DM Inspired
Daniel Murray's protfolio
RESEARCH
Research was undertaken in the form of a 3 by 3 fractional factorial print experiment , this experiment was to find the biggest effects on surface finish and dimensional accuracy in each axis, to be able to optimise for speed whilst still retaining detail. A test piece was modelled with features to specifically test attributes of accuracy and surface finish, some of these features included stepped overhangs to test surface finish at different angles, spikes going thick to thin as the smaller the circle the more chance in error and a collection of measurable features so that dimensional accuracy can be measured. Print speed, Nozzle temperature and Line width were chosen for the variables of the experiment because they have the largest effect on accuracy and surface finish in relation to overall print speed.
The full factorial would have taken 27 prints to completely, therefore a fraction factorial test was chosen as 9 prints better fitted the time scale. These 9 prints took 4 days due to working around working hours (9 to 5), even though the print test print took 2.5 hours to print.
After the experiment the conclusion was made that the fastest way to print and still retain accuracy is to change the slicer setting to a print speed of 70mm/s, Nozzle temperature to 180c and a line width of 0.5mm. the experiment also concluded that the fastest way to print and still retain surface finish is to change the slicer setting to a print speed of 90mm/s, Nozzle temperature to 180c and a line width of 0.3mm. This experiment also highlighted the ways which the printer process can be sped up, to speed up the time wasted on failed prints, the dormant time between prints and the actual time it takes to print a model.
PROJECT AIMS
-
To reduce the overall time to print 9 models by 25%, from 4 to 3 days.
-
To be easy to fit and need no extra skills to operate.
-
To cost less than £80 as the Ender 3 is an affordable printer at £160.
-
To not loose quality compared to the 3 by 3 factorial test taken in the research.
-
Health and safety issues must not decrease with this upgrade
CONCEPT GENERATION
After observing a factorial experiment, it was noticed that there were 3 areas where the print time of the Ender 3 could be reduced, these are: the speed of the actual print, the time wasted from print failure, and the dormant time in between prints.
To speed up an individual print the upgrade could be a bigger nozzle, a direct drive system, or even reducing the weight of the Y movement. This would allow the printer to print faster but retain quality.
Time could be saved on prints that have already failed, by using a camera that monitors the print and with an AI system, decides if the print has failed or not. A series of smaller modifications could also prevent against the most common reasons of print failure, such as bed adhesion, a device to measure remaining filament, or a device to check the nozzle is not clogged. When these are triggered it would notify the user by either sounds or digital alerts.
To reduce the dormant time in between prints, digital alerts could be adopted to notify when a print is finished. Another method to reduce the dormant time in between prints, is a conveyor system that would take prints on and off the bed automatically. The advantages would mean there is no time between prints, and the Ender 3 could function without an operator.
Concepts were impartially marked against weighted criteria; these criteria were formed from the specification for the project. Speed and improving the limitations of the printer were the most weighted criteria.
The conveyor belt system was chosen because the results of the weighted matrix show it scored the highest rating, even though areas of weakness in the concept that needed to be developed were identified. The initial concept for the conveyor belt would be difficult to install because the frame for the printer would have had to be changed and even though it does not take any extra skill, the average user wouldn’t want to change the frame in case of permanent damage.
DESIGN DEVELOPMENT
With very little constraints on the concept at this current stage of development, it was completely redesigned to make everything user centred. It was designed so that the print surface would double over the existing heated bed, therefore both the frame and bed would not need to be changed to upgrade the printer.
Material choice for the printing surface was the next decision to make as it directly influenced the quality of print. Criteria for the material was loosely based on the specification but mainly looking at quality of print and good adhesion, affordability, and durability.
Leather scored the highest rating due to it being the overall best suited print surface material for the conveyor belt. PEI, the material currently used by the Ender 3, was scored second because it costs 9 times as much as the leather. Although having a better durability, it is made expensive by the size of which is required. If cost were not an important factor in this project, PEI would be the better choice.
Developments were made to the design to save material, costs, and time when printing the parts. The rollers would have taken 536g, 179.62m, and 38h to print. Most of the material is on the inside of the roller and therefore not needed. By splitting and hollowing out the rollers into parts, they can be printed without the need for inside support structure. The new design for the roller takes 234g, 78.32m, and 17h to print.
The linkage between the parts of the roller were tested to get a tight fit with no play in rotation. Play in the linkage would result in missed steps.
After testing the driving roller with the leather print surface, steps were being missed as even with the tensioning mechanism there was not enough grip to move the surface. After trying a few different methods, the method of lining the printed roller with felt was chosen, to increase the coefficient of friction.
SOFTWARE
The idea for this project is to speed up the overall time in a 3^3 factorial. The printer printing one after the other, can be achieved in a slicing software with a long Y axis and having each model spaced down the bed printing consecutively.
Another way this can be done is with a Gcode sending software. Prometheus was a coding project that was undertaken at the same time. This is the way Gcode is being sent to the printer.
FIRMWARE
The hardware works in such a way that there is not a lot to change in the firmware, this is because it uses the existing motor and electronics. The ability to move the bed completely around is stopped by the axis limits. This can be changed in the configuration.h file by defining the Y_MAX_POS value to larger value than the bed circumference. By setting the Y_MIN_POS to minus the max bed size, it allows the user to move the bed a full rotation in either direction after homing.
The standard homing sequence in marlin, uses an endstop switch to home each axis. This design of conveyor belt printer doesn’t have an end stop in the Y axis, therefore an error is given when trying to home using G28. In the Gcode.cpp file in the libraries of marlin, Gcodes are defined using a function case system. Changing the code that is called when the G28 is triggered, allows Z and X to be homed as normal and the Y axis to be zeroed using the G92 function.
Once the printer is homed and can move to every point on the belt, the Y axis must be calibrated against the roller by changing the steps per mm. This is calibrated in marlin in the configuration.h file by changing the DEFAULT_AXIS_STEPS_PER_UNIT variable. The 70mm roller changes the steps per mm to 30.476.
The firmware already has existing safety measures built into it, and none of these measures have been changed. The new limits have been changed so that these measures would still work under the new conditions.
HARDWARE
Fitting this upgrade is no harder than fixing and maintaining the Ender 3. It could become a commercial product as a package that contained the bed, timing belts, and gears with downloadable .stl file for the rest of the parts. These could be fitted together using nuts and bolts. A copy of both the marlin firmware and Prometheus software could also be provided and integrated together, so that when the printer is plugged into the Prometheus software for the first time it flashes the marlin firmware onto the printer.
When fitting the upgrade, there is no chance of burning or electrocution as fitting the upgrade can be done while the printer is off and unplugged. When the upgrade is in operation there is a chance of entrapment, entanglement, or pinching. The side plates are designed to lower the chance, but it is still a possibility. There are a lot of forces going through this upgrade as belts are tensioned in opposite directions, so if not tensioned correctly problems could injure the user.
LATEST IMAGES
