG, to keep up the current pace of research at the university, we need to design and print components which are stronger in critical areas while still using cheaper material in less critical areas. Printing with versatile materials like Nylon, carbon reinforced nylon, poly carbonate, PVA (water soluble) and poly-carbonate is currently impossible at university with the current setup mainly because of the reasons related to the printer setup, design restrictions and process parameters. This project work aims to look at the Prusa i3 Mk3s at the university along with the Multi Material Upgrade Kit in achieving the versatility in terms of choice of materials, colors as well as stability, consistency and repeatability.

1.2.1 Current setup, its limitations and possible work (Problem statement)

Currently, the only possible type of materials which yields successful prints out of the Prusa i3 MK3s are PLA and PETG. Moreover, out of the two printers, only one is in working condition. Hence, the primary focus of this thesis work would be divided into two main aspects; the printer itself and the MMU.

In terms of printer, the main aim would be to get the printer working, setting a benchmark for each of the materials in focus, suggest necessary design/parameter changes if applicable and develop a robust testing method where consistency and repeatability are ensured by reducing the failure rate of prints. The parameters, to name a few, such as feed-rate, extruder temperature, bed temperature, retraction setting, nozzle size, type and heat-break material since all plays a signicant role in getting a successful print and tends to dier from one material to the other.

(8)

The MMU in possession of the JTH currently has problems related to jamming and re-traction when printing with various materials including water soluble materials. Various knowledge bases including the Prusa community forum has reported problems ever since MMU 2.0 has launched where it has worked awlessly for some whereas it has been a problem for the others pertaining to the issues related to IR sensor lament auto-loading and unloading to the unit. The key issue needs to be identied as of what causes this inconsistency and what remedial steps needs to be taken in-order to rectify the same. Even though there are numerous solutions in terms of modications or upgrades in the 3D printing community, the reliability of the upgrades or their eectiveness vary from one machine to another. We reckon to face these reliability issues by upgrading and introducing modications which suit the multi material unit in such a manner so as to make it functional and reliable.

2 Purpose and Research Questions

The problems stated above leads us to formulating the research objectives and purpose of the work conducted in which we aim to answer the following questions.

How can the printer be redesigned/upgraded in its current setup so as to enable the support of printing materials of versatile nature other than generic PLA and PETG.

How does the various parameters like printing speed, temperature, feed-rate and retraction vary and aect for each material.

How can the MMU 2.0 be redesigned so as to get a robust and consistent print output, printing generic materials while also implementing engineering grade ther-moplastics.

How eective are the current mods and redesigns available in the community and if any further redesign/suggestions are possible existing solutions.

Limitations

The following can be projected out as the limitations of the proposed thesis work The topic is focused on just one brand of printer. Each printer has dierent

perfor-mance parameters and numerous variables which dier hence the data and results obtained from this work may not be applicable to machines of other brands. Software aspect of the machine would remain unchanged. Other than the slicer

(9)

3 Theoretical Background

3.1 FDM Technology: Timeline and its history

Additive manufacturing or 3D printing is a production process by which an object is produced in an additive fashion, layer by layer. A digital 3D model, made using the CAD software or even by scanning , is sliced into individual layers, which then supply the tool path code for a 3D printing machine. Depending on the exact technology implemented, the machines uses a specic process to recreate the model in the physical world from the base slice to the top, until the object is complete [21].

While Additive manufacturing has been in existence in one form or the other since 1981, Hideo Kodama at Nagoya Municipal Industrial Research Institute was the rst person to describe a layer by layer approach for manufacturing which is considered as a predecessor of SLA(Stereolithography); Chuck Hull is widely regarded as the inventor of the technology when in 1987, commercialized his rm's (3D systems) rst printer SLA1 (Figure 1)[22] [5]which utilized a UV laser to cure photosensitive resin layer-by-layer so as to recreate the CAD model.

Figure 1: SLA-1

The company also introduced Stereolithography(STL) le format which is used even today in proprietary software supplied with the 3D printers to generate g-code. The process of rapid prototyping is divided into three main categories based on the raw material used which are liquid-based, powder based and solid-based according to the material used in the process[19]. Essentially, the process of rapid prototyping has ve stages which are pre-processing, machine processing and post processing[2.1] (Figure 2).

(10)

Figure 2: Five stages of Rapid Prototyping

Fused Deposition Modeling or FDM is a type of additive manufacturing process which falls under the category of solid based rapid prototyping which has been widely accepted form of prototyping because of its simplicity and cost eectiveness. This has been,in fact adapted at JTH in terms of numerous brands of machinery and forms the basis of the project-work conducted.

The technology of 3D printing, although prevalent since early 1980's, it wasn't until 1989 when Scott Crump founded the company namely Stratasys and led a patent for a technology called Fused Deposition Modeling thereby commercializing this method of prototyping. While the technology continued to develop where more companies emerged developing new methods of RP like SLS, binder jetting and color printing emerged.

(11)

Figure 3: First FDM patent led by Scott Crump(1981)

While the technology started to evolve rapidly, in 2005, Dr. Adrian Bowyer at Univer-sity of Bath inspired the concept of Rep-rap (short for replicating rapid prototyper)[6]. This open source concept was based on the philosophy of 3D printing the parts of your own printer and assembling it yourself. Come 2008, Darwin, the rst Rep-rap style printer was released by Dr. Bowyer and his team which lead to immense popularity of desktop styled 3D printers which was easy to build, assemble and operate at a time where room sized printers were the only option available. Moreover, the patent for Fused Deposition Modeling Technology owned by Stratasys expired in 2009, which led to many hobbyists and smaller organizations replicating their own versions of FDM technology which led to the variety of choice of machines there is today in the market[6].

(12)

Figure 4: Dr.Bowyer with the Darwin

FDM is essentially a process based on extrusion where a thermoplastic, primarily in the form of spools, is heated to its melting temperature and deposited onto a substrate, usually a heated bed platform of the 3D printer (Figure 5)[2.1].

The reason for FDM being so popular are the lower costs and user friendliness of this technique. Compared to other additive manufacturing techniques , FDM is a lot cheaper. When compared to the powder used in SLS and liquid in SLA, the cost of spools of lament used in FDM technology are much cheaper. For this reason, FDM is more appealing to both companies and individual users. The range of choice of materi-als is another appealing factor for FDM. Printing with multitude of materimateri-als or color simultaneously is also possible with FDM which proves it to be another advantage.

(13)

Figure 5: FDM process: (1) Heated extruder, 2 (molten thermoplastic), 3 (Substrate to which the molten thermoplastic is deposited on)

The two main design variations of these machines available in the market today are Cartesian and Delta models (Figure 6)[7].

Cartesian

Pros Cons

Better surface nish due to rigid axes. Increased weight due to more moving parts. Huge user base & community support. Space constraint/Small build volume. Easy to operate & maintain. Not t to print tall objects.

(14)

Delta

Pros Cons

Excels in printing in height. Bowden style drive limits choice of materials. Higher print speed. Increased speed results in lesser precision.

Lighter in weight. Harder to print objects in horizontal direction. Table 2: Pros and cons of Delta format FDM printers

Figure 6: Cartesian and Delta type of FDM printers[7]

The main dierences in design lie in the movement axes; the Cartesian has a moving bed ,usually with the help of belts driven by a stepper motor and an idler at the other end whereas the Delta conguration has a stationary bed while the extruder head movies in a non-linear fashion when compared to the Cartesian version. The machine in focus with which this project work was conducted is a Cartesian version.

(15)

3.1.1 Other formats CoreXY printers

Even though a type of Cartesian style printers,the main distinguishable fact of coreXY printers are their unique movement mechanism. The lateral movement of these machines are quite complex and is driven by two long timing belts. But their advantage is that they can handle much faster printing speeds, the reason being less number of moving parts which are also lighter. Also, the vertical z-movement is done entirely by the build plate and always in a downward direction. The timing belts can be a real issue over time as slacking and fraying can lead to print quality and other issues such as misalignment.

Figure 7: Schematic of a coreXY machine Scara 3D printers

Scara stands for Selective Compliance Articulated Robot Arm. It usually has a robotic arm which performs the motion of a printing. This printer can much likely be compared to robotic welding machine or a pick and place robot; main advantage being the exibility of the machine. These type of machines are mainly used in and industrial scale for construction and similar large scale projects.

(16)

Figure 8: A SCARA 3D printer in action[2.1] Polar Printers

Polar printers combine the two movements;namely circular movement of bed and linear movement of the head in order to execute a print. They utilize their build area to a maximum since their bed rotates when compared to cartesian or delta printer. Polar printers, usually has only two motors to power each axis, since it makes use of angle and length when compared to cartesian or delta which typically requires three.

(17)

Figure 9: A Polar format printer[2.1] 3.1.2 Machine in Focus and decision of choice

Prusa i3 MK3s was released as a successor to its previous models namely mark3, mark 2 and mark 1 respectively. Based on Adrian Bowyer's international open source Reprap project Prusa3D has become one of the leading organizations in producing aordable machines with qualitative results.

(18)

Figure 10: Prusa i3 Mk3s

This leading edge is achieved by having a plethora of sensors and features which are explained in short below:

Direct Drive extrusion system:

A major advantage of Prusa i3 Mk3s is the direct drive extrusion system where the extruder motor forms a part of the hot-end (Figure 11). Since the extruder motors are mounted above the hot end, the distance of travel of lament from extruder to hot-end is minimum which leads to better extrusion and retraction which in turn results in less stringing and oozing of molten laments. The bowden setup is where the extruder motor is mounted on the printer body on the frame and a PTFE tube runs to the hot end. Even though weight of the hot end is reduced in this setup, the range of laments that can be used such as exible or abrasive laments are reduced because of the travel distance and also chances of binding in the tubes are a real possibility.

(19)

Figure 11: Direct vs Bowden Extruder Setup[8] Filament Sensor

The Mk3s uses optical sensor which is triggered by a mechanical lever. Hence, the optical properties for transparent laments such as Polycarbonate or Glassber does not inuence a false trigger by the sensor. The installation of the multi material unit poses a few changes to the optical sensor which is further discussed in detail.

Hot end and Build platform

The hot end used in the Prusa Mk3s is supplied from a third party supplier based in UK namely E3d[9]. The specic model is E3D V6 hot end is a 24V hotend which comprises of aluminium heat-block,thermistor and 40W heater cartridge(g.12).

Figure 12: V6 hot end(assembled)[9] The build platform

(20)

was developed in 2010 which has been developed ever since and the current iteration is a 24V heated bed which can attain a temperature up-to 120 degree Celsius. The Mk3s has a magnetic Mk52[20] heated bed upon which a exible spring sheet can be placed for the molten thermoplastic to be deposited upon.

P.I.N.D.A Probe

The Prusa INDuction Autoleveling probe or P.I.N.D.A probe works by sensing a change in magnetic properties of the material in front of it. The P.I.N.D.A probe is an air core inductor and its inductance changes when the steel (high magnetic permeability) gets close to the tip of the probe thereby switching the output. The P.I.N.D.A probe also has a thermistor embedded in it, which compensates for the temperature drift thereby preventing the probe to trigger at dierent height because of dierent temperatures. The robe is mounted on the same assembly mount as the extruder and forms the part of the extruder assembly unit.

Figure 13: Extruder assembly with P.I.N.D.A probe EINSY RAMBo motherboard

The motherboard used in the printer features Trinamic 2130 stepper drivers which makes the machine silent when compared to other machines in the market today. It can also detect missed steps which means shifted layers can be detected and also makes it possible to print at high speeds up to 200mm/sec.

Slicing software

The slicing software used for G-code generation is Prusaslicer. Based on the open source slicer Slic3rPrusaslicer is the in-house slicing software developed by Prusa research which has predened print proles for all the models released by the manufacturer.

(21)

Figure 14: Prusa Slicer Homescreen

Apart from the pre-congured proles the user can also edit all three aspects which are print settings, lament settings and printer settings which are detailed below. The slicer software can also be switched between three active user modes which are simple,advanced and expert mode which gives the user more options to tune with depending upon the mode selected. The set of parameters available in prusaslicer (in simple mode) are stated in brief below to give a novice user a basic understanding as of how to carry out slicing of a 3D model to print it with Mk3s.

Parameters determining a successful print 1. Print settings

The major set of parameters which are in the print settings are: Layers and Perimeters:

Layer height and rst layer height determines the height of each and rst layer height of molten material to be extruded and deposited on the bed respectively. The option of perimeters generates the set number of minimum perimeters for each layer. Extra perimeter layers will be generated for slopes if the set number of layers are inadequate. Horizontal shells have the option of setting number of top and bottom solid layers ac-cording to user specication. Standard available layer heights in the slicer software are 0.15mm, 0.2mm and 0.3mm in case of MMU2s congured i3Mk3s.

(22)

Figure 15: Print settings in simple mode Inll

Inll,as the name states is the amount of material to be extruded internally. It is ex-pressed in percentage and in increments of 5 ranging from 0% to 100%. The inll pattern for both top and bottom layers can also be set according to various options available which are rectilinear, grid, gyroid and cubic to name a few.

(23)

they do not fail. Supports have the option for being generated automatically where the software determines when the support to be generated after certain threshold angle of overhang.

Figure 17: Skirt and brim settings

Figure 18: Support settings 2. Filament Settings

The lament settings are categorized into lament, cooling and lament overrides. The lament option lets the user to adjust the extrusion and bed temperature for rst and the subsequent layers respectively. Statistical information like cost and density can also be entered so as to gather information on how much lament per print is used and also its cost.

(24)

Filament overrides has the option of setting the retraction length, which is the recoil movement to prevent the oozing of the material. Hence, when a retraction is triggered the the lament will be retracted according to the length specied. The other option in this category is raise Z, which raises the Z axis by the amount specied every time a retraction is triggered.

Figure 19: Filament Settings 3. Printer Settings

The printer settings for MMU congured MK3s has individual settings for all 5 extruder selections including individual retraction speed,length and Z lift. Apart from retraction settings for individual extruders minimum and maximum layer height limits can also be assigned in this section.

(25)

Figure 20: Printer settings for MMU2s

A wide array of settings are available in advanced and expert mode which will need to be used in implementing engineering grade materials which are discussed further below. Moreover, expert settings unveil more options

3.1.3 3D printer assembly, calibration and testing.

The machines are available either as completely assembled ready to print out of the box or as a kit in a DIY(Do-It-Yourself) setup. The machines in the university are of the latter kind and hence the project began with the assembly of printer following the guidelines specied by the manufacturer.The guidelines are specic pertaining to dierent sections of the printer build[21]. The kit has comprehensive build instructions which helps the user to build the structure, the axes and the extruder in a systematic way so as to gather the understand the functioning of each component in the machine.

(26)

Figure 21: Various stages of printer assembly Calibration and Testing

The printer requires various calibration and self tests namely the perpendicularity check for x,y and z axes, hot-end and bed temperature checks, print and hot end fan checks where essentially the printer ensures for the proper functionality of all the con-nected components.

Once the self test is completed, the printer then searches for the height calibration points at the four corners of the bed after which the live z-height needs to be set. This bed leveling procedure is enabled with the help of P.I.N.D.A probe and a majority of the FDM printers in the same category as Mk3s lack this feature which has proven to be a major advantage since this manual trial and error method of bed leveling has been eliminated.

Figure 22: A schematic of theoretically perfect rst layer

A perfect rst layer is achieved when the molten material deposited has enough squish to remain on the build plate without being lifted while also not being too squished so as to not to obstruct the ow of material from the nozzle. The heated build plate ensures that

The live z-calibration is an inbuilt method whereby the distance between the nozzle and the bed can be saved into the EEPROM(Electrically Erasable Programmable Read

(27)

Figure 23: Live z-calibration

Various stages of live z calibration from left to right in the order of correct layer height. Majority of the printers in the market in the same bracket do not have automatic bed leveling and has to do bed leveling manually; this is an important appeal to many of the users which eliminates the method of manual trial and error bed leveling.

4 Method and Implementation

4.1 MMU Unit

4.1.1 Introduction

The Multi Material Unit or MMU2s was introduced as the successor of MMU1(Figure 24). The most notable dierences of MMU2s from MMU1 is that the number of lament choices were raised from four to ve and also the multiplexer unit (the assembly on the extruder for lament entry) on the extruder side was eliminated. This helped in reducing the bowden drives from fours to just one. Also, the MMU1 had four individual stepper motors for each of the lament to be driven into the multiplexer unit on the extruder side. This individual motor for each lament has been eliminated to one motor for driving all laments with the help of a selector motor driven by an idler shaft.

(28)

Figure 24: An assembled MMU1(left) and an assembled MMU2s(right) MMU1 MMU2s

No. of materials 4 5 Direct Drive No Yes No. of extruder motors 4 1 Bowden tubes to hot-end 4 1 Print recovery No Yes

F.I.N.D.A No Yes Manual Controls No Yes

Table 3: Major hardware dierences of MMU2s compared to rst generation MMU1 4.1.2 Working Principle

The MMU2s has three motors; namely pulley,selector and idler motor. Apart from the three motors, it also has a lament sensor namely F.I.N.D.A; which is essentially a

(29)

to the extruder. The extruder and the selector is connected with the help of a PTFE tube which guides the lament back and forth between the printer and multi material unit. The multi material unit has a separate control unit which helps it to communicate with the printer and the unit is powered by tapping into the same power supply as that of the printer.

Figure 25: MMU2s and its major components.

There are also three push buttons on the module where the left button moves the selector to the left while the right button moves the selector to the right and the center push button can be used to load the lament.

(30)

Figure 26: Pulley Shaft with gears on MMU2s

4.2 Existing issues and possible solutions

Even though marketed as an ready to use unit after assembly, the MMU2s has had its fair share of problems ever since its release. The most prominent issues faced during out test

Filament loading and unloading failures due to non triggering of Infrared Sensor. Filament loading failures(Figure 28) happens when the infrared sensor does not trigger which leads to the extruder gears not gripping them an hence resulting in unsuccessful lament loading. Incorrect calibration of IR sensor is one of the main reason for load failure to occur. Calibration, in this case, is mechanically done by adjusting the position of the infrared sensor holder on the extruder.Also at present, there is no visual indicator so as to know if the infrared sensor is triggered correctly when the lament enters the extruder assembly other than going into the support menu and checking the sensor status. An ideally successful lament load and F.I.N.D.A sensor should show 1 on display which means they are triggered successfully(Figure 27).

(31)

Figure 27: A correctly calibrated infrared sensor showing 1

Apart from the incorrect calibration of the sensor as mentioned above, hairy lament tips can also lead to load failures. The hairy lament tips will lead to triggering F.I.N.D.A since the hairy tip will still remain lodged in the chamber . Filament loading and unload-ing speed can also inuence to an extend, the successful or unsuccessful material loadunload-ing onto the extruder. Another aspect is the PTFE(Poly Tetra Fluoro Ethylene) tube which guides the lament from the MMU2s to the extruder. The manufacturer supplies the tube with the dimensions 2mm I.D(internal Diameter) and 3mm O.D(Outer Diameter). If the lament tip is not of perfect shape, this tight tolerance of the tubes can cause issues in unloading leading to lament being unable to be retracted into the MMU2s unit. Any kink in the guide tube can also aect the travel of the lament through the tubes.

Figure 28: An MMU2s load failure displayed on MK3s

Badly shaped lament tips on unload between a color swap mid-print which sub-sequently leads to a loading failure.

Hairy strings on the lament tips(Figure 29)[22] leads to F.I.N.D.A still detecting the presence of lament while in practice, it is just the hairy tip of the lament. Bad lament tips can occur due to insucient cooling moves before a lament color change, low extrusion temperature or even moisture in the lament.

(32)

Figure 29: An illustration of proper and improper lament tip

F.I.N.D.A remains triggered even though canal chamber is empty and free of la-ment or lala-ment strings

F.I.N.D.A probe essentially has two components to perform its function. The probe itself and a metal ball in the selector chamber(Figure 30). The metal ball moves up as the lament enters the selector to reach the extruder through the PTFE (Poly Tetra Fluoro Ethylene) tube thereby triggering the probe. Correct amount of clearance between the probe and the ball is required so that in idle state the probe does not trigger an output. The ideal distance between the probe and the ball is 1.5mm which can be calibrated using hex key. The false trigger can also happen if the a piece of lament gets stuck inside the selector by any chance.

(33)

Figure 30: Cross section of selector chamber with F.I.N.D.A and metal ball When functioning properly, the ball should fall back into the slot if there is not lament present, but improper internal surface nish of the chamber or inadequate clearance between the ball and probe will lead to the F.I.N.D.A being triggered falsely. Moreover, the selector is built in such a way that there is no possibility to see if there is any residue of lament caught up in the chamber or the metal ball is back in place when there is no lament moving through the chamber.

Idler barrel cracking

The idler barrel is connected to the idler motor which rotates the barrel and engages the lament to be loaded onto the extruder(Figure 31)[19]. The barrel is connected to the motor via a shaft which is held onto the motor shaft by two screws. The idler barrel is prone to cracking where the screws are tightened onto the motor shaft. A better idler barrel assembly; either made of engineering grade material or one of a better design could possibly be the solution to this problem.

(34)

Figure 31: Idler barrel assembly of MMU2s

4.3 Hardware and software changes implemented

4.3.1 Hardware upgrades and their eectiveness The selector

The selector unit of MMU2s is mounted on a lead screw(Figure 32) which is being driven by the selector motor to enable the choice of available ve laments according to the need The selector also has two smooth rods to slide across as the lead screw moves in order to enable the selection of laments. The selector also houses the F.I.N.D.A probe on the top to sense the lament when it enters the selector though a 2mm opening.

(35)

Figure 32: A fully assembled selector mounted on MMU2s

A major drawback of this selector is that there is no possible way to know if there are any hairy strings, lament dust or even the metal ball is back in its slot if the F.I.N.D.A still remains triggered while there is no lament since the design is fully enclosed in nature(Figure 33). Hence, a false trigger of F.I.N.D.A would require either of the two following solutions listed below.

1. In case of lament dust or strings, the user will be required to remove the F.I.N.D.A probe extract the metal ball and clean out the chamber.This would also mean the F.I.N.D.A probe and the ball would need to be re-calibrated.

2. If the metal ball in the chamber is not moving up and down freely, this would also require disassembly of the F.I.N.D.A probe.

(36)

Figure 33: Side proles of the stock selector

Hence, diagnosing the false trigger of F.I.N.D.A due to reasons listed above is not possible without removing the F.I.N.D.A probe or disassembly of the selector in case of a lament stuck in the selector. This would prove to be a major inconvenience since this would require entire disassembly of selector aspect of the unit.

The design upgrade, in this case, implemented was an updated version of selector with an opening on the side prole[20] (Figure 34)so that user can visually see the movement of the metal ball as well as clear out the chamber of lament strings, dust or possibly a piece of lament caught up in the chamber.

(37)

Figure 34: Test print of updated selector(white) installed on MMU2s unit. The manu-facturer supplied version(orange) in the background

This updated version makes quick troubleshooting over false trigger of F.I.N.D.A caused by residue of laments or because of the metal ball very eective without requiring any disassembly or re-calibration of F.I.N.D.A afterwards and completely eliminates the problem on the F.I.N.D.A side of the unit. This modied version of the selector also has a provision for inserting a magnet so as to ensure the metal ball falls back into its place when there is no lament passing through so as to ensure there is no false triggers(Figure 35). The magnet used in the testing unit, was a 2x6x10mm rectangular magnet. The selector has a blade which enables to cut the lament if a lament load fail occurs con-secutively three times. Although supplied with kit, the the blade functionality is not yet incorporated into the rmware. But, observations were made that when the lament was attempted to cut the lament would be bent by the blade rather than cut. But, the real functionality cannot be tested until the blade is activated.

(38)

Figure 35: Schematic of modied selector with magnet slot The idler barrel

As stated earlier, another relevant hardware issue among various users of MMU2s unit in the community is the idler barrel cracking[21] at the point where the barrel is mounted on motor shaft(Figure 36)[22]. The screws are to be tightened neither too tight or too loose. If not tightened enough, the barrel will not be seated properly and this can lead to loading failure wheres over-tightening can lead to cracking of the part. This issue can be rectied either by printing another barrel of a tougher material or redesigning the stock idler to a better version.

(39)

Figure 36: MMU2s idler barrel mounting point susceptible of cracking

The approach taken was that to reprint the the idler with an engineering grade plastic while also reinforcing the shaft connector of the idler. The width of the shaft connector was increased by 4mm on the base, thereby not hindering the rotation of the barrel while reinforcing the connection between the barrel and motor shaft(Figure 37) and was successfully installed in the unit (Figure 38).

(40)

Figure 37: Reinforced Idler(left) and the stock idler (right) for comparison The Idler barrel was again printed using PrimaSelect NylonPower Glass Fibre which provided enough stiness against potential cracking while being the idler being secure enough. The increased width means that the stock M3 X 10 screws used for securing the idler onto the motor shaft was replaced with M3 X 12 screws(Figure 36). The reinforced idler also provided a snug t for the bearings so that lament remained centered to the bearing while being loaded and unloaded thereby avoiding any misalignment.

(41)

Figure 38: Reinforced idler barrel installed on the unit M-10 pass-through coupler for lament entry

The stock MMU2s unit uses a screw compression type tting for PTFE tubes at the entry point(Figure 39). Even after tightening the screws in place, the PTFE tubes tends not to be locked in place and can be pulled out easily. Moreover, continuous compression of PTFE tubes over time can form a kink on the tubes leading which would lead to friction in the tubes against lament entry which is not ideal.

Figure 39: The original rear PTFE entry vs the M10 adapter PTFE entry The manufacturer supplies PTFE tubes with internal and outer diameters of 2mm and 3mm respectively and hence any deformity in the form of kinks can result in fric-tion or obstrucfric-tion of the lament being loaded. The M10 rear entry adapter[19] with pneumatic coupling solves this issue since the PTFE tubes are locked in place with the

(42)

couplers(Figure 39) and do not just rely on the compression of screws. The tubes are locked in place with the help of pneumatic couplers which prevents it from falling out of place.

PTFE tubes, PTFE pass-through adapter and custom PTFE guide

The stock PTFE tubes was replaced with tubes of better dimensional tolerance; 3mm and 4mm Internal and Outer diameter respectively so as to ensure that even if the lament tips have unusual tips, it can avoid snagging on the stock 2mm I.D tube causing unload failures.

The PTFE pass-through guides the tubes into the extruder gears which was replaced to suit an M10 coupler so as to ensure that this coupled with the wider PTFE tubes would make sure there is enough clearance even if the lament tips were not up to the mark as expected. Also, the larger diameter of the PTFE tting along with introducing the thread on the printed part instead of an metal insert, reduces joints and friction spots in the lament path by having smooth guide way as far as possible down to the extruder.

Figure 40: Stock pass-through guide(left) v/s the M10 version(right) PTFE entry aligner

A custom PTFE aligner frame was designed to rectify the issue of lament entering at an angle and hitting chamfered edge of the hot-end PTFE tube. The frame is a simple clip-on design which makes sure the PTFE is perpendicular to the lament sensor cover thereby ensuring the lament enters the extruder straight and not in an angled manner. The aligner frame was redesigned from the initial design to relocate the hole to t better and also to relocate the hole on the frame.

(43)

Figure 41: Various design iterations LED indicator for Infrared Sensor

As mentioned earlier, the calibration of the infrared sensor is purely a mechanical pro-cess which involves adjusting the lament sensor cover while there are no visual indicators other than going into settings and check the sensor status switch from zero to one; one being the value when the sensor detects a lament.

Figure 42: The infrared sensor in Mk3s

The infra red sensor is triggers an output when there is an obstruction between the U channel present on the sensor. The obstruction, in this case, to detect the pres-ence of lament, is done by the extruder idler gear tower which moves inwards as the lament is gripped blocking the infrared beam thereby triggering an output. Figure 43 (highlighted in red) shows the ideal position of the extruder idler gear position when the lament presence is detected correctly. The extruder assembly mount screw (highlighted in orange) should be tightened moderately in order to allow the free movement of the idler gear assembly. The idler tower has a small notch at the end of the idler assembly

(44)

which blocks the infrared beam. The state of sensor and its correct calibration is critical for the multi material unit to function correctly.

Figure 43: Extruder idler gear assembly position during presence of lament. To make the process of calibration easier and also to ensure the proper functioning of the infrared sensor, a visual indicator in terms of LED coupled with resistor was soldered and attached to the sensor. A 330 Ohms resistor coupled with a 3V LED was soldered between the +5V and OUT pins. The anode of the LED was connected to the +5V pin while the cathode coupled with one end of the resistor and the second probe of resistor connected to the OUT pin. The resulting solution ensured that for every successful lament load, the use had a visual indication by means of the LED without having the need to go to sensor stats in settings to view the sensor info.

(45)

4.3.2 Software changes implemented

Apart from the various hardware changes implemented in the unit, there were also some software changes implemented mainly to tune and achieve the best possible lament tips. These changes, combined with the hardware changes were essential in order to utilize the maximum possible reliability out of the machine. Majority of the changes in the Prusaslicer were done in the lament settings(Figure 45) under tool-change parameters with single extruder multi material printers since the objective being nding an optimal settings in shaping the tips. The various tool changing parameters that denes the shape of a lament and the changes made are briefed below:

Figure 45: Stock lament settings in the slicer software v/s the changes made in the slicer(highlighted)

Minimal purge volume on wipe tower

After a tool change, the exact position of the newly loaded lament inside the nozzle is unknown and the lament pressure may not be at a stable state. Before the actual extrusion of the newly loaded lament resumes on the inll print object or on a sacricial wipe object, the printer will prime the specied amount of material in the wipe tower to produce successive inll extrusions reliably. This can also be adjusted for sharp transition in color changes if color bleed is detected. Color bleed happens when the previous material color happens to bleed into the currently loaded material due to residual material present in the nozzle.

(46)

Loading speed denes the speed used for loading the lament onto the wipe tower and loading speed at the start is the speed used at the very beginning of the phase. Unloading speed and unloading speed at the start

Similar to loading speed and loading speed at the start, the unloading speed denes the speed used for unloading the lament on the wipe tower. This speed does not aect the initial part of unloading after ramming. The unloading speed at the start is the speed which is used to unload the tip of the lament immediately after ramming

Ramming settings

Ramming denotes the rapid extrusion just before a tool change in case of a single extruder multi material printer. The purpose of ramming is to properly shape the end of unloaded lament so it does not prevent insertion of new lament and can itself be reinserted later. This phase is important and dierent brands or types of laments can require dierent extrusion speeds to get a good tip shape. Hence, the extrusion rates are variable during ramming. Ramming line width and line spacing denotes the width and spacing of the rapid extrusion respectively.

Figure 46: Ramming settings (stock v/s modied settings) Number of cooling moves

The number of cooling moves determines how many times the lament is cooled by being moved back and forth in the PTFE tubes.

First cooling move speed and last cooling move speed

These are the speeds which sets the speed of rst and last cooling moves respec-tively. In case of both these speeds, they gradually accelerate beginning at the

(47)

4.4 Implementation of engineering grade materials in MMU2s

4.4.1 Thermoplastic:Adura X

Adura X is a brand of engineering grade plastic developed and marketed by a Scandi-navian organization namely add:north based in Sweden[20]. Adura X contains carbon ber additives which makes it lightweight, dimensionally stable,tough, impact and tem-perature resistant and ideal for engineering applications in an industrial setting.

Carbon ber laments are composites where carbon bers have been added to a base material to enhance certain characteristics. Carbon ber laments are dierent from car-bon ber materials. In a conventional carcar-bon ber material, long continuous carcar-bon ber strands are spun and later weaved into a sheet. In case of carbon ber laments, however, carbon ber is chopped an milled before being mixed with a base material(Figure 47).

Figure 47: Schematic of manufacturing Adura X

In case of Adura X, the base material is Nylon and the resulting lament, hence, has the added characteristics of microscopic carbon ber particles while retaining the characteristics of base material. Printing with laments having additives like carbon ber can be abrasive on the stock brass nozzle supplied with the printer since carbon ber does not melt like base material, hence nozzle made from a much tougher material, preferable hardened steel is always recommended. In our case, we decided to make use of PrimaSelect hardened steel nozzle[21] which was in already in possession and installed on the machine prior to commencement of all the tests.

Testing Adura X and ndings

The testing of the lament began by using the parameters and settings recommended by the manufacturer (Figure 48). The sample test model chosen to print was 3DBenchy.

(48)

Figure 48: Recommended manufacturer settings for Adura X

This model is often used for bench-marking and/or reviewing a lament since the model includes a number of dicult to print features including; overhanging curved surfaces, smooth surfaces, symmetry, planar horizontal faces, large, small and slanted holes, low-slope-surfaces, rst layer details and tiny surface details[21].

Four test prints were conducted by using the recommended parameter settings from the manufacturer which yielded less than ideal results. Three prints failed owing to a clogged nozzle while the fourth failed owing to improper bed adhesion(Figure 49).

(49)

Figure 49: Test prints failure owing to high retractions(left) and inadequate bed adhe-sion(right)

After the rst test print failure, the settings were scrutinized again and the decision was taken to reduce the retraction length and speed. To mitigate the eect of print lifting from bed, the print bed temperature was raised from the recommended value of 75 degree Celsius to 95 degree Celsius. The manufacturer recommended value of retraction length 1.2mm at a retraction speed of 40mm/sec was found to be the possible reason for clogs. The reason being, with a lament full of microscopic carbon ber laments that does not melt like the base materials, there is an increased risk of clogging. Other settings such as the speed, layer height, inll,The bers can build up and form clogs in the nozzle. Hence, for the test prints which followed the retraction speed and length was reduced as follows:

Test Print Retraction Speed(mm/sec) Retraction Length(mm) Result

1 40 1.2 Failed

2 38 1 Failed

3 36.5 1 Failed

4 33 0.9 N/A(Print lifted from bed mid-print)

5 30 0.8 Success

Table 4: Retraction length and speed for the test prints

The fth attempt with the above mentioned settings yielded a successful print albeit some artifacts which is clearly visible(Figure 50).

(50)

Figure 50: First successful print with Adura X

Further tests prints were successfully conducted and completed with various perimeters and extrusion temperatures ranging from 265 degrees till 290 degrees.

The conguration settings for Adura X was therefore successfully tested in MMU2s setup and the conguration is saved for future use in the printer.

(51)

in printing parts which require to be sti, rigid and high stiness/weight ratio(table.4). Since it has additives in the form of glass ber, brass nozzle can be worn out easily and a hardened nozzle is recommended.

Table 5: Mechanical properties of NylonPower Glass Fiber[22]

The manufacturer recommends the nozzle sizes 0.6 or 0.8mm to reduce the clogging although the use of 0.4mm nozzle while testing proved to be successful but with reduced retraction length and speed similar to that of ADuraX.

Testing NylonPower Glass Fiber and ndings

Since the only provided information for printing was recommended speed and tem-peratures for printing and bed, keeping that as a base line, following test settings were created and tested(Figure 51). The retraction settings were kept at a length of 0.8mm at 35 mm/sec since this material, too had additives in the form of glass bers.

(52)

Figure 51: Prusa Slicer settings indicating key changes made for Nylon Power Glass Fiber The MMU2s upgraded components such as Idler barrel, selector assembly and la-ment sensor cover with M10 adapter was printed successfully using these settings and congurations are saved in slicer for future use.

4.4.3 Implementation of TPU(Thermoplastic PolyUrethene) in MMU2s

Thermoplastic PolyUrethene is the most commonly used polymers for exible materials. The brand in focus is Easyprint Flex95A; 95Abeing the shore value of the material. The exibility of the material can be controlled by the inll. For instance, with less inll the resulting print will be exible, almost collapsible structure, that you can squeeze together but it will always return to its previous shape. If the inll is more, the resulting print will be more like hard rubber.

By default, prusaslicer does not have the option to slice or print models in TPU for the multi material mode. The justication provided for this is that exible materials like TPU cannot be be used successfully in multi material print where the tool changes happen which requires constant loading and unloading from multi material module to the printer. Moreover, the material being exible, there are possibilities of material having very stingy tips during cooling moves which can further lead to loading or unloading issues. Printing TPU in conjunction with other materials in multi material mode would also require further tuning of extruder idler pressure since exible material can easily bond with gears and cause a jam in the extruder motor gears. Hence, the existing

(53)

Slow loading and unloading speeds

Adding a 3 second delay after unloading in order to give the material enough shrinkage time in prints where constant swapping between TPU and other materials can happen.

Figure 52: Sliced model of a gear knob showing multi material combo of PLA(bottom) and TPU(top)

The settings for print; namely speed,inll after a material change, in case of a combi-nation of materials can be individually assigned as shown in the gure above(Figure 52). The tool-changing parameters, in case of TPU is listed below with main focus on reduced speed for both loading and unloading as well a delay as stated earlier after unloading in order to allow the material to shrink back to its original shape.

5 Results

5.1 Results pertaining to the MMU2s

The MMU2s with many of the modications installed, ranging from modied selector, stier idler barrel, M10 adapter PTFE entry and m cover was tested both for reliability of the modications as well as the reliability of the setup as a whole.

(54)

Figure 53: The current setup of MMU2s including all the modied parts

Prior to installing the modications/upgrades, the attempted test prints yielded less than ideal results including hairy lament tips, inconsistent loading and unloading fail-ures, false trigger of F.I.N.D.A probes to name a few.

Software changes implemented led to improving the lament tips to a level that no more failures due to stringing laments tips were detected. Also, false trigger of lament probe on the MMU2s unit was also eliminated as a result of this upgrade.(Figure 54).

(55)

selector and rear PTFE tube entry, glass ber infused nylon for idler barrel and PTFE pass-through cover.

The upgraded version of the selector made it easier to inspect the F.I.N.D.A in op-eration and also to clear out any lament debris which can be possibly accumulated in the chamber over time. Also, in case of a lament jam, the same can be cleared without requiring the whole disassembly of the selector.

While the PTFE aligner clip, PTFE tube with increased diameter and M10 PTFE pass-through adapter aided in achieving consistent loading and unloading of laments regardless of the materials tested; the combination of both software and hardware would require thorough testing involving numerous tool-changes.

Numerous models were printed out in order to test the eectiveness of the installed modications using the all possible ve choice of materials.

Figure 55: A test print[23] involving 545 tool-changes(highlighted) with a duration of 25 hours

(56)

In terms of reliability, the multi material module coupled with the Mk3s was able to be produce reliable prints with less or no major issues. Many more models(g.) were printed out taking the cumulative tool change count above 3000 during the testing period which can certainly be considered as progress in terms of establishing a reliable multi material unit.

5.2 Results pertaining to the thermoplastics

Both Adura X and NylonPower glass ber, was further tested by printing various test models; in case of the former; test hooks models with varying number of perimeters (from two till ten) were printed with the aim of tensile strength test while the glass ber material itself was used for printing modications for the multi material unit.

Figure 57: In clockwise direction:Test prints having various perimeters(2,4, and 10) Combination of the above mentioned materials was also attempted and the model was printed successfully using the previously used settings for the same(Figure 58).

(57)

Figure 58: Combination of Adura X and NylonPower glass ber

This combination of materials was possible because of the reason that the base material for both the laments were same(Nylon). Moreover, the temperature range of extrusion was well within the achievable range for both the materials. Hence, two materials within the same temperature range can be printed in multi material mode or two materials having additives can be printed as long as the base material for them is the same. Figure 58 shows two samples printed in this manner. The rst sample shows signicant color bleed of Adura X. This can be avoided by increasing the purging volume so as to prevent the same. The second model was printed with an extrusion perimeter overlap of 100% so as to make sure proper bonding of dierent two materials.

While TPU was also implemented successfully, a combination test of PLA and TPU was carried out which yielded successful results after varying the settings at the interface where two materials meet;namely extruder overlap width. Sample on the left in gure 59 shows us the combination of PLA(green) on the bottom and TPU(blue) on the top for a gear pin design. Sample 2 depicts the same print as sample 1 but in single mode printed out of TPU; while sample 3 is again a combo of PLA(green) and TPU(blue), the most important point for sample 3 being that the multi material part is printed in place and not an assembly.

(58)

Figure 59: Successful TPU prints in single and multi material mode

The successful testing and implementation of engineering grade materials by tuning the software parameters such as retraction distance and speed helps us in answering one of the research questions which was how to tackle the issue of printing materials other than the generic PLA and PETG. The software tuning coupled with hardware improvements such as hardened steel nozzle and more powerful extruder cooling fans makes printing of abrasive and engineering grade thermoplastics possible. The parameters like retraction speed, distance and temperature (both extruder and bed) and its impact on the print output was also tested and analyzed to understand their impact on the dierent types of materials tested; especially materials with abrasive additives.

More than the redesign of MMU2s itself, the consistency and reliability of the unit was increased by taking into consideration both hardware and software aspects. While few of the parts such as selector and PTFE rear entry holder was replaced and reprinted with stier material, software settings was also tuned and tested with in order to utilize the MMU2s to its maximum potential. Even though there are numerous suggestions and upgrades available in the community, the installed modications were kept to a minimum since the degree of success varies slightly.

6 Scope of future work

Now that the multi material unit is reliable, in terms of producing both visually pleasing prints using PLA and high strength prints using engineering grade thermoplastics, fur-ther future work can be conducted in terms of mechanical testing of samples produced for dierent overlap parameters with two dierent materials or implementing the indus-trial grade materials like PEEK(Polyether Ether Ketone) or PEI(Polyetherimide) which would require upgrading and also reworking both hardware and software aspects of the machine. Hardware aspect would be to replace the hot end assembly like heat block and thermistor to withstand high temperatures while software aspect would be manipulating the rmware so as to cater the thermal runaway and temperature limits. Possibility of

(59)

the wastage of material during tool-change can also be an area to be explored which would require changes in both hardware and software.

The blade holder assembly is another avenue which can be tested. The latest rmware for the printer (version 3.9.0) was released during the nal stages of compiling this report which reportedly enabled the blade functionality. The full functionality and the eec-tiveness will need to be tested and upgrades or redesigns to be suggested accordingly.

This project work can be considered as laying the groundwork for the many of the possibilities which can be achieved with this machinery , few of which are stated above.

References

[1] Zhang J, Jung Y-G. Additive manufacturing: materials, processes, quantications and applications . Cambridge, Massachusetts: Elsevier; 2018

[2] https://www.asme.org/about-asme/media-inquiries/press-releases/3d-systems-rst-3d-printer-named-historic-mechanic

[2.1] https://3dsourced.com/3d-printers/types-of-fdm-3d-printer-cartesian-delta/ [3] C.K. Chua, K.F. Leong, C.S. Lim,Rapid Prototyping, Principles and

Applica-tions(WorldScientic Publishing Co. Pte. Ltd, Singapore, 2003)

[4] Kumar LJ, Pandey PM, Wimpenny DI. 3D Printing and Additive Manufacturing Technologies. 1st ed. 2019. Singapore: Springer Singapore; 2019.

[5] Jones, R., Haufe, P., Sells, E., Iravani, P., Olliver, V., Palmer, C., & Bowyer, A. (2011). RepRap the replicating rapid prototyper. Robotica, 29(1), 177191. Cambridge University Press.

[6] https://3dsourced.com/guides/history-of-3d-printing/ [7] https://all3dp.com/2/fused-deposition-modeling-fdm-3d-printing-simply-explained/ [8] https://www.fargo3dprinting.com/advantages-disadvantages-direct-bowden-extrusion/ [9] https://e3d-online.com/prusahotend [10] https://shop.prusa3d.com/en/3d-printers/180-original-prusa-i3-mk3-kit.html# [11] https://manual.prusa3d.com/c/Original_Prusa_i3_MK3S_kit_assembly#_ga=2.5395188.613732401.1585680769-1782458164.1579095932 [12] https://blog.prusaprinters.org/multi-material-upgrade-2-0-is-here/#_ga=2.14825343.895535118.1589719782-1782458164.1579095932 [13] https://prusa3d.com/downloads/manual/prusa3d_manual_mmu2s_eng_1_00.pdf?2

(60)

[14] https://help.prusa3d.com/es/article/mmu2s-setup-and-inspection_5819 [15] https://www.thingiverse.com/thing:4104419 [16] https://www.3dprima.com/se/laments/primaselect-nylonpower-glass-bre-1-75mm-500g-natural/a-23425/ [17] https://forum.prusaprinters.org/forum/?wpfs=barrel+crack [18] https://help.prusa3d.com/es/article/mmu-loading-failed_5343 [19] https://www.thingiverse.com/thing:3233579 [20] https://addnorth.com/en/shop/product/ANAX15BLA [21] https://3dprintingindustry.com/news/3d-benchy-torture-test-pushes-3d-printers-limit-103662/ [22] https://www.3dprima.com/laments/a-23425 [23] https://www.thingiverse.com/thing:3068256 [24] https://www.prusaprinters.org/prints/9209-spyro-reignited-trilogy-mmu2s-remix