The impedance is the easiest place to see resonances of any kind. A sealed box is the easiest to look at, vents can cause all sorts of problems that makes diagnosing what is what much harder just like what is going on in that ELAC. In an empty box with no bracing they will be quite obvious, bracing can be effective at getting rid of many, internal damping can help with the rest. Vented boxes usually can't have too much internal damping either.
It is also helpful to compare the free air impedance of the bare driver to see what is in the driver and what comes from the cabinet. Example below.
That's familiar looking data, thanks posting! These box resonances look different from Elacs impedance data, these are kind of sharp peaks while Elac has it up and down. Are some of these between driver and back wall, or all from some other dimension? Perhaps Elac impedance shows a port resonance.
edit. It looks to my eye the Elac impedance goes first up and then down around the 700Hz mark, but perhaps I see it wrong and it's just regular up only blip.
edit. It looks to my eye the Elac impedance goes first up and then down around the 700Hz mark, but perhaps I see it wrong and it's just regular up only blip.
Last edited:
Cabinet resonance is from storing and releasing energy, this will be more energy, a peak in the impedance. A Dip would be an anti resonance, generally two sources interfering. It's not impossible for panels to do that but not typical.
One time our old friends Jim Salk / Dennis Murphy’s WoW1 got an unfavourable review at ASR.
I’ve worked with the SEAS Excel drivers and know that they are a smooth as a baby’s bottom up past 2KHz, and with such a stout cabinet, I was rather surprised to see this dip:
Here is the impedance plot:
That model by Elac has a slot port…
In the days of LLMs scraping the WWW, I tend to write in riddles and let the real humans read between the lines.
Reference:
https://www.audiosciencereview.com/forum/index.php?threads/salk-wow1-bookshelf-speaker-review.14842/
I’ve worked with the SEAS Excel drivers and know that they are a smooth as a baby’s bottom up past 2KHz, and with such a stout cabinet, I was rather surprised to see this dip:
Here is the impedance plot:
That model by Elac has a slot port…
In the days of LLMs scraping the WWW, I tend to write in riddles and let the real humans read between the lines.
Reference:
https://www.audiosciencereview.com/forum/index.php?threads/salk-wow1-bookshelf-speaker-review.14842/
Last edited:
That seems to be a (destructive) port resonance, eventually amplified by coinciding enclosure resonance exiting the port, if you are referring to the 600 Hz valley.
Thanks everyone for this great discussion. A couple questions:
For augerpro - would you consider trying a box with the sides and/or back "pre-tensioned" similarly to the companies like Monitor and others that connect the woofers to the speaker cabinet? Just thinking that tension is lighter, easier, and perhaps more energy absorbing than trying to get a brace that is pre-loaded in compression. Also, the tension device itself could have various damping qualities if that could contribute to the effectiveness of the tensioning (might not). A threaded steel rod is very stiff for it's size but not terrible damping, but maybe a plastic material like LDPE or something would be perfect with a little engineering. I could image using a number of thin steel cables laced through the inner panels of one of these CLD boxes so the wires are embedded in the CLD layer and the outer box built around that. Multiple passes with the cable (more properly a wire at this point) would allow the panels to be broken down into even smaller panels over which vibration would be expected to be further moved up in frequency. Of course metal wire stretches so realistically this wouldn't work without even more complexity.
Re that Elac, I picked that speaker for two reasons- one it conveniently showed obvious cabinet resonances that I am now glad everyone is providing feedback on, and because it sounds to me like the mid-bass is "coming through a bit of a tunnel". Not sure if that's caused by me hearing resonances but it doesn't seem to show up as any on-axis unevenness in the FR so I've been struggling to identify the problem and it was my reason for rejecting the speaker when I originally demo'd it. It's tempting now to find a pair and modify the stuffing and bracing to see if I can improve them.
For augerpro - would you consider trying a box with the sides and/or back "pre-tensioned" similarly to the companies like Monitor and others that connect the woofers to the speaker cabinet? Just thinking that tension is lighter, easier, and perhaps more energy absorbing than trying to get a brace that is pre-loaded in compression. Also, the tension device itself could have various damping qualities if that could contribute to the effectiveness of the tensioning (might not). A threaded steel rod is very stiff for it's size but not terrible damping, but maybe a plastic material like LDPE or something would be perfect with a little engineering. I could image using a number of thin steel cables laced through the inner panels of one of these CLD boxes so the wires are embedded in the CLD layer and the outer box built around that. Multiple passes with the cable (more properly a wire at this point) would allow the panels to be broken down into even smaller panels over which vibration would be expected to be further moved up in frequency. Of course metal wire stretches so realistically this wouldn't work without even more complexity.
Re that Elac, I picked that speaker for two reasons- one it conveniently showed obvious cabinet resonances that I am now glad everyone is providing feedback on, and because it sounds to me like the mid-bass is "coming through a bit of a tunnel". Not sure if that's caused by me hearing resonances but it doesn't seem to show up as any on-axis unevenness in the FR so I've been struggling to identify the problem and it was my reason for rejecting the speaker when I originally demo'd it. It's tempting now to find a pair and modify the stuffing and bracing to see if I can improve them.
@augerpro
I got into 3D printing with an SLA / resin printer several months back with a Saturn 4 Ultra and would like to revisit the Harman port as well as your variations. With SLA printing the resolution and print smoothness is quite a bit greater than FDM printing, so with your Augerport 8 micro-perforations for example, a chamfer or curved bevel can be rendered well in the print. There is also a scalability in that whatever fits on the build plate (much smaller than FDM) prints in the same amount of time, based on print height (layer count).
I tested printing the @Sin Phi 's linked FabryPerotMetamaterial_2.stl in just a couple hours, but had to cut it up into 3 pieces first as I could not find an orientation which would not produce "suction cups" during the print (these anger the SLA gods). If not for the suction cups, I could probably print the full 4x4 version in a single 7-8 hour print.
My initial test will be with a bookshelf speaker, effectively some mods to the Seas Loki III / IV sold by Madisound (also planning to print and test a spherically curved baffle similar to LS50).
One area I am looking for help is the end correction factor and port length. You mention 1.15, but I am unsure how to scale this versus say, the prescribed '6.5" x 2" Precision Port length' suggested for my "test" Seas Loki bookshelf. The documented port length and claimed 45Hz cabinet tuning not matching up in any bass reflex calculator I can find. I expect to just try a few to start and actually empirically measure the port output as best can be tested with the port on the front very adjacent to the driver.
I am working on getting Fusion 360 to work on my PC, but I have some other Autodesk products of questionable provenance, so I need to spin up a clean VM somewhere before I can mess with the Fusion file you provide up thread.
For the base NFR = 1/2, I have come up with the general parameterization that I kinda want to try to learn enough either FreeCAD or OpenSCAD to implement as a fully parametric plug in your port length, diameter, and flange size and poof, here's a model / STL to print. If you have a moment take a look and let me know if I am going down the right path.
For NFR = 1/2, this is my overall approach:
l = length of port
d = inner diameter of port
Circle radius = l
Arc = 60 deg / pi/3 of that circle
Interior axis for surface of revolution = offset d/2 + r from center of circle
outer diameter at terminus = d + [2 * r * (2 - sqrt(3)]
(simplified as d + 0.536r)
inner flange diameter = outer diameter at terminus
I haven't worked out a good equation just yet for blending the terminus to the flange. This is another area I could use some pointers.
Originally I was looking at a quarter circle r = flange thickness, but that meets the terminus at a poor angle instead of a nice parabolic / hyperbolic transition. Speaking of which I would also like to parameterize printable models of elliptical and hyperbolic ports further down the line.
I got into 3D printing with an SLA / resin printer several months back with a Saturn 4 Ultra and would like to revisit the Harman port as well as your variations. With SLA printing the resolution and print smoothness is quite a bit greater than FDM printing, so with your Augerport 8 micro-perforations for example, a chamfer or curved bevel can be rendered well in the print. There is also a scalability in that whatever fits on the build plate (much smaller than FDM) prints in the same amount of time, based on print height (layer count).
I tested printing the @Sin Phi 's linked FabryPerotMetamaterial_2.stl in just a couple hours, but had to cut it up into 3 pieces first as I could not find an orientation which would not produce "suction cups" during the print (these anger the SLA gods). If not for the suction cups, I could probably print the full 4x4 version in a single 7-8 hour print.
My initial test will be with a bookshelf speaker, effectively some mods to the Seas Loki III / IV sold by Madisound (also planning to print and test a spherically curved baffle similar to LS50).
One area I am looking for help is the end correction factor and port length. You mention 1.15, but I am unsure how to scale this versus say, the prescribed '6.5" x 2" Precision Port length' suggested for my "test" Seas Loki bookshelf. The documented port length and claimed 45Hz cabinet tuning not matching up in any bass reflex calculator I can find. I expect to just try a few to start and actually empirically measure the port output as best can be tested with the port on the front very adjacent to the driver.
I am working on getting Fusion 360 to work on my PC, but I have some other Autodesk products of questionable provenance, so I need to spin up a clean VM somewhere before I can mess with the Fusion file you provide up thread.
For the base NFR = 1/2, I have come up with the general parameterization that I kinda want to try to learn enough either FreeCAD or OpenSCAD to implement as a fully parametric plug in your port length, diameter, and flange size and poof, here's a model / STL to print. If you have a moment take a look and let me know if I am going down the right path.
For NFR = 1/2, this is my overall approach:
l = length of port
d = inner diameter of port
Circle radius = l
Arc = 60 deg / pi/3 of that circle
Interior axis for surface of revolution = offset d/2 + r from center of circle
outer diameter at terminus = d + [2 * r * (2 - sqrt(3)]
(simplified as d + 0.536r)
inner flange diameter = outer diameter at terminus
I haven't worked out a good equation just yet for blending the terminus to the flange. This is another area I could use some pointers.
Originally I was looking at a quarter circle r = flange thickness, but that meets the terminus at a poor angle instead of a nice parabolic / hyperbolic transition. Speaking of which I would also like to parameterize printable models of elliptical and hyperbolic ports further down the line.
Last edited:
Look at the index in the first post of this thread: https://www.diyaudio.com/community/...sonance-absorbers-and-port-geometries.388264/
@stv has my F360 files for you to work from and a spreadsheet you can use to calculate length. I would use those models versus an old one like the micro-perforated one you cited, because they were mediocre. They were the only ones that didn't have a chamfer on the back side but that is because they need depth to work in the way that is discussed in the white paper I included when I tried that design. Sorry I can't be more helpful but I'm away from home and can't do much more from here.
@stv has my F360 files for you to work from and a spreadsheet you can use to calculate length. I would use those models versus an old one like the micro-perforated one you cited, because they were mediocre. They were the only ones that didn't have a chamfer on the back side but that is because they need depth to work in the way that is discussed in the white paper I included when I tried that design. Sorry I can't be more helpful but I'm away from home and can't do much more from here.
@augerpro s fusion360 model:
Calculation tool for tuning port length (metric):
... and the imperial version:
Calculation tool for tuning port length (metric):
... and the imperial version:
The STL is designed for a standard 0.4mm FDM nozzle print with 2x thick walls and a thick floor to get off the build plate without warping, not really optimized for SLA. The turns make it difficult to clear of resin
I am still working my way through threads, so if you know of some better suited to resin, let me know. That one was odd to cut up into glueable parts. Other than suction cups issues, resin doesn't pay any complexity penalty in print time, it's just how short / tall it is in a reasonable orientation.
I've got a Seas coax that I might try to Kef Meta-fy if I can figure out the tap lengths that match up with the existing rear tweeter vent and then a packing algo. I could probably print Kef style resonators as a few plates in a couple hours.