questions on Ultracortex M4: 3D-Print / Mechanics [resolved]
Alexander_Zuest
Switzerland
Hi
I work with the Ultracortex M4 Headset which my university purchased within the Neurotechnologist Starter kit URL: shop.openbci.com/collections/frontpage/products/d-i-y-neurotechnologists-starter-kit?variant=13043169493064. We purchased the "Print it yourself" version of the kit.
As soon as I received the package, I started printing the parts according to the settings specified in the assembly guide. I used multiple different PLA printers. Jet the results in terms of part quality were quite insufficient. There are two main issues. Firstly, the cover part for the CYTON has multiple solid-state joints, those were printed incredibly bad and broke of very easily. The second problem was encountered while printing the first half of the headset grid. Some parts broke up during the printing process.
I then discussed the results with a professor who is specialised in additive manufacturing and I had a closer look at the part geometry. We came to two conclusions.
The headset should be printed as one part with ABS and with the inserts in place. The flat and vertical connecting surfaces of the two-part-assembly are not suited for mechanical loads. If possible, the part should be printed as a whole and else the surfaces should be angled for better mechanical connection with glue. The ABS printer should have high enough precision to print the inserts as part of the grid part. (ABS printing remains significantly more expensive than PLA but should be safe to deliver the desirable result.)
The cover part for the Cyton is extremely thin, further the solid-state joints have sharp edges which should be avoided when constructing solid state joints. Those edges also were, where all our parts broke off. I then started a mechanical simulation of the robust case from the development kit. (I wanted to simulate the standard part, which I printed, jet the simulation program had a problem with .stl file during mesh generation because of bad part quality and I could not find a .stp of the same part.) The simulation showed a minimal safety factor of only 0.143 (See Figure below). A safety factor of around 1.1 is desirable.
The simulation is of course slightly simplified and only meant to identify the location of the problem(s). The biggest simplification is that the mechanical properties of the part are currently set as solid PLA (100% infill). Even so, this should show that there is a not negligible problem with the solid-state joints in this part.
I searched the forum for similar problems, but I did not find an answer which would solve my problems.
Based on the above described situation I have the following Questions/ Requests:
Is there a file with the complete headset (front and back) and the inserts in place?
(So we could use it for ABS print)Does someone have experiences with printing the part with ABS?
- Did someone else experience problems with the printing of the cover part? Especially the solid-state joints breaking off?
Thank You for your support!
Alex
Comments
Alex, hi.
Have you looked through our 'Headware' category here on the forum, or used the 'Advanced Search' about the breakage you are seeing? I turned up this forum thread regarding printing tips:
https://openbci.com/forum/index.php?p=/discussion/1174/difficulty-3d-printing-tips-for-printing-from-aaron
Aaron Trocola is our printing and dimensional design wizard / expert, so I'm mentioning @ThreeForm (his handle) here, for possible replies to your ABS questions. @Conor Russomanno may also have insights. As far as I know, many many Ultracortex M4's have been printed in the two part format and successfully glued together.
ThreeForm / OpenBCI does have nylon printed versions available at Shapeways, using SLS, Selective Laser Sintering.
https://www.shapeways.com/shops/ultrabci
https://www.shapeways.com/materials/versatile-plastic
Best regards, William
Alex, Thanks for the feedback. We do extensive testing, but it's always helpful to hear user experiences. Thousands of M4 headsets have been produced, and we have not recieved any reports of this part breaking or experience that ourselves, unless the print is defective. The most likely cause of the failed print is the material temperature being too low, under-extrusion, or expired material (from moisture/UV). I recommend PETG filament, which is what the production headsets switched to a couple years ago. It has better toughness and resilience while being less prone to warping than ABS.
I'm travelling at the moment, but I'll be happy to address your points about the M4 design when I'm back at my workstation.
@ThreeForm @wjcroft Thank you very much for your answers and support!
Regarding the material selection I will consult our specialist again and we will evaluate if we will use SLS, PETG or ABS. We will also keep your tips in mind, thank you!
In terms of the part breaking, it is possible that my PLA prints were not of sufficient quality and I’m aware that many parts have been printed and are working properly, jet I think one should not ignore the simulation results and take a look into the part again.
@ThreeForm Would it be possible to share the part below (Size M) so we could print it ourselves? This would be extremely helpful. I would like to start with signal acquisition as soon as possible and hope to not longer delay the progress of my thesis.
https://shapeways.com/product/WKM73GNJM/m4-ultra-headset-frame?optionId=63113712&li=shops
Thank you and best Regards
Alex
Alex, hi again. Can I suggest that you investigate local printing with the PETG? You cannot use SLS without very expensive printers. The SLS process used by Shapeways (for nylon) is very different than heated filament based additive manufacturing. Thus the dimensions and tolerances for the SLS process can be much finer. For example the design features in the SLS file (pre-threaded holes) will not print in filament. There may also be support structures missing, that are not required in SLS, etc.
https://www.google.com/search?q=selective+laser+sintering
Regards, William
Alex, here are responses to your other points.
The headset should be printed as one part with ABS and with the inserts in place.
As William pointed out, the threads do not print well at an angle. Even using soluble support, the results are rough and inconsistent using any FDM printer. Without soluble support the threads are jammed with material or drooping features. Try printing an insert at an angle if you'd like to test this. Powder-based systems don't have this trouble, so the SLS nylon frames are a good solution.
The flat and vertical connecting surfaces of the two-part-assembly are not suited for mechanical loads.
You may have printed the parts sideways. Those vertical surfaces should be horizontal - flat on the bed. This ensures they are very flat and ideal for joining using a thin cyanoacrylate glue. The inserts align and reinforce that connection. There's lots of glue contact area so the result is a robust structure. The large contact area on the bed (the perfectly flat division between the halves) ensures the structure is stable once the diagonal bars connect (about an hour into the print). If your printer has poor adhesion and small parts break off in the beginning, use a brim (not a raft or excessive support).
The ABS printer should have high enough precision to print the inserts as part of the grid part.
Angled threads didn't work coming from any FDM printer. I assume that by "ABS printer" you were referring to the Stratasys printers with a heated build chamber and soluble support, which can print ABS without warping. I printed samples on a uPrint, and the threads were rough and inconsistent (action on each bolt varied greatly).
(ABS printing remains significantly more expensive than PLA but should be safe to deliver the desirable result.)
I didn't bother to calculate cost of material, since the uPrint cost over $30k and is still not big enough to print half of a small headset. The larger machines are something like $120k. This design is optimized for regular desktop FDM machines like Prusa, Makerbot, Ultimaker, Lulzbot, Gcreate, CR-10, etc. You can buy a suitable FDM printer for less than $200, or simply buy the high quality SLS version as William suggested. That version was made specifically for industrial production and is not yet available for download.
The cover part for the Cyton is extremely thin, further the solid-state joints have sharp edges which should be avoided when constructing solid state joints
When inspecting a 3D model for printing, keep in mind that the extrusion nozzle typically has an opening of 0.4mm, so details at or below this scale can't be evaluated the same way as large, homogenous plastic parts. When printing thin walls, the ideal thickness is usually an even integer multiple of the nozzle diameter. In this case 1.6mm was the thickness of the original case, and the Robust case is 1.85. 1.6mm walls print very fast (linear movement only with no infill) and 1.85mm, with the right settings, causes a third "stirring" pass that helps strengthen the bond at the cost of printing time. Edge sharpness also has a different meaning in this context because a 0.4mm nozzle hitting a 90 degree turn on a 1.6mm wall versus going around an extremely small round corner makes almost no difference, but in practice it means that the extruder slows down (due to accelerations limits), and extrudes a tiny bit of extra material. This makes a wall less likely to split because it has a slight bulge at the end. Still, zoom in on the part and you'll see that very small chamfers, about 04.mm, have been added to smooth most edges and eliminate that effect where it is not desirable.
I then started a mechanical simulation of the robust case from the development kit...
I understand that your use of ANSYS was only for a general geometry analysis. The visualization does help show what I would describe as "relative stability", but the software doesn't know anything about the forces the part might experience, It lacks the mechanical context of the part. As you noted, the material definition ignores the mesostructure of 3D printed parts, and the variables that influence the strength of printed polymers. In the results, some of the image can always be made red because the variable (input force or palette scale) is arbitrary and is usually adjusted to make a colorful picture. A more informative image would be like this one, from an FAE analysis that defined the body of the case cover as fixed, and force being applied to the inside of the tab (deformation is highly exaggerated).
Notice that the inside corner of the base of the tab is not 90 degrees. It is angled to spread the force over a larger area. The tab only needs to move about 0.25mm to release the cover. The amount of force needed to do this was dialed in over many generations of testing on many machines. The slots were shallower before, but were extended to where the cover had exactly the right feel to snap on and off. The board was designed before the case, and those headers are not in a convenient location, but the design functions as required despite not looking great under a microscope. In fact, a small amount of weakness (flexibility) is desirable there because the tab is sort of rotating around the approximate axis shown in red in your ANSYS inspection. The function of the cover is to protect the board. It needs to stay on, but be easy to remove. It needs to print fast with little material. We determined it does so successfully. While there was no issue with breakage, PLA does have a small amount of "creep" over time and can become looser after a few hundred cycles. This is why we switched to the more resilient PETG. The tabs should look about like this:
I do appreciate that you took the time to do the simulation and consider these details. Software that approximates the results of such a messy process as FDM is apparently very tricky, when the scale is anywhere near that of the additive elements as defined by the layer thickness and nozzle diameter. Without going into much detail, I'll say good rule of thumb is to not expect realistic results from simulation for anything less than 10x the element scale (a couple of mm), and the simulation must at least support anisotropic material definition to reflect the weaker z-axis bonding.
Let me know if you have any further questions and I'll be happy to help. We can switch to DM or email if you need more assistance specific to your equipment/situation.
P.S. - Anyone working with the M4 Dev Kit should download the whole archive from this link. Due to some weird quirks of GitHub, files often download incorrectly. A "right click->save as" downloads an HTML to view the file, not the file itself). The STEP version of the Robust Cover is in the same folder as the STL. The file extension is ".STEP" instead of ".STP" so make sure your OS is associating ".STEP" to your CAD program correctly.
P.P.S. - I noticed the guy on the other thread wanted to try custom MeshMixer support. YMMV, but I use this method, so I'll post supported STL files there just to see what the results are.
Hi Alex,
I apologize in advance if my question is stupid, I'm just starting to learn 3D printing. Why did you decide to use PLA and then replace it,
https://www.craftyhangouts.com/best-dual-extruder-3d-printer/#nitem_14
with ABS? After all, there is also SBS and PETG. I apologize in advance if my question is stupid, I am just starting to learn 3D printing.
And if you decide to use PLA, why not combine it with ABS? I read that there are 3D printers with dual extruders.