Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Metal Matters SLM Printer
#11
More updates. I've been working on combining multiple arrays. I've decided that a single 72W array just won't cut it even with the previously planned efforts to reduce the spot size. The scanner I have planned for this system will require a longer focal length / larger spot size so the 'less is more' approach will have to be shelved. In the photos below I have combined two NUBM38s using a single prism. I opted not to use mirrors as I want to keep the form factor small and not have to manage the heat mirrors produce as a result of inefficiency. I have what I suspect to be uncoated prisms on hand but have ordered a prism with an AR coating. So far the losses I have measured are on par with the typical losses you would see from an uncoated lens (8-10%). I am a little confused by this as the incident beam is about 30 degrees away from the surface normal meaning there should be some loss due to reflection. I did notice some light reflected but didn't think to measure the individual rays at the time. I did setup the prism with a single NUBM44 to measure the loss however, I will compare the results when the new prism arrives.

[Image: 20210715-212255.jpg]
[Image: 20210715-212807.jpg]
[Image: 20210715-222224.jpg]
[Image: 20210715-222907.jpg]
Reply
#12
Recently I've needed to address an issue related to the build piston. Up until now I've been using ASA plastic which has been suitable for testing but was never a long term solution. As heat creeps into the bore, mainly through the leveling screws, the plastic supporting the screws softens resulting in the print colliding with the spreader. In the past I've managed to get around this issue by using glass to separate the leveling plate from the substrate but as the laser power has increased this has become less effective.

There are a few solutions to this that I can think of; one option may be to cast the piston using conventional metal casting techniques or perhaps even a fine grade of cement - or alternatively get the piston machined. I'm reluctant to turn my hand to casting as from what I've seen fine details are hard to recreate, the process itself is long winded and requires skill. CNC'ing the piston is probably the most logical path but unfortunately it's quite expensive. Since there are two components to the build piston (the piston itself + a yet to be water cooled leveling plate) I suspect that after a couple of iterations the cost will quickly get out of hand. The quotes I've had for the piston alone to be machined are around $180 - 200 USD with some modification. This probably isn't too bad and would improve if they were to be made at scale, but it lead me to think about an old project I had since put to one side.

I've been interested Laser Metal Wire Deposition (LMWD) as an alternative to the Laser Powder Bed Fusion (LPBF). It has its limitations but given its relative simplicity and resemblance to FDM, if it were able to provide functional metal parts it would be a huge asset, particularly in this instance. Several months ago I made a brief attempt at recreating the process but failed due to a few oversights. The design used a 21G syringe to feed 0.25mm SS316 wire into the path of focused NUBM31. The spot size of the laser was too small and the power too low. The syringe itself was too long and flexible making it unstable. It turned out to be a learning exercise.

In a recent video I mentioned that the spot size of the dual NUBM38 laser head is around 0.8mm. After measuring it with the beam profiler it appears to be over 1mm. The tracks vary from 0.8mm to 1.1mm depending on translation speed and generally suffer from a higher radius of curvature at a narrower width. Because of this, I've decided to repurpose it for the sake of a LMWD project as the spot size will sacrifice too much detail for the sake of the LBPF printer. If I am able to print metal using this wire based technique I may be able to produce the build piston for myself. Assuming that this is possible, it would enable water cooling to be directly integrated into the piston, minimizing the foot print of the leveling plate.

I've made a couple of attempts at testing the process, initially just validating whether it was possible with ~144W and a second attempt with a slightly more robust setup. Unsurprisingly it was much easier to produce metal tracks with this method. The most problematic aspect of the design illustrated below is the angle which the metal is introduced. So far I have only been able to deposit metal away from the wire feed and to the left and right of it. Traveling 'into' the weld pool has been difficult for reasons I'm not 100% certain on. I suspect that with such a low angle of entry the wire is unmelted wire is preheating and melting before it gets to the laser, grabbing the substrate and moving it, disrupting the print. To get around this issue I'm working on a design that should resemble a coaxial approach, allowing wire to be fed perpendicular to the substrate, hopefully resolving the issues with directionality. That'll be a topic for another post...

https://www.buymeacoffee.com/metalmatter...ternatives

[Image: 57729574a7029747586a7aa32e6061fe.jpg]

[Image: 880c88548e4a7280f52f0ce832d6aa26.jpg]

[Image: e6d9afa340ecdd557d4da3c9839a6bf6.jpg]

[Image: 1772247ba7da1d5d47e62680c14afbf5.jpg]
Reply
#13
Since the last video I put out there have been relatively few yet one quite major change regarding this effort. In that last video I quoted the spot size of the laser as being 0.8mm.. I was wrong about this. I was initially basing this on the width of the metal tracks but grew a little suspicious after noticing the width varying significantly when adjusting printing speeds. After said video, I later setup the beam profiler to find that the beam was as big as 1.2mm (based on xy movements). Again, this was wrong but lead me down the path of rethinking things, and also later revealed that the CCD of the profiler was saturated and was suffering from flaring pretty badly. Before realizing this I decided that if the spot size were as large as I thought it was, this wasn't going to do. I bought a new laser, this time a 200W 915nm fiber unit with a fiber diameter of 0.2mm. Using the aspheric lens I have on hand, when focused this should achieve a spot size of 0.3mm. The tighter focus plus additional watts will make for some respectable translation speeds/better accuracy. 

I was also wrong about the power I quoted. I had quoted the standard specs of the lasers x 2 (144W) but in actuality both units were running at 20% + over their rated capacity. This was due to the PSU regulating in CV mode which hadn't been correctly set, allowing 8.6A to pass versus the recommended 7A. Once I had adjusted the power to the rated specs the performance of the head wasn't quite what I had hoped for. It was beginning to resemble the performance of the NUBM31 (< 15mm/s) and hence motivated me to buy a new laser. I suspect that part of the issue comes from the losses across the lenses (prism + focusing lens + window) which only leaves around 115W (20 more than the '31). Combine this with the difficulty of aligning each spot perfectly and you have a lot of displaced energy, lacking the peak intensity required to maximize translation speeds. 

All is not lost however. The larger spot size can be utilized by the LMWD printer (I've now created a separate thread for this). It turns out the spot size is close to 0.55-0.6mm based on several passes over a bare substrate which accommodates the wire size well. In the meantime I am waiting for a new pair of safety goggles to arrive and am looking for a suitable PSU for the fiber laser. 

For interests sake, TheBackyardScientist used the same fiber laser as the one I've purchased in one of his videos. He has the higher spec model however, 106um vs 200um ($1500 vs $800). Check it out here:

Reply
#14
Work resumes with this project. In truth, I've been passively working away on getting this printer back up and running since the LMWD video. The effort has been largely scattered across rebuilding the piston assembly to prevent any movement of the substrate and better tolerate heat soak, and integrating the new laser and power supply.

The build piston is still of a very 'DIY' nature. It makes use of a 54mm hole saw loaded with substrates to help increase thermal mass and to provide elevation for the substrate that will serve as the base of the print. The cantered blades of the hole saw also work well to prevent any rotation as they bite into the edge of the substrate but will flex enough to allow the substrate to be swapped. However, I imagine with repeated heat cycling that this won't be consistent and may require some 'adjusting'. I have mentioned in the past that I wanted to get this part of the printer machined, and still intend to do so, but as of this moment I am not sure whether I should working toward cooling the substrate or preheating it. If the substrate is cooled it will yield all of the issues that conventional metal printers face (stress/warping due to rapid fluctuations in heat) as well as require more input from the laser. If the substrate is preheated then it will most likely be a scratch build as nothing will be able to be made from plastic.

As for the laser, I ended up using 6 high flow copper heatsinks (3 top, 3 bottom) to create a cradle to keep the unit cool. I've yet to put the unit through its paces but if there's a chance the laser can operate without a refrigerated chiller I would think this is it. For comparisons sake, the NUBM31 I used in my early videos displaced a similar amount of heat (~190W) over a much smaller area, although could tolerate 70C versus 55C. This lower temperature threshold is one of the reasons I have avoided fiber coupled diode lasers until now. 55C isn't too bad but fiber coupled units often have lower thresholds, as low as 30C.

The power supply is now integrated into the build. I had bought this unit under the pretense that it has rise time control. It does and it doesn't. What it actually has is a series of presets which are bound to a time scale where the trigger time for each preset can be adjusted, allowing the PSU to transition between settings over the specified interval. However this feature does not work with remote triggering which is very disappointing. I have found that despite this, the constant current response is very good and doesn't allow for overshoot meaning I can simply enable/disable the supply without being conscious of rise time. Unfortunately these diode lasers are very sensitive to transients/ESD as it only takes around 2V/25uA of reverse biased voltage/current to kill the laser hence the need for a good quality PSU.

There's quite a bit more to discuss but I'm going to leave that for the next video. Here's a short video of me testing the laser at 1A, measuring the loss of the gas enclosure window. The display is blown out but it reads 15.5W uncovered and 13.5W covered.

Reply
#15
Reply
#16
Alright, I'm going to throw a post up here for those that are still checking here from time to time. I took a few weeks off after the last video, waiting for materials and taking some time to reflect. I've since setup the printer up on its own table, created its own gantry, eliminated the window into the chamber, changed the recoater design, changed some of the electronics, and have had to tackle a few issues in regards to the laser (I'm becoming more proficient at terminating the fiber - ha!). Long story short, the printer is up and running again. I'm currently working on finding an optimal config. At the moment it's operating at around 70mm/s - 100mm/s at 70% power. I've also been modifying the surface tension of the steel in attempt to control the bead profile with some success. I have around 400GB of footage and climbing, and will have a video out in the next couple of weeks.. Sorry for the absence

[Image: 20220207-223931.jpg]

[Image: 20220309-181935.jpg]
Reply
#17
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)