Printed Vacuum Pump Muffler Quiets The Lab


[Joel] made a brilliant improvement to his shop. If you think about it, most folks would hear a loud vacuum pump and either tolerate it or put in some ear plugs. But [Joel] heard a loud vacuum pump and thought: hey, I can fix that! His solution was to design and print his own muffler.

He did a bit of research on the topic and found that design complexity runs the gamut based on the application. For instance, you don’t want to affect the airflow of a vehicle’s exhaust too much or you will take a horsepower (and efficiency) hit for it. In this case the vacuum pump making all the noise has a relatively low airflow so that is not a concern. What he ended up doing is designing a baffle that will help cushion the vibrations in the airy by piping it through a maze of channels. The end result drops from about 92 dB to 82 dB. That might not seem like much, but decibel measurements aren’t linear so it ends up having a great effect. Hear for yourself in the video after the break.

46 thoughts on “Printed Vacuum Pump Muffler Quiets The Lab

      1. I did so too. A simple soft Nalgen bottle with in and out tubing. It 1) stops the sound 2) increases the effective pressure 3) prevents water return. Essentially – exactly acts like a big capacitor.. :-) smootthing out some residual AC ..

  1. I`d like to take this opportunity to correct a slight misconception here:

    Although the decibel scale is logarithmic (i.e. the amplitude is proprortional to 10 raised to the decibel value), the human ear also has logarithmic response so the volume you perceive IS proportional to the decibel scale – that`s the whole point of using it.

    Therefore, although -10dB means a tenfold decrease in amplitude, the perceived volume only decreases to 82/92~89% of its initial value – a decrease of about 11%.

    1. I believe a -10 dB difference doesn’t mean a tenfold decrease in amplitude (of pressure), but a tenfold decrease of power. I think the amplitude of pressure corresponding to a -10 dB difference would be a reduction factor of sqrt(10). Somebody please correct me if I’m mistaken.

      On the perception front: As you said, the human ear’s response to volume has a logarithmic response, but a -10 dB difference will correlate to the perceived volume being approximately halved.

      1. It is correct that the amplitude is the square root of the power, however, normally when we say `dB` we mean `dB SPL` which is proportional to the amplitude (not the power) and is the scale to which our ears response is proportional.

        It`s still a great hack of course and plenty worthwhile, I just thought it was worth clarifying an oft-misunderstood concept, for the benefit of everyone :o)

        1. God damn it, you`re right – I justed busted out the old textbooks and our ears ARE proportional to intensity and not amplitude, just like you said. Crap now I`ve polluted these comments with a mistake, exactly what I was trying to prevent!

          1. Right I`m getting well confused now:

            On one page it says that dB SPL is the log of the power ratio and one the next it says that it falls off with 1/r – thats a contradiction. How can power go with 1/r in three dimensions!?

          2. So your use of the word “proportional” might be an oversight, or it might be intentional and contributing to the confusion. Our hearing is not proportional to intensity or amplitude. Our senses generally respond logarithmically to stimuli, and with hearing volume specifically they certainly respond logarithmically.

            On the issue of the relationship between dB and power or amplitude: dB can be expressed as a relationship of power ratios or as a relationship of amplitude ratios. With the power ratios you multiply the base 10 log of the ratio by 10. With the amplitude ratios you multiply the base 10 log of the ratio by 20. Checkout Where it has the squared ratio of pressures, just think of that as power–since it’s a ratio, the same terms in the numerator and denominator would divide out and equal one.

            As for the 1/r relationship: wikipedia addresses this in the same article at

            “No”, I didn’t just edit wikipedia. : )

          3. See that`s the thing I f now don`t understand as Wikipedia says that dB SPL falls off with 1/r but if dB SPL is a measure of intensity then it must go with 1/r^2 or it violates conservation of energy!

            I wish I could figure this out.

          4. Long ago, I spent an entire two weeks on log functions for electronics. At the end of it I was more confused about it than when I started!

            Just roll with it and smile. You aren’t alone ;-)

    2. Importantly though, 85dB is the general maximum sound level you can listen to for protracted periods of time without having hearing damage.

      I think the guidelines were 85dB for 8 Hours as the cut off, so this hack is VERY useful.
      Still 82dB is louder than I’d like to endure for any period of time then again I listen to music at about 65-75dB

    3. However, in this case you can clearly hear that the majority of the energy removed was higher-frequency sound, which is perceptually louder than lower frequency sounds. Even though it’s a ~ 10% cut like you say, it’s perceptually a huge difference.

      1. He did it wrong by using rounded corners. The point is to reflect the sound backwards in the channel at every turn so the energy would get absorbed on many more reflections, so the maze needs to have sharp 90 degree turns. Rounded corners act as waveguides.

        Each lengh of channel between turns must also be off-tune to the main frequency of the noise so it won’t develop standing waves.

        1. Technically, to do it “right” you need to properly model the wave channel to dissipate as much sound energy as possible, given the structure. There are programs that can handle acoustic modelling in that kind of environment, although they are rather expensive.

          Source: My dad used to do R&D for a muffler company, and showed me the ropes. He also modelled the church that we installed a sound system at.

          1. Dunno about industry standards. I’ve tried out different things for muffling fan noise in PCs by ducting them, and what works best is hard turns and flat baffles in front of vent holes to reflect the sound back.

    1. Typical 3D printing requires the area to be printed to be supported by material underneath. The lid would only have such support above the baffles, which is how the baffles were printed.

      There are ways around this by using a filler that can be melted or otherwise removed via solvent, but that’s a patent minefield so you don’t see it used often.

    2. It could be done with a redesign. With an FDM printer, you can create a “bridge” between to towers or walls, as long as it is a straight line. However, you can’t create a curve without support under it. So in this case, you would need to modify the design a bit, namely 2 changes: 1) The exhaust ports would need to be rectangular and span the width of the channel underneath it. 2) The hose nipple would need to have a rectangular boss underneath it, if only one layer. If the author would upload it to Thingiverse, I could make the said modifications to make it a one piece print.

  2. So how does he “know” that all those channels do anything at all?

    I’d like to see a comparison between his box, and a simple empty/hollow box. My guess is the extra tube and the expansion chamber is all that’s making it less noisy, and the channels do little to nothing to add to that effect.

    1. What is odd about it is that the sound would also go through the side and not somehow only travel in a straight line along the airflow, so he’d need to make the sides of the box sound absorbing for it to properly work I would think.
      And yeah I think a silicone tube would adsorb better than (semi-)hard ABS and be more effective, especially if you bend it a few times.

      But I’m no expert since I tried so many things to reduce the sound of various noise-making machines and the results were always so-so rather than a great success.

  3. Hah, the short video already made me want to scream from the annoying sound, and it might be less piercing with the muffler but it’s still very very annoying.

    I think ti’s also the size of these things, the smaller the noisier, but small means you can put a whole enclosure over the thing with some sound insulation.

  4. In addition to lowering the sound volume, this (and maybe most) muffler seems also to lower the pitch. That’s what gets to me in noise from such sources: the treble-frequencies. This printed muffler seems to filter those a bit. I’d like to try something similar to this for my 2hp, 30-gallon “oil-less” compressor. I think I’d have to use conventional fabrication methods, though. I hate all oil-less compressors. Piston-type, belt driven ones are much easier on the ear–but not the pocketbook!

  5. I wonder if soft foam rubber milled out would have been better for the inside baffles of this thing? Acoustic baffles are usually soft to absorb some of the sound energy a little everytime it strikes a surface, rather than just passing it along.

    1. Take a look at a muffler for a lawn mower or snow blower. It’s nothing more than a box containing heat-resistant foam, an inlet and some sort of screened outlet (usually just an array of holes in the outlet side).

  6. Wouldnt creating a rather rough surface on each of those baffles help even further?
    There was a trick that some mini sumo builders did was covering the front/sides of their robot with a fluffy or furry material. This would really hurt the opponents that used sonar.

  7. Should be stuffed with some fiberglass (or polyfil). You can then convert (some of) the sound adiabatically to heat. Just like in a real muffler ;)

    (Or in some speaker boxes where it’s desirable to dissipate the rear wave, like a transmission line)

  8. I am currently having similar noise issues with a small diaphragm pump that I have integrated into a medical device that I am designing for a client. Just moments ago, I designed a very basic muffler like yours and then started printing it on my uPrint SE Plus. I then Googled various key words looking for details on such a design and was pleasantly surprised to see your blog showing a drop in decibels from 92dB to 82dB. This is very promising, and with a little more design work, I might be able to beat that. My pump noise is currently at 76dB using a small porous filter that merely fits onto the barbed pump inlet.


Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.