DSP Best Practices - Page 3

Heh heh, Terry said "homomorphic."

In best Jack Benny impression, "Now cut that out!"

I want a homomorphic lens on my projector...

You have that large space between the false front wall and the actual front wall. You could pack the parts that are not needed for sound travel with insulation or other absorbent material. That would go a long way towards killing resonance nulls. Then do the same thing with a false wall in the back, and you're home free. For bass trapping, you don't need "acoustically transparent" cloth as much as for the higher frequencies. You could probably cover your traps with that wall velvet, and they'd still work OK. Totally integrated!

Does your seating platform extend to both side walls and the rear wall? If so, what are the dimensions of the platform, do the supporting joists run front to back or side to side and what is the spacing between the joists?

The slots will make the platform's absorption characteristics tuned to a specific frequency and Q...you could be absorbing the wrong frequency. Generally a bad plan unless you know what you're doing. It's too bad your platform doesn't extend wall to wall and to the back wall. There would have been a solution in that.

Are you willing to put an extra pair of subs in back ? Does what you have fit in the columns? What space is available in front stage for existing subs?

Indeed you can cut in some pegboard and make a helmholtz resonator in your columns - that can be easier to tune - just fill in another hole if you are off key. Dennis even has a how to hint if I recall in the archives of HomeTheaterBuilder magazine if they are still available.

Doesn't gamma curves in video qualify as homomorphic processing already?

Even at higher frequencies fabric does not need to be "acoustically transparent" because if it absorbs a little more on its own no harm is done. What really matters is that the fabric not be reflective.

--Ethan

Yes, that's what I meant. A plastic tarp or vinyl cover wouldn't be good for high-frequency absorbtion, but might be OK for a bass-only trap.


Yes, MDF is way too dense. You could build your columns out of vertical members with spacing (like an octagon), and clad them in cloth (which will prevent the insulation from escaping). Then fill them with insulation to get bass trapping. An alternative is to wrap them in 703 or 705 board, or SoundSelect black fiber mats, and put cloth outside that. 703/705 doesn't really have much particular matter that escapes, compared to regular R-13.

If anything the holes are just damping the wrong frequencies or are more broadband, though sometimes when the damped phase of the modes interact they do non-intuitive things! If I recall the Auralex design it was a broadband version with different volumes for each hole? As a coke bottle changes blown frequency depending how much you drink - same principle.

Portholes work on the same principle a port in a speaker does - you ever hear what happens if you stuff a sock in a speaker port? The mass of air in the hole needs free to vibrate in the hole - otherwise it is not a Helmholtz. Pegboard is the same principle - smaller holes but more of them.
Maybe Dennis knows his own article citations - I would have to dig thru the mag stack....you could easily retro in the corner cutout register vent he was suggesting though unless it is in the walking path (your wife does not want to catch her heels in a vent!)

Doubling up subs on the front stage makes it worse not better - you would just be driving the modes even harder - to combat length modes you need to have a pair in back - if they were smaller they could fit in the back columns - but I think your walkway is in the corner so subs cannot go there if they are as you say big bastages The best analogy is a kids swingset - you get mom&dad pushing in the direction of the wave the kids goes faster and higher - but if one pushes the wrong way - the kids takes a spill (kills the modal wave)

Anyways I just checked my SubSpace calculator for an all corner sub placement often a great solution for keeping your SPL up whilst killing modes. The problem is you are in the null of the third length mode but at the peak of the 2nd/4th modes- you still get the dip in between them because they are so strong http://krasmuzik.biz/portfolio/Kellog_4corner.pdf . You can mitigate that by moving back or forth off center and/or dropping them with a pEQ or killing the 2nd with quarter wall placement http://krasmuzik.biz/portfolio/Kellog_4quarter.pdf and mitigating the 4th with lower crossover and placing your LCR'S in the null or near at 1/8 length.

It is likely different subs will have different magnitude/phase response - so the results will not be as perfect - but the most important thing is the subs be all the same polarity - so you can always resell them if it did not work. You also have to be careful about picking one of the Harman patterns - here is one that does not work http://krasmuzik.biz/portfolio/Kellog_4half.pdf - it mostly kills the 1st/2nd modes - which also kills the dip between the 2nd/4th - and thus exposing the deep null at the third! I think the same could happen with the 4 corner solution if you tried to kill the 2nd mode with a pEQ - the dip goes away exposing the null you are still in.



broadband just dampens - it does not eliminate peaks/nulls.

Actually the color near the subs is more pastel - showing it is more white - you are mostly seeing the subs near field response there. It tends to show the color of the modes that it is driving as you start to move off the sub - you see the quarter wall subs plot is pale green near the subs as it is driving the 4th mode hard which is in the green spectra. But you don't want to sit so close to a sub so that one is maybe not good.

The colors in the spatial plot show what modes dominate the spectra - the idea is to get all your seats in similar colors so you have the similar response - then you pEQ the remaining issues. If seats are in different colors - you make the money seat better and risk the cheap seats getting worse with pEQ.

Can you get the subs out to 1/6th wall? That would place them in the 3rds null and guarantee it does not come back to haunt you!

Did not work - i need to think why not!

http://krasmuzik.biz/portfolio/Kellog_4sixths.pdf

(sorry for breaking off my edit - train of thought from your reply!)

I see the answer in your plot- you also have a tangential mode at 60Hz coming from the 2nd length and first width mode! You have two nearby nulls - I only had one null because my perfect model perfectly aligned the tangential and axial frequency - your room is not so perfect a model and they shifted - exposing two nulls! So killing the 3rd length mode - just exposed the tangential null! You can see you have to be a multidimensional thinker/troubleshooter to solve modal issues! I had to write the modal calculator because even simple math I lose my train of thought in the middle of the thought process... Note to self - make chart background grey so I can see the yellow lines representing tangential frequencies!

Moving the subs to the fifth fraction brings us back to the 0dB dip instead of deep null so that problem is solved- but now you have a problem being on the null with the first length/width tangential.

http://krasmuzik.biz/portfolio/Kellog_4fifths.pdf

Any possibility at all of moving forward away from half length and getting out of the null? It will be like a free bigger screen upgrade - and you will get a better soundstage angle!

I am done for tonite- too many dimensions to consider!

"I think the same could happen with the 4 corner solution if you tried to kill the 2nd mode with a pEQ - the dip goes away exposing the null you are still in."

Oops that will not happen with the four corner solution - it kills off the 3rd mode by being in opposing phases - so its null will not get exposed again.. But it may expose the other 50Hz tangential null if you reduce the 2nd with pEQ. SO really only the 4 on quarters which kills the 1st, 2nd, and 3rd length modes and 1st width mode will work for sure. It kills the 2nd by being on the null - the 1st/3rd are killed by opposing phases. So if it is not the same sub you may need to fiddle with the back subs gain/phase so 60Hz indeed is killed.

What location is that plot?

Don, your long, skinny room is perfect for a "Double Bass Array". Terry (long time no talk) you should check this one out too. Short version, you can achieve "perfect" sub response in a room with no EQ and no room treatments. At this point I can't imagine using anything else in a dedicated (expensive) theater. The cost is lots of sub drivers but the drivers (either mounted IB or in some cheap boxes) can be cheaper than a couple of commercial subs and/or extensive bass trapping.

What you do is build an "infinite planar array" of drivers on the front wall. That means the drivers are spaced no more than 1/2 wavelength apart and the driver-boundary distance is 1/2 the driver-driver distance. This totally eliminates vertical and horizontal modes and nulls. How about longitudinal modes and nulls? You build an "active bass trap" on the back wall, i.e. an identical driver array. The rear array is wired with reverse polarity and it's delayed by the time for sound to travel the length of the room. So the bass passes the listeners and just dies when it reaches the back wall, being cancelled by the out-of-phase, delayed drivers.

Here's the thread in the DIY section. Note that he starts running into problems around 80 Hz because his horizontal driver spacing is too big with only 4 drivers. Using more drivers to get the horizontal C-C distance smaller would fix that. The pic is with 1/24 octave smoothing -- pretty dang impressive if you ask me!

http://www.avsforum.com/avs-vb/showthread.php?t=837744

Edit: moving the pic to the next post to make this one easier to read.

Cool idea! Wish I'd thought of it. I have played around with simulating the 1/2 wavelength grid of subs, and also the cancelation from the opposite wall using a time delay. But I never had the sense to limit this to just one dimension, which is the secret to getting a pseudo-planar wave and thereby not having to deal with the other 2 room dimensions.

Gonna start modeling this, perhaps as soon as tomorrow!

Regards,
Terry

Active LF cancellation/diffusion is an area my thesis advisor Dr. Avis has been working on - it really is not practical as a product because of all the drivers and custom build required - much like IB would never make sense as a product. But if a DIY's time is free then it could be done. It think it makes more sense to do once LF planar speakers become more common - though in-wall subs from Triad and others certainly make it more feasible for custom.

I had a similar idea for a push/pull soffit transmission line sub years ago - rather than using the electronic delay you use the soffit as acoustic delay - what comes out of the soffit end is out of phase since it is coming from the back side of the driver with that coming down the room from the front side of the driver. Of course like the IB - such a design is only possible with DIY - nobody is going to be shipping a custom built soffit sub! I have never been a successful DIY speaker builder - it requires wood working skills I just do not have...I never got round to modeling it - just a sketchpad idea in my book.

Placing 1/4, 3/4 points is exactly what the Harman method is about - you kill off the first three modes that way - though it boosts the 4th mode - this is usually over the crossover especially in narrow short rooms. Combining that with active cancellation is indeed an interesting idea - you essentially are making the room into a 1D duct problem..

Terry do you remember a thread on studiotips where someone linked to the idea of a subwoofer array for similar idea to remove the modes impact on TL testing allowing lower frequency testing?

Your room is not that tall - it only starts to go height modal near the crossover - you could put the subs at mid height and that takes care of that. Even your width only has one mode under crossover. Seems this would work in your room with just front/back center of the wall subs - rear sub delayed and out of phase. Of course that is in your door - not very practical - you can't leave the door open to prototype it as that changes things! So a pair up front and pair in back on quarter width and half height is better keeps your virtual width distance within the half wavelength (~7' for 80Hz)

Of course the beastly subs you have will not work - you need to do an inwall design. But if you were doing that then the quarter point side walls solution already kills off all the modes without any pEQ if you lower your crossover to avoid the 4th length mode. But I would have to assume your side walls are not well damped the way the tangential mode is biting you - so seems this other idea should work assuming the math works out and there is not some hidden oops in it that the DIY types not being acoustic engineers are not seeing...