Ernest Christley wrote:

Here's a question for you audiophiles on the list.  How much effort is 10dB of muffling worth?

I've expanded on my experiments with phononic bandgap filters, and I think I've fairly conclusively demonstrated up to 15dB of attenutation in the frequency ranges between 8kHz and 12kHz.  I used the run-up video that Mike Wills provided for us a few weeks back (thanks Mike), and I was a little more methodical than last time.

For those that don't remember, a phononic bandgap filter is created by geometric arrangement of dissimillar materials, with the goal of filtering specific vibration frequencies.  It appears that 1/2" diameter tubes, arranged in a grid with 1" centers does a nice job of attenuating frequencies in the 8kH to 12kHz range.
I set up my simulated muffler...a wooden shell with PVC filter elements.   I stuck a couple speakers in the inlet side, and a microphone on the outlet side.  In this setup, there is going to be all sorts of losses from the conversion from an electronic signal to sound in the speaker, then conversion of sound to an electronics signal in the microphone.  Comparing the original to the recorded muffled sound, which I did in the first experiment, is not valid.

What I did this time instead, is to compare several recordings.  Each with one less row of filter elements than the last, until I was measuring an empty box.  The empty box measurement is the control that lets me know if I'm doing anything with the array of tubes.  What I found was that each row of tubes provide a significant attenuation in the frequencies between 8kHz and 12kHz.  The ones that give us the weed-wacker sound.  I determined this by taking a time sample across each of the 5 recordings and having Audacity plot the spectrum.

I have an xcf file created by The Gimp ( a free image editing software).  The xcf format allows you to keep an image in layers.  Each of the plots is in a separate layer, and by playing with the opacity individual layers, you can easily see where the attenuation occurs.  The file is 465kB so I can't post it to the list, but if anyone is interested, I'll be happy to send it direct.  I'll do a writeup with links to the audio files "real soon now".
I have attached a flatten image showing the 0 row run, against the 4 row run.  The 0 run is the lighter purple shade.  There are some some frequencies where the  4 row would be louder....around  4kHz and 7kHz, but I think the attenuation of those frequencies around 10kHz would be most beneficial.


------------------------------------------------------------------------

Hi Earnest,

10-15dB is a very respectable reduction. A 10 dB change is a 10X power change and a perceived doubling or halving in volume to the human ear (dB's are a logarithmic power measurement).

However, I doubt that working on the 8k-12k range is going to have much effect on the perceived noise from a rotary, for a couple of reasons. I can't find the emails & info from Monty Roberts' testing a few years ago, but IIRC, the problem areas are much lower in frequency. Look at the ~1k-2khz area & the

~3.5k-4khz areas. Those areas are much stronger, and they are in the 'zone' where human hearing is most sensitive. When Monty did his testing, the stuff above 4khz was harmonics radiating off the muffler itself (the muffler 'rings' like a bell). The muffler wasn't enclosed in a cowling.

So here's what I think. The 8khz & above stuff is barely audible to people over 40, or who have rock concerts in their history, or fly airplanes for any length of time. Additionally, stuff in that frequency range hates to turn corners and is easily absorbed by relatively low mass materials. When you take those two factors into consideration together, something as simple as enclosing the muffler in a housing that won't sustain a resonance (think fiberglass cowling) will kill just about all of it. The stuff under 4khz, though, is a bit harder to deal with, and that spot down below 2khz is really tough. As I mentioned earlier, those are much more likely to sound offensive, too.

If you can suppress the 1k-2khz range, the higher freqs will likely take care of themselves because they are harmonics of the lower freq stuff.

Remember the discussion about Paul Conner's engine with the stock exhaust manifold? The cast iron was massive enough to absorb all those higher order harmonics and the FG cowl finished the work, even with the exhaust being dumped out the cowl with just a little 8-10" stub pipe off the manifold. It wasn't 'quiet', but it was very pleasing to listen to, like a small block V-8. Of course, the downside for a non-renesis 13B is the loss of power because you don't get a tuning boost from a header. The only other pleasant-sounding rotary plane that I've personally listened to (other than turbo'd engines) is Dennis Haverlah's Renesis, and it uses the stock Renesis exhaust manifold. If you look at the way it's constructed, it's made of 3 layers of material, apparently for noise suppression and heat shielding in one package.

If Monty reads this, maybe he can repost his test results & expand on or correct what I've written here.

It might be helpful to load a tone generator program & listen to various frequencies to get an idea of what each frequency sounds like. If you use Audacity or a similar sound editing program, it has a tone generator built in.

Charlie