Home Theater Forum and Systems banner

1 - 20 of 33 Posts

·
Registered
Joined
·
48 Posts
Discussion Starter #1
Long time listener, first time caller. I'm getting into modeling subwoofers for the first time. I have done a bit of research and might know enough to be dangerous (probably in a bad way). I'm detail oriented and want to understand why I'm doing what I'm doing. I'm a mechanical engineer working in the aerospace industry by trade, and do a lot of work in measurements (from sensor to data acquisition to post processing). My life at work is inundated with FFTs, Campbell diagrams, etc.

Long story short, my home theater subwoofer crapped out. The amp started popping the speaker randomly. Its very old and I bought it used from a friend many years ago. I'd like to replace it with a DIY subwoofer. Ok, so few pieces of info to start.

- Will be used for TV, home theater, and music. TV is at a very low level, usually when the kids go to bed. Home theater is a lot of kids movies, and action movies. Music, I like all genres. The Pandora stations go from country, to rock, to pop, to classical and everywhere in between.

- The room is on a slab, with bedrooms above it. The room is L shaped, and is roughly 27'x23' with 8' ceilings. The room is open another half level (quad level house). Relatively large space.

- Price point is under $1000.

I started out looking at only sealed enclosures, planning to build a box and brace it properly. My first thought was to use a Dayton UM15-22 driver in a ~3.0 cubic foot box. I modeled the box in a 3D CAD program, and then passed it into ANSYS (finite elements analysis program). I ran modal analysis on the box, to find the natural frequencies of the enclosure itself. It was immediately aparent that even with 1.5" walls, there were flat panel modes under 250 Hz. So, I went and added internal bracing thinking that would take care of the problem. It did not, it moved the panel modes way out, but the braces themselves now had natural frequencies under 250 Hz. After going through about 10 revisions and coming up with a box that didn't have any natural frequencies under 500 Hz, it weighted ~200 lbs w/o driver. To me that seemed ridiculous. Onto other geometries that may work better.

After that I decided to look into either cylindrical enclosures or spherical enclosures (I do vacuum infused carbon sandwich panels for race cars as side work, so interesting geometries are in my wheelhouse). Quickly I dismissed the spherical enclosure, and started looking at cylindrical enclosures. That lead me to find a lot of information on Sonotubes and LLTs.

I modeled the sonotubes in ANSYS (the FEA program) to find the natural frequencies of the sonotubes, and like I expected, I found many under 500 Hz. The cylinder flexes along its axis and the mode shapes look like ovals. That then put me down the path of internal bracing. Imagine an I-beam curled into a circle, placed in the inside of the tube. That worked well, no modes under 500 Hz.

Here I am now, knowing the how I'll build the enclosure, and what methods of stiffening the enclosure I'll use, I need to know WHAT to build. I downloaded WinISD Pro and started playing a bit. Here is where I'm a bit stuck on what to do, and where I'd very much appreciate any advice and guidance you could give.

WinISD screen shots coming in a few, I'm working on compiling them into a coherent story.
 

·
Registered
Joined
·
48 Posts
Discussion Starter #2
First, I have only been looking at the Dayton UM15-22, for no other reason that my own ignorance to what is available in my price range of <$1000 for the entire build (amp, sub, enclosure, misc.....)

Here is my driver details. I think I have it input right. Dual 2 ohm VCs. I plan to run them in series for a 4 ohm load. Can someone please confirm I have this entered correctly.
Capture by Kevin, on Flickr

My first look is at sealed boxes. I want to know if I'm going about making these comparisons in a fair way. What I have done for the sealed boxes is created two, one with a 0.707 alignment, and one with a smaller enclosure. For the larger enclosure I set the input power such that the driver excursion was at the Xmax limit. For the smaller, I set the input power at 800W, the RMS power handling of the driver. Then I changed the enclosure volume until Xmax was reached.

The transfer function
Capture3 by Kevin, on Flickr

The driver excursion
Capture4 by Kevin, on Flickr

The SPL
Capture5 by Kevin, on Flickr



Now to figure out what this means. One question is my methodology correct. I basically pushed the designs to the point of hitting a limit (whether it was Xmax or RMS power handling). Here are my conclusions.

1. For an equal amplifier capable of supplying at least 800W RMS, the smaller enclosure easily outperforms the larger one. For a sealed enclosure, why would I think about going with a larger enclosure? Save money and buy a smaller/cheaper amp?
 

·
Registered
Joined
·
48 Posts
Discussion Starter #3
Adding in an LLT type design into the mix. 300 liter box, 15 Hz tune, 6" vent, ~24.5" long. Xmax was reached at 440W with this design. This one is the purple one in the plots.

Transfer fuction
Capture6 by Kevin, on Flickr

Cone excursion
Capture7 by Kevin, on Flickr

SPL
Capture8 by Kevin, on Flickr


So, Here are my questions about the LLT type design.

1. Excursion exceeds Xmax under 15 Hz. Is this a concern? I'm thinking it is, but not sure how to get around it, other than making a very small enclosure. In that case I'm not looking at an LLT enclosure now.

2. With setting the input power to 440 W, you can see in the SPL graph that this enclosure outperforms the sealed boxes under ~30 Hz, but above that the sealed box surpasses it. I assume that is because the sealed sub is being driven at 800W, while the LLT design is limited at 440W due to Xmax limitations. Does this mean that the UM15-22 is not an ideal sub for an LLT design? Other 15" driver suggestions for LLT.

Where should I go to look now? What else should I be looking at? Am I intrepeting these results correctly?

Thanks!
 

·
Registered
Joined
·
168 Posts
Hello Kevin, interesting way of tackling your new project. I built a few subs myself and I am familiar with FEM/Ansys. I may help with the modeling aspect. It would be interesting if you could stick accelerometers to the walls to validate the analytical approach :).

A few questions to establish common ground:
Are you modeling the enclosure with shell or solid elements? What are you using for Material, its Young's modulus and density (with units)? What types of bracing have you tried?

Other random thoughts:
- Assuming your modes are correct, an harmonic analysis in which you include material internal damping and additional damping (e.g. bitumen pads) may still show the wall deflections are negligible. For instance, modal analysis with the bitumen pads may lower the modes due to added mass without adding much rigidity but the net effect is positive since the modes are better damped.
- Of course the braces being made of the same material would vibrate at similar frequencies as the enclosure (Hardwood braces or birch ply may help with this.). Note that the braces vibrating within the enclosure would not radiate sound directly outside.
- Straight braces across opposing faces will not lower the modes modes but only couple the sides. Still, the mode shape is much less susceptible to excitation by internal pressure pulses.

Too late to look at WinISD files right now... Have you thought about building two subs instead of one?
 

·
Registered
Joined
·
48 Posts
Discussion Starter #5
I am modeling the solids, and in ANSYS using the default Workbench mesh, which I believe is primarily a tet element. The material properties I looked up for MDF. I will have to pull out the properties tomorrow and update this thread.

I did a pure modal analysis, not a forced response analysis. Given that the displacements I get out are all relative displacements. In the real world, those responses could be large or minuscule. W/o entering a forcing function I won't be able to tell. In my professional life, I don't like to predict vibration amplitudes, I simply design the hardware to have modes outside of my operation range.

As for bracing methods tried on the box shaped enclosure. I was working quickly and not saving my results for each iteration.

- No bracing, single 3/4" thk wall.
- No bracing, double 3/4" thk wall.
- No bracing, triple 3/4" thk wall.
- No bracing, triple 3/4" thk wall, with the middle panel perforated with 1/2" holes to about 50% solidity, to act like a core of a composite panel.
- Double thk 3/4" wall, 4" ribbing internal. No connections between opposite panels.
- Double thk 3/4" wall, internal bracing between all 4 sides.
- Double thk 3/4" wall, internal bracing between all 4 sides, as well as bracing extending down to the baffle side, and the side opposite of the baffle. I had this iteration working pretty well. I was fairly happy with the results. Here is a quick view of roughly what it looked like.


This one did not work well. The two ribs between the side panels had a mode of their own well under 500 Hz.
Capture by Kevin, on Flickr

This one worked better. Its not the final version of this type design, I don't have a pic of that on this PC. Its similar except there are several braces added to stiffen up the main bracing. Its a slippery slope, that's for sure.
Capture2 by Kevin, on Flickr

And the last one I have on this PC is the perforated version. This worked pretty wel, but was heavy with triple thick 3/4" walls.
Capture3 by Kevin, on Flickr


For the sonotubes, I have a few versions of bracing, but the final configuration which works well is the I-beam type design with a circular I beam internal in several axial locations. I also found panel modes in the end caps. To fix that I modeled in some transition rings, such that the inner volume at the end caps looks similar to what you'd see in an air tank, there the ends are hemispherical. That completed pushed those panel modes well past 500 Hz.
 

·
Registered
Joined
·
168 Posts
The design in the 1st picture could be improved by connecting your two "A" ribs to the left and right panels and having a brace between them. This will kill their mode and support the left/right sides as well. Usually 1in MDF with perforated "shelves" is just what you need. I should be able to show you examples if needed.
 

·
HTS Moderator , Reviewer
Joined
·
3,335 Posts
I'm no pro in WinISD and subwoofer designing, nor do I have education in your field, so a few of your points are going over my head. I did find two item I can comment on though, so hopefully that proves to be of some assistance...

So, Here are my questions about the LLT type design.

1. Excursion exceeds Xmax under 15 Hz. Is this a concern? I'm thinking it is, but not sure how to get around it, other than making a very small enclosure. In that case I'm not looking at an LLT enclosure now.
Yes, that is a problem. Under port tune the driver is essentially running in free air, so there's nothing but mechanical limits to hold it back. The best way to mitigate damage is to put in a high pass filter that will severely limit any output below the port tune. That shunts the signal going to the sub in the first place, meaning the driver is far less likely to exceed its design constraints.


- The room is on a slab, with bedrooms above it. The room is L shaped, and is roughly 27'x23' with 8' ceilings. The room is open another half level (quad level house). Relatively large space.
Even in LLT trim a single 15" driver will not be able to do much in a 5000 ft^3 space. Factor in the other half level - which has to be considered when calculating the overall space - and you should strongly consider duals. If you can do without the very lowest octave perhaps a horn design might work, although they're far more complex to design than a bass reflex alignment.
 

·
Registered
Joined
·
48 Posts
Discussion Starter #8
The design in the 1st picture could be improved by connecting your two "A" ribs to the left and right panels and having a brace between them. This will kill their mode and support the left/right sides as well. Usually 1in MDF with perforated "shelves" is just what you need. I should be able to show you examples if needed.
I agree, that design could have been optimized, but the second design's first mode was much higher in frequency, so I decided to spend time on that one instead.

As far as material properties, here is what I used for the modal analysis.

Density: 0.028935 lbs/in^3
Young's Modulus: 500,000 psi
Poisson's Ratio: 0.1

The density used was an average of several stated values I found. The Young's modulus seemed to be between 500,000 and 600,000 psi, so I took the lower of the two. For Poisson's ratio, I had a much harder time finding that spec. I found a listing in a paper somewhere online. I'm not sure its correct. Can you confirm these values are valid for MDF, or not?
 

·
Registered
Joined
·
48 Posts
Discussion Starter #9
Jim, Thank you for your advice. As far as a HPF to limit, how is this best accomplished? Analog RLC circuit placed after the amp, but before the driver? Digital (DSP) in the amp? Digital (DSP) in the receiver? I'm pretty much a newb when it comes to this stuff. I realize there are options, but I'm not sure what is common place, and/or the easiest.

I was looking at Dayton SA1000 amp, saw that it have a 18 Hz, 2nd order HPF with a Q of 0.8. I entered that into WinISD, and that really killed the low end, making the LLT design marginal as compared to a sealed enclosure. I assume there are way to get different HPF filters.

Also on that note, are there any other amps I should be looking into, that would fit into my price range (<$1000 driver + amp + construction of enclosure)
 

·
Registered
Joined
·
48 Posts
Discussion Starter #10
Some more details on my enclosure modeling/analysis.

Here are some more details on my box type design.
Capture by Kevin, on Flickr
Capture2 by Kevin, on Flickr

Meshed.

Capture3 by Kevin, on Flickr

First mode, 510 Hz, acceptable to me. Its a bracing mode shape.
Capture4 by Kevin, on Flickr

A few iterations of sonotube. This one is a double wall, with MDF rings between the two. Didnt' work out well.
Capture5 by Kevin, on Flickr

The last one I did was this one. Single wall. I beam style stiffeners. Hemispherical cap on the other end.
Capture6 by Kevin, on Flickr

Modal looked good. This first mode was at 475 Hz. Good enough I'd say. Its a nodal diameter = 2 response. Could push that out higher by increasing I-beam stiffness, but I don't think its necessary.
Capture7 by Kevin, on Flickr
 

·
HTS Moderator , Reviewer
Joined
·
3,335 Posts
Jim, Thank you for your advice. As far as a HPF to limit, how is this best accomplished? Analog RLC circuit placed after the amp, but before the driver? Digital (DSP) in the amp? Digital (DSP) in the receiver? I'm pretty much a newb when it comes to this stuff. I realize there are options, but I'm not sure what is common place, and/or the easiest.

I was looking at Dayton SA1000 amp, saw that it have a 18 Hz, 2nd order HPF with a Q of 0.8. I entered that into WinISD, and that really killed the low end, making the LLT design marginal as compared to a sealed enclosure. I assume there are way to get different HPF filters.

Also on that note, are there any other amps I should be looking into, that would fit into my price range (<$1000 driver + amp + construction of enclosure)
My preference would be in the DSP stage of the subwoofer amp. That direct correlation between driver and the output stage where the tuning occurs is perhaps the best way to go about it (rule of thumb is the fewer the influences in the signal chain the better off you'll be).

The Dayton amps are not favorites of mine. I have direct experience with 2 SA-1000's, 1 SPA-1000 and 2 SPA-500's and in every case they hummed (the reason for 2 SPA-500's was due to the fact I had one fail). On paper they seem to be quite nice, but I wasn't terribly impressed myself.

I'm surprised a 2nd order HPF at 18Hz made such a drastic change in the curve. 3Hz below that - port tune - and the output from the driver should have started dropping faster than 12dB per octave all by itself. I assumed the natural roll off would have made the HPF-induced one superfluous. Perhaps someone with more DIY experience can enlighten us on why that occurred.
 

·
Registered
Joined
·
48 Posts
Discussion Starter #12
I may have mis-spoke about the differences. I went back and looked at the models.

I modeled three designs. All UM15-22 driver.

1. Sealed. 3 cubic feet. Shown in yellow.
2. Vented. 9.5 cubic feet. Tuned to 15 Hz. No filters. Shown in grey.
3. Vented. 9.5 cubic feet. Tuned to 15 Hz. 18 Hz HPF filter (2nd order butterworth). Shown in Magenta.

For each configuration , I set the power level to acheive Xmax. The difference between the two vented configurations is the 18 Hz HPF, as well as a change in power to keep Xmax the same between each configuration.

First is transfer function. Grey is vented, no filters, magenta is vented w/ 18 Hz HPF, yellow is sealed.
Capture8 by Kevin, on Flickr

Then cone excursion. Grey is vented, no filters, magenta is vented w/ 18 Hz HPF, yellow is sealed.
Capture10 by Kevin, on Flickr

Finally SPL. Grey is vented, no filters, magenta is vented w/ 18 Hz HPF, yellow is sealed.
Capture9 by Kevin, on Flickr
 

·
Registered
Joined
·
48 Posts
Discussion Starter #13
My preference would be in the DSP stage of the subwoofer amp. That direct correlation between driver and the output stage where the tuning occurs is perhaps the best way to go about it (rule of thumb is the fewer the influences in the signal chain the better off you'll be).

The Dayton amps are not favorites of mine. I have direct experience with 2 SA-1000's, 1 SPA-1000 and 2 SPA-500's and in every case they hummed (the reason for 2 SPA-500's was due to the fact I had one fail). On paper they seem to be quite nice, but I wasn't terribly impressed myself.
Given that, do you have an amp suggestion which contains a DSP, which would be suitable?
 

·
HTS Moderator , Reviewer
Joined
·
3,335 Posts
Given that, do you have an amp suggestion which contains a DSP, which would be suitable?
For an internal amp the best I know of are from SpeakerPower. Inexpensive they aren't. An external amp I've heard mentioned frequently is the Behringer iNUKE 3000DSP, but that one I have no experience with. Because it sells for less than $300 I don't know how good it is though. I wish there was more info I could provide but not being a DIY guy means I only know things from the periphery.
 

·
Registered
Joined
·
168 Posts
I have a few comments to share about the box design that I'll keep for later since it should be one of the last steps. A straightforward decision path may have to look like this:

1. Define objectives in terms of SPL, extension etc... Decide between 1 and 2 subs.
As Jim mentioned, given your room size and budget, you may have to compromise on the last octave of extension. Another thing to consider is the output capability of your actual/projected main speakers. You also need to be aware of the advantages of multiple subs.

2. Decide on driver and tuning.
UM15 does seem popular. Requires large box due to high Qts, mostly due to high Qes. Ultimately this choice will depend on objectives.

3. Define preferred amp/EQ/LPF strategy. An internal amp will influence enclosure design.
Here it is the eternal integrated vs. separates dilemma. Both have pros/cons. I personally prefer separates for permanent setup which your family does not have to worry about. Something like AVR-> MiniDSP -> Pro/ClassD-amp -> Sub.

4. Select internal amp (if req'd), external amp can be selected later.
5. Box design, including FEM etc.
About the box design, I’m pretty much aligned with you in terms of properties except Poisson’s ratio which I found more around 0.33. In any case going from 0.3 to 0.1 changes modal frequencies by about 3%. Your wall thickness is probably overdesigned but we should review it later when the internal amp question is sorted out. Also the bracing in your preferred/latest solution may restrict the airflow and create reflections back to the cone. Think along the lines of http://www.parts-express.com/dayton-audio-15-ultimax-subwoofer-and-cabinet-bundle--300-7097, but improved. Is there any particular reason you selected that box shape?
 

·
Registered
Joined
·
48 Posts
Discussion Starter #18
Thanks for the advice on how to methodically go about this.

1. Define Objectives. That is a tough one. I don't have any real experience to reference. Based on reading and research here is what I have come up with. For the volume of my space ( 5000+ ft^3), I have decided two independent subs will be best. I would like to reach down to 15 Hz for the purpose of home theater usage. As far as SPL, I don't have a frame of reference, so I can't really say.

My mains are currently likely limiting. They are a set of JBL speakers I picked up from Best Buy some 10 years ago. I would like to slowly upgrade all of my speakers, so I've chosen at this point, to ignore the fact that my new subs will be far more capable than the current mains can match.

2. Decide on Driver and Tuning. I've played around with modeling some various subs in my price range, and found that the Stereo Integrity HT18 seems to model well and meet my needs. My current modeled driver/enclosure consists of the following:

- HT18, dual 2 Ohm version.
- 9.7 cubic foot enclosure
- 14 Hz tune
- 5.5625" round port, 29.5" long.
- HPF to protect driver at low frequences. Modeled as an 11.5 Hz butterworth filter, 4th order.

3. Define Amp/Eq/LPF Strategy. I have tenatitively picked out the Behinger iNuke NU6000DSP. Since it toes not have an HPF capable of 11.5Hz, I have planned to use a MiniDSP. I currently have an email into their tech support at MiniDSP to verify that the 2x4 device can support an HPF of that frequency.

4. Select Internal/External Amp See #3.

Enclosure Design. I have worked on this only from the standpoint of designing a shape that will fit in my area. At this point I think a Sonosub type design would fit best, but that is not concrete at this point.



So here is where I'm at. i'd like for someone to take a look at what I've selected and let me know if I'm going in the right direction. From my research I think so, but I'm no expert, that's for sure. Here are a few screen shots of my concept.

Capture by Kevin, on Flickr

Capture2 by Kevin, on Flickr

Capture3 by Kevin, on Flickr


Capture4 by Kevin, on Flickr

Capture5 by Kevin, on Flickr

Capture6 by Kevin, on Flickr
 

·
Registered
Joined
·
48 Posts
Discussion Starter #19
Also, I want to confirm. Port volume is not included in enclosure volume? So I take my interior volume, and subtract the port volume to get enclosure volume?
 

·
HTS Moderator , Reviewer
Joined
·
3,335 Posts
2. Decide on Driver and Tuning. I've played around with modeling some various subs in my price range, and found that the Stereo Integrity HT18 seems to model well and meet my needs. My current modeled driver/enclosure consists of the following:

- HT18, dual 2 Ohm version.
- 9.7 cubic foot enclosure
- 14 Hz tune
- 5.5625" round port, 29.5" long.
- HPF to protect driver at low frequences. Modeled as an 11.5 Hz butterworth filter, 4th order.

3. Define Amp/Eq/LPF Strategy. I have tenatitively picked out the Behinger iNuke NU6000DSP. Since it toes not have an HPF capable of 11.5Hz, I have planned to use a MiniDSP. I currently have an email into their tech support at MiniDSP to verify that the 2x4 device can support an HPF of that frequency.
You shouldn't really set your HPF less than port tune because that has a tendency to destroy drivers during spirited listening sessions. Below port tune the driver "unloads" and is essentially running in free air, with nothing to slow its motions but mechanical limits. If you continue to pump signal to it at full strength you're likely to exceed xmax, and perhaps even xmech. Bad/expensive things generally occur when that happens. You might want to consider 15Hz for an HPF instead. That way by the time you hit port tune the HPF has already started to throttle back the signal strength.

Also, I want to confirm. Port volume is not included in enclosure volume? So I take my interior volume, and subtract the port volume to get enclosure volume?
Correct; port volume is not part of enclosure volume.
 
1 - 20 of 33 Posts
Top