Would it be hard to try to make it a giant maglev bearing with some magnets and the correct shape of the outer ring ? Or if you could use the drive coils themselves to hold the ring while running.
@@Maaniic I suspect magnetic field of maglev bearing will interfere with mechanics of fan itself, although I guess you can minize it by making them rather tight fitting (below 1mm distance), but reducing magnet's power (maybe even make it 2 rings on opposite sides at 45° incline). All that said, it will take quite a lot manual labor to place all the magnets, since you would probably need to place hundreds of them PS fans like it probably would cost quite a lot more (even if produced at scale) due to much copper and magnets being required, as well as more complicated bearings. Not to mention that lateral speed at outer edge it much faster than at inner edge, making it significantly harder to make bearings quite, on the other hand if one could nail down bearings design really well, he might potentially achieve quieter fan by removing the hub supports
@@MaaniicMaglev requires superconductivity to operate, our current tech relies on either very low temps at normal pressure, or super high pressure at more reasonable temperatures, neither of which is very practical. It would make for a great show though.
A Dutch team in the ugh hyperloop competition used aluminium and magnets to levitate. It worked. Im sure one can also make a magnetic bearing with just magnets.
It would definitely be a high dollar project. 120x120 ball bearings are like $40-$60. Not sure how much ceramic ones are but possibly more expensive. I think the best thing to do is make one out of CNC parts and then optimize the blade design first.
@@fuzzyturtlez8994 wouldnt even be worse in terms of friction with that big bearing? I actually think this is probably a worse fan design than what we have today.
@@AndrewTSq It is worse. If you look at swept area you lose more space on the outside compared to what you gain on the outside. Then the amount of loading on the bearing is nowhere near max capacity. So smaller balls if possible would be nice to have.
I think using steel ball bearings would quiet it down quite a bit, because of the resonance difference. The poly-crystalline structure of ceramic is very brittle, hard, basically making it noisy. Using a steel ball bearing along with a dry lubricant like graphite would drop the noise immensely. Maybe a resin printed channel for the ball bearings to ride in would help a bit too. This project is AWESOME!! I would love to see you take your awesome idea farther, and I am amazed at your 18 pole handcrafted motor's performance.
A good PTFE base lubed would be better but dry lubes wont help with noise as much as a liquid or paste type lube although they would slow it down. Trying some of each type to compare would be a good video idea to go with other improvements.
Def some sort of lube on the bearings would help. I was thinking something like a grease, but know that would make a tremendous mess. Graphite is a good idea, though it will still be hard to keep whatever lubricant you use on the bearings and not all over everything else.
I remember when that tip driven fan came out. I thought it would be great, but it was so expensive. It never really took off because it had too many down sides. The bearing was in the hub, which was why it still had the hub, but that was far superior to having the bearing around the edge like yours. The reason the hubless design was used for the boat thruster is because water isn't as compressible as air. It's actually needed in water.
Even though air is more compressible, wouldn’t having a clear channel of flow reduce the amount of work it has to do? No specialist at all just wondering
@@devangoad you could see in the smoke test that indeed air wad flowing through the hole. But extending the blades to nearly touching would probably be even better at airflow and static pressure.
@redcrafterlppa303 which is probably why the hubless fan does not take off. You are just trading off a centralise hub with a hub spread across the the rim.
@@wahidpawana424 I think technically the hubless could push more air but the large baring is killing the efficiency. Maybe if tech evolved some more a levitating version might make the hubless unbeatable and a common appearance.
- both ideas are good, the corner wheels can easily use point/captured ball or other type of cone bearing - plain bearings for very little frictional torque. - the ring may need to be a little more substantial - but testing different types (of blade sets) or tip to root twist would be interesting... (also blade breadth from root to tip may be an interesting variable regarding efficiency vs effective pumping - even using a sweep from tip to root to induce extreme core velocities... hmmmm.) Another benefit of using thermoplastics is the blade units could be printed straight (zero twist) - at the designed "tip pitch" - and the angle set or varied through heat setting - the optimal twist may not be a simple linear twist..
1. Replace the bearings with small sealed bearings as wheels in the corners. 2. Build the frame as 4 parts(cut in half from front to back and side to side) so they can clamp the bearings in and then the fan hub/motor assembly.
@@JETWTF The problem is that the bearings will still need to travel the whole circumference of the outer rim, which translates to much greater speed and friction than if it was a standard hubbed fan. That's not to say your ideas won't help - they would definitely reduce the noise (4 balls instead of many), but the core of the issue remains - the high friction and energy loss. I was thinking about a pair of magnetic bearings instead, but interference with the engine's coils might be an issue.
But what kind of wheels would they be? If they were rubber wheels it would certainly make it ultra-quiet, but the rubber would wear off extremely fast.
I'm quite impressed with this. As you noted late in the clip, maybe it does need to be run in a bit before it reaches it's full potential. A very small amount of machine oil or possibly 'white lithium grease' on the ball race may help to reduce noise. The fact that the whole unit is 3D printed, including the 'iron' cores is what I find really interesting, the boost in performance if you used real iron cores instead could be significant although a LOT more time consuming to set up. I'm guessing there are a few metal impregnated plastics available.
There's some 3d printing material that you can kiln/burn and you end up with a 100% metal print. The tomatoe rocket engine yt guy that I can't remember his name (...Integza) used some in his last rocket video.
@@KareemFloat farmers use graphite powder in their hoppers being pulled by tractors, when they're fertilizing or whatnot, for this exact reason to keep the hopper lubricated. should definitly be considered
@@michaellew1297 the issue isnt the grease breaking it down, but rather being too sticky to be worth doing along with it also holding on to debri and making it turn to a polishing compound. i have yet to run a printed robot/rc where that wasnt the issue. edit: btw this is actually something hes tested and tried in this channel before with a multi part fan and ran into those issues and also landed on graphite being the best option
a few drops of lightweight silicon grease would reduce the bearing rattle a lot. you could print little air scoops on the inward slope that would cool the bearing as well. great design. the center open region might be creating the lower FPM reading. i can envision the air packet leaking off the tips. or go with the latest 'bound tip' from MIT and have the blades meet in the middle in some convincing symmetry like a spiraled needle. I started a freeCAD course a while ago, inspired by your enjoyment, i'm not sure i got past lesson 2 before i got distracted by QGIS and rest servers. I might have been the poster child for ADHD if they had that in the 70s :) carry on, I will live vicariously through your tinkering.
It might, but a large part of the noise is just going to be from the surface finish and a loose fit. Think about riding a bike with really thin tires on cobblestones, every bump in the rolling surface will jostle things around. The best finish you can get off a printer is going to be like the surface of the moon compared to the ground finish you normally get in a bearing raceway, and you're not going to be able to hold tenths to get a snug fit and low eccentricity either. It's a really cool project and proof of concept despite all that though.
A couple of things you could potentially try to reduce the noise is 1. Reduce the amount of bearings contained within the race. In turn should also reduce friction making it spin faster. 2. Use a small amount of grease like you did on that planetary gear fan. That might initially increase friction but once it’s worked itself around it should be lovely and smooth. Plus quieter
Nice build! I remember buying the TMD hubless and thinking it was a no-brainer for my heatsink. I ran it for many years and retired that PC only a few years ago; I'm sure I still have the fan in working condition. Would be happy to send it to you if it's of interest.
Yeah my thoughts exactly. This is one channel I randomly seem to find in the feed & content is always quality, interesting & entertaining. Though this video is so so so more advanced than normally seen. It's amazing what has been achieved. At age 43 I have a degree in industrial design achieved in my very early 20's, during study of CAD we had our projects 3D printed in wax (yes common wax) at a cost of $500 - $900NZD no bigger in size to a computer fan. The company did choose one design to be free of charge, mine was chosen which was totally unexpected by all, as my attendance to class was not of a seen acceptable level though it didn't effect my competence to finish. Off topic a little there sorry. This does entice me a little more to purchase a 3D printer at some point in time, when funds improve or it doesn't cost the earth to only live. As no employment was ever secured from my degree, these memories are but tiny dreams I'm still able to remember. Great channel, keep up the great work. Andy, NZ.
This is so impressive! If you can reduce the bearing noise, there’s SO much potential. You then have complete freedom with fan shape to optimise noise/airflow/static pressure. Please make more variants of this as it’s a fantastic idea
@@_B.C_ Yep, magnetic bearings would be the best choice.. but in a fan that small, it's going to require some incredible engineering. It can be done...
It is remarkably well balanced. I expected it to just up and go when it was freespinning on the table but no, it just stood there and spun away. Very interesting to see what can be done with this concept.
I love this idea. I’ve been thinking that the hub blocks so much potential. I think if this was done professionally or even with resin/injection molded it would be so much quieter and shake less. It may still be good to have the blades touch in the middle for strength and static pressure
My first thought is that the linear velocity at the centre is low enough that you’re not actually missing out on much having the hub there. But who knows, maybe it works like those Dyson fans that draw in more air than they move by themselves. Either way it would be cool to see a refined version.
He could make some steel molds and start mass production of them at that point. Injection molding is always better. Worked in that industry for a few years. Injection molds are perfect
Really one of the coolest hardware channels on here. So cool seeing the effort and thought involved. You're a great communicator for us non-engineers 😛
I think it can be done with mag lev so it's much quiter. And less friction, and i bet there is crazy efficient balde design for hubless fan. It might actually be very good.
Awesome design. One slight design change you should make is to make the blade section easily removable so you can play with different blade configurations
@@agn855 Might be able to just out the blades, thin the wall as much as possible, then put some short dovetails on the inner surface. I think any solution that allows for removable blades is going to lose some wall diameter, but I feel like there's not much that could be done about that.
Yeah, he will def need to experiment with the pitch of the blades to see what moves the most air. A way of balancing it and quieting the bearings would help a lot as well.
@@erictheepic5019 or maybe a screw in hub. Where a hub with a shallow long winded winding is holding to the bearing and a blade ring with a positive screw end to screw into the hub. This way you wouldn't lose anything compared to the static design.
The best build yet. That print quality is amazing. And while the fan is loud as heck, it worked decently, which is good for being a completely custom design. I want to see you run water-cooling tubing through the hole in the middle.
Awesome design! I guess the bearing friction is the limit on RPM, not the moved air. So you could increase blade count and make the blades way steeper to move more air at lower RPM!
I used to re wind dryers for grain bins. You legitimately did a better job than probably 75% of the people I worked with. Amazing. I would like to see this with a non-ferrous metal race for those bearings. I think you might have something here... Especially for applications in air coolers where this would eliminate the dead space occupied by the hub.
Been thinking of doing this for ages! Great job! Wondering if lubricating the bearings would make it much quieter. Eventually, making a metal channel for the bearings with tight tolerances and lubrication would make it quieter and spin faster (less friction)
From what i have seen of 3D printed things they are not really smooth so a lot of noise will be coming off the rough surface of the bearing races and are the ceramic balls smooth enough .
The hole in the middle is leaking air pressure, it works on the boats because the water rams at it as the boat moves, but for air it would benefit from different impeller design that has blades connected in the middle of the fan
I love seeing how happy you are at the end of the video! Creating something and having it turn out better than expectations, theres nothing else like it!
One of the most interesting videos I've seen recently. Thanks! The fact that this is fully 3D-printed, except for the windings and the metal balls, is really mind blowing! I've always thought that rim-based bearings would be the way to go for a 3D-printed design and you've proved that to be the case. This might be a blueprint for 3D printed bearings in the future.
The blur you edited in at 7:30 made me literally laugh out loud haha! Great video my dude! That's a really cool design! I bet if you were to machine it out of aluminum and or steel and with a little bit of lubrication you could probably bring that noise down significantly! With a little bit more enginenerding that could be a pretty sick case fan!
😂😂😂 I didn’t catch that during my viewing, thanks for tagging it. It’s was a sweet gag, glad I got to see it. Thanks again and I tip my hat to you sir/madam.
If it wasn't for the shear pain of loading bearing every time, I'd love to see this go "full fan showdown". Some different blade designs on it could be really interesting.
I had a fan like the TMD on a CPU cooler in 2005-6. I cant remember the brand. It was so very quiet! I loved it until the center bearings failed many years later. I LOVED THAT FAN!!
I own one of those fans! lol I bought it as a curiosity and it remains as that. For a raw prototype, this was amazing! Well done! Even though it wasn't record breaking in air flow, I noticed immediately that the flow was exceptionally smooth without the usual back-pressure some fans exhibit. I believe that refinement of the blade pitch and overall design should provide significant improvements. 1.) anti-magnetic stainless steel or nickel ball bearings - contact Berry Bearings or some other company 2.) cage made from PTFE will reduce chatter in the raceway, may require cutting from flat stock. 3.) use metal impregnated resin to make the raceways or try to find actual naked raceways of required size 4.) final polish pass on raceways / blades and a very small dab of Teflon Butter on the cage (optional) 5.) and finally, you'd need to do some critical balancing on the "rotor", mostly due to varying amounts of glue and small deviations in resin fill. There was a visible oscillation of the bearings from off balance stress (I'm guessing) as you passed through certain speed thresholds. Other thoughts: maybe go with fewer poles as the torque required is not significant. I will say it seems to have a very respectable startup at low settings. If available, consider using small round bar magnets as the retention method could be printed into the rotor. Consider down sizing the wire gauge and increasing the number of windings. This will increase the impedance and flux density while reducing the current required to operate the device. Certain devices, no matter how exactly they are designed or how precisely they are made are, at some point, subject to resonance. One such item is an X-Ray tube that uses a rotating anode inside a glass vacuum envelope (tube). This is a 3 to 5 lb chunk of molybdenum with an iridium coating on the active surface. If allowed to slowly speed up to it's optimal operating speed (10,000 rpm) it will develop a resonance between 600 and 3000 rpm with enough kinetic force to shatter the glass envelope. Boom goes $15,000 X_Ray tube. It has a very specialized startup system that boosts the RPM just before it hits resonance. (on an odd note, the GE CT8800 CT Scanner used a turret bearing from a tank and the rest of the scanning gantry was designed around that. lol)
i recognize the determination to engineer. small mods can always improve . someone has to buy you a bag of graphite powder and some lithium for the desk. :) Thanks for another inspiring video
Awesome to see a TMD-like hubless fan being made. Great proof of concept and being fully 3D printed is amazing. If you leave it running for awhile to wear it in and then hit the bearings with some grease I wonder how much noise and friction reduction there will be. After future iterations on the design are done I'd love to see the community fan blade designs for this.
Dude, this is one of my favorite experimental fans you've made! Great job engineering the motor and bearing system. I'm shocked at how well it worked, and it seems like the motor you made has plenty of power. I think you could really beef up the blade design and get a killer fan out of that with only minor tweaks to the design. Making it quiet would be almost impossible without molded plastics, incredibly tight tolerances, and some sort of lubrication lol, but I would love to see you make a v2 of this fan.
Well damn, a prototype that actually just plain works, that's always nice to see. Well done. I could definitely go for some more experiments on this little devil, starting with improving the acoustic performance let's say 😅
Glad to say I've been on board since the beginning of this series. It's been really cool to see the evolution of this project over time as well as the community participation. Cool stuff :)
that turned out great! good job sir. you could design a ring/race with v slots to allow changing of fan blades, and do a season of showdown with people designing blades for this thing edit: you might look up halbach array if you want to improve performance. it's a way to get more strength out of your magnets by the way you arrange them, and they're efficient because they have a strong and weak side, so the force is directed and concentrated towards your armature. to fit the magnets in this arrangement you might need to reduce the number of coils, but you'll get a side effect of bigger coils, so hopefully more torque. by the way, if you want lower speed but stronger torque, you can go down to a smaller wire and do way more turns of wire. if you do that, you could design some really high pitch blades with close spacing that go to the center and probably get good static pressure. if you go larger wire but fewer turns, you'll have less torque but higher speed. would be more for cfm in that case, so less blade pitch as for bearings, i wonder if it's possible to do magnetically levitating bearings. that would solve the bearing noise issue oh and i have an idea. what if you leave slots in the chassis and blade assembly for the raceways. then print the raceways in something harder like ASA which my bambu prints quite nicely. if you set it to 100% walls and no infill, and print it flat on the build plate, the grain goes with the direction of the circle, so you shouldn't end up with perpendicular lines that cause the bearings to make noise when they cross them. and maybe even add a light abrasive or even acetone and preprocess them so they end up really smooth. could probably add graphite for lubrication as i'm not sure how ASA or similar would stand up to something like lithium grease. chemistry guys will have to learn us that bit but i think oils usually break plastics down, so that was why i had this thought, but i could be wrong
Doesn't really make much sense beyond looking cool though. Since the blade velocity near the center is so low its not producing much pressure/flow there anyway. If anything you're losing static pressure as air isn't being prevented from back flowing by the solid hub. And the bearing on the edge means you have to have slightly shorter blades for the frame. Meaning you get less efficient blade surface area in the same 120mm package. Also, at high RPMs this bearing is going to be more difficult to tolerance, which would require more expensive materials and more engineering hours, due to the extra force of the faster velocity.
You know I can actually see a very useful application for this. If your cabling needs to go down the same pipe as your in-line blower then having one of these with an open ring in the middle to run your cable through would make these applications much more simple! This is such a neat idea!!
I think the reason why the original tip driven fan had the hub was two fold: first, the fan doesn't push much air through there, simply because there's no velocity there. And so you might as well keep the hub to keep the air from accidentally moving back through the center. And once you do, then it's a great place to put a regular fan bearing.
Yeah. By doing it on the outside he's increasing the area and moving mass of the bearing significantly. Since he already has permanent magnets in place, I wonder if there is a way to magnetically float the center?
@@ventilate4267 Connecting the fan blades in the middle would lower the overall flow. In the fog segment, you can see that there is actually air going through the center, which is sucked in/carried with the surrounding air. If you would connect the blades, you would obstruct this path without any acceleration of the air from the blades in that area. Therefore many ubless props only use short stubby blades and leave the rest empty.
The fitment just shows your engineering prowess, good job. I am not sure what it could be used for considering the sound, a maglev bearing would be cool lol. I also like the flow characteristics.
The maglev bearing could be the best feature of this design. I wonder if some of the maglev trains designs could be used that would not need complicated setup with sensors and controllers.
I think this is a solution looking for a problem. Already in a fan design the center is the least impactful to the airflow. So then you have to make trades. Driving it from the outside requires faster speeds, more bearings, more friction, a better gear ratio, and more mass. Dropping the efficiency. And really what we're after here is most is efficiency, among things like quietness, form factor, size, etc. Otherwise all pc fans would just use a 30,000rpm motor, draw 10 amps, and sound like a wind tunnel. Not saying innovation is a bad thing, but optimizing one parameter at the expense of many others usually doesn't pan out
So I designed a hubless impeller for watercraft quite a while ago, it's rather more useful because water doesn't compress like air does. I couldn't even attempt to get the parts made to the necessary tolerences back then, so I never persued it. I've been wanting to try it on an air fan for a while though, so this is super cool to see. Tip one, bearings do not have circular tracks in the race. THe bearing rides on two microscopic ridges on each half, so it's clear at the bottom and on the sides, like it's on rails. This reduces friction dramatically. Bearings also have no clearance, the ball is touching both races all the time. That's where your noise is coming from. The clearances necessary to do this are crazy tight, you have no chance of a printer getting anywhere close. Maybe print undersized and lap to fit with a disk? Tip two, I would make the magnet pockets load from the inside, so the centrifugal and magnetic forces are pulling the magnet INTO the pocket, as opposed to trying to pull them out. Shouldn't need glue, but it's probably helpful to stop vibration and wear. Suggestion, I would test making the pole pieces for each coil much wider, approaching a full ring as opposed to small bars. I think in this design you want there to be a minimum gap between each pole and between the poles and the magnetic ring? I havent worked with motor design in decades so I don't remember much lol.
You don't need a center-hub for a mag-lev fan. However, you need to place the magnets exceedingly precisely, the magnets have to extremely similar in strength, and have a very exact shaped magnetic field for it to work without the hub flying out at high RPM. Either that or find the strongest permanent magnets you can. That way the field might be strong enough to contain the hub at any RPM. Not sure if the Neodymium magnets are strong enough for that though, and stronger rare-earth magnets are ridiculously expensive...
@@The_Keeper I guess I should have been more clear, no you do not NEED a hub, but doing it without one would be exceedingly difficult outside of a lab with very precise equipment...Also it would the worlds most expensive computer fan...
I think that connecting the fan blades to a hollow axial cylinder in the middle could increase efficiency by eliminating vortices at the tips of the blades, while still allowing air to pass through.
Dude, this is BRILLIANT. I really want to see this refined and iterated to it's best performance. You have a really good series in your hand right now.
A much cheaper option for the bearing balls would be airsoft BBs. I've used some cheap white ones that have an amazingly high roundness spec with good results. Edit: You'd have to modify the CAD model a bit though, as they're 6mm diameter.
you are supposed to add all the ball bearings on one side and then move the core over. And then space them out and then place the cage. A little grease makes this easier since it keeps the bearings n place. Larger balls and smoother raceways would make it quieter.
- there are various ways these things are done, depending on architecture - your "suggested" method is common for deep grove bearings - with welded cages.. Designing integrated raceways can be a limiting factor - as clearances may be too tight (balls too large)- and access limited for conventional methods of assembly.
I suspect a large reason for all the noise is the surface smoothness of the running surfaces for the ball bearing. Throw in some imbalance in the rotating assembly for good measure as it looked and sounded like it was imbalanced. Definitely a cool proof of concept. Certainly opens up room to look at the effect of hub size on flow/flow distribution.
I wanna see how far the blades and bearings can be optimized. I would also like to see one specifically for airflow and another for static pressure. This could actually be a competitive product because it could be made to have no gaps.
The most important part of a fan is the outermost diameter. The further you go in the less it matters. This is why you stuff the motor and bearings in the center. This fan design will fundamentally always move less air for a given size/power.
@@Anyone700 That makes sense to a point but respectfully I am going to have to disagree. Although the first part seems true because as you go inward the rotation slows. My thinking is that the fact there is no motor in the center means less obstructions, both from the motor and the little arms that often hold it in place. I am fully ok with being proven wrong though.
@@DoubsGaming my reasoning is that putting the motor on the rim makes the fan a type 2 lever, hence need a drive that is low torque but high speed. A electrical motor is full torque all times. so if we use the motor in a type 2 lever it is actually less efficient than the hub drive. from an efficiency stand point hub drives are less fitting for the task
@@DoubsGaming the central part can't create significant air pressure, so the engine is placed here. "Hubless" is good for directing airflow, but can't deliver the same air pressure with an equivalent form factor, which is more important. The structural complexity is insane for a product of this type. There is no way how it can compete with "classic" design (except style).
@@Timberwolf69 The issues with the old TMD fan: 1) The TMD was ONLY made in 70 mm Fan size (Most Fans smaller than 80 mm gets loud)! 2) They were running it at 5800 RPM 3) I don't think they used the quality bearings that we have today, especially nice ones like MagLev Bearings
Dude you are amazing you eyeballed a hub bearing and fan blade geometry in a hub-less fan you pulled from a drawing on the web eyeballing and guestamating at the measurements you were able to nail industry standard CFM rates (better actually 50 CFM is pretty good) and did it all with a 3d printer and i didnt hear it fall out of balance till right at the end. Thats so cool!!!!
Every radiator have a gap between fan and actual fins. Air need space to spread and calm down before it enters the fins. In case of radiators Pressure and bulk airflow matters, not area. less area even better, less area usually means higher pressure - for example 120mm fans have more static pressure than 140mm fans, and radiators for 140 fans have less fins density in general (to reduce resistance)
I think you'd get a lot of noise reduction with some kind of silicone or other lubricant on the bearings, I'd love to see a Round 2 with something like that. I think finding the correct balance of film strength and just overall viscosity of the lubricant will be the largest challenge. Using actual petroleum bearing grease would likely be too thick to allow good movement-- but maybe I'm mistaken, maybe without a centralized motor like you might be able to generate more torque to overcome the resistance of the grease. Beyond that though using something like WD40 which has little to no adhesion or film strength I think would probably provide a negligible (if any) benefit. Silicone lubricant is frequently used on plastics and would probably provide a good balance between too much and too little adhesion and viscosity.
@@tacticalcenter8658 if you use a light oil I doubt it. Gotta remember that it's not a perfectly smooth metal ball against another piece of metal, it's on plastic. I'm certain that if the oil was able to get in-between the more porous plastic and the ball, it'd improve performance, and noise. In theory, in my head. I could totally be wrong though.
The usefulness of oil in bearings is a hotly debated topic, realistically my personal guess is a small amount of a light thin oil, would probably help.But I have no experience with ceramic and plastic bearings to my knowledge. But last I've heard it's very much design, application specific, along with things like material choices. But yeah it's hard to say, generally speaking of you lube a bearing it's using a small amount of a thin lubricant. That being said even properly lubed I still expect this to be able to wake the dead.
I'm quite confident that this design has future. Like many people already said, this thing could definitely use some lube, and I'd personally go with some liquid and thick one, maybe some Krytox 205g0, like keyboard enthusiasts use for their keyboard switches, and all the surfaces that have friction should be as smooth as possible, some polishing should help. With these relatively low-effort and low-cost improvements, I hope we'll see a v2 soon!
Close the gaps between the blades in a way that forces the air through the middle hole. Can you make a video trying to take advantage of this build and send concentrated air through the middle hole. How fast can you make the air go.
What you need to test is a 6 or 9 pole design as that will increase peak velocity. Then add steel ball bearings and races with steel wire guides to reduce rattle and friction. Should up to double speed and airflow.
The big noise maker here, as is more than likely obvious, is the ball bearings. If you were to switch to a maglev design, it would be a lot quieter. Also, the fan blades could be optimized significantly which would almost certainly give you better CFM at a given RPM. Also, as mentioned elsewhere, balancing everything out would also help reduce noise and vibration. I'd be very interested to see how far you can take the design overall. Nicely done so far :)
My biggest concern is that the maglev may interfere with the rotators. But if they're not strong enough, they won't do their job properly. What you need is 2 maglev tracks sloped to 45 degree to cup the v-edged fan unit. Make sure this channel is perfectly smooth as catches on startup or shutdown would be disastrous. Try to keep the space as tight as possible and of course add lubricant, and send me royalties on the patent! XD
You can use 4 small off-the-shelf ball-bearings in the corners to hold the impeller in place, instead of the large peripheral one. Also, only 4 windings in the corners (as in the old video) will be sufficient to move the impeller, but will allow to increase the diameter of the impeller especially in the periphery where it works best and should significantly increase the output.
I've made bearings out of pla. They benefit from running in at low speed for 30 mins or so and some lube. I use silicone grease and it will work a lot lot better and quiter. The thing I like about hubless is that its actually naturally an air multiplier. The centre part drags more air through than the swept distance.
An idea might be trying slightly thinner gauge wire and more winding on each core section to have a higher impedance. Might be able to increase your speeds doing that. Also with some tweaks to the bearing design then I think this could make an excellent unique fan
The problem with thinner gauge wire is increased risk of burning out the coils, so lower voltage tolerance. This might work for a low power, high torque fan? Bearing in mind I'm no electrician and my response is based on logical assumption alone.
Very cool build! Although, I wonder if the wobble it seems to have in the start of it's movement has any affect on it's final numbers. It feels like it has more potential in it's design. Would that un-evenness have made it slower? Would solving that make it quieter?
Hell of a good job with creating this. In terms of performance I can not really see benefits compared to traditional center hub. The most amount of air is being moved on the outer part of the fan, not the center. Here the outer space is being taken because of the motor parts, so fan diameter is smaller, like some argb fan. Still very impressed for your design that is very thin especially for a 3d printed prototype. For example when you look at small, 30 or 40mm fans you will see that hub takes a lot of space. Sometimes even 70% of the fan. It would be possible to make smaller hub with less powerful motor, but increasing blades in that inner space close to the center is not so beneficial. At least not enough to compensate for power of bigger and stronger motor. 😂
Okay, this is the first new kinda fan I've seen on this channel that I actually wanna see. Noctua needs to get on this right now, I feel like there's a lot of potential here. It just looks futuristic seeing a hubless fan and I can see enthusiasts using these in gaming PC's
With the hub gone it gets rid of a huge obstruction that takes up probably 30 plus percent of the fan area. In the smoke test you see the smoke actually center itself concentrating the air to the center. The center looks to be a lower pressure but that can be fixed with different blade profiles. With a better bearing solution this can only get better.
The reason for the noise is because of a non-perfectly symmetrical fan assembly, if you pay close attention at a moderate RPM, you will see the fan blade assembly is shifting resulting in the noise. Your fit is a bit loose, but this is normal with ball bearings you can try adding a lubricant that is a bit extract viscous to cushion the vibration to absorb much of the noise of the loose bearings rattling on imperfections of your fan and frame.
Bravo! Very well done! Here are a couple of ideas for you to try: #1 Arrange your magnets in a halbach array to optimize the magnetic field #2 Eliminate mechanical bearings and go with magnetic bearings. Your noise problem will be solved! ( not sure how, but no doubt you will research both of these ideas and figure it out ) Cheers, John
The tmd fan shown at 0:33 is the same one used on a heatsink i used for a socket A(462) motherboard i recently configured, looks and sounds like a jet engine, cool to see how the fan operates and someone taking the design further.
I like what you do. One of many often overlooked aspect of forced air cooling is air speed. FASTER IS NOT BETTER. If the air isn't spending enough time around the aluminum heatsink or whatever device is being used it will not soak up enough heat and there comes a point where the air is moving too fast and becomes less effective.
I have made a couple of these hubless fans myself. Let me help you fix the noise. All you need to do is change the orientation of your copper wiring by making it sideways rather than inwards and rotate it so that they face towards each other, duplicate it so that it is now found on both shallow & deep ends with empty room between them and then allow for magnets to be free flowing between these copper wire spools. There you go, Noiseless rotation. When it boils down to air flow, they perform a lot better by their fins taking after the shape of a rose flower petal (or a Pringles chip at a rotation) and coming real close to overlapping one another.
unfortunately because air is a lot less viscous than water, it might end up just backflowing or stalling in the center, a hub might actually be slightly beneficial if it's got decent aero
Interesting fan. I look forward to seeing how this improves airflow in the future. The question is how to support the fan, which could be magnetic floating, ground effect, minimum center bearing, etc. An interesting thing might be to use the ground effect. Centrifugal force compresses air and uses that compressed air in ground effect bearings. There are many other possibilities, such as installing a rail on the outer ring of the fan and supporting the rail with about three wheels.
I noticed that the older, TMV fan was not truly hubless. There's a spindle hub on which the fan is supported while it is rim driven, while your noisy beast is truly hubless. The spindle mounted / rim driven design has opportunities for quieter operation. With multiple spindle mounted / rim driven units you could produce something akin to a jet compressor.
Love the idea and the result. Hats off! The only problem is that the fan is most efficient at the outer edge and least efficient at the center, kinda like a hard drive. Unless you can make the motor and bearing ultra thin, the center driven fan will be more efficient.
This concept is awesome and you got a working prototype with a 3D printer, some magnets and some wire. Great Job! One recommendation I would make is to make an alternative frame that runs on a much simpler mechanism, like a drive sprocket with a belt and a stand alone motor. That way you would be able to test out multiple iterations of blade design without having to attach magnets to every one, and once you found a winner you could make a proper one.
Just chiming in to say that that segment from 10:58 to 11:40--especially 11:22 to 11:40--was awesome. That's a camera angle you should.use for all future fan showdown videos.
That was really fun! Please make that the next fan show down season. There are so many components that can be tweaked to get the best air mover. Not sure how, but you could come up with a point system, for ie 2000 RPM = 2000 points, 91db sound = minus 100 points per 10db. , mili amp = minus points, air flow = plus points. Style points, easy to print, etc. Thanks for sharing your project! :-)
I have a Baratza Sette 270 coffee grinder that works kinda similar to this. The center conical burr is stationary, and the ring burr on the outside is the part that spins. It comes in a self contained gearbox that rotates around the center burr. It’s quite effective for espresso with very little retention, but not great for airflow I’d imagine.
Temperature and magic smoke loss is inversely proportional to surface area. Moving the “hub” to the ring increased its area by some factor or maybe an order of magnitude (I am not going to calculate this). So you can probably increase the amperage by quite a bit before the magic smoke escapes.
Okay, this is awesome. Imagine a balanced fan with better bearings and optimized blades.
Would it be hard to try to make it a giant maglev bearing with some magnets and the correct shape of the outer ring ? Or if you could use the drive coils themselves to hold the ring while running.
@@Maaniic I suspect magnetic field of maglev bearing will interfere with mechanics of fan itself, although I guess you can minize it by making them rather tight fitting (below 1mm distance), but reducing magnet's power (maybe even make it 2 rings on opposite sides at 45° incline). All that said, it will take quite a lot manual labor to place all the magnets, since you would probably need to place hundreds of them
PS fans like it probably would cost quite a lot more (even if produced at scale) due to much copper and magnets being required, as well as more complicated bearings. Not to mention that lateral speed at outer edge it much faster than at inner edge, making it significantly harder to make bearings quite, on the other hand if one could nail down bearings design really well, he might potentially achieve quieter fan by removing the hub supports
@@MaaniicMaglev requires superconductivity to operate, our current tech relies on either very low temps at normal pressure, or super high pressure at more reasonable temperatures, neither of which is very practical. It would make for a great show though.
Let's create a competitor for the brown company
A Dutch team in the ugh hyperloop competition used aluminium and magnets to levitate. It worked. Im sure one can also make a magnetic bearing with just magnets.
As a "Proof of concept" type project, this is amazing. The problem is, now I would like to see what it's capable of with refinement.
It would definitely be a high dollar project. 120x120 ball bearings are like $40-$60. Not sure how much ceramic ones are but possibly more expensive. I think the best thing to do is make one out of CNC parts and then optimize the blade design first.
@@fuzzyturtlez8994 wouldnt even be worse in terms of friction with that big bearing? I actually think this is probably a worse fan design than what we have today.
@@AndrewTSq It is, takes more magnets and coils, plus has more friction, which is the enemy of a fan lol
Not much of a "proof of concept" given that it already existed and was proven to work. But okay.
@@AndrewTSq It is worse. If you look at swept area you lose more space on the outside compared to what you gain on the outside.
Then the amount of loading on the bearing is nowhere near max capacity. So smaller balls if possible would be nice to have.
What I want to see is a community driven series where we try to build the BEST hubless fan together.
Yessssss please
Give me fans or give me deaf
@@puerlatinophilus3037 ha, nice
Yes please
yes please :) what an idea
I think using steel ball bearings would quiet it down quite a bit, because of the resonance difference. The poly-crystalline structure of ceramic is very brittle, hard, basically making it noisy. Using a steel ball bearing along with a dry lubricant like graphite would drop the noise immensely. Maybe a resin printed channel for the ball bearings to ride in would help a bit too. This project is AWESOME!! I would love to see you take your awesome idea farther, and I am amazed at your 18 pole handcrafted motor's performance.
+1 for another go with some dry lube or graphite lube, such as Drop'L Do
A good PTFE base lubed would be better but dry lubes wont help with noise as much as a liquid or paste type lube although they would slow it down. Trying some of each type to compare would be a good video idea to go with other improvements.
If he could turn the channels out of aluminum that would be ideal
@@corpsemasterB312 I agree, although he did state that he wanted to keep it as 3D printed as possible.
Def some sort of lube on the bearings would help. I was thinking something like a grease, but know that would make a tremendous mess. Graphite is a good idea, though it will still be hard to keep whatever lubricant you use on the bearings and not all over everything else.
I remember when that tip driven fan came out. I thought it would be great, but it was so expensive. It never really took off because it had too many down sides. The bearing was in the hub, which was why it still had the hub, but that was far superior to having the bearing around the edge like yours. The reason the hubless design was used for the boat thruster is because water isn't as compressible as air. It's actually needed in water.
Even though air is more compressible, wouldn’t having a clear channel of flow reduce the amount of work it has to do? No specialist at all just wondering
@@devangoad would it be same working principle of bladless fans?
@@devangoad you could see in the smoke test that indeed air wad flowing through the hole. But extending the blades to nearly touching would probably be even better at airflow and static pressure.
@redcrafterlppa303 which is probably why the hubless fan does not take off. You are just trading off a centralise hub with a hub spread across the the rim.
@@wahidpawana424 I think technically the hubless could push more air but the large baring is killing the efficiency. Maybe if tech evolved some more a levitating version might make the hubless unbeatable and a common appearance.
You should try the fan with some lube in the bearings.
Im wondering if Graphite lubricant would help or make more heat or not.
I would try silicone lubricant
I'd suggest silicone grease...
Some thin lithium grease
Vaseline. Very tiny amount.
1. replace the bearings with just wheels in the corners
2. build the outer ring with 3 dove tails so that we can submit press fot blade sets
- both ideas are good, the corner wheels can easily use point/captured ball or other type of cone bearing - plain bearings for very little frictional torque.
- the ring may need to be a little more substantial - but testing different types (of blade sets) or tip to root twist would be interesting... (also blade breadth from root to tip may be an interesting variable regarding efficiency vs effective pumping - even using a sweep from tip to root to induce extreme core velocities... hmmmm.)
Another benefit of using thermoplastics is the blade units could be printed straight (zero twist) - at the designed "tip pitch" - and the angle set or varied through heat setting - the optimal twist may not be a simple linear twist..
1. Replace the bearings with small sealed bearings as wheels in the corners.
2. Build the frame as 4 parts(cut in half from front to back and side to side) so they can clamp the bearings in and then the fan hub/motor assembly.
@@JETWTF The problem is that the bearings will still need to travel the whole circumference of the outer rim, which translates to much greater speed and friction than if it was a standard hubbed fan. That's not to say your ideas won't help - they would definitely reduce the noise (4 balls instead of many), but the core of the issue remains - the high friction and energy loss.
I was thinking about a pair of magnetic bearings instead, but interference with the engine's coils might be an issue.
This
This soooo much
Please do this
But what kind of wheels would they be? If they were rubber wheels it would certainly make it ultra-quiet, but the rubber would wear off extremely fast.
I'm quite impressed with this. As you noted late in the clip, maybe it does need to be run in a bit before it reaches it's full potential. A very small amount of machine oil or possibly 'white lithium grease' on the ball race may help to reduce noise. The fact that the whole unit is 3D printed, including the 'iron' cores is what I find really interesting, the boost in performance if you used real iron cores instead could be significant although a LOT more time consuming to set up. I'm guessing there are a few metal impregnated plastics available.
plastic and grease dont tend to work well from my experience. graphite powder is his best bet as it's very slick without any viscosity.
There's some 3d printing material that you can kiln/burn and you end up with a 100% metal print. The tomatoe rocket engine yt guy that I can't remember his name (...Integza) used some in his last rocket video.
@@KareemFloat PLA is not affected by grease, and greasing 3d printed parts is commonly done. (Plastics are not plastics.)
@@KareemFloat farmers use graphite powder in their hoppers being pulled by tractors, when they're fertilizing or whatnot, for this exact reason to keep the hopper lubricated. should definitly be considered
@@michaellew1297 the issue isnt the grease breaking it down, but rather being too sticky to be worth doing along with it also holding on to debri and making it turn to a polishing compound.
i have yet to run a printed robot/rc where that wasnt the issue.
edit: btw this is actually something hes tested and tried in this channel before with a multi part fan and ran into those issues and also landed on graphite being the best option
a few drops of lightweight silicon grease would reduce the bearing rattle a lot. you could print little air scoops on the inward slope that would cool the bearing as well. great design. the center open region might be creating the lower FPM reading. i can envision the air packet leaking off the tips. or go with the latest 'bound tip' from MIT and have the blades meet in the middle in some convincing symmetry like a spiraled needle. I started a freeCAD course a while ago, inspired by your enjoyment, i'm not sure i got past lesson 2 before i got distracted by QGIS and rest servers. I might have been the poster child for ADHD if they had that in the 70s :) carry on, I will live vicariously through your tinkering.
I don't think the bearing should have any chance to get dust in them, especially with oil on them.
It might, but a large part of the noise is just going to be from the surface finish and a loose fit. Think about riding a bike with really thin tires on cobblestones, every bump in the rolling surface will jostle things around. The best finish you can get off a printer is going to be like the surface of the moon compared to the ground finish you normally get in a bearing raceway, and you're not going to be able to hold tenths to get a snug fit and low eccentricity either.
It's a really cool project and proof of concept despite all that though.
A couple of things you could potentially try to reduce the noise is
1. Reduce the amount of bearings contained within the race. In turn should also reduce friction making it spin faster.
2. Use a small amount of grease like you did on that planetary gear fan. That might initially increase friction but once it’s worked itself around it should be lovely and smooth. Plus quieter
Nice build! I remember buying the TMD hubless and thinking it was a no-brainer for my heatsink. I ran it for many years and retired that PC only a few years ago; I'm sure I still have the fan in working condition. Would be happy to send it to you if it's of interest.
Idk if he will respond but I should. I'm sure he'd show it.
gotta love this so it will be bumped up
currently the 3rd places in top comments, hopefully James will read it
Bump.
123 likes POG
I still have two from my old dual Pentium III system.
This may have been one of the most impressive things I've ever seen you do
Yeah my thoughts exactly. This is one channel I randomly seem to find in the feed & content is always quality, interesting & entertaining. Though this video is so so so more advanced than normally seen. It's amazing what has been achieved. At age 43 I have a degree in industrial design achieved in my very early 20's, during study of CAD we had our projects 3D printed in wax (yes common wax) at a cost of $500 - $900NZD no bigger in size to a computer fan. The company did choose one design to be free of charge, mine was chosen which was totally unexpected by all, as my attendance to class was not of a seen acceptable level though it didn't effect my competence to finish. Off topic a little there sorry. This does entice me a little more to purchase a 3D printer at some point in time, when funds improve or it doesn't cost the earth to only live. As no employment was ever secured from my degree, these memories are but tiny dreams I'm still able to remember. Great channel, keep up the great work. Andy, NZ.
This is so impressive! If you can reduce the bearing noise, there’s SO much potential. You then have complete freedom with fan shape to optimise noise/airflow/static pressure. Please make more variants of this as it’s a fantastic idea
If he’s already using magnets, why not magnetic levitation?
@@_B.C_ That was the first thing that came to my mind, too.
@@_B.C_ Yep, magnetic bearings would be the best choice.. but in a fan that small, it's going to require some incredible engineering. It can be done...
It is remarkably well balanced. I expected it to just up and go when it was freespinning on the table but no, it just stood there and spun away. Very interesting to see what can be done with this concept.
I love this idea. I’ve been thinking that the hub blocks so much potential. I think if this was done professionally or even with resin/injection molded it would be so much quieter and shake less. It may still be good to have the blades touch in the middle for strength and static pressure
My first thought is that the linear velocity at the centre is low enough that you’re not actually missing out on much having the hub there. But who knows, maybe it works like those Dyson fans that draw in more air than they move by themselves. Either way it would be cool to see a refined version.
He could make some steel molds and start mass production of them at that point. Injection molding is always better. Worked in that industry for a few years. Injection molds are perfect
Really one of the coolest hardware channels on here. So cool seeing the effort and thought involved. You're a great communicator for us non-engineers 😛
Absolutely!
As others are saying, it would be cool to have like a mini-season of the fan showdown with this fan design.
This is really impressive no matter how it performed. The way the light through the red fan made the fog look blood red was also a pretty cool effect.
I think it can be done with mag lev so it's much quiter. And less friction, and i bet there is crazy efficient balde design for hubless fan. It might actually be very good.
Love this! Loudest most over engineered fan, that's almost impractical... That's why I keep coming back to this channel!!! ❤️❤️❤️
I still think the pretty one with me multiple gears was more over engineered
Awesome design. One slight design change you should make is to make the blade section easily removable so you can play with different blade configurations
I highly doubt that it’s only a slight change, but thought exactly the same. Kudos
Just put on some lube it might be softer
@@agn855 Might be able to just out the blades, thin the wall as much as possible, then put some short dovetails on the inner surface. I think any solution that allows for removable blades is going to lose some wall diameter, but I feel like there's not much that could be done about that.
Yeah, he will def need to experiment with the pitch of the blades to see what moves the most air. A way of balancing it and quieting the bearings would help a lot as well.
@@erictheepic5019 or maybe a screw in hub. Where a hub with a shallow long winded winding is holding to the bearing and a blade ring with a positive screw end to screw into the hub. This way you wouldn't lose anything compared to the static design.
The best build yet. That print quality is amazing. And while the fan is loud as heck, it worked decently, which is good for being a completely custom design. I want to see you run water-cooling tubing through the hole in the middle.
Awesome design! I guess the bearing friction is the limit on RPM, not the moved air. So you could increase blade count and make the blades way steeper to move more air at lower RPM!
Cool build, and actually spun well.
Would be cool to see a version that allows press-in fans for a community series.
I used to re wind dryers for grain bins. You legitimately did a better job than probably 75% of the people I worked with. Amazing. I would like to see this with a non-ferrous metal race for those bearings. I think you might have something here... Especially for applications in air coolers where this would eliminate the dead space occupied by the hub.
I think even a non-ferrous race would slow it a LOT as the free-spinning neodymium magnets induce opposing eddy currents as they fly past it.
The blades in the hub area are mostly dead space anyway though. Since the blade velocity is much lower near the center than the edges.
Been thinking of doing this for ages! Great job! Wondering if lubricating the bearings would make it much quieter. Eventually, making a metal channel for the bearings with tight tolerances and lubrication would make it quieter and spin faster (less friction)
Was thinking lube too and alot of it. Might even make the fan faster with less resistance in the bearings.
From what i have seen of 3D printed things they are not really smooth so a lot of noise will be coming off the rough surface of the bearing races and are the ceramic balls smooth enough .
A grease would likely make it quieter but is likely to slow the fan down due to friction. Under light loads lube adds to friction.
That is a beautiful piece of engineering, very nicely done! I also have trouble using any of my other printers after getting an X1CC…
The hole in the middle is leaking air pressure, it works on the boats because the water rams at it as the boat moves, but for air it would benefit from different impeller design that has blades connected in the middle of the fan
So cross bar blades with a taper from edge to core? Like a cross?
This needs a two sided fan like a turbo. A large extended air scoop type blade forcing air into a sucking black hole vacuum fan on the exit side
@@DominicRyanOsborne yes
I love seeing how happy you are at the end of the video! Creating something and having it turn out better than expectations, theres nothing else like it!
One of the most interesting videos I've seen recently. Thanks! The fact that this is fully 3D-printed, except for the windings and the metal balls, is really mind blowing! I've always thought that rim-based bearings would be the way to go for a 3D-printed design and you've proved that to be the case. This might be a blueprint for 3D printed bearings in the future.
The blur you edited in at 7:30 made me literally laugh out loud haha!
Great video my dude! That's a really cool design! I bet if you were to machine it out of aluminum and or steel and with a little bit of lubrication you could probably bring that noise down significantly! With a little bit more enginenerding that could be a pretty sick case fan!
😂😂😂 I didn’t catch that during my viewing, thanks for tagging it. It’s was a sweet gag, glad I got to see it. Thanks again and I tip my hat to you sir/madam.
If it wasn't for the shear pain of loading bearing every time, I'd love to see this go "full fan showdown". Some different blade designs on it could be really interesting.
I bet he could make come kind of adapter. Hopefully a hubless showdown is incoming.
@@8g00gl yeah, outer rim and the inner part could be separate and snap together.
Dude! This is next-level DIY fan design. I had one of those TMD fans waaaay back in the day (I think I might still have it somewhere, actually!)
I had a fan like the TMD on a CPU cooler in 2005-6. I cant remember the brand. It was so very quiet! I loved it until the center bearings failed many years later. I LOVED THAT FAN!!
I own one of those fans! lol I bought it as a curiosity and it remains as that.
For a raw prototype, this was amazing! Well done! Even though it wasn't record breaking in air flow, I noticed immediately that the flow was exceptionally smooth without the usual back-pressure some fans exhibit. I believe that refinement of the blade pitch and overall design should provide significant improvements.
1.) anti-magnetic stainless steel or nickel ball bearings - contact Berry Bearings or some other company
2.) cage made from PTFE will reduce chatter in the raceway, may require cutting from flat stock.
3.) use metal impregnated resin to make the raceways or try to find actual naked raceways of required size
4.) final polish pass on raceways / blades and a very small dab of Teflon Butter on the cage (optional)
5.) and finally, you'd need to do some critical balancing on the "rotor", mostly due to varying amounts of glue and small deviations in resin fill. There was a visible oscillation of the bearings from off balance stress (I'm guessing) as you passed through certain speed thresholds.
Other thoughts:
maybe go with fewer poles as the torque required is not significant. I will say it seems to have a very respectable startup at low settings.
If available, consider using small round bar magnets as the retention method could be printed into the rotor.
Consider down sizing the wire gauge and increasing the number of windings. This will increase the impedance and flux density while reducing the current required to operate the device.
Certain devices, no matter how exactly they are designed or how precisely they are made are, at some point, subject to resonance. One such item is an X-Ray tube that uses a rotating anode inside a glass vacuum envelope (tube). This is a 3 to 5 lb chunk of molybdenum with an iridium coating on the active surface. If allowed to slowly speed up to it's optimal operating speed (10,000 rpm) it will develop a resonance between 600 and 3000 rpm with enough kinetic force to shatter the glass envelope. Boom goes $15,000 X_Ray tube. It has a very specialized startup system that boosts the RPM just before it hits resonance.
(on an odd note, the GE CT8800 CT Scanner used a turret bearing from a tank and the rest of the scanning gantry was designed around that. lol)
i recognize the determination to engineer. small mods can always improve . someone has to buy you a bag of graphite powder and some lithium for the desk. :) Thanks for another inspiring video
That is super cool! I would love to see you make a V2 with a different blade design to try and get better flow.
Awesome to see a TMD-like hubless fan being made. Great proof of concept and being fully 3D printed is amazing. If you leave it running for awhile to wear it in and then hit the bearings with some grease I wonder how much noise and friction reduction there will be. After future iterations on the design are done I'd love to see the community fan blade designs for this.
Love it! With better blade geometry, closing the "air hole" in the middle, and usuing a magnetic bearing... This is sick
Dude, this is one of my favorite experimental fans you've made! Great job engineering the motor and bearing system. I'm shocked at how well it worked, and it seems like the motor you made has plenty of power. I think you could really beef up the blade design and get a killer fan out of that with only minor tweaks to the design. Making it quiet would be almost impossible without molded plastics, incredibly tight tolerances, and some sort of lubrication lol, but I would love to see you make a v2 of this fan.
Well damn, a prototype that actually just plain works, that's always nice to see. Well done. I could definitely go for some more experiments on this little devil, starting with improving the acoustic performance let's say 😅
Being used to prototypes on other channels, I expected it to be scraping and for it to seize up or catch on fire within minutes.
@@user2C47 Not just other channels. I myself have fried a LOT of prototypes on almost anything imaginable.. :P
Glad to say I've been on board since the beginning of this series. It's been really cool to see the evolution of this project over time as well as the community participation. Cool stuff :)
that turned out great! good job sir. you could design a ring/race with v slots to allow changing of fan blades, and do a season of showdown with people designing blades for this thing
edit: you might look up halbach array if you want to improve performance. it's a way to get more strength out of your magnets by the way you arrange them, and they're efficient because they have a strong and weak side, so the force is directed and concentrated towards your armature. to fit the magnets in this arrangement you might need to reduce the number of coils, but you'll get a side effect of bigger coils, so hopefully more torque.
by the way, if you want lower speed but stronger torque, you can go down to a smaller wire and do way more turns of wire. if you do that, you could design some really high pitch blades with close spacing that go to the center and probably get good static pressure. if you go larger wire but fewer turns, you'll have less torque but higher speed. would be more for cfm in that case, so less blade pitch
as for bearings, i wonder if it's possible to do magnetically levitating bearings. that would solve the bearing noise issue
oh and i have an idea. what if you leave slots in the chassis and blade assembly for the raceways. then print the raceways in something harder like ASA which my bambu prints quite nicely. if you set it to 100% walls and no infill, and print it flat on the build plate, the grain goes with the direction of the circle, so you shouldn't end up with perpendicular lines that cause the bearings to make noise when they cross them. and maybe even add a light abrasive or even acetone and preprocess them so they end up really smooth. could probably add graphite for lubrication as i'm not sure how ASA or similar would stand up to something like lithium grease. chemistry guys will have to learn us that bit but i think oils usually break plastics down, so that was why i had this thought, but i could be wrong
I _really_ wanna see more of this. This needs to be developed into an actual option on the market.
Doesn't really make much sense beyond looking cool though. Since the blade velocity near the center is so low its not producing much pressure/flow there anyway. If anything you're losing static pressure as air isn't being prevented from back flowing by the solid hub. And the bearing on the edge means you have to have slightly shorter blades for the frame. Meaning you get less efficient blade surface area in the same 120mm package.
Also, at high RPMs this bearing is going to be more difficult to tolerance, which would require more expensive materials and more engineering hours, due to the extra force of the faster velocity.
You know I can actually see a very useful application for this. If your cabling needs to go down the same pipe as your in-line blower then having one of these with an open ring in the middle to run your cable through would make these applications much more simple! This is such a neat idea!!
I think the reason why the original tip driven fan had the hub was two fold: first, the fan doesn't push much air through there, simply because there's no velocity there. And so you might as well keep the hub to keep the air from accidentally moving back through the center. And once you do, then it's a great place to put a regular fan bearing.
Yeah. By doing it on the outside he's increasing the area and moving mass of the bearing significantly. Since he already has permanent magnets in place, I wonder if there is a way to magnetically float the center?
would probably work better if the blades touched at the center or you made a smaller fan in it like some have done in the rest of the showdowns
Molecular entrainment keeps the air all going the same direction. It's the same principle as an air ejector or a peri jet water eductor.
@@ventilate4267 Connecting the fan blades in the middle would lower the overall flow. In the fog segment, you can see that there is actually air going through the center, which is sucked in/carried with the surrounding air. If you would connect the blades, you would obstruct this path without any acceleration of the air from the blades in that area. Therefore many ubless props only use short stubby blades and leave the rest empty.
The friction of the air will just pull more air with it. (its what @morgan5941 said)
In fact he could make the blades smaller and it will still work.
Now I want to see a whole season of fan showdown dedicated to making fans for this setup!
The fitment just shows your engineering prowess, good job. I am not sure what it could be used for considering the sound, a maglev bearing would be cool lol. I also like the flow characteristics.
The maglev bearing could be the best feature of this design.
I wonder if some of the maglev trains designs could be used that would not need complicated setup with sensors and controllers.
I was thinking maglev bearings too
I think this is a solution looking for a problem. Already in a fan design the center is the least impactful to the airflow. So then you have to make trades. Driving it from the outside requires faster speeds, more bearings, more friction, a better gear ratio, and more mass. Dropping the efficiency. And really what we're after here is most is efficiency, among things like quietness, form factor, size, etc. Otherwise all pc fans would just use a 30,000rpm motor, draw 10 amps, and sound like a wind tunnel.
Not saying innovation is a bad thing, but optimizing one parameter at the expense of many others usually doesn't pan out
So I designed a hubless impeller for watercraft quite a while ago, it's rather more useful because water doesn't compress like air does. I couldn't even attempt to get the parts made to the necessary tolerences back then, so I never persued it. I've been wanting to try it on an air fan for a while though, so this is super cool to see.
Tip one, bearings do not have circular tracks in the race. THe bearing rides on two microscopic ridges on each half, so it's clear at the bottom and on the sides, like it's on rails. This reduces friction dramatically. Bearings also have no clearance, the ball is touching both races all the time. That's where your noise is coming from. The clearances necessary to do this are crazy tight, you have no chance of a printer getting anywhere close. Maybe print undersized and lap to fit with a disk?
Tip two, I would make the magnet pockets load from the inside, so the centrifugal and magnetic forces are pulling the magnet INTO the pocket, as opposed to trying to pull them out. Shouldn't need glue, but it's probably helpful to stop vibration and wear.
Suggestion, I would test making the pole pieces for each coil much wider, approaching a full ring as opposed to small bars. I think in this design you want there to be a minimum gap between each pole and between the poles and the magnetic ring? I havent worked with motor design in decades so I don't remember much lol.
Would love it if you and the community optimize the hell out of this. Better blade design will no doubt increase airflow.
Yes! Finally someone made a fan like I've wanted. Except I was hoping for a maglev type effect with no friction and thus reduced noise.
I think you would need a center hub to be able to do that...the bearings are what is holding the blade assembly to the rest of the fan...
You don't need a center-hub for a mag-lev fan.
However, you need to place the magnets exceedingly precisely, the magnets have to extremely similar in strength, and have a very exact shaped magnetic field for it to work without the hub flying out at high RPM.
Either that or find the strongest permanent magnets you can. That way the field might be strong enough to contain the hub at any RPM.
Not sure if the Neodymium magnets are strong enough for that though, and stronger rare-earth magnets are ridiculously expensive...
@@The_Keeper I guess I should have been more clear, no you do not NEED a hub, but doing it without one would be exceedingly difficult outside of a lab with very precise equipment...Also it would the worlds most expensive computer fan...
i absolutely want to see revisions of this, try lubing the bearing to bring the noise and friction down
I think that connecting the fan blades to a hollow axial cylinder in the middle could increase efficiency by eliminating vortices at the tips of the blades, while still allowing air to pass through.
1:00 that fan is still super awesome! I'd put those on my case in a heartbeat!
Dude, this is BRILLIANT. I really want to see this refined and iterated to it's best performance. You have a really good series in your hand right now.
A much cheaper option for the bearing balls would be airsoft BBs. I've used some cheap white ones that have an amazingly high roundness spec with good results.
Edit: You'd have to modify the CAD model a bit though, as they're 6mm diameter.
you are supposed to add all the ball bearings on one side and then move the core over. And then space them out and then place the cage. A little grease makes this easier since it keeps the bearings n place. Larger balls and smoother raceways would make it quieter.
- there are various ways these things are done, depending on architecture - your "suggested" method is common for deep grove bearings - with welded cages..
Designing integrated raceways can be a limiting factor - as clearances may be too tight (balls too large)- and access limited for conventional methods of assembly.
@@kadmow My balls are also (too large)
I suspect a large reason for all the noise is the surface smoothness of the running surfaces for the ball bearing. Throw in some imbalance in the rotating assembly for good measure as it looked and sounded like it was imbalanced.
Definitely a cool proof of concept. Certainly opens up room to look at the effect of hub size on flow/flow distribution.
You sir just opened up a whole new relm of fan showdown designs! There was hardly any turbulence after the fan! Now to quiet the bearings.
I wanna see how far the blades and bearings can be optimized. I would also like to see one specifically for airflow and another for static pressure. This could actually be a competitive product because it could be made to have no gaps.
The most important part of a fan is the outermost diameter. The further you go in the less it matters. This is why you stuff the motor and bearings in the center. This fan design will fundamentally always move less air for a given size/power.
@@Anyone700 That makes sense to a point but respectfully I am going to have to disagree. Although the first part seems true because as you go inward the rotation slows. My thinking is that the fact there is no motor in the center means less obstructions, both from the motor and the little arms that often hold it in place.
I am fully ok with being proven wrong though.
@@DoubsGaming my reasoning is that putting the motor on the rim makes the fan a type 2 lever, hence need a drive that is low torque but high speed. A electrical motor is full torque all times. so if we use the motor in a type 2 lever it is actually less efficient than the hub drive. from an efficiency stand point hub drives are less fitting for the task
@@DoubsGaming the central part can't create significant air pressure, so the engine is placed here. "Hubless" is good for directing airflow, but can't deliver the same air pressure with an equivalent form factor, which is more important. The structural complexity is insane for a product of this type. There is no way how it can compete with "classic" design (except style).
There’s a reason the other guys design that you shown still had a small hub and that’s to have as little mechanical parts to reduce noise.
That fan still was quite noisy, albeit in a higher pitch.😵💫😣
@@Timberwolf69 The issues with the old TMD fan:
1) The TMD was ONLY made in 70 mm Fan size (Most Fans smaller than 80 mm gets loud)!
2) They were running it at 5800 RPM
3) I don't think they used the quality bearings that we have today, especially nice ones like MagLev Bearings
@@kamenriderblade2099 I think the RPM would suffice to make any fan loud...
But those small fans usually are the ones with the highest pitch...
@@Timberwolf69 When PC Hardware was young and many manufacturers didn't know better. What a time to be alive.
@@kamenriderblade2099 Yeah, the first fan I put on a CPU cooler back in the days was about that size... But it never ran that fast.
Would love to see an evolution of this design.
Dude you are amazing you eyeballed a hub bearing and fan blade geometry in a hub-less fan you pulled from a drawing on the web eyeballing and guestamating at the measurements you were able to nail industry standard CFM rates (better actually 50 CFM is pretty good) and did it all with a 3d printer and i didnt hear it fall out of balance till right at the end. Thats so cool!!!!
That thing moves air very nicely! I was impressed when I saw the smoke test.
This is the super cool project dude! It might be interesting for radiator fans that would benefit from having more service area being cooled!
Every radiator have a gap between fan and actual fins. Air need space to spread and calm down before it enters the fins. In case of radiators Pressure and bulk airflow matters, not area. less area even better, less area usually means higher pressure - for example 120mm fans have more static pressure than 140mm fans, and radiators for 140 fans have less fins density in general (to reduce resistance)
I think you'd get a lot of noise reduction with some kind of silicone or other lubricant on the bearings, I'd love to see a Round 2 with something like that.
I think finding the correct balance of film strength and just overall viscosity of the lubricant will be the largest challenge. Using actual petroleum bearing grease would likely be too thick to allow good movement-- but maybe I'm mistaken, maybe without a centralized motor like you might be able to generate more torque to overcome the resistance of the grease. Beyond that though using something like WD40 which has little to no adhesion or film strength I think would probably provide a negligible (if any) benefit. Silicone lubricant is frequently used on plastics and would probably provide a good balance between too much and too little adhesion and viscosity.
Did you oil the bearings at all? It doesn't sound like it 😂 also im sure it would get quieter over time
Oil actually slows down bearings.
@@tacticalcenter8658 if you use a light oil I doubt it. Gotta remember that it's not a perfectly smooth metal ball against another piece of metal, it's on plastic. I'm certain that if the oil was able to get in-between the more porous plastic and the ball, it'd improve performance, and noise.
In theory, in my head. I could totally be wrong though.
The usefulness of oil in bearings is a hotly debated topic, realistically my personal guess is a small amount of a light thin oil, would probably help.But I have no experience with ceramic and plastic bearings to my knowledge. But last I've heard it's very much design, application specific, along with things like material choices.
But yeah it's hard to say, generally speaking of you lube a bearing it's using a small amount of a thin lubricant.
That being said even properly lubed I still expect this to be able to wake the dead.
@@Ayane13b no its a fact in actual experiments.
@@bobbybobman3073 I'm not arguing anything about the usefulness of oil in bearings. Just the fact the oil slows friction.
I'm quite confident that this design has future. Like many people already said, this thing could definitely use some lube, and I'd personally go with some liquid and thick one, maybe some Krytox 205g0, like keyboard enthusiasts use for their keyboard switches, and all the surfaces that have friction should be as smooth as possible, some polishing should help. With these relatively low-effort and low-cost improvements, I hope we'll see a v2 soon!
Why waste krytox for few tests of 3D printed fan ? Usual silicone oil with ptfe powder will be enough
Close the gaps between the blades in a way that forces the air through the middle hole. Can you make a video trying to take advantage of this build and send concentrated air through the middle hole.
How fast can you make the air go.
What you need to test is a 6 or 9 pole design as that will increase peak velocity. Then add steel ball bearings and races with steel wire guides to reduce rattle and friction. Should up to double speed and airflow.
The big noise maker here, as is more than likely obvious, is the ball bearings. If you were to switch to a maglev design, it would be a lot quieter. Also, the fan blades could be optimized significantly which would almost certainly give you better CFM at a given RPM. Also, as mentioned elsewhere, balancing everything out would also help reduce noise and vibration. I'd be very interested to see how far you can take the design overall. Nicely done so far :)
My biggest concern is that the maglev may interfere with the rotators. But if they're not strong enough, they won't do their job properly. What you need is 2 maglev tracks sloped to 45 degree to cup the v-edged fan unit. Make sure this channel is perfectly smooth as catches on startup or shutdown would be disastrous. Try to keep the space as tight as possible and of course add lubricant, and send me royalties on the patent! XD
would love to see a resin printed version with tighter tolerances. dual bearings would also be nice
Enclosed bearings with lubrification. Or some way to use air pressure when spinning to center the rotor?
You can use 4 small off-the-shelf ball-bearings in the corners to hold the impeller in place, instead of the large peripheral one.
Also, only 4 windings in the corners (as in the old video) will be sufficient to move the impeller, but will allow to increase the diameter of the impeller especially in the periphery where it works best and should significantly increase the output.
I've made bearings out of pla. They benefit from running in at low speed for 30 mins or so and some lube. I use silicone grease and it will work a lot lot better and quiter. The thing I like about hubless is that its actually naturally an air multiplier. The centre part drags more air through than the swept distance.
An idea might be trying slightly thinner gauge wire and more winding on each core section to have a higher impedance. Might be able to increase your speeds doing that. Also with some tweaks to the bearing design then I think this could make an excellent unique fan
The problem with thinner gauge wire is increased risk of burning out the coils, so lower voltage tolerance. This might work for a low power, high torque fan? Bearing in mind I'm no electrician and my response is based on logical assumption alone.
Very cool build! Although, I wonder if the wobble it seems to have in the start of it's movement has any affect on it's final numbers. It feels like it has more potential in it's design.
Would that un-evenness have made it slower? Would solving that make it quieter?
Hell of a good job with creating this.
In terms of performance I can not really see benefits compared to traditional center hub. The most amount of air is being moved on the outer part of the fan, not the center. Here the outer space is being taken because of the motor parts, so fan diameter is smaller, like some argb fan. Still very impressed for your design that is very thin especially for a 3d printed prototype.
For example when you look at small, 30 or 40mm fans you will see that hub takes a lot of space. Sometimes even 70% of the fan. It would be possible to make smaller hub with less powerful motor, but increasing blades in that inner space close to the center is not so beneficial. At least not enough to compensate for power of bigger and stronger motor. 😂
Okay, this is the first new kinda fan I've seen on this channel that I actually wanna see. Noctua needs to get on this right now, I feel like there's a lot of potential here. It just looks futuristic seeing a hubless fan and I can see enthusiasts using these in gaming PC's
With the hub gone it gets rid of a huge obstruction that takes up probably 30 plus percent of the fan area. In the smoke test you see the smoke actually center itself concentrating the air to the center. The center looks to be a lower pressure but that can be fixed with different blade profiles. With a better bearing solution this can only get better.
Did you try greasing the balls?
😂😂😂😂😂
The reason for the noise is because of a non-perfectly symmetrical fan assembly, if you pay close attention at a moderate RPM, you will see the fan blade assembly is shifting resulting in the noise. Your fit is a bit loose, but this is normal with ball bearings you can try adding a lubricant that is a bit extract viscous to cushion the vibration to absorb much of the noise of the loose bearings rattling on imperfections of your fan and frame.
this is by far my favorite one out of all and only can get better from so cool
Bravo! Very well done! Here are a couple of ideas for you to try:
#1 Arrange your magnets in a halbach array to optimize the magnetic field
#2 Eliminate mechanical bearings and go with magnetic bearings. Your noise problem will be solved! ( not sure how, but no doubt you will research both of these ideas and figure it out )
Cheers, John
The tmd fan shown at 0:33 is the same one used on a heatsink i used for a socket A(462) motherboard i recently configured, looks and sounds like a jet engine, cool to see how the fan operates and someone taking the design further.
Well done!! I would love to see you continue improving on this design and that journey.
Had one of those TMD fans in my old Athlon XP system back in the day. Always liked the idea.
I like what you do. One of many often overlooked aspect of forced air cooling is air speed. FASTER IS NOT BETTER. If the air isn't spending enough time around the aluminum heatsink or whatever device is being used it will not soak up enough heat and there comes a point where the air is moving too fast and becomes less effective.
I have made a couple of these hubless fans myself. Let me help you fix the noise. All you need to do is change the orientation of your copper wiring by making it sideways rather than inwards and rotate it so that they face towards each other, duplicate it so that it is now found on both shallow & deep ends with empty room between them and then allow for magnets to be free flowing between these copper wire spools. There you go, Noiseless rotation.
When it boils down to air flow, they perform a lot better by their fins taking after the shape of a rose flower petal (or a Pringles chip at a rotation) and coming real close to overlapping one another.
This fan has potential. I love that it allows the air to follow into the center. If you remove the friction and get better balance, you have a winner.
unfortunately because air is a lot less viscous than water, it might end up just backflowing or stalling in the center, a hub might actually be slightly beneficial if it's got decent aero
Most interesting project yet, carry on with this line of development.
Interesting fan.
I look forward to seeing how this improves airflow in the future.
The question is how to support the fan, which could be magnetic floating, ground effect, minimum center bearing, etc.
An interesting thing might be to use the ground effect.
Centrifugal force compresses air and uses that compressed air in ground effect bearings.
There are many other possibilities, such as installing a rail on the outer ring of the fan and supporting the rail with about three wheels.
I noticed that the older, TMV fan was not truly hubless. There's a spindle hub on which the fan is supported while it is rim driven, while your noisy beast is truly hubless. The spindle mounted / rim driven design has opportunities for quieter operation. With multiple spindle mounted / rim driven units you could produce something akin to a jet compressor.
Love the idea and the result. Hats off! The only problem is that the fan is most efficient at the outer edge and least efficient at the center, kinda like a hard drive. Unless you can make the motor and bearing ultra thin, the center driven fan will be more efficient.
That was awesome. Looking forward to future episodes and ideas.
This concept is awesome and you got a working prototype with a 3D printer, some magnets and some wire. Great Job!
One recommendation I would make is to make an alternative frame that runs on a much simpler mechanism, like a drive sprocket with a belt and a stand alone motor. That way you would be able to test out multiple iterations of blade design without having to attach magnets to every one, and once you found a winner you could make a proper one.
Just chiming in to say that that segment from 10:58 to 11:40--especially 11:22 to 11:40--was awesome. That's a camera angle you should.use for all future fan showdown videos.
That was really fun! Please make that the next fan show down season. There are so many components that can be tweaked to get the best air mover.
Not sure how, but you could come up with a point system, for ie 2000 RPM = 2000 points, 91db sound = minus 100 points per 10db. , mili amp = minus points, air flow = plus points. Style points, easy to print, etc.
Thanks for sharing your project! :-)
I have a Baratza Sette 270 coffee grinder that works kinda similar to this. The center conical burr is stationary, and the ring burr on the outside is the part that spins. It comes in a self contained gearbox that rotates around the center burr. It’s quite effective for espresso with very little retention, but not great for airflow I’d imagine.
Temperature and magic smoke loss is inversely proportional to surface area. Moving the “hub” to the ring increased its area by some factor or maybe an order of magnitude (I am not going to calculate this). So you can probably increase the amperage by quite a bit before the magic smoke escapes.