How "real" are internal standing waves?

[dscline]

I am going to re-build my sub, which is going to result in a much taller enclosure. It will be a tall slender column with multiple drivers, divided into two sections. This each section will have internal dimensions of approx. 13.5x13.5x39.5". I know to prevent standing waves, you need to stay below the quarter wavelength of the frequencies produced. The quarter wavelength for the 39.5" dimension is about 85hz, and my sub will be crossed over at 80hz. But due to crossover slopes, it will still produce some output well above 85hz.

Would there be a real issue here, or is this so close to the edge that the actual output in frequencies that can cause standing waves in this enclosure will be too low to cause any mentionable problems? FWIW, this will be built using Rythmik servos, wo I suspect the servo system will somewhat attenuate any issues within the box from being reproduced "through" the driver. But I'd still like to hear any thoughts on the issue.

Thanks!

 

[Collo]

Your wall-to-wall resonances are pressure node to pressure node, and are thus half wavelength, not quarter. This doubles the resonant frequency.

Driver-to-wall resonances are velocity node to pressure node and are quarter wave. The most troublesome one is the driver-to-top (or bottom)-wall for tall boxes, or the driver-to-back wall for deep boxes..

Here's yours.....

An externally hosted image should be here but it was not working when we last tested it.

With your sub crossing over at 80hz, I don't think you'll have any dramas. I would add some damping on the faces that are furthest apart to cover harmonics, but thats just me...

 

[dscline]

Well ****, typed up a big reply, got an error that my attachment was too big, with no option to recover what I'd typed. Let's try this again:

Thanks for your help, but I inadvertently omitted a crucial detail: my drivers will be mounted at the very far end of the long dimension. My understanding is the driver-wall resonance is still quarter wavelength based. Thanks for the tip on boxnotes... I hadn't seen that software before. I'll skip the screen capture I did of it, but it suggests a resonance at 100hz. I modeled one of my two sections, which has one driver on the front, and one driver on the back, at the very bottom of the enclosure. The other half (separate section) will have a driver at the very top of the back panel, and one on the top panel. So I suspect it will be slightly worse, since middle of the driver in the top panel will be the full 39" away from the bottom panel.

So I guess my best bet is to focus on damping the ends opposite the drivers. I've seen varying opinions of damping the sealed Rythmik subs. Some like no damping at all, some like just a thin lining of the walls (recommended by Brian Ding @ Rythmik), and I think maybe I've seen one person who stuffed it. Brian suggests against using too much damping material, as its non-linear response (relative to air) can be detrimental. So I think my best bet may be to not line all the walls, just focus all the material at the end opposite the drivers. Sound reasonable?

 

[Cal Weldon]

If the file is too big, can't you just push the back button? I just posted an oversize pic like you and it worked here for me.

 

[dscline]

No, when I tried to go back, it said the page had expired.

 

[GM]

Driver-to-wall resonances are velocity node to pressure node and are quarter wave.

Sorry, but these resonances are reflection generated, ergo half WL, same as a plane wave or vent tube.

GM

 

[planet10]

To expand on what GM is saying.

(assume that one dimension is longer than the rest) If a pipe is open or closed at both ends it will have a half-wave resonance. If it is open at one end it will be a quarter-wave resonance.

The half wave resonance will show up on the impedance plot. The bump just under 100 Hz (i'm guessing the broad bump just above 100 Hz is from not enuff holes in the holey brace). Box is end loaded and about 4.5 x 8 x 35" sealed

dave

 

[dscline]

Hmmm, ok, so is Boxnotes wrong? It says: "Standing waves occur between the driver and walls, where the distance represents 1/4 wavelength. You can also get standing waves between opposite walls, where the distance represents 1/2 wavelength" The output values shown in the software also suggest quarter wavelength resonances for the driver, and half wavelength for opposing walls:

An externally hosted image should be here but it was not working when we last tested it.

I guess what I can't wrap my head around is whether the driver would represent an "open" or "closed" end. It's not completely open, but it is essentially just a membrane between the interior and exterior. I would think that would impose an impedance shift, and maybe land somwhere in between depending on the Vas of the driver?

I certainly hope I don't need to worry about quarter wavelength resonances... just trying to make sure before it's built.

Thanks for all the help!

 

[Collo]

Originally posted by GM
Sorry, but these resonances are reflection generated, ergo half WL, same as a plane wave or vent tube.GM

The driver acts as an open end and the opposite wall acts as a closed end, so the resonance is quarter wave. As shown in this experiment at Rutgers University (New Jersey)


Damping opposite ends of the longest section of the box is the best approach. Unfortunately at low frequencies, most damping materials don't work very well. Given that the walls are pressure zones rather than velocity zones, any damping material that relies on air movement will not help.

If you look at treatments to be used for room walls, the most effective ones use a membrane that converts pressure into motion, which can then have energy removed by friction. An example of this can be seen with the Modex traps. This approach suggests that some closed cell foam, such as that used in those thin camping beds, would work better than normal stuffing.


When posting a long message, I try to remember to copy the message to the Windows Clipboard before I hit the Preview / Submit button

 

[dscline]

When posting a long message, I try to remember to copy the message to the Windows Clipboard before I hit the Preview / Submit button

So do I. Unfortunately, sometimes I fail.

Given that the walls are pressure zones rather than velocity zones, any damping material that relies on air movement will not help. If you look at treatments to be used for room walls, the most effective ones use a membrane that converts pressure into motion, which can then have energy removed by friction.

And this is where I'm having difficulty. The problem with membrane type systems is they tend to have their own resonant frequency, and therefore their attenuation tends to be most efficient at a specific frequency. I'm afraid anything like that would introduce it's own coloration, as it would be difficult to devise something that would target the actual resonant frequency in question. The end result could be worse than the problem. I know absobative type bass traps can be made more efficient by moving them further into the room, where they can have an effect where there's still motion. I wonder if I would be best served by a section of stuffing a few inches away from the far end wall?

 

[GM]

The driver acts as an open end and the opposite wall acts as a closed end, so the resonance is quarter wave. As shown in this experiment at Rutgers University (New Jersey)

I believe you're misinterpreting it somehow, but I'm unable to concentrate enough right now to offer any rebuttal beyond the attached sims.

Anyway, measuring the standing waves between the driver and floor or to a perpendicular side wall should have the same harmonic structure as in the box since it's just a tiny room.

Same 33.48 Hz Fs driver in end loaded 40" high aspect ratio sealed cab and TL showing 1/2 WL Vs 1/4 WL resonances:

GM

 

[planet10]

Just read Part 2 of Joe dA's most recent article in aXp. He does a specific example of tracking down the cause of an impedance curve bump -- a 1/2 wave resonance.

dave

 

[Collo]

It seems commons sense to me that a driver operating in sympathy with a standing wave would act as an open end.

Given the lack of specific evidence on the net (at least that I can find), I guess the easiest way is to do an experiment......

An externally hosted image should be here but it was not working when we last tested it.

Using REW to test the response outside, away from reflections would establish a baseline.

Adding a reflector and some damping around the enclosure to minimise any waves off the baffle (which would be half-wave) should cause the driver-to-reflector standing waves.

I predict a fundamental who's wavelength is 4 times the spacing between the driver and the reflector, followed by a series of odd harmonics.

If I'm wrong, the fundamental will have a wavelength that is twice the spacing, followed by a series of even harmonics.

It will take a few days to organise.

Meanwhile, I would be interested in a link to the article you mentioned (Part 2 of Joe dA's most recent article in aXp)

 

[GM]

Hmmm, ok, so is Boxnotes wrong?

I guess what I can't wrap my head around is whether the driver would represent an "open" or "closed" end.

Any reflection based on 1/4 WL is wrong TTBOMK, so just double the frequencies associated with the driver or wait for some definitive measurements before making any design decisions.

Normally neither since it would be either vibrating or being held at a rest position, so will be an active 'passive' radiator, i.e. the driver moves in and a wave travels out and bounces off a rigid parallel surface back towards the driver. This takes time and time = phase = frequency = relative intensity, so when it returns to the driver it's 180 deg out of phase, ergo its intensity has been reduced by the amount of time its been traveling, so its frequency is much higher and when it bounces off the driver's vibrating membrane their outputs sum in complex ways for a short period that causes the driver to 'miss a step' (the blip in the impedance) and further delay (damp) our wandering traveler into a still higher frequency before it heads back out to make another lap. If the driver is being held at rest, then the motor will damp any attempt to deflect it.

So from this we see that we either want to keep the driver close enough to boundaries to push any standing waves high enough to be easily damped or use non parallel walls or at least an acoustic ratio to create as uniform a particle density as practical. See the sim of the same driver in a golden ratio cab with an optimized driver location. Even though these are just 1D sims, the difference is dramatic, so imagine how major a 3D difference would be.

Anyway, just pushing them above the XO point or even the XO's BW doesn't necessarily work because they're reflections that if the driver is truly at rest it will act as a passive radiator with output that will comb filter with the mains output. Obviously, if the sub/whatever driver is acoustically far away, then there can be some audible time smearing in this BW.

GM

 

[rythmikaudio]

The driver acts as an open end and the opposite wall acts as a closed end, so the resonance is quarter wave. As shown in this experiment at Rutgers University (New Jersey)

Correct me if I am wrong, the mode due to the standing wave is largely because of the enclosure dimension and in that case, close end model fits best. In addition to this standing wave, the first reflection and it is phase relation with the direct wave can cause additional cancellation and reinforcement. SO it seems to be the problem has two parts, one is invariant and the other is variant.

Damping opposite ends of the longest section of the box is the best approach. Unfortunately at low frequencies, most damping materials don't work very well. Given that the walls are pressure zones rather than velocity zones, any damping material that relies on air movement will not help.

For now, I would agree damping is the best approach. To make it effective, try to keep the dimension of the enclosure small so that the nodal frequency is in the range that material is more effective. On the other hand, servo has a damping that is 3x of the nonservo (because for equivalent circuit purpose, one can imagine the servo system has a 1ohm Re, instead of 3ohm physical Re). However, the improved damping is mostly below 100hz (even though actual measurement does show effectiveness above 100hz as well, but I don't have explanation). So it is more effective to combine with regular damping material. However, I always recommend against stuff the entire enclosure with damping material as it tilts the balance of sound. Only stuff the ends. Enclosure's rigidity is also important.

Physically i can imaging, by putting various divider in the middle can reduce the amplitude of the mode, not completely remove it, and its effectiveness goes back to the fundamental debate of audio: whether it is less audible to keep the problem localized narrow band with higher Q, or to have multiple problem with lower Q. The chance of hitting the former is definitely lower and how does not play in our actual perception of sound quality.

 

[GM]

PWT experiments applies to dscline's app, so heavily damping the closed end opposite the driver will yield the best results and the side wall standing waves are too low in intensity to bother damping IMO, though lightly lining the walls around the driver shouldn't hurt its LF output.

GM

 

[dscline]

Thanks everyone for the input. I had no idea this would be such a gray topic. Sounds like my best bet is to continue with my plan to focus the damping material I would have lined the walls with towards the end opposite the the drivers. Seems like the best "real world" example and solution I can find to my application is this one linked at the boxenotes page. His results seem impressive, so I may take a very similar approach: closed cell foam at the opposite end wall, and some fiberous damping material attached to one of the braces near that wall. I think pulling the fiber away from that wall may make it more effective, since it works better with velocity rather than pressure. I would normally be a bit leary of adding closed cell, as I'd fear it would color the sound (and not be really sure how it affects internal volume), but fortunately the far end wall is only 13.5x13.5", so it won't take too much. And fortunately the servo drivers will also help dampen any nasties.

There seems to be quite a few people perfectly happy with their much taller low tuned sonosubs, so the effect can't be HORRIBLE. I would still like to minimize it to the extent possible.

Thanks again!

 

[Collo]

The rig.....tests were actually done with the pillows removed......

An externally hosted image should be here but it was not working when we last tested it.

The results.....

An externally hosted image should be here but it was not working when we last tested it.

Measurements were taken with the board at distances of 600mm to 1300mm in 100mm increments. The meter was left at a fixed distance of 260mm from the subwoofer.

It was expected that peaks corresponding to standing waves would be present, and that they would change in frequency as the board was moved. The expected frequencies are shown for both quarter-wave and half-wave possibilities.

As can be seen, no such phenonena are present, indicating that standing waves were not generated.

- Perhaps the reflecting board was not large enough, although doubling the size did not change the result.
- There may not have been time for the resonances to develop, although the REW sweep duration was quadrupled from the default value.
- The SPL may not have been high enough, although the test was carried out at around 90dB.

The idea that standing wave testing can be done external to the sub may have to be abandoned, and some internal testing done instead.
Accordingly, the next stage will be to construct a sub with a variable wall. Stay tuned.....

 

[Collo]

Someone pass me a tissue, I'm having an "Egg on Face" moment!

The new test equipment was a great success, with the driver to wall resonances clearly visible.

Unfortunately for me, they're HALF-WAVE, rather than quarter-wave as I had predicted.

The new rig....

A 6.5inch woofer was mounted at the end of a 6 inch pipe. This meant that the entire end of the device was driver, with no wall present. At the other end, a baffle with an SPL meter mounted through it, could be moved into the pipe to test different spacings

The results...

Testing was done with spacings of 500mm, 400mm and 300mm.

As can be seen, the peaks found correspond to multiples of half-wave resonances.
No evidence of quarter wave resonances was found.

A re-write of boxnotes (and sonosub.exe), is underway.

 

[GM]

Great work! Thanks for the effort.

BTW, with your original test rig you should have gotten your answer with a deep notch for 1/2 WL Vs a dip if 1/4 WL had you placed the mike at midpoint.

GM