Made further progress with acousto and just about ready to switch from getting things working to running off a set of simulations to examine and discuss. However, I think I have probably made a mistake in lumping together 1) methods development with 2) a group project to design an on-wall speaker given I am likely to be the only one wanting to actively participate in both. Although I think a few are interested in developments no-one has yet to suggest participating in a task.
Is anyone interested in actively participating in a group project to develop an online speaker today? If not, I suggest putting it aside and raising the topic again in a few months after the simulation side is better sorted.
In order to handle the details of the design I was looking to put/pull together the following:
1) 3D acoustic radiation code (e.g. BEM or possibly FEM) to simulate the acoustic near field around a speaker in detail.
2) Nonlinear lumped model to simulate nonlinear distortion in order to help size suitable drivers for the project.
3) FEM/BEM model of sound radiation from the cabinet.
I am just about in a position to run off simulations for 1) although I am considering a switch to better support the longer term or, possibly, just fixing the acousto source code to better support our usage. I am in a position to run off 3) with open source code having done so before but was less than satisfied with the damping models available. Want to improve the damping models but it is not a requirement. Is anyone aware of open source code for 2)? If not I would estimate it taking about a weeks work to put together (a week of 8 hour days not working a few hours in free time).
Is anyone interested in actively participating in the methods development? If not, it would be easier and quicker to develop on my computer rather than setting up, maintaining and supporting an open repository with all the effort that entails. Will set aside until it is needed by group activity.
Once developed for linux is anyone interested in adapting the software to run on the Windows and Apple platforms?
Is anyone interested in actively participating in a group project to develop an online speaker today? If not, I suggest putting it aside and raising the topic again in a few months after the simulation side is better sorted.
In order to handle the details of the design I was looking to put/pull together the following:
1) 3D acoustic radiation code (e.g. BEM or possibly FEM) to simulate the acoustic near field around a speaker in detail.
2) Nonlinear lumped model to simulate nonlinear distortion in order to help size suitable drivers for the project.
3) FEM/BEM model of sound radiation from the cabinet.
I am just about in a position to run off simulations for 1) although I am considering a switch to better support the longer term or, possibly, just fixing the acousto source code to better support our usage. I am in a position to run off 3) with open source code having done so before but was less than satisfied with the damping models available. Want to improve the damping models but it is not a requirement. Is anyone aware of open source code for 2)? If not I would estimate it taking about a weeks work to put together (a week of 8 hour days not working a few hours in free time).
Is anyone interested in actively participating in the methods development? If not, it would be easier and quicker to develop on my computer rather than setting up, maintaining and supporting an open repository with all the effort that entails. Will set aside until it is needed by group activity.
Once developed for linux is anyone interested in adapting the software to run on the Windows and Apple platforms?
Hi Andy. I would enjoy participating in a group project, but I feel I have nothing useful to offer at this point. I have no experience with boundary element methods or acoustic simulation software. So I am relegated to the role of observer, rather than participant.
Do you think the above would be useful? I have my doubts. When I compare the harmonic distortion and IM distortion performance of a variety of, say, 8" woofers... they are all different, very different in fact. Which means any model which aims to predict distortion based on driver size is likely to be wrong. Perhaps I am not getting your point... ??
j.
Do you think the above would be useful? I have my doubts. When I compare the harmonic distortion and IM distortion performance of a variety of, say, 8" woofers... they are all different, very different in fact. Which means any model which aims to predict distortion based on driver size is likely to be wrong. Perhaps I am not getting your point... ??
j.
The purpose is again to provide a quantitative method for assessing high fidelity for the 3 likely configurations. Nonlinear distortion is important in setting the limits for a driver in the sense that when it becomes noticeable the driver is being used outside it's high fidelity operating range. This is not straightforward to determine reliably from typical plots of distortion vs frequency for swept sine waves, multi-tones and the like. Distortion vs SPL plots are a bit more useful but rarely presented but even then the dynamic vs average influence of music and, to some extent, the driver parameters and crossover frequencies/slopes is missing. A nonlinear lumped model would enable the inclusion of this missing physics to be assessed both quantitatively and by auralisation.
The recent appearance of Klippel-type data online should provide much of the empirical information needed for such models. I suspect there may be one or two issues on the thermal side but assembling a model and checking it against existing published data looks doable (but speaking as someone that hasn't actually done it!). It will almost certainly be open to further development but should allow us to size drivers for the 3 configurations in a reasonably justifiable quantitative manner.
Like the 3D BEM and FEM simulations this would be a method to address details that DIY folk currently don't consider in a quantitative manner in their designs unlike engineers in the larger speaker companies. It is complementary to the fast efficient linear design methods currently used by DIY speaker folk and likely to be of interest to only a small proportion prepared to put in significantly more time and effort into design before building their speakers. Making it a good fit for a group design project (when the software is usable!).
So no it won't be useful for many but it should be for a few. It is relevant to this project because the cardioid-type radiation pattern approach is likely to be noticeably poorer in this respect compared to the other two.
Last edited:
Just my opinion: These endless simulations without practical problem are boring. Could be more interesting to make some prototype and continue simulations if there is problems to solve or want to develop it better.
Last edited:
Any progress on this project? Hopefully there has been some activity behind the scenes. It is a very interesting project, so I hope you guys plan to move forward...
j.
j.
Found this while searching for something else. Might be some useful information for this thread.
http://www.diva-portal.org/smash/get/diva2:1016965/FULLTEXT02.pdf
http://www.diva-portal.org/smash/get/diva2:1016965/FULLTEXT02.pdf
Real life has intruded on hobby time over the last couple of months but a bit of progress has been made with the 3D BEM and FEM code although not the nonlinear lumped code. The background task has got a bit bigger though because I have opted to drop open source code in favour of my own code. I don't intend to actively work on the design side of on wall speakers until all simulation stages are working reasonably well although I may post one or two articles earlier.
Thanks for the link which looks interesting although I have yet to read it.
This is also a discussion that may be useful, or at least informative...
https://www.audiosciencereview.com/...gn-via-shape-and-topology-optimization.23453/
https://www.audiosciencereview.com/...gn-via-shape-and-topology-optimization.23453/
Hi guys, did you finish this project? working on onwall speakers right now (similar like in the description in the first post of this topic - home theatre LCR speakers, closed box with 2x SB acoustic SB23NRXS, Lavoce MAN061.80 as mid and Dayton RST28F in WG300 waveguide) and it will be nice to have a chance to gain from your experiences.
Boxes are integrated in the cabinet below tv - see photo below - making everything even little bit more challanging (diffractions - but the frame in front of the speakers solve most of it and tweeter response is very flat). The biggest issues are caused by SBIR - front of the speaker is about 38 cm from the back wall. Second problematic reflection is from the floor (will be solved by absortive coffe table).
I can share *.mdat horizontal measurements and geometry details - are you open to help me with that? I am interested in sim vs real word measurements comparison and latter on I will be interested in tips how to compensate at in the passive crossover design.
Boxes are integrated in the cabinet below tv - see photo below - making everything even little bit more challanging (diffractions - but the frame in front of the speakers solve most of it and tweeter response is very flat). The biggest issues are caused by SBIR - front of the speaker is about 38 cm from the back wall. Second problematic reflection is from the floor (will be solved by absortive coffe table).
I can share *.mdat horizontal measurements and geometry details - are you open to help me with that? I am interested in sim vs real word measurements comparison and latter on I will be interested in tips how to compensate at in the passive crossover design.
Attachments
A possibly timely prompt. Some progress in the background but, as ever, things have changed. What may be relevant to this thread is that I recently spent some time getting acousto to compile and run with current software rather than ones from decades ago in order to comapre with the results from some of my own software. I have also spent some time recently working out how my software might be runnable on Windows. Prior to this prompt I had intended to hack acousto until it worked after a fashion for my purposes which I have largely done.
Anyway, acousto seems to have been abandoned by the original authors and it wouldn't take much to fork it, rework the api to better serve our purposes, fix the parallel processing (it uses deprecated and currently removed features) and get it running on Windows as well as the unix variants it was originally written for. The 3 approaches to on wall designs mentioned above could serve as the initial test cases.
If people other than me are interested please post to the the thread and I will check if what I think needs doing is as straightforward as it seems and is aligned with what others with an interest think needs doing.
Hi,
thanks for reply. For me, practical implementation is what really matters. My observation is, that it is beneficial to have 2 woofers and maybe 4 woofers will be beneficial - aim is to control directivity in problematic region. Also, crossover works better with higher order filters. From directivity point of view, mids region is the problematic one - directivity is too wide. My idea was to compensate it it by pushing that region little bit down on axis. Will see, how well it works - parts for CO are already ordered.
Check simulations below - there is horizontal simulation (like 1,6m from speaker) and on axis at MLP (like 4m). Both measurements were done in the final position (onwall). I did not measure verticals, but my idea is, that in vertical plane, woofers will play with very wide directivity and compensate the drop of the power response in the 200-450Hz region.
My design use little bit different onwall speaker approach - usually it is beneficial to make as shallow bafle as posiible and problems in 600Hz+ region solve by absorption. This was not possible in my scenario - speakers are hidden in the bookshalve with 400mm depth. So I am trying to adress SBIR problems with multiple speakers playing all (woofers + mid) in the problematic region (200-400Hz) and counting with the floor bounce (speakers are low)
thanks for reply. For me, practical implementation is what really matters. My observation is, that it is beneficial to have 2 woofers and maybe 4 woofers will be beneficial - aim is to control directivity in problematic region. Also, crossover works better with higher order filters. From directivity point of view, mids region is the problematic one - directivity is too wide. My idea was to compensate it it by pushing that region little bit down on axis. Will see, how well it works - parts for CO are already ordered.
Check simulations below - there is horizontal simulation (like 1,6m from speaker) and on axis at MLP (like 4m). Both measurements were done in the final position (onwall). I did not measure verticals, but my idea is, that in vertical plane, woofers will play with very wide directivity and compensate the drop of the power response in the 200-450Hz region.
My design use little bit different onwall speaker approach - usually it is beneficial to make as shallow bafle as posiible and problems in 600Hz+ region solve by absorption. This was not possible in my scenario - speakers are hidden in the bookshalve with 400mm depth. So I am trying to adress SBIR problems with multiple speakers playing all (woofers + mid) in the problematic region (200-400Hz) and counting with the floor bounce (speakers are low)
Attachments
As Televisions get larger and cheaper, the HT market segment is likely to gain a lot more ground with IMO on wall speaker solutions becoming more and more prevalent in the mid to high end users. I also think these larger sets pose a problem for the dedicated center channel where placement is restricted to very low in the sound stage……..a pair of on walls that can create a viable phantom center image will emerge as the mainstay in years to come.
With average dwelling sizes already smaller than the US, we’re catching up over here with a cultural markets
shift to rental/apartments of smaller size and multi integrated living space floor plans. The commercial HiFi market segment is drying up fast as large footprint tower speakers are looking more and more like Members Only jackets and track suits.
With these large TVs on wall, there’s a typical mounting depth to profile of 4-5” so at least on one side of each speaker, designers have the option of leveraging the smooth surface screen as an extended baffle……what to do on the outer edge presents problems with as many solutions as one’s creativity will allow. Best not to exclude the emergence of Spatial Audio which is becoming a real thing……with ambient listening dominating most listeners time, systems that create a pleasing and yet convincing background for other activities will also garner a greater market segment.
So keep on brain storming guys……affordable on wall low profile solutions are on the horizon if you can design one that’s convincing and still images well.
With average dwelling sizes already smaller than the US, we’re catching up over here with a cultural markets
shift to rental/apartments of smaller size and multi integrated living space floor plans. The commercial HiFi market segment is drying up fast as large footprint tower speakers are looking more and more like Members Only jackets and track suits.
With these large TVs on wall, there’s a typical mounting depth to profile of 4-5” so at least on one side of each speaker, designers have the option of leveraging the smooth surface screen as an extended baffle……what to do on the outer edge presents problems with as many solutions as one’s creativity will allow. Best not to exclude the emergence of Spatial Audio which is becoming a real thing……with ambient listening dominating most listeners time, systems that create a pleasing and yet convincing background for other activities will also garner a greater market segment.
So keep on brain storming guys……affordable on wall low profile solutions are on the horizon if you can design one that’s convincing and still images well.
Yes, I am also surprised there is so little good quality diy onwall speakers projects. There are some profesional solutions out there (=very expensive) and majority of people use normal speakers and simply put them onwall - not very good choice
Ad WG-300 - yes, I know, but did not know that when I started this project. WG300 has its weakpoints and some tweaks and tunning was needed, but unfortunatelly there are almost no good waveguides on the market and 3D printing has its own problems...
So crossover is done and I have some first impressions.
Any ideas what to focus on? Distortion profile is clear, so is burst decay. Shoud I tilt down fr. response more?
- fantastic concert like dynamics
- sound is bright (metal sounds, female voices...), too forward (sounds clean, but like this kind of sounds is louder, then others)
Any ideas what to focus on? Distortion profile is clear, so is burst decay. Shoud I tilt down fr. response more?
Attachments
Our perception of high frequency level is significantly influenced by beam width and the level of high frequency absorption in the room. Some rooms have soft furnishing, thick curtains, thick carpets,... leading to large amounts of high frequency absorption and speakers with a wider beamwidth tending to sound better balanced. The same speakers in a room with wooden floors, leather furniture, blinds instead of curtains,... is likely to be perceived as having high frequencies at too high a level and a mild downward slope of the on-axis response preferably. Alternatively a speaker with a strongly narrowing high frequency will sound better balanced in a hard room and a bit dull in a soft room.