Taming the bark of an Avenger

[subscriber]

Thanks Bob

They sound different.  Not as dramatic as I expected.  I wonder if a standard continuous background noise would help, to serve as comparator.

[subscriber]

So,  after playing around with labyrinth designs, I decided that I needed to adapt one to the "Plume" LDC that Bob has already made and tested.

It is the actual "Felt Plume" design, but with an integral "digital felt casing".  Less messing about than putting the Felt Plume LDC together; hopefully with a similar effect.

See images below for what the Plume LDC with "digital felt" looks like.  It has a full thread to work with the 2-stage in-shroud air stripper Bob has been using.

The STL can be downloaded from here;  https://we.tl/t-8yRubxFS2E .

[Back_Roads]

Thanks Bob

They sound different.  Not as dramatic as I expected.  I wonder if a standard continuous background noise would help, to serve as comparator.

 Something like white noise as a background sound perhaps

[JimD]

You might want to try unequal sized chambers in your felt type moderator designs.  I had half curlers in my P35 shroud and went to one full curler and one half curler instead of 3 half curlers because the sound level (measured) was lower.  Putting the half curler closer to the barrel exit seemed to make it  about 1 db quieter (average of 3 measurements).  The three half curlers measured an average of 85.1 db (with the meter back by me for ease of measurements).  Long then short measured 82.46 and short then long measured 81.4. 

I have said several times that I do not trust people's impressions of the sound of their rifles.  If I know the person well and know their impressions are similar to mine and consistent with measurements, I would trust their impressions.  On line forums are very valuable but they don't normally give us that sort of knowledge.  But I would also agree that measurements need to correlate with impressions or they require further investigation.  The meter does not have the issue with subjectivity that we humans deal with but they have other limitations that can cause them to give data that isn't useful.  They may be more sensitive to high frequencies which do not travel far, for instance.  Or sample in too short or too long a time interval.  But as long as the data my meter gives me matches my impression, I trust the data.  I wish comments about "back yard friendly" would consistently come with a measurement.

Data on sound versus frequency and duration is great although I am not confident I know exactly what to do with it.  But as an engineer I love data, the more the better.  I wonder if any of the difference in Bob's measurements might be attributed to the weighting scale used in the software.  Weighting scale A drops a 31.25 Hz signal by almost 40 db.  It weights higher frequency noise heavier.  Z has no weighting above 8khz, not sure about lower, and C only drops a 31.25 Hz signal by 3 db.  If the sound of our guns has significant low frequency content, an A weighted measurement would be a lot quieter than a C weighted measurement.  My old radio shack meter allowed me to switch but it doesn't work any more. 

[subscriber]

You might want to try unequal sized chambers in your felt type moderator designs.

Thanks, Jim

The images below shows that the chambers are all different in shape and length - depending on how you want to measure them for length and volume.  Yes, there are a number of circumferential bands, but half of them are reinforcements, and cone supports, rather than full partitions.

You are correct about measuring sound and reporting it, not being a simple matter.

[WhatUPSbox?]

Bob, thanks for posting your results, the zip file worked well.
I pulled it in to Audacity to see the trace. Below is the display (I hope you don't mind my posting your results).
The first image shows the overall trace. I copied it to the second track so I can see both linear and db scales.
The second and third images zoom in on the first and seventh shot respectively. The light blue highlight is 200 ms and is used for the spectrum window.
Note, Audacity measures db from full scale as reference so they are negative and only useful for relative comparisons within a trace.

Disclaimer: My knowledge of Audacity is at the push-the-buttons stage so don't assume any rigor in the results. 

[TorqueMaster]

Bob, thanks for posting your results, the zip file worked well.
I pulled it in to Audacity to see the trace. Below is the display (I hope you don't mind my posting your results).

Don't mind at all, I posted it so all interested can view it.  Appreciate any input or opinions on the data collected. 

I found the db view to be very telling -- the envelope of the sounds, the XY "area" it takes up,  gives some perspective on loudness.  The inset images of the spectrum window is also telling. 

Subscriber said "They sound different.  Not as dramatic as I expected.  I wonder if a standard continuous background noise would help, to serve as comparator."

Listening to the clip, they do all seem about the same "loudness" -- far from how it sounds in person.  I wonder if the differences would be more pronounced (and realistic) with a real amplifier (or headphones) rather than playing on my pc's tiny built-in speaker...

I should have given a disclaimer -- some shot sounds are slightly truncated because I was hitting a metal target 35 yards away, which I edited out.  The waveforms appear to be mostly played out at that time, so not much info is lost.  I also had very minor audio clipping according to Audacity, so I need to place the recording phone slightly further away.  It was roughly 6 inches from the muzzle on these.

Constant background sound -- is a metronome clicking enough, or add pink or white noise...?

And yes someone mentioned sound level meter settings -- fast or slow, a, c, or z weighting, are all going to affect the db numbers.  IIRC, c and z were better for impulse sounds...  One of them a, or c, better reflects human hearing.   I think I was using C Fast.   Now with this, I'm just recording the sound, so it's the limitations of the phone's microphone tainting the information, no added weighting.  Probably would be more accurate with an external microphone intended for all frequencies, rather than the built-in which is probably best suited for voice.

Nonetheless, I'm always amazed how much instrumentation the common man literally has at one's fingertips.

Hoping I can get the felt plume and digital plume printed up soon, for the next round of testing.  Are the sound level meter apps, or spectrogram apps really needed with the raw recording and the tools within Audacity to analyze the recordings?  Guess I'm asking what's most useful...?  If any are not, or redundant, I'll skip them.

[WhatUPSbox?]

Bob, thanks for posting your results, the zip file worked well.
I pulled it in to Audacity to see the trace. Below is the display (I hope you don't mind my posting your results).

Don't mind at all, I posted it so all interested can view it.  Appreciate any input or opinions on the data collected. 
Glad to hear it. You did all the tedious work of collecting the data. 
I found the db view to be very telling -- the envelope of the sounds, the XY "area" it takes up,  gives some perspective on loudness.  The inset images of the spectrum window is also telling. 
To me the db may be a little misleading on duration, I find it harder to pick where it transitions to background sound amplitudes. The linear plot highlights the first order of magnitude drop. The answer is probably in between. On the spectrum shot you can zoom in to say the top 15db (Image for first shot). I'm guessing the roll off at the high frequency may be your mic.
Subscriber said "They sound different.  Not as dramatic as I expected.  I wonder if a standard continuous background noise would help, to serve as comparator."
I'm not sure injecting (digitally) a standard sound provides a reference. I think it would have to go through the same mic and ADC and also be carefully positioned. I've been using a second airgun in a stable configuration (P17 with a small LDC) to provide a sanity check datapoint inside a dataset.
Listening to the clip, they do all seem about the same "loudness" -- far from how it sounds in person.  I wonder if the differences would be more pronounced (and realistic) with a real amplifier (or headphones) rather than playing on my pc's tiny built-in speaker...
Probably some over ear headphones would help.
I should have given a disclaimer -- some shot sounds are slightly truncated because I was hitting a metal target 35 yards away, which I edited out.  The waveforms appear to be mostly played out at that time, so not much info is lost.  I also had very minor audio clipping according to Audacity, so I need to place the recording phone slightly further away.  It was roughly 6 inches from the muzzle on these.
I've been using 10-15 ft away on grass to reduce any geometry, mic pointing errors or local reflections
Constant background sound -- is a metronome clicking enough, or add pink or white noise...?

And yes someone mentioned sound level meter settings -- fast or slow, a, c, or z weighting, are all going to affect the db numbers.  IIRC, c and z were better for impulse sounds...  One of them a, or c, better reflects human hearing.   I think I was using C Fast.   Now with this, I'm just recording the sound, so it's the limitations of the phone's microphone tainting the information, no added weighting.  Probably would be more accurate with an external microphone intended for all frequencies, rather than the built-in which is probably best suited for voice.
The NOISH app description stated that the Apple hardware was pretty good, probably applies to the iPad as well
Nonetheless, I'm always amazed how much instrumentation the common man literally has at one's fingertips.
Yes! The instrumentation available at hobby pricing and free software are amazing
Hoping I can get the felt plume and digital plume printed up soon, for the next round of testing.  Are the sound level meter apps, or spectrogram apps really needed with the raw recording and the tools within Audacity to analyze the recordings?  Guess I'm asking what's most useful...?  If any are not, or redundant, I'll skip them.
For all the detail available in Audacity, when I compare the RMS of a portion of a shot trace to the same length portion of a second shot, I don't get the db difference I get from SLM readings. No matter the length of the sample. So I have not found the recipe for getting a solid number for the reduction an LDC provides. 

Great Work printing the different designs and collecting the data. Thanks.

[subscriber]

Thanks Bob and Stan,

The chances of saturating the mic at 6" seems rather high.  Capturing shot sound from closer than the industry standard 1 meter (3.28 feet) seems contra-indicated.  As the ears that might object are probably further away, my vote is for 15 feet; or 30 feet; or 60 feet.  This providing the mic is not right at a loud trap.  A steel plate would make for a loud trap...  The mic could be off to the side (a longer distance), level with the muzzle, to stay away from the trap.

Another reason for a longer mic to muzzle distance is that a loud high frequency component is absorbed by the air more than the rest of the spectrum.   I would like a better idea of what the guy next door hears than what would be deafeningly loud, 6" from the muzzle.

The background sound suggestion includes live noise being recorded, along with the shot sound in real time.  Not injected digitally afterwards.  That way, the constant background sound also serves as checksum to verify that the recording level is not auto adjusting.  If the recorded background sound is softer directly after every shot, then the shot sound is causing an immediate dip in recording level.  A floating recording level makes the results real hard to interpret...

A metronome at 2 Hz, placed a distance from the mic such that it is not louder than the shot should work.  A constant tone or white noise would work as well.  Just something that records clearly, and serves as evidence of recording level drift or distortion (if any) triggered by the shot sound.   

Bob's latest recordings make the difference in shot sound length obvious.   That suggests the shorter sounds are actually peaking at a lower level, despite what the recordings captured; probably due to clipping, or possibly also to residual auto-levelling.  Clipping would be a crude form of auto levelling.  Clipping would distort the sound, but I am not sure one would notice, unless it is music or voice.

[subscriber]

I was taking a nap yesterday, when another idea came to me. 
I call it the real digital sponge.  It took me over 8 hours to generate this - starting with the "Plume" LDC design: 

The wall of the LDC consists of multiple perforated tubes - except for the outer one.  The perforations facing the LDC ID  are heavily chamfered, while the back sides of the tube wall at the holes are "flat".   Perforations are staggered so they do not face each other directly between "layers".  The outer wall ID also has little cones to encourage flow into the annular space between tubes, but not out of it.

1. The idea is for the facets facing the LDC ID to scatter sound waves.
2. Then, for the differential hole edge conditions to result in air that flows more freely towards the outer wall, than back from the outer wall towards the LDC bore.

I have removed cone support ribs and added new ones, to free up internal volume.  Else the baffles are the same as the recent LDCs with "Plume" in the part name.

The STL can be downloaded for 7 days from:  https://we.tl/t-rehW0JQxVW

[WhatUPSbox?]

Wow, some very interesting, detail designs.

They bring several basic questions to my mind.
Where is the sound generated? Is the bulk of the sound generated at the initial air expansion and turbulence at the muzzle? At each of the air stripper features? As the turbulent air hits the outer wall? Probably all of these to varying degrees and it influences the kind of sound absorption or scattering features that are important.

The sound wavelength is a foot at about 1.1 Khz and an inch at about 13Khz. As surface roughness features become significantly smaller than that, the scattering is reduced. I'm not sure about the perforations.

On my printer (Ender 3) printing freestanding features at the 1mm scale gets challenging, especially with PETG. Though my printer (or the user) is not optimized.

Where is the Physics/engineering student in need of a senior project idea? 

[subscriber]

Stan,

Where does the sound come from?  I think all the factors you mention come into play.

To help contemplate the question, start with a bare muzzle.  There, one large pressure drop acts like an empty oil drum being smacked with a single hammer blow.  Partitioning off multiple smaller baffle volumes in an LDC, creates multiple smaller "bangs".   Each less intense than the ones before - in very fast succession.  These bangs occur inside the device, and so are shielded from outside.  But they also excite the structure to transmit noise directly at the structure's resonant frequencies.

The air may resonate at wavelengths you mentioned on, at one atmosphere.  At the barrel muzzle the speed of sound in air would be higher because the pressure is higher than one ATM.  Also, the air acts like the hammer whacking the oil drum.  My perforated tubes are an effort at decoupling the smaller bangs inside from the outer wall; and creating a poorly reflective surface - essentially what felt and hair curlers do.

The last textured interior design I had tested, did not do much.  It was too fine I believe, and so my last few have been scaling up the texture.  The way I generated the faceted holes is  flexible, and a result of a method, more than typing in a dimension.  I was aiming at 2.5 mm with a 45 degree per side cone.  Something that can be 3D printed.  See images below.

The perforated tubes work a bit like an IC engine muffler.  There, the small holes filter pressure spikes, but their collective area is large enough to allow flow, to and from what amount to collectors; or plenums.  Rather than flow through, my perforations act as a capacitor, fed by a leaky diode.  There, flow resistance acts as damping in both directions.   If the holes or their collective area are too small, then very little flow would enter the wall space.  If the holes are too large, then the intermediate walls might as well not be there.

As for the limits of printing resolution, I expect the sharp edges and points to round over and distort.  Being able to print a 1 mm pitch thread is an indication of the limits.  I would be happy if the cone and dimple features have 80% fidelity to the CAD model.  As long as the smallest aspects don't cause more loose "fibers" or other trouble, I don't mind if they are ignored.  I sliced this design in Cura and looked through it.  Did not see any hint of trouble.

You are right, this would make an interesting design of experiments project for a student or three.

[TorqueMaster]

Whoa -- another design!  I'm getting way behind

I had a little daylight with no rain this evening, did some rushed recordings.  I forgot the metronome/white noise background idea, but did verify the android app was set and stayed on the +1db gain, the lowest it stays on, rather that the "auto" gain setting. 

Tested no cap, stock cap and original 2-piece plume -- at 10 feet downrange from the muzzle -- and again at 10 feet to the side (90 degrees) of the muzzle.  On all of the downrange shots, Audacity shows audio clipping -- they sound different but I don't believe Audacity tools can make significant db evaluations.  I tried them all at exactly 150ms, truncating off my target hits -- I think I saw 3db between no cap and the other two, which is bogus. 

The 90 degree shots are vastly different!  Again at 150ms, I think I saw 9 db difference between stock and plume and 13 between no cap and plume.  Now we're getting somewhere!  The 90 degree no cap recording also showed clipping, so it may actually be higher.

I also tried the plume with the mic downrange at 9ft, 6ft, 3ft, and 0ft.  I'm a couple feet above the mic, so the distance and angle are both changing.  The 9ft and "0ft" both show audio clipping.  6ft and 3ft did not.  So I'm thinking there is a sweet spot between too close/too far, and too directly in front of the barrel, and too far off angle.
As far as annoying neighbors, directly downrange is probably the loudest and most useful, I need to move further back, until "stock" doesn't audio clip.  I'll probably move out to the field, with no (significantly loud) target hits and not much for sound to reflect off of.

Its annoying, that I take readings, go inside, download to pc, evaluate...maybe I can find an app that shows the clipping immediately.  Else the next test will be shoot, back up 10 feet, shoot, back up 10 feet, shoot...and evaluate the recording for clipping.  I think I have room to get 50 yards between me and the mic, and still have 50 yards behind the mic where the shot hits the underbrush.  I'll have to use my algebra-fu to figure out how much distance I need behind to not eff up the shot sound, which may linger for up to half a second it appears.

I may also try to map the phone mic response at various frequencies and plot a curve.  I have an analog sound level meter I trust (mostly) for comparison.  With that curve, my simple mind says we could apply the inverse curve to the recordings, and "flatten the mic."  Then apply whatever weighting, A or C, or not, to get "realistic" db changes between tested LDCs... For now however, just a getting a distance/angle with no clipping is top priority.

MPG is an .mp3 and the ZIP is a .wav

[subscriber]

Very good info, Bob

Clipping needs to be avoided, else all readings top out at that level...

My latest inspiration is for something other than the Avenger.  So, no more surprise designs; I hope

[subscriber]

Considering how the inner wall tubes are material deficient due to the many close hole cones, I decided to link the tubes to the outer wall by continuing the 6 longitudinal ribs that support the baffle cones.  A small change, but one that might reduce the potential for the inner walls to "flap".  Image below.

All other details are the same as above.

Updated STL file here:  https://we.tl/t-KCMSjEETpa

[subscriber]

I was not happy with an even number of ribs intersecting an odd number of coned holes in the tubular LDC wall.  So, I increased the holes size from about 2.5 to 3.5 mm, and reduced the number of holes by one, for an even 12 around the circumference. 

As the texture is even more aggressive and the remaining material in the perforated tubes even less, I decided to use 12 full length longitudinal ribs to stiffen the perforated tubes.

I promise not to post another design after this one (until there are test results), unless someone points out something that needs fixing.   I sliced the STL and walked through it to see if anything looked broken.  Looks fine to me.

The STL  can be downloaded from here for 7 days:  https://we.tl/t-SyyoPcLAzs

[TorqueMaster]

I haven't taken time to closely scrutinize the newer designs, but they look reasonable and printable.  I hope to get some printing done this (rainy) weekend.  I'll be sure to do this most recent version of the "sponge" and try to do the digital felt and actual felt as well.  Either print or give feedback why it won't print or failed print. 

EDIT -- Is there an order of printing preference on these?  If I can only get one or two done before I test again?  -- ENDEDIT

Cura has a new version 5.0 beta -- the awesome sauce it provides is variable print line widths, which should allow faster, more solid prints, with a lot less (hopefully none) hopping around on each layer to fill in the gaps left by constant line widths.  Ideal for prints like some of these with lots of thin walls of varying thickness.  The hopping around, on PETG especially, tends to leave blobs and strings -- a lot more than just straight-line printing does.  At some point I'll have to take time to try it.  Meanwhile, I'll keep using my old version that "ain't broke."

And hopefully, with an optimal recording distance, we can see and hear the differences of each style.

I got to thinking -- the recordings I attached to the last message -- the 90 degree recordings are most representative (so far) of what the shooter hears.  Plume is dramatically quieter than stock, stock a lot quieter than no cap at all.  So depending on one's need, i.e. not disturbing neighbors, or just a quieter shooting experience, both are useful to know...downrange AND what the shooter experiences.

[WhatUPSbox?]

Bob, can you provide a little more detail on how you are recording the data?
Are you recording a sound file (with a phone?) and then transferring it to a PC and Audacity? Or are you recording with the iPad running the ios version of Audacity?
For the 90 degree case, there is no downrange distance, the mic is 10 ft left or right of the muzzle, Is that correct?
For both the downrange and the 90 degree case, is the mic aimed at the muzzle? For the downrange case, have you tried different mic orientations?
For the 90 degree case, it may be interesting to try the moving blanket wrap to see if the sound is coming from the front aperture or caused bey the shroud/LDC walls.

Thanks for the data.

[subscriber]

Bob,

If there is an order to making and testing, then Felt, Sponge (latest); then digital felt.

Ahead of the muzzle is loudest, but unless there is a tall backstop (or strong wall), or a very long distance in between, we tend not to shoot towards neighbors.  Mostly, I think neighbors are off to the side. 

That said, measurements and sound recordings taken almost directly in front of the muzzle from 50 yards away; and perpendicular to the barrel, level with the muzzle a similar distance, should provide a very good picture of the differences between LDCs.  That will improve understanding, to support better designs.

[TorqueMaster]

Bob, can you provide a little more detail on how you are recording the data?
Are you recording a sound file (with a phone?) and then transferring it to a PC and Audacity? Or are you recording with the iPad running the ios version of Audacity?
Recording a sound file on Android phone, the app has an option to set a constant gain, rather than "auto" gain.  Then downloading to PC where I use Audacity.

For the 90 degree case, there is no downrange distance, the mic is 10 ft left or right of the muzzle, Is that correct?
Correct.  Muzzle was about 2 feet higher than mic in both cases.

For both the downrange and the 90 degree case, is the mic aimed at the muzzle?
Yes?  Mic was 10 feet away, aimed toward the muzzle in downrange case.   For 90 degree, the mic didn't move, but I and the rifle rotated 90 degrees and took a step or two back, so the mic was now pointed perpendicular to the muzzle/LDC tip.
 
For the downrange case, have you tried different mic orientations?
No, it was a quickie test, mostly to verify the phone wasn't changing gain on me, and to see if I could get some better recordings that weren't clipping.  I sort of failed on the gain since I didn't have a reference sound running.  Background noise seems pretty steady though.

For the 90 degree case, it may be interesting to try the moving blanket wrap to see if the sound is coming from the front aperture or caused bey the shroud/LDC walls.
Gotcha, yes the wrap test is still on the menu.  Probably better if the mic is pointed mid-rifle, or at the LDC rather than at the very tip?

Thanks for the data.

Glad it's of interest -- I hope to have recordings that are more meaningful soon... shooting for early next week. ha ha

That said, measurements and sound recordings taken almost directly in front of the muzzle from 50 yards away; and perpendicular to the barrel, level with the muzzle a similar distance, should provide a very good picture of the differences between LDCs.  That will improve understanding, to support better designs.

@subscriber -- Ok, got the order if I can't get to them all right away.  I have a fair amount of property/distance behind the house, no neighbors back there.  If I read you right you want the perpendicular recording to be 50 yards off to the side of me and rifle?  The 90 degree recordings I just did of stock configuration and especially the plume were already quite low at only 10 feet to the side... 

I'd like to find a downrange distance where the stock configuration doesn't clip, for starters.  Possibly closer for comparing LDCs, as long as none of them clip.  I understand the consistent distance idea, but my thinking is also to use the full scale of the recording to better highlight small differences.  Thoughts on that?