I read on https://www.johnfoy.com/faqs/whats-wrong-with-3m-companys-dual-ended-combat-arms-earplugs-version-2-caev2 (mirror):
One side of the earplugs acted provided protection from loud noises, such as explosions or gunfire, while still allowing soldiers to hear and receive orders.
How can one design earplugs so that they provide protection from loud noises, such as explosions or gunfire, while still allowing the user to hear voices?
Tricky.
The spectrum of small fire arms (guns & rifles) is about 150Hz to 2500Hz (with a peak SPL at 900Hz-1500Hz). This overlaps a lot of the speech spectrum. Larger weapons sit lower: below 400Hz with the bulk below 100Hz.
Passive air plugs have a distinct lowpass characteristics: good attenuation at high frequencies but much less protection at lower frequency, primarily due to residual air leaks. That's why custom fitted plugs work a lot better than "one size fits all" models.
The main problem with speech intelligibility with ear plugs is not overall level, but the titled frequency response which affects mostly consonants.
I'm guessing the ears plugs in question were designed with one side being as sealed as possible and the other with having an intentional high frequency leak to flatten out the total attenuation and get more of the plosives and fricatives. Maybe you can design this leak with a non linear behavior that closes at high levels.
Both approaches are not great: ear plugs don't do well at low frequency especially if they are not fitted and any attenuation spectrum that maintains speech intelligibility will not offer a lot of protection for against gun fire noise. I'm guessing that a fair bit of benefits of this particular plug came from a creative marketing department (which the military is surprisingly vulnerable to).
Alternatives:
Active noise cancellation can be quite good at low frequency and adaptively deal with some amount of leakage. However, they generally don't have enough output to cancel a gun shot and will simply clip.
Ear plugs alone are unlikely to give sufficient protection: the combination of over the ear protector and ear plugs is much better. The over the ear portion can have a speech detector and selectively pass sound and can also partially compensate for the frequency tilting of the passive earplug underneath.
Gun fire and explosions are insanely loud: anywhere from 130dB to 190dB. Fortunately they are also were short but some amount of damage is unavoidable. It's not just guns: the inside of a navy sea rescue helicopter with the door open is REALLY loud and I fried a few microphones on my first measurement attempts.
In your question you asked about the 3M Combat Arms Earplugs, I skimmed there patents and documentation. Here is roughly how they work.
The current gen Combat Arms Earplugs have a rocker to switch between high and low attenuation, the side with the small hole is the low attenuation setting:
.
They provide data tables with open and closed attenuation figures and standard deviations, I plotted then below:
And here is the plot of the attenuation vs peak sound level for impulses:
The earplugs are covered by two main patents, the first is US6148821A Selective nonlinear attenuating earplug whish uses a slightly different continually rotating but presumably similar to the production devices scheme to select between an air channel that goes through an "acoustic valve" vs completely closed, like traditional earplugs.
All the magic is in the acoustic valve covered by US6068079A Acoustic valve capable of selective and non-linear filtering of sound diagram below.
It has two main parts, the dual cavities that presumably act as a Helmholtz resonator (here is a paper on a similar coupled cavity resonator) to generate the low pass effect above, coupled with tiny perforations on the ends that due to there scale interact via viscus effects, to cause capillary resistance in addition to having to physically vibrate the disc in the center, which must physically move to transmit the sound. As the amplitude of the sound increases the plastic must start to deform more and more, as this occurs it resists the deformation causing it to not move as much as the air does, causing nonlinear acoustic effects that result in attenuation that's proportional to the sound level.
I know almost nothing about acoustics and have no idea if this is physically plausible, but my first thought upon reading the question and presuming it hypothetical was "fill a soft shell with a non-Newtonian fluid."
My thought process: it flows and allows wave propagation with minimal force but acts solid with sharp force, so the concussive force of a loud bang would be blocked/absorbed by the solid phase, while the relatively low impact of a human voice ("DON'T YELL!") would allow transmission of sound through the liquid phase. In the context of the referenced lawsuit, I could certainly imagine such a substance degrading with time, sub-optimal storage conditions, etc. and/or having horribly low tolerance for deviation from an ideal composition, easily resulting in many "flawed" batches.
Feel free to correct my science, but please be gentle. I love to learn, and I figured I'd share this odd-ball idea in case it had any scientific -- or, at worst, comedic -- value.
The latest Apple earbuds have a setting that works like this. Most Active Noise Cancelling (ANC) earbuds have a pass-through mode operated by a touch control on the earbud (so you can talk to someone when they come by to take your order). In this mode the closed-loop ANC is still active, but the “outward-facing” mic is added to the signal. So it’s just a matter of making the pass-through mode level-sensitive (pass-through enabled at low SPL, and gradually disabled for larger SPL’s) As an amateur musician, I find this feature useful, although it would be nice if the threshold were adjustable up to rock-n-roll levels.
Protecting against loud gun-shots while still passing intelligible speech suggests a non-linear function that reduce levels at gun-shot levels (and spectrum) more than it does speech-levels (and spectrum). One could imagine passive purely acoustical means to achieve this, or active electronics (for cancellation as well as reproduction) combined with passive acoustics for attenuation (sort of like noice cancelling earbuds with "speech mode").
Electronics works well for attenuating low frequency noise, while acoustics works well for attenuation high frequency noise.
Our ears are capable of adapting to a wide range of levels. From whispers to shouting. If the goal is to remove/reduce the damages from really loud sound, perhaps a (spectrally sensible) linear attenuation can shift some kinds of gunshots down into the "not so damaging range", while still maintaining speech in the "possible to hear" range? Particuarly if the buds have nice spectral characteristics (flat or high-pass rather than the usual lowpass)? Just speculating here.
It seems to me that this can only be solved with electronics support.
It starts by having a very good sound isolation function. As example, one brand of hunter headsets has 26dB of isolation in their product as sound protection. The amount of attenuation should be enough to protect against hearing damage. This amount of attentuation does not allow you to hear spoken instructions.
An electronic circuit with variable amplification is added with an outside microphone and a small speaker inside. The circuit is made to not create sounds higher than a safe level presented to the ears. The ones I have used even allows me to set the headsets to actually amplify sounds, allowing me hear soft sounds even better than without headsets, ie listening to sounds from aninmals moving in the wood. External sounds above a certain sound level are not amplified, even decreased in volume.
It seems to me that the military cheated out on supplier proper protektion. They furnished the soldiers with a choice of good protection but not hearing orders or low protection giving rise to damages.
