This build will attempt a large 3D printer (nominal 2ft by 2ft by 2ft build volume) using primarily v-slot components for all three axis of motion (X,Y,Z). This configuration is that of a gantry robot.
About 4 years ago I purchased a Solidoodle 3 3D printer to hop on the 3D printer bandwagon or perhaps more appropriately join the stream of lemmings headed for the nearest cliff. My real interest was modeling (not the run way kind) a cartesian 3d printer's errors. Towards this end I wrote 2 Matlab programs recycled to some extent from my Applied Mathematics graduate project titled "Master's Project 2009 Error Analysis of Robot Arm Kinematics". The first was a Print-Test driver and the second one was a Monte Carlo simulation of the printer errors. The Print-Test driver provided the ability to run 30 tests of escalating build volume and error estimate assumptions. Test-30 was for a 609mm cube and the results are provided here. This was the genesis for the idea for a 2ft=609mm cube 3D printer.
Officially my build started April 1,2017 but the frame shown was put together about 8 months prior. The frame is made from rail, that is , actual aluminum stair railing from my back yard that was replaced with recycled wrought ironing from my sister's front porch remodeling. On top is fixed Makerslide rail primarily trying to use up the Makerslide rail before I start using my preferred V-slot rail.
Over the last several years I purchased V-slot rails and a smaller number of Makerslide rails and their respective required wheels and a few large plates. I 3D printed many Openbuild connectors and such in anticipation of starting the build. I purchased both Nema 17 and 23 stepper motors and a Ramps 1.4 kit and of course a large assortment of metric nuts and bolts required for the V-slot rails. Submitting a "build in progress" has given me the impetus to take the build seriously and devote a lot of time to it.
Stepper Motor Selection Requirements
The printing end (hot-end) of a 3D printer has an expected range of accelerations and velocities for each x-y-z direction. These velocities and accelerations are the driving requirements for any 3D printer, large or small. Along each of the axis, there is an effective inertia composed of both mass and rotational inertia. Each axis's inertial mass is due to the weight of structural components and the weight of any stepper motors. Each axis's rotational inertial is due to the stepper motor's rotor's rotational inertia. It is assumed that each axis is driven by one stepper's ability to generate torque. Torque is defined as the force acting tangentially at the stepper's output shaft or a lesser force acting at the stepper's pulley. This force must be able to accelerate and drive at required speed the axis's inertia loads including the stepper's rotational inertia. Detent torque and starting force (to overcome each axis's starting friction) will be integrated into the calculations. The power of each proposed stepper is to be computed and constrains which torque speed combinations are possible. Stepper motor specifications usually include a torque-speed graph which includes two curves defining the operating region for the stepper. The first "pull-in" curve identifies the set of all maximum torque-speed points for which the stepper will not lose synchronization during starts and stops. The second "pull-out" curve defines the set of maximum torque-speed points for which the stepper will not lose synchronization along as the stepper does accelerates or de-accelerates slowly. The region between these curves is called the "slew" region (TBS-figure).
This build has been more like a design in progress. Since v-slot components make it easy to assemble components into subassemblies such as the x-carriage it is just as easy to disassemble if a better design seems evident.
I recently assembled a carriage with a x-axis stepper and a z-axis stepper to control the 20 by 20 mm z-axis hot-end arm. I'm designing this on the go so it's no surprise that it's "better" i.e. more build volume will be available if I use 80 by 20 mm V-slot on the x-axis ends. This will elevate the x-axis rail so that the carriage assembly will clear the aluminum railing frame as depicted here. Before I go any further I'm going to step back and look at the mass of the different axis and what size Nema stepper motors are required. I have been assuming that the z-axis can be handled by a Nema-17. My trial x-z carriage uses a Nema-17 which must lift up and down without gearing the 20x20 hot-end arm which at present I am questioning. It should be noted that a Bowden extruder will be used to eliminate a stepper motor on the z-axis. My plan was to use a single Nema-23 stepper to drive the y-axis. I would set it up the way Solidoodle 3 did it. If you think about it the Solidoodle 3 is close to a Gantry design at least for the x and y axis. I need to establish my stepper motor torque requirements and do some testing prior to committing to a final design. I googled the specs for my Nema-17 and Nema-23.
Comparing the Nema-17 holding torque specs with information from Question: how much weight can my stepping motor lift? and Understanding Torque it became questionable if the Nema-17 was powerful enough to lift and hold the current z-axis of length 108 cm. A python program was written to determine the force available at the pulley for Nema-17 and Nema-23 steppers. Additionally, calculations were performed to determine the effective loads as seen by each z,x and y stepper motors. The results indicate z-axis Nema-17 can hold the 1.27 lb gravity load but must in addition be able to accelerate the z-axis inertial mass corresponding to it's weight of 1.27 lb but additionally must counter the external gravity force of 1.27 lb.. The weight of the x-axis is 3.64 lb and 5.59 lb for the y-axis but their steppers must move their inertia equivalents. Further analysis is required to determine if the steppers can meet the acceleration requirements. To assist in validating the viability of the planned Nema 17 and Nema 23 stepper selection usage in this design the following reference documentation was gathered:
- http://openbuilds.org/attachments/s...8/?temp_hash=a331dc35ee0939e81a649c07dc2f56a6 Step Motor Basics Guide
- TBS (To Be Supplied) ...
1,2,3 May 2017
The current status of the build is that the frame is now rigid with no sway. The makerslide/v-slot rails have been squared properly on top of the frame. The frame actually is out of square but that only affects reduces the x-y plane working area slightly. Measurements were performed to determine the starting force required to accelerate the frictional and inertia loads of the x,y and z axis assemblies from a dead stop. These forces were measured towards the ultimate goal of computing torque requirements. To these torques will be added the detent torque values from the stepper specs. Requirements that need to be defined include the following:
- motion profile (trapezoidal) requirements for each axis - get parameter ranges from Marlin firmware
- velocity: steps/mm X200 Y200 Z200
- acceleration: mm/s^2 X500 Y500 Z20
6 May 2017
- load inertia for each axis, F=ma, calculate masses from weight measurements- need stepper rotary inertia values.
- speed/torque curves for specific Nema-17,23
- acceleration torque, frictional torque
- determine maximum acceptable inertia ratio 5:1 typical (TBS-reference)
To date , torque-speed curves for the specific steppers have not been found. There are several options:
So far, the reference stepper selection tutorials and guides all require Torque-Speed curves. Of course they do. Somehow this data must be obtained or estimated.
- try to cross reference the steppers to ones for which data sheets can be obtained, so far no luck with this approach - "free data sheet site" didn't respond to a non-company
- take some torque-speed measurements - found some web content along this path
- model mathematically the steppers and run simulations to obtain the graphs - found an article describing this approach
- since power = torque * speed , if stepper power is constant can generate plots - doubtful this is correct
- ...... test steppers under real conditions, observe their performance
A selection of long metric screws were ordered today to allow for a rework of the x-axis carriage. A more symmetric, robust design is possible. The new x-carriage design will straddle the y-axis v-slot rail providing one side to accept the stepper x-axis drive and the other side will accommodate the z-axis stepper.
7 May - 4 June 2017
The x-axis trolley has been reworked as shown here and here. The new z-axis assembly as described below with be mounted to the x-axis trolley.
The z-axis is being reworked using a 1000mm acme 8mm diameter threaded rod. It will work like a linear actuator with a minimum of 24 inches of travel. The part driven by the threaded rod will be a 20x20mm v-slot extrusion with the hotend attached. V-slot wheels will constrain the motion of the extrusion as depicted by the z-axis mockup. It was required to mill a special threaded rod block mount to interface with the 20x20mm extrusion. As part of the z-axis redesign, a 20x20mm drill guide was assembled with the goal of connecting 20x20mm extrusions directly without always using the the full or have blocks. V-slot wheels are used in pairs to constrain z-axis motion. One can be fixed and the other uses an eccentric nut for fine contact adjustment.
A 24x24x0.25in carbon steel plate was purchased for the heat bed based on the fact that the coefficient of linear thermal expansion for aluminum is approximately twice that of steel's coefficient. This decision was made at the expense of increased heat bed power requirements. The specific heat capacity of aluminum is 0.220 Btu/(lb-degF) verus 0.120 Btu/(lb-degF) for carbon steel but the density of steel is 7850 kg/m3 versus 2720 kg/m3 for aluminum (6160). To heat two identical heat beds composed of steel (40 lbs) and alumimum (14 lbs) it requires 4.8 Btu/(degF) verus 3.08 Btu/(degF). Aluminum power requirement is 64% of steel.
heat bed calculations/simulations
Current coarse calculations indicate that to heat the steel plate up to 100 degrees celius requires about 480 Btu. At 3412 Btu/hr per 1kw conversion identity it would take about 8.5 minutes to warm the plate. But this does not include the ongoing convection losses which I surmise are significant. A computer model is needed. The results may indicate a need to use an aluminum bed because of a wall socket's power limitations.
heat bed simulation
Large cartesian gantry style 3D printer
This will be a fixed bed, v-slot implemented X,Y,Z axis gantry style 3D printer.
- Build License:
- CC - Attribution - CC BY
Reason for this BuildThis build is to attempt a build without rods, acme screws. I will use just v-slot (and some makerslide) rails and components.