Building a Small and Quiet Air Compressor [4K] 

Yes that is my experience too, press fits in 3D prints hold for a day or so.

Thanks for mentioning that. This was something useful to learn about which I didn't know about before

Thanks

One thing i didn't catch, what assures the timing of the pistons?

Thank you. And the only thing important is that they are 180° out of phase which is guaranteed by the holes and the steel pins in the crank shaft.

True, I am indeed looking for buffer tank and one of those small fire extinguishers or a small propane bottle might do the trick. Thanks for the idea.

​@@DiffractionLimited1.5 L sparkling mineral water bottles do decent job for pressures below 3- 4 bar. They can rip at about 5 bar. 0.5 l bottles rip at 11 bar. I pumped bottles 1.5 L by foot pump to 2-2.5 bar. It is a bit tiring job if smth is leaking.

Though about that too. But the valve head might get quite warm so a resistant filament material needs to be used + It should be air tight, which can be done by submerging the print in a special sealing fluid as I learned recently. The rubber sheet is a good tip and in the valveheads shown here a similar flat rubbery material is used. The construction is actually quite simple.

I wouldn't get your hopes up on this one. There are basically three variables you can tinker with: Power, price and noise. You can only choose two. Basically a quiet powerful compressor is going to cost you (rotary compressors). A cheap powerful compressor is going to be loud. And lastly, a quiet cheap compressor is going to find you lacking in power.

​@@patrickd9551this is very well said good sir :)

@@patrickd9551 Years ago, i built a 40l/min 6 bar compressor for

@@TeslaAtomsWell it depends on your application, but 40L/min is basically nothing for most applications, perhaps except paintbrushing. But you've proven my point exactly, because my $500 compressor does 350L/min, makes a lot of noise, and I still can barely run my plasma cutter on it. (those require massive amounts of air)

@@patrickd9551It really depends on the usecase! I use my compressor exclusively to supply air for tool changes on my VMC, as well as mql (minimum quantity lubrication). 40L is the perfect match for this!

It may annoy you, but the tilting motion seems pretty crucial to getting the seal tilting along with it in a reliable fashion. 🤔

@@AnnaVannieuwenhuyse sure... but you don't need a tilting seal for a compressor. This design maybe, but other designs exist

Yeah I've clicked on three random points and saw the demo. Which part impressed you specifically? I wasn't really impressed by pressuring a small piece of tubing. That is if you have a basic understanding of compressors.

Hard to tell, I also have to gain some experience with this thing because I havent done anything similar before.

that would work for solid machined parts, yes.

@@AnnaVannieuwenhuyse considering the pistons are running inside a metal cylinder, and the o-rings dont really have any relative movement to the 3D printed pistons, then yeah, Id say it would work quite reliably with continuous oil supply. Maybe a gravity fed oiling system running the compressor upside down and a filter of some kind at the exit to prevent oil contamination of the supplied air.

I am currently working on some changes on the air bearing to make it easier to manufacture - hence the delay... but it will come. As for the fridge compressors: Yes they are quite low noise and relient, but all the ones I know are filled with oil. Quite a lot of it can make it to the air outlet. But with a filter they might be a good alternative.

Eventualy I guess they would, when running at higher rpm and at high pressure. However I think a small fan at each cyclinder would help quite a lot in that case.

@@DiffractionLimited it’s a beautiful design. Well done!

I plan to use it on pretty low duty cycle. But yes long term lubrication is an open problem. Ideas for improvments are very welcome. As for the heat I hope that some of it is still in the gas as it leaves the compressor and eventually I might add a small fan or two if it becomes a problem.

I think in that case you would want to use a classical seal at the piston and a dedicated intake check valve in the valve head. Seems much easier than sealing the entire compressor body.

Nice, let me know how it worked out or add an issue to the github if you find some problems. I actually migh do a video on the parts you mentioned. As for the Cylinderpipes a good source are gassprings. You hace to open them up (which can be dangourous without the proper technique due to very pressure and oil) and then slice off rings. By design the inside is made to be a very smooth and a durable surface for seals. As for the seals them self search for 'Piston U-Seals' make sure they do not have a steel core. Also be aware that they have to be slightly oversized for the cylinder bore to seal properly. Good luck with your build !

Hello, I googled around the last days, but until now I was not able to find the mentioned seals in the needed dimensions. Do you remember, where you bought your parts? Also the cylinder heads. The 3D-printed parts worked out so far, the tolerances are a bit snug. Maybe a version of the parts with some larger tolerances would be helpful for a lot of people.

@@Ulm26mAs mentioned in the video I got the cylinde heads and seals from a cheap air compressors for car tires and alike. But I also bought some U-Seals from ebay as replacement. The OD can be a bit larger so you have some tolerance and should be able to finde some appropiate seals.

Bit I think the best order would be to look for cylinder - then for the seals - then adapt the CAD accordingly and then print the parts.

Yes sure. Then it would only have to work against 1 bar not 10 so you could increase the volume or reduce the belt reduction ration. However you would reach a very deep vaccuum because ther will be a certain dead volume within each cylinder with this design.

Thanks. I plan to add the inner workings of the check valves to the CAD in near future and maybo make a supplemental video to this one too, where I show how htey work. Its essentially just a springloaded rubber pad on a hole.

@@DiffractionLimited Thats awesome thank you for the reply! I will subscribe and wait to see it!

It is a 3 phase brushless DC motor (BLDC Motor) !

@@DiffractionLimited It definitely looks just like a big ol' stepper motor, but I believe you.

It should scale well, but you may need a thicker crank shaft and one or more supporting shaft bearings in the middle at some point. But it would also run smoother with more cylinders.

@@DiffractionLimited do you have any idea of its output?

This depends on many things, but lets say the motor runs at a max speed of 3000rpm. Then we have a 4:1 reduction ratio and get 14ml per rev - this would yield about 10l/min which is 0.35cfm

yerah that threw me off too

Oh is it an AI voice? I just thought the dude was really nerdy or something lol

Well then this thing might be for you. How would you measure it? Would the pressure over time given the tank volume be sufficient?

@@DiffractionLimited You would measure it with a... Flow Meter. One of those things that looks like a glass tube with a ball floating in it.

Thank you, but I think its far from being a product and I'd rather see the community use and improve upon it for free...

You mean the speach or the actual audio? Note that the speech is generated with the 'bark' text to speach model and not me talking.

@@DiffractionLimited I did realize later on that you were probably using text-to-speech, but yes the actual speech. Honestly very impressive how well those sound nowadays, the actual voice itself sounds like a proper voiceover except for those weird artifacts and spikes in the audio. Maybe it's possible to remove those in post?

Thank your for the constructive feedback. I am not sure though, that the Problem you are describing applies to this construction, so lets quickly run the numbers (SI-Units) for an unpressurized cylinder (worst case) to see whats going on: * 1000rpm crankshaft speed is about 17s rev per seconds. * The vertical travel of the piston is 0.01m. * The vertical movement (y) of the piston can be approximated by a sinusoidal motion: y = sin(17*t*2*π)*0.01m * The accelleration is the second derivative of the position: y" = -114.093 sin(34 π x) m/s² => so we have a maximum acceleration of 11.6G * Assuming the piston weights about 100g (its probably way less) we get a maximum accelleration force of F = 0.1kg*114m/(s*s) F = 11.4N [calculation was corrected as pointed out by @giantisopod] Overall I see your point, however the resulting force of 11.4N is still way lower than the 400N compression force due to the compressed air, so I dont think it will be a problem overall nor that the concept is fundamentally wrong. Time will tell if anything fails due to elongation forces and I will keep you up to date... Thanks again for the comment thoug :)

@@DiffractionLimited Did you mean 7.3G rather than 73G? Actually, upon closer inspection, I'm not sure about your value for the second derivative, either. But it was nonetheless interesting to see how you arrived at your conclusion. I'm not an expert at any of this, but it makes sense to me, I'm just not sure about the numbers.

@@giantisopodI was surprised about the high G number too, so I double checked and it actually seems right. Used Wolfram for calculations so I dont think there is an error in the second deriv.

​@@DiffractionLimited The unit for y'' is m/s², and 1 G ≈ 10m/s², so I think you are missing a decimal point :) Wolfram Alpha gives me -114.093*sin(34*π*t) as the second derivative.

@@giantisopodOhh you are totally right. I must have messed up something very bad. I am correcting the comment. Thank you !

You are right, the motor looks very similar to a stepper motor but actually is Brushless Direct Current Motor (BLDC) with the same formfactor (NEMA23). Such a motor is well suited for the application.

Interesting. Free air delivery is expected to be about 10l/min which is 0.35cfm (at 3000rpm of the motor). However the piston travel and cylinder bore could be changed for more volume/more pressure.

@@DiffractionLimited Yeah the bore and stroke would need to be adjusted. I also wonder about the volumetric efficiency at the intake. Sizing the right diameter to certain rpm for peak efficiency. No different from a ICE.

@@DiffractionLimited So typically as a general rule of thumb for piston compressor, FAD will be 2/3 of your theoretical CFM (bore * stroke @ rpm). So free air delivery would be around 0.23cfm or 6.51 l/min.