Designing a Quiet Air Compressor

[Steamchick]

Bentwings suggests silencing of intake noise.... This is best achieved with a Helm Holt Resonator. When you know the volume of air per cylinder stroke/suck, and the frequency (from rpm) the calculations are simple and found on the web.
As revs and displacement are constant, a single resonator, correctly sized and positioned, will considerably reduce intake noise to "quiet"....

K2

 

[vederstein]

Progress continues to be had.

The pump, when pressurized backwards will work as an air motor. Today I fit the pistons and initially put the pump together (albeit without the cover plate).

For a test, I indeed pressurized the thing and ran it as a motor. It runs and I found some leakage. But at least I know the porting isn't completely garbage.

(Yes, my workbench is currently a complete mess...)

...Ved.

 

[davidyat]

Lookin' good Ved.
Grasshopper

 

[vederstein]

A bit more analysis. I wanted to work out the actual motion of the piston. As the piston rotates around the central axis, it also moves towards and away from that center axis. Doing a bit of math in a spreadsheet shows the piston moves in an ellipse:


The X position relative to rotation angle is a sine wave and the Y position relative to the rotation angle is a cosine wave:

Summing up these vibrations relative to the rotation angle and you get:

The next graph takes a series of derivatives of the X+Y position graph with respect to rotation angle to get velocity, acceleration, and jerk relative to the rotation angle. As you can see, the acceleration is fairly low and jerk is minimal. The should result in a stable rotating system.

Isn't mechanical engineering fun?!?!

...Ved.

 

[Toymaker]

Progress continues to be had.

The pump, when pressurized backwards will work as an air motor. Today I fit the pistons and initially put the pump together (albeit without the cover plate).

For a test, I indeed pressurized the thing and ran it as a motor. It runs and I found some leakage. But at least I know the porting isn't completely garbage.

(Yes, my workbench is currently a complete mess...)

...Ved.

Listening to your spinning pump, what's the source of the low frequency (maybe 5 or 6 Hz) thump, thump, thump sound? Almost sounds like a compression stroke, but the pump's rpm seems far too fast for a piston to be the source of the sound. I don't see anything in your graphs to explain the sound.

BTW, your workbench looks a lot like mine

 

[Zeb]

(Yes, my workbench is currently a complete mess...)

Me three. An empty desk is a sign of an empty mind.

Bentwings suggests silencing of intake noise.... This is best achieved with a Helm Holt Resonator. When you know the volume of air per cylinder stroke/suck, and the frequency (from rpm) the calculations are simple and found on the web.
As revs and displacement are constant, a single resonator, correctly sized and positioned, will considerably reduce intake noise to "quiet"....

K2

One of my first tasks in a lab was moving compressor frequency coupling outside of normal operating ranges on a device. We used a piston and moved it along a cylinder to dial in the volume needed. Never thought I'd see that again!

 

[Steamchick]

Maybe some of these will interest you?
https://newt.phys.unsw.edu.au/jw/Helmholtz.htmlhttps://www.calctool.org/waves/helmholtz-resonator
http://www.apsimplepsaltery.com/appendices/helmholtz/
https://www.lautsprechershop.de/tools/t_helmholtz_en.htm
https://github.com/statikkkkk/helmholtz-resonator
http://www.vk2zay.net/calculators/helmholtz.php
There are plenty more if you research the web.
K2

 

[vederstein]

the source of the low frequency (maybe 5 or 6 Hz) thump,

I have a rotational interference that I need to figure out. Along the rotation, it binds a bit at one point. It can be several things, the worst being the crankshaft slightly bent (perhaps stress relief from the machining operation).

At this point I don't know, but it's certainly something I cannot ignore.

...Ved.

 

[Toymaker]

I have a rotational interference that I need to figure out. Along the rotation, it binds a bit at one point. It can be several things, the worst being the crankshaft slightly bent (perhaps stress relief from the machining operation).

At this point I don't know, but it's certainly something I cannot ignore.

...Ved.

In the past, I've manage to solve a few mechanical problems by taking a video, and then carefully observing the playback in either slow motion or stepping through the video frame-by-frame. Your cell phone typically takes video at 24 or 30 or 60 frames per second (FPS) which should more than fast enough to observe the location of your mechanism when it produces that sound. The hard part will be synching the sound with the video frames. If you have an audio receiver that flashes a light when it hears a beat, then you could simply note the mechanism position at each light flash, which would show up even in frame-by-frame playback.

 

[Steamchick]

Interesting idea Toymaker. Makes me want to buy such a phone. I just have a device that talks to people... But I have a camera that takes video so I can use that I guess? Not sure about the blinking light audio device though. I'll have to find one of my machines with a problem, then see what I can do?
I last saw a high speed film used for such analysis when a 3 second test failed, and the 1000frames per sec film showed where the failure was initiated. I just never though I could do that sort of analysis in my own workshop. Thanks for the good idea!
K2

 

[Toymaker]

Interesting idea Toymaker. Makes me want to buy such a phone. I just have a device that talks to people... But I have a camera that takes video so I can use that I guess? Not sure about the blinking light audio device though. I'll have to find one of my machines with a problem, then see what I can do?
I last saw a high speed film used for such analysis when a 3 second test failed, and the 1000frames per sec film showed where the failure was initiated. I just never though I could do that sort of analysis in my own workshop. Thanks for the good idea!
K2

Back in the '60s companies made a device that flashed a bright light on every beat of sound or every loud noise; they were meant to be used at small dance parties where the group didn't already have an expensive audio-visual sound system. The better devices had an input sensitivity knob that could adjusted to select what sound lever would cause the light to flash.

Most smartphone cameras will take video at 30 fps, better phones will go up to 60 fps, and high end smart phones will take 120 fps. Doing a little math: If your engine or compressor is spinning at 60 rpm, one rev per sec (rps), a video shot at 120 fps would show 120 frames per revolution. Double the rpm to 120 (2 rps) and you get 60 frames per revolution. At 600 rpm (10 rps) you still get still get 12 frames per revolution.

If the RPMs aren't too high, your typical smartphone camera can produce pretty good slow motion video.

 

[Steamchick]

Just bought a second hand Olympus mu 1200.. From its age I assume it may be 24 fps or something? Manual doesn't say.
K2

 

[vederstein]

I found the source of the binding. The rear shaft bearing wasn't perfectly centered on the rotational axis. I bored the housing out a little, and the binding went away.

Now the bearing fits a bit too loose. I'll compensate with some Loctite 638.

I then hooked up the pump to my lathe for a trial run. There's more internal channel leakage than I would like.

It's ok though. A prototype almost never works perfectly the first time.

I'm so close to having the original design finished, that I'll go ahead and finish the thing. I'll get some performance data then decide where to go from there.

...Ved.'

 

[johnmcc69]

Nice work Ved!
It's fun building the prototypes.

John

 

[vederstein]

A quick of a primer on sound pressure and the decibel scale. The dB scale is a logarithmic scale. Therefore every decade number change is ten times the level of the previous whole number.

Assume you have something at 50 dB and another at 60db and another at 61db. Going from 60 to 61dB is the same intensity change as going from 50 to 60 dB.

Therefore while a quiet room in your home is around 40dB, average street noise is around 70dB. Even though the number change is 75%, because this is a logarithmic scale, the actual sound pressure change is more like 999 times (999,000%) more.

Below are three graph of the same data. The first, graphs the data (source at the bottom of the post) as a regular, linear chart with the sound pressure in w/m². The second graphs the same data on a logarithmic scale. The third graph is the same data converting sound pressure (w/m²) to dB. The log scale is much easier to read the lower pressures, but you must remember that each decade is ten times more than the previous decade.

So when I post some performance data on the air pump and compare it to existing equipment I have around, please remember that sound decibels are a logarithmic scale and it's anything but linear.

...Ved.


Data Source: Sound - The decibel scale

 

[vederstein]

I got all the pa;rts finished, bolted the pump to the motor and ran some tests. The results are pathetic.

The pump only generates about 10 psi. Obviously there's a large amount of internal leakage.

My barb hose fittings were discards from a work project. I wonder if they're sized for metric tubing because the tubing I'm using don't fit onto the barbs as well as I would've liked. There's possible leakage there.

I always knew there would be leakage between the crankshaft and bore to which the crankshaft resides. Perhaps it's far far greater than the 15-20% I was expecting.

The motor in the video below runs at 1725rpm. When I drove the air pump off my lathe at about 1000 rpm, it developed about the same pressure.

Not only that, the pump is louder than my California Air Tools air compressor. The CAT compressor averaged around 63.2 dB @ 3ft whereas my air pump averaged 76.1 dB at equivalent distance.

I'll need to contemplate what's going on and make some design changes.

...Ved.

Video below:

 

[Toymaker]

I got all the pa;rts finished, bolted the pump to the motor and ran some tests. The results are pathetic.

The pump only generates about 10 psi. Obviously there's a large amount of internal leakage.

My barb hose fittings were discards from a work project. I wonder if they're sized for metric tubing because the tubing I'm using don't fit onto the barbs as well as I would've liked. There's possible leakage there.

I always knew there would be leakage between the crankshaft and bore to which the crankshaft resides. Perhaps it's far far greater than the 15-20% I was expecting.

The motor in the video below runs at 1725rpm. When I drove the air pump off my lathe at about 1000 rpm, it developed about the same pressure.

Not only that, the pump is louder than my California Air Tools air compressor. The CAT compressor averaged around 63.2 dB @ 3ft whereas my air pump averaged 76.1 dB at equivalent distance.

I'll need to contemplate what's going on and make some design changes.

...Ved.

Video below:

Ved, your compressor is producing quite a bit of vibration, as shown in the video by the pliers dancing around the table top, which by itself will produce a lot of noise. Also, the relatively thin plywood table top will act the same as the cone in a speaker, amplifying the vibrational noise from the compressor. I'm guessing your compressor will be a lot quieter when mounted on some rubber vibration isolators.

Question: are you concerned that all the lubricating oil inside the cylinders will be flung towards the cylinder heads by centrifugal force, where the oil will remain unless you have an effective oil pump to remove it?

 

[Steamchick]

Hi Ved, An "observation", or whatever you call it when I hear something and make an opinion. (Audiovation???).
I hear the same "pulsations" from the following, but of varying degrees:
1: An old fridge compressor, where the air intake is inside the closed chamber of the compressor, with only "leakage" (very narrow long slot) being the way for air-in and sound-out: Very quiet, hardly different to your motor.. When I lift the lid it becomes moderately loud.
2: Another old fridge compressor. Air intake is a small pipe with a tin on the end, with a bit of porous foam in it - about as loud as your compressor inlet noise.
3: My 2CFM compressor: rated at 90dBa! NO intake silencing on it - You cannot hear the motor for the intake air pulsations... (It has a reed valve built into the piston crown, and open port - just an air-cleaner - on the crankcase.
4: My 500cc motorbike. The intake noise appears to be "half-way" (?) between the Exhaust noise and the engine noise. Exhaust will be ~85dBA?, total engine ~75dbA? So maybe air intake is ~82dBA?

If you fit Helmholtz resonators (it is a constant known frequency of pulsation - easy to design and tune) you should easily reduce the intake noise by more than 3dBA, making the major noise the mechanical noise?
Most of the intake noise comes from the timing of intake valve opening, so if you are using reed valves, try a lighter spring? then the valve will open at less air pressure Difference and make less noise?
K2

 

[Toymaker]

I'm not sure if this will help, but just to compare noise levels, I uploaded a very short video with sound of my wobble plate compressor. The recording was made using my smart phone. The video is the last post (#26) on my wobble plate air compressor thread. Wobble Plate Compressor Video

 

[vederstein]

are you concerned that all the lubricating oil inside the cylinders

The thinking was twofold:

1. As the piston went to bottom dead center, the oil on the back end of piston would get pushed out of the cylinder and then work its way into other areas.
2. When the pump stopped, the oil would then fall onto the center bearings.

As for the other comments following my test run, they're all good comments. I fully understand that you all are only trying to help. But there's a bigger issue:

The Air Pump Don't Pump Air!

At this stage in the game, I'm so far off from my original intentions, that I really need to evaluate the cross leakage. I'm not even really worried about volume right now. A noisy air pump that actually pumps air is better than a quiet air pump that doesn't do squat.

Thus far I'm thinking of:

• Replacing the possibly ill-fitting barbed fittings
• Making a new crankshaft to incorporate 100 series (.103 thickness) o-rings instead of 000 series (.070 thickness) to have more tolerance for air gaps/mis-alignment.
• Making something akin to springed motor brushes made of either teflon or brass that would press up against the crankshaft for port sealing.

Please understand that I'm not trying to be an ***. Yesterday was quite the disappointment.

...ved.