Transcript:
I 3D printed a CPU cooler mount for my new workstation and almost wrecked it that way. Let’’s find out more. Guten Tag everybody! I’m Stefan. And welcome to CNC Kitchen. The last two years I’ve been editing all of my Youtube videos on my Razer Blade 2017, which worked okay since it is a gaming laptop but still had limited resources and got quite loud as soon as I started stressing it. So, at the end of last year, I decided to invest in a new workstation for editing and maybe some gaming I decided to go for a Ryzen 7 2700X, 32 gigs of DDR4 and an RTX 2070. Yes, Ryzen. Instead of Threadripper. Just because I didn’’t want to have an electric heater in my office and energy costs are something to consider here in Germany. I still had a big tower cooler from my old workstation left over, So I thought I save myself a couple of bucks and not buy a new one. Before making my purchase, I just quickly researched and noticed that at least the current version of the Thermalright HR-02 was compatible with the AM4 socket, The Ryzen 7 uses. Well, when I received everything I noticed that my 7 year old cooler Didn’’t have the right bracket for mounting and also there weren’t any AM4 conversion kits available anymore. So I had the option to either use the boxed cooler, which had RGB lighting but was kind of loud as soon as you stressed. The PC buy a new CPU cooler or make my own CPU cooler mount. And who would I be if I didn’’t use my 3D printers for that task? So a while back. The German filament maker FiberThree. Send me a roll of quite expensive, but crazy, strong, carbon fiber Nylon filament. This stuff is printed at 285°C and is twice as strong as everything. I’ve tested so far in my filament tests. Even the layer adhesion seemed to be at least acceptable. I also sneaked a test sample in one of my temperature tests, where it didn’’t fail, even at 140°C Therefore, I thought this would be the perfect material for that application. The CPU cooler bracket doesn’’t only need to be strong. It also needs to be stiff and capable of handling the temperatures next to the CPU. Just on a side note. I noticed that 75% of you. Watching right now are not subscribed to the channel. So if you find that content helpful, do me a favor and subscribe and hit the bell. It’’s highly appreciated. The CPU cooler mount for my Ryzen 7 consists out of standoffs that are screwed into the mainboard. These provide clearance to the components next to processor. The bracket itself, which was a piece of sheet metal, sits on top and has threaded holes in the middle. When the huge cooler is now mounted on CPU, an additional metal bracket presses it down via two additional screws. Since you apply a significant amount of force on the mainboard, there is another bracket on the backside of the mainboard that adds rigidity and prevents deformation of it. So my task was it to design two new cooler mount brackets. One for the back and much more importantly, the one in the front. Before I started with the designs, I took pictures of both sides of the mainboard to use them As reference images in Fusion360, They can be imported and then scaled to size with the use of one reference dimension. The back bracket was simple and I basically used the same space as the already existing one only quite a big, thicker than the metal one in order to make it as stiff as possible. The front one was a bit trickier. I again made it quite thick to add rigidity with cutouts at the locations where the bolts will sit. Since I didn’’t want to cover any of the parts below, especially in the middle, the remaining strip of material was very thin, which, as you will see later turned out to become a problem. I added hexahedron shaped pockets on the lower side. Where later two nuts will sit? In order to avoid printing problems, I left a thin membrane at that location so that during printing, the whole hole can be bridged and no supports are necessary. The remaining material can then simply be drilled out. The other big cutouts on the top were for clearance of the heat pipes, and I printed a couple prototypes in standard PLA until everything fit perfectly. As I said in the beginning, I’’ll be printing the brackets in FiberThree’s PA-CF Pro, which is fiber, reinforced nylon and since the fibers will Orient in printing direction. You’’ll also get the best part strength. If you Orient the part in a way that the printed material is in line with the loads. So I changed the infill angle from 45 to 0 degrees to increase the stiffness even more. Since Nylon is know for it’’s moisture problems, I dried the material over night in my food dehydrator and then directly Put it into my dry box out of which I later printed. Bed adhesion is usually not a problem with fiber reinforce materials since the carbon strands mostly eliminate warping. I printed it on my Original Prusa i3 MK25 using a hardened steel nozzle directly on the Pei sheet that I treated with Magigoo PA beforehand. I glued two M3 nuts into the prepared pockets and everything was ready to assemble. Everything fit fine, and I made sure to equally tighten both bolts that are used to clamp the cooler down because there wasn’’t a mechanical end. Stop present anymore. The workstation booted and temperatures looked great Perfect. So thanks for watching and until next time… [scratch]. Wait, remember that I said I almost killed my PC with this bracket, Yeah? So temperatures were fine for the next couple of days even with longer. Uhh, rendering sessions until one day the machine crashed, and I already had a fear. What happened. I’’m very happy that CPUs nowadays have a thermal shutdown functionality. When I built my first PC just at the turn of the century, this probably would have meant magic smoke and the need for a new processor. So already when I opened the case, I was able to feel that the CPU cooler was loose and after taking a closer look at the bracket, it was clear that it had severely deformed. So what went wrong? I think this was a nice case of the creep phenomenon. What is creep. Creep is the tendency of a material to permanently deform under the influence of mechanical stresses. The interesting point here is that creep happens at stresses where the material should still behave elastically, which means that after a deformation, the material should spring back to its original shape. So for example, pretension should not reduce over time. The speed of creep is usually dependent on the applied stress, the temperature and a couple more. So even though the temperatures were far away from the melting point of the material and probably not yet at the glass transition point. This permanent deformation happened over the course of a couple of days until the clamping force between CPU and cooler were too small and they separated Well. You know the rest? So back to the drawing board. What can we do to prevent that? First using plastics for this application is probably not the best thing in the first place. But since I want to make it work changing, the design is probably the way to go. Creep will probably always be a problem, but with a better design, the pretension should last over months or years, giving you the possibility to regularly. Check everything. So you remember this small lip? We had at the highest loaded section. Well, even though the carbon fiber reinforced nylon is really strong and we didn’’t load it all the way to it’’s yield point. The stresses were just too high. So I simply reinforced that location, covering up some of the capacitors below, But I made sure that there was still some clearance. Back on the Prusa printing, the new bracket out of the CF-PA, unfortunately failed every time. At that point, I thought the material was the problem and switched to another filament. Turns out the Prusa firmware was the problem and had the bug in one release that the nozzle temperature dropped randomly from time to time leading to jams in my nozzle. Since I didn’’t know that at that point, I used my Nano. Diamond reinforced PLA that I also reviewed in the past and was perfect for this application for two reasons. First, it is twice as stiff as normal PLA and thus just as stiff as the Carbon Fiber Nylon and the base PLA they use can be annealed in the oven with only minimal warping. This way, the part is heat resistant up to 140°C and the dimensions do not alter. Out of curiosity, I also annealed regular PLA and take a look at that difference. So I glued two more nuts into my bracket. Assembled everything, and this now has been running for the last two months without any issues and the cooler still sits tight on the CPU. With the wider bracket, the stresses are now lower. So creep in the material should happen a lot slower. I don’’t particularly recommend that you try this out on your own. But with some precautions, even 3D prints can be used for such a high demanding application. Still, the better option probably would have been to either. Try to make the old bracket work on the new socket or just spend the money on a new cooler. What do you think? Have you ever used 3D printing to Mod your PC? Do you have any idea what I could have improved? Leave a comment down below. Thanks for watching everyone. I hope you learnt something. If you enjoyed the video, then please hit the like button and subscribe to the channel. If you want to support the me in making these videos, then consider becoming a Patreon or support me in other ways. Also take a look at the other videos on my channel AUF wiedersehen. And until next time.