Wednesday, March 22, 2023

Rotoforge 3/21/2023

    In my last update I mentioned that we "solved a problem" with the welding of the metal feed stock wire to the walls of the rotating BLDC shaft. This turned out to be a better, but not wholly effective solution. In time the glass fibers in the liner break down and create alumino-silicate dust due to heat and abrasion against the hollow shaft walls, and the wire feed stock itself, particularly the foot that forms when the wire presses against the rotating die/nozzle. 

Figure 1. A closeup of the business end of Rotoforge with the die/nozzle removed. This reveals the foot, AKA "flash" that has formed where the wire contacts the rotating die, as well as a protrusion where the material was previously extruding at the bottom of the image. The damaged state of the glass fiber liner is also apparent.

The glass fiber liner is relatively effective at preventing contact between the wire and the rotating shaft wall, particularly when the liner is impregnated with 10W-40 motor oil. However, because the liner is not rigid it does not effectively constrain the growth of the "foot" or "flash" as it is sometimes referred to in the literature. Basically:

  1. When a material contacts a rotating surface heat is generated from friction. 
  2. This heat softens the materials.  
  3. At some temperature the material strength will decrease(the material will soften enough) to a point that the material begins to flow like a fluid. 
  4. Once the material begins to flow, fluid mechanics demands that it must flow through the path of least resistance.
  5. Conservation of mass demands that the material that flows through the path of least resistance must be replaced by other material that has been softened sufficiently to flow as well.

In Rotoforge, because the motor shaft is of a larger diameter than the wire, there exists space for material that has been softened to the flow-able state to extrude into the space between the wire and the shaft. This reduces the flow rate through the nozzle/die, and causes problems with destroying the glass fiber liner, and eventually with welding/galling to the walls of the rotating shaft which jams the motor. This is similar to the unconstrained flash growth that occurs in the literature and is described broadly as being the result of extrusion of plasticized material under load. as in figure 2.

Figure 2. a helpful diagram from Vilaca et. al showing the formation of a viscoplastic region in a piece of rod being friction surfaced onto a plat, and subsequent extrusion of flash from the contact region when the material is unconstrained under load and heat form friction.

There are several ways to approach this problem. 

  1. Increase the wire diameter to reduce the space for the flash to form
  2. Reduce the shaft internal diameter by making a custom shaft
  3. Inserting a tightly sealed stationary guide tube through the shaft which presses against the rotating die and prevents the wire from contacting shaft walls, and constrains the flash/foot.

The first method is somewhat impractical due to the undesirable scale up of cross sectional area of the wire, and scale down of the extrusion pressure at a constant force as a result.  Perhaps a more powerful extruder with better grip, like the proper extruder. To obtain the required tolerances between wire and existing stock shaft, we would need a wire with OD of ~2.2mm which gives a cross sectional area of ~3.8 square millimeters. Even with the best off the shelf bowden sytem, no more than 100 newtons of force is available. This principally provides enough pressure for extrusion, but only under ideal conditions. In reality, force will be spread out over the whole area of contact between the wire and shaft. 

 A tight fit with the shaft and wire in the current configuration also poses a galling/welding problem, as 1045 carbon steel(the shaft material) is readily wet by hot aluminum, zing and other metals, and thus, is likely to seize under the high RPM and temperature conditions in the Rotoforge printhead. 

 The second option will require acquiring a lathe and doing substantial custom design work. I will say no more about it for the time being.

The third option is what I decided to attempt recently to overcome the reliability problems with the glass fiber liner and various lubricants.  This required the removal of the retaining screw for the motor bell to fit the stationary guide tube, and thus the supporting of the motor bell with an external thrust bearing structure.  The insertion and support of the guide tube itself was also somewhat tricky.  some images of the guide tube assembly and the whole print head assembly with external thrust bearing are provided in figures 3 and 4. 

Figure 3. The pneumatic bowden fitting with guide tube installed. The guide tube is made from 13 gauge luer lock needles from mcmaster. The needle's luer lock head is cut short, leaving just a small collar to retain the assembly inside the pneumatic fitting. This collar is soldered in place to the inside of the pneumatic fitting. 

The luer lock-pneumatic bowden fitting assembly screws into the gantry mount plate and holds the needle floating in the middle of the rotating hollow shaft.Figure 4 shows the complete assembly with external thrust structure...

Figure 4. The complete rotoforge printhead assembly with external thrust bearing structure and stationary guide tube inserted. the thrust structure is an off the shelf Z bracket from mcmaster along with standard bearings and mounting hardware.

It was important to make sure the length of the guide tube allowed it to press tightly against the inside of the die/nozzle (acorn nut) to form a seal. This tight sliding contact fit also necessitated more careful concentric drilling of the acorn nut die/nozzle, due to the unfavorable interaction (cutting/shearing) that the die orifice had on the guide tube end in operation.  The setup for which I show in figure 5. In short, I took advantage of the low run out and high RPM capability of the BLDC motors to drill highly concentric die holes on my mini mill...This technique has been elucidated previously in the course of the reprap project and elsewhere as general machinists knowledge. The BLDC spins the part to be drilled, while the bit is held stationary like on a lathe. Slow well lubricated drilling with the aid of a bubble level for part leveling enables highly straight concentric holes to be drilled in the die/nozzle. The CNC mill here is a genmitsu pro 3020.

Figure 5. setup for using the BLDC motor and a mill as a highly concentric support and drilling tool for making straight concentric die/nozzle without a lathe.

Unfortunately, even this stationary rigid guide tube approach failed due to flash formation and galling. Essentially, what appeared to happen was that the flash/foot would form inside the stationary guide tube at the entrance to the die, material would begin to extrude through the die, and then some amount of play in the external thrust structure would open up the gap between the die and the guide tube allowing more material to flow into the seal, and eventually this material would cool and gall against the stainless steel guide tube. Occasionally breaking off pieces of the guide tube end or the flash from the feed material ultimately halting material feed due forming a plug of metal galled inside the guide tube.  Figure 6 shows an example of a flash that caused an extruder jam.

 

Figure 6. the end of the hollow shaft with the die removed, showing a galled plug of aluminum metal filling the end of the guide tube, resulting in an extruder jam.

Figure 7 shows what the situation looked like once I cleared the jam by pulling the guide tube out and tugging on the wire with a pair of pliers to break the galled material loose.

Figure 7. flash has been galled off by contact with the stainless steel guide tube end. The guide tube has been retracted to free to motor, and this is what remained after the tube was broken free.

Given this latest reliability failure I have opted to go the custom shaft route. My kind, caring and ever supportive partner, Sam, has provided me a lovely Proxxon FD150E micro lathe to do the work with as a birthday present. 💖

Fortunately there is some preexisting reasoning for making custom nozzles/dies for elevated temperature handling and resisting wetting/galling of semisolid or molten aluminum and other metals. Chris from the hackaday aluminum FDM printer project was inordinately helpful in providing advice on where to source molybdenum alloys he used for the project.  Thus, from his advice and further research on the matter a new plan for the next couple of weeks (once the lathe arrives) has emerged...

1.) Acquire 1/4" Titanium-zirconium-molybdenum alloy rods from mcmaster. or TZM, which are ~96-98% molybdenum. I intend to turn these down on a lathe into replacement shafts for the BLDC motors to solve several problems at once.
TZM is used in hot work tooling for die forging, extrusion, molten metal handling, and injection molding, it is not soluble in molten metals, is resistant to wetting by them at elevated temperatures, and is relatively self lubricating and resists galling similarly to bronzes but at 2X-3X higher temperatures and applied loads and some information about its wide variety of useful properties can be found here.


2.) By custom making the molybdenum into press fit shafts for the BLDC  we can also eliminate the clunky external thrust bearing and replace it with a simple flange on the molybdenum shaft end.  this saves weight, and reduces rotating assembly losses due to friction in additional bearings and mechanical contacts...My experience with the external thrust bearing has mostly been an unstable and lossy one due to all the additional mechanical interfaces.

3.)We can also control the exact tolerances of the interior of the shaft, and thus how closely it conforms to the wire feed stock. This is important, because it is necessary to have a tight fit, with low friction between the wire and the shaft interior to facilitate a stable lubricant film (as in a fluid bearing) and to restrict back flow of plasticized material at the die orifice in the form of flash.  Additionally, having the die orifice be integral to the shaft, IE machined into it, instead of threaded on as with the acorn nuts, should eliminate problems due to play in the threads under high extrusion loads, and with the die unthreading during use. In general, a machined part saves mass, provides better tolerancing and durability and eliminates mechanical slop in the system which should help us better control the flow of plasticized material at the die/nozzle orifice.


Thus has been the basic course of experiments in the last few weeks... interspersed with a visit to the lunar and planetary sciences conference in Houston Texas.  Met plenty of fascinating new people, scientists and engineers alike there, many of who may have some valuable insights in the future. 

Such is the plan for the next few weeks of experiments. 

If you made it this far, thanks for reading and look forward to some additional youtube updates and blog post updates soon where I will be making this custom molybdenum die/nozzle/motor shaft! 

I have been hemming and hawing on whether I should ask for support from my readers / community for these projects... they get rather expensive and take a great deal of time. I am not yet comfortable asking for people to support the project without concrete an reliable printing though. 

I suppose you all could tell me in the comments below. I would love to know how you feel about it and if you think this project is worth investing in. 

 

 

 

Rotoforge 2023-11-30 Cold Working and Other Problems

 7 months of implementations later... I thought I grasped the physics. I was wrong. well not completely, but at least 50% wrong, which might...