Tub trap fallacy

[DanDan]

Master Handbook of Acoustics Pages 227 -229.
Here we find a room with a strong long primary mode at 47Hz. On the Waterfall plot it appears to ring on for almost a second. A tuned Helmholtz TUB trap is introduced and the decay diminishes to around 300mS. Surely an almost unbelievable spectacular improvement, particularly as it comes from just one Trap. This section of the book comes from Doug at ETF. I emailed him to verify. He reckons the example was real, although 'performed by someone else'.
The result is so spectacular I undertook to replicate the example. I have a room, generally very good, but with a long 34Hz sub boom. I constructed a TUB trap, bigger than the original.

I went through a long process of calculation and experimentation. I will not present it here for reasons which will become obvious. I could detect Helmholtz resonance easily. I could also detect the widening of the Q with the addition of absorbent fibre in the TUB.
I placed the TUB in a corner where the 34Hz booms was loudest. I readjusted for perfect resonance in situ. ( A sliding port)
I noted a decrease in SPL near the tub of 3dB or so.
However I could detect no significant change in the mode. NONE.

I am not showing details here but I did this fairly carefully. I was quite excited at the prospect of these single (recycled) traps sorting out those long primary modes which we all struggle with. I wanted this to work and to share my tunable design with y'all.
I did enjoy this little adventure, clearly I need to get out more!
I did also learn some calculation and measurement tricks and techniques, flexed a few of the old grey cells.
Perhaps the originator of this example will see this and explain?
In the meantime I hope this post will save others from wasting time and effort.

Best Regards, DD
Sound Sound - Homepage

[spm_gl]

Where did you place the tub? It needs to be in an anti-node of the mode to be effective. Do a quick check here: hunecke.de | Room Eigenmodes Calculator

And don't expect miracles from helmholtz, their effective absorption area is around 2.4x mouth area.

Edit: We have an array of helmholtz resonators in the corners of our tracking room, they work very well there. But it's an entire array, floor to ceiling in two corners.
Edit again: Sorry, I just read your placement.

[andrebrito]

Are you sure the tub is acting in the correct frequency ? What methodology have you used to build this ?

[spm_gl]

BTW, I sold my copy of the "master handbook" after just one read, there are better, more precise books.

[DanDan]

Thanks for you interest guys. I would love it if someone found a flaw in my method or results, however I think spm has nailed it, and the idea of one Trap (as in the book) is simply not true.
spm- My instinct told me that it would take a lot of anything to affect Bass. However in the book, confirmed by Doug, it was just the one Trap. Their frequency was 47Hz, mine is 34, but my Tub looks a lot larger. I haven't come across that 2.4 factor before, where did you get that? It does of course ring true! I cannot agree about MHOA, I really like that book and others by Mr. Everest. While his writing style is indeed simple, I have never found him to be imprecise. The section of the book was hosted by Doug of ETF, the experiment was done by someone else, who clearly didn't give Doug the full detail, or else gave him simply false information.
Some details-I fed the room with 34Hz, tuned for max level with a Hardware Generator, no increments. DPA 4061 microphone readis the level in ProTools. I place the TUB in a corner which seemed to have the highest level (quite uncomfortably loud in fact). There were significantly different results depending on the exact position and orientation. On it's side, pointing into the corner, gave a tiny alteration in the Waterfall. Other positions had even less effect.
The port tube can be slid in and out because that lid is 40mm thick.
Resonance in another room was established very clearly with a 10cm long port tube.
I cut this down to 9 or so. The combination of the 9 plus up to 2cm from the lid depth made it possible to get exact resonance. This Resonance was detected by maximum level inside the TUB. That DPA is tiny and can go anywhere. .
I went at this a few times so I am reasonably sure of my methods and results.
But any ideas or queries are welcome, and I will do any test anyone wants, before I get rid of all this stuff.
DD

[spm_gl]

To be efficient, a resonator needs covering over the mouth. But not just any, you need a specific flow resistance Rs of approximately Rs=2,2 S/(f*V) [kPa s/m] with S= mouth area in cm², V = resonator volume in dm³, f resonant frequency.
Great to have a shelf of literature, I'm still searching for the effective area reference...

[spm_gl]

Okay, got it. "Schallschutz & Raumakustik in der Praxis" (German books are usually better) tells me, the equivalent absorption area depends on volume and frequency. Goes from 0.2 to 3.0. There's an equation too, but no equation editor :-)
Edit: found a simplified formula: A = 6*10^-5 Vfk [m²] with V volume, f frequency, k "positioning factor"
k=1 for wall, k=2 for edge, k=3 for corner. (That's 6 times (10 to the power of -5))

[DanDan]

spm thanks for those formulae etc. Can one use a formula editor in Word or such and cut and paste? At the start I tried all sorts of calculations. The one at mhsoft gave a result which related to measured reality. Others did not. Terry over at studiotips got a port length twice my number. So I went entirely empirical. The resonances I am detecting are pretty sure footed.
I haven't come across that notion of placing some gas flow resistance in the mouth of the port before. I will try it with various densities of fluff, fabric, etc.
I think it is fairly clear that the error here is in the notion that one trap would affect such low bass. As you said an array would be needed. Much like any other bass device.
DD

[JohnPM]

Dan, when you put the tub in the corner which way up was it? Putting the mouth pointing downwards (obviously raised off the floor 6 inches or so!) can give better results, from recollection. BTW, did you get my reply re the impulse responses you sent me? They were not normalised, so the dynamic range is too small to run an accurate analysis, either need to save them normalised (preferred) or failing that try saving as 24-bit if that is an option. There is a visible effect even with the original files, though.

[JohnPM]

Here's a couple of waterfalls with and without the tub, the limited S/N doesn't help these though.

[DanDan]

Perhaps DanDan is doubly doubly wrong? Those plots seem to indicate a shortening of the decay. However the longer one seems to hit a brick wall in front. I would rather see it extend out and trail away nicely. I have seen similar things in FuzzMeasure. I usually don't accept them as valid. Changing the length of view window and threshold usually delivers a pair of Waterfalls that both end up tailing off. When I adjusted the plots of this particular case in that manner, I am seeing a small lump of new resonances, slightly higher, after the main lump has died down. Again a bit hard to believe. A delayed but frequency shifted resonance? I will tweak the files and send them to John for forensic analysis.....
Hi John, thanks for that. I have sent you two more files by Hotmail not PM, just minutes ago. Normalised and Minumum Phase. Don't know what MP means but the manual recommends it when exporting for use in external software. I tried the Tub upside down, supported to provide wide access to to the port. Indeterminate result, with no EDT and indeterminate Waterfalls. Upright, even less interesting. On it's side, pointing into the corner caused some disturbance of the waveform and sometimes, sometimes, a change in EDT from 490mS down to 230mS. Hard to accept any results unless they repeat, but at 34Hz bets are off. Looking forward to what you come up with from the tweaked files.
Best Regards, DD

[jayfrigo]

I wouldn't expect that placing a single unit in the corner would be remarkable. The corner tunnel-vision you see on the internet is great for a DIY starting place, but it gets overstated.

Just because the bass builds up predictably due to the pressure increase where multiple boundaries intersect, this doesn't mean that the corner is where you need to eliminate bass. You don't listen in the corner after all. You want improvement at the listening position, not the corner.

If the mode is a front-to-back one, and you sit in the center of the room, then a trap in the corner will not necessarily do anything for you at the listening position. The standing wave still exists between the front and rear surface, and you'll still have trouble at the listening position.

I have no idea about these big ol' drums made into Helmholtz resonators. I've never done that. However, I have used various Helmholtz devices and can confirm that they do work. They have some advantages and some disadvantages, but properly built and placed, they do perform their function.

[jayfrigo]

Originally Posted by spm_gl

To be efficient, a resonator needs covering over the mouth. But not just any, you need a specific flow resistance Rs of approximately Rs=2,2 S/(f*V) [kPa s/m] with S= mouth area in cm², V = resonator volume in dm³, f resonant frequency.
Great to have a shelf of literature, I'm still searching for the effective area reference...

That's not necessarily true with all resonators. Things like slat/slot traps, for example, should have the resistance at the openings. It can help with certain things in other designs as well. However, resonators can still work with open mouths. As you get smaller and smaller, like the micro-perfs for mid absorption, you start to get viscous losses at the mouth as well. However it doesn't necessarily need to be a micro-perf to work without the porous absorbtion. After all, those big clay pots in the back of very old churches didn't have any owens-corning available.

[spm_gl]

My mothers silk scarf made a significant difference in our tracking room But I must admit that is subjective, we didn't have time to make measurements then.
I've come across "resonance ringing" in literature, meaning a narrow-band helmholtz can amplfify frequencies close to the absorption frequency. But I don't have experimental proof for that.

[spm_gl]

Here's a quick pick of our corner helmholtz array. I can't find the construction pics, so for now you'll simply have to imagine lots of different sized holes behind the dark blue stripe.
Point is, we needed to cover the mouths with a different fabric, the best compromise was a silk scarf.

[DanDan]

Hi John, can you extend the time line of the Waterfall that seems cut off? I don't understand the level issues, all is fine in FM2. The programs must be working in different ways. Unless that cut off decays nicely on a longer axis, I would regard it as false. I have seen such anomalies which present themselves later and later when I extend the time axis. They move with it! Of course I don't accept them as data. I will ask Chris to take a look, and maybe run my samples in FM3.
DD

[DanDan]

Jay, it is remarkably unremarkable, unlike the example in MHOA which claims a dramatic reduction in the mode by one trap only. We may be sruggling with possible inadequacies of measurement here. I may call over a friend with FuzzMeasure 3. If this experiment shows some promise I would be prepared to consider an array. I can get these tubs free.
I am intrigued by your comments about location of the device. I assumed the best place for the Tub was in a back corner where the SPL of that room mode is highest. Note this experiment is all about empirical. I hit the room with 34Hz and found, as expected, a null in the room centre, very high level at the back, more so in the back corners. I assumed that is the best place for the TUB. Do you have a suggestion for another, maybe better, location?

DD

[avare]

Thanks everyone for a great topic and thread.

Andre

[DanDan]

Cheers Andre. Some tidying up- John of REW has looked at the exported impulse responses and there does appear to be quite an improvement. However there are anomalies, which I have also seen in FM. The absolute level of the export is not compatible with REW so he cannot get a better picture, partcularly a longer time axis.
I am currently sending the file to Chris at FM. Let's see what he makes of the strange Waterfalls with his much more recent version of FM.
Best, DD

[liscio]

I threw together some quick shots of the waterfalls without, and with, the treatment applied. They're attached to this post.

I made sure to set a 1000ms duration for the waterfalls, and matching analysis windows (~1000ms). You can see a bit more of the mode's tail in these shots.

[spm_gl]

We still can't see the complete mode without treatment, but the effects are visible. Still far from drastic though.
Dan, do you think you could make a second test with 2, and perhaps 4 tubs, one in each corner?

[Brainchild]

Looks hell of significant. Those plots strongly recommend Helmholtz absorbers for modal ring. Plots for an array would be great.

[DanDan]

Thanks for doing those pics Chris. There is a lot more detail in those, presumably due to increased power in FM3. However, IMHO neither looks better to me, and I cannot clearly identify a 34Hz component, with a change or not in it. I wish I could but I am not seeing it.
Chris would you oblige some more please- Lengthen the axis to 2000 or even more mS. See if any anomalies show up. Try as far 3000mS. I have found that the anomalies travel out the time line with these extensions. Ghosts! Then lets go back to to 1500mS, which should show all of that mode.
I think the -65 might be too much. My room is not that quiet! My settings tend to be -40 or -45.
Please also try for third octave information decay times?
spm- I will think about that. I can get the tubs free but a carpenter associate made that lovely lid, and I paid him. I don't think the results so far merit further work but I will consider it. I do note your earlier comments regarding your own array.
DD