ROAR18

View attachment 624693

The blue line is with the microphone in the mouth of the ROAR12.
The red line is with the microphone at the exit of the tapped pipe section at the bottom of the front-resonator.

If the measurement was taken with the microphone in the mouth of the horn, that could tend to exaggerate its LF performance. I'm guessing that the purple line is a 1M measurement of the same horn? If so, the faster drop-off in FR is noticeable when you compare both lines on the graph. For my TH measurements, I've found that backing the mike off a bit so that it's about the same distance from the mouth as the effective radius of the mouth gives results that are a little more realistic without letting too much of the room get into the measurement. Some experimentation is required to get the best results though.

Measuring FR at various points *inside* the horn can be a illuminating exercise. I did the same thing with my THAM-like build, and the results eventually led to the "dogfood duct" "fix" I came up with to deal with the midbass notch in the THAM's response.

The ROAR12's "notch" is a pretty sharp one, but it's up @200 Hz, so it's a likely a bit less problematic than other alignments that have such a notch appearing considerably lower.

Interesting point on the symmetrical loading of the driver, BTW. I find it curious though that apparently none of the techniques used in car audio drivers to strengthen the driver's cone all the way up to its edge (where asymmetrical loading IMO will most likely lead to failure) haven't found their way to the pro audio subwoofer world. These include flattening the edge (Image Dynamics) or folding it back (JBL, Infinity), or even changing the entire geometry of the cone (Kicker Comp series). Maybe the problem isn't that significant?

What's the lowest resonant frequency of the ROAR12? What about its impedance curve? Comparing those two against the HornResp sim for it should give a pretty good idea of how close the sim matches the built system.
 
If the measurement was taken with the microphone in the mouth of the horn, that could tend to exaggerate its LF performance. I'm guessing that the purple line is a 1M measurement of the same horn?

Yes, the purple line is at one meter from the mouth.
The driver is brand new and raw and I expect more LF performance once the driver has a few hours of a few hundred watts of workout. My neighbors protest whenever I fire up my old trusty QSC-amp for some nice punchy bass, so it might take a while before I can do proper measurements.
 
Hi, martinsson, am i right suggesting this kind of cutting for 1,5x1,5 meter sheets?

With my 18" drivers this bin could provide very good power conversion efficiency.

Frequency response can be easily corrected with Driverack PA. First parametric filter tuned up to -3,8 dB at 93 Hz, so there are a plenty of headroom for bass tweak.
 

Attachments

  • 1.png
    1.png
    47.6 KB · Views: 845
  • 2.png
    2.png
    45.5 KB · Views: 850
  • 3.png
    3.png
    56.1 KB · Views: 826
  • 4.png
    4.png
    18.7 KB · Views: 772
  • 5.png
    5.png
    25.4 KB · Views: 767
  • 6.png
    6.png
    20.5 KB · Views: 279
  • 7.png
    7.png
    18.7 KB · Views: 178
Last edited:
Hello BesPav

Thank you for taking an interest in the design, remember that the design is not yet built and so I cannot verify it's performance, but in theory it seems very promising, and the ROAR12 certainly made an impression if one can extrapolate on that (subjective impression only).

It's a bit difficult to make out the numbers in your cut sheet images, and I'll have to double check with the drawing detail dimensions, as far as I can see though the layouts seems good with as little waste a possible.

What kind of drivers do you have? from the simulations they seem very capable, I'm guessing 18SW115 or similar?
 
remember that the design is not yet built

Let me try.
:xfingers:

I cannot verify it's performance

Doesn't worry, we can compare design ideas and performance graphs, but the most critical corner - the sounding can't be picked up from any kind of graphs or measurements.
So we have no other way than build and listen.

It's a bit difficult to make out the numbers in your cut sheet images

Yup... That's very simple cutting software, so let's check with school arithmetics.
960 mm (height of outer panels) + 564 mm (internal panels width) = 1524 mm

We just need cutting table with 1-1,5 mm thick saw, one cut and we're in.
;)
Assume ply sheets are really square and precisely cutted at the factory.

What kind of drivers do you have? from the simulations they seem very capable, I'm guessing 18SW115 or similar?

This is RCF LF18X401, you can pick datasheet from official site here:
RCF - LF18X401

You are right, it's a hair lower or mostly similar to 18SW115 in performance terms, in the ocean of fake 18LW1400 they both are crusaders.
Last winter one of our local suppliers announced stock clearing and offer those drivers for my company for around $200. We could not resist and pick up large batch.
 
Let me try.
If you decide to go ahead - thank you BesPav, it means a lot to us, it is not common that someone builds an unverified design, especially not one requiring this much effort and such an expensive type of driver.

Regarding the RCF LF18X401 in a rear mounting application, given the shape of the interfacing basket edge I have concerns that they might not seal properly towards the panel, which is very important that it does, do you share these concerns?

@ Mrcy - Apologies to the thread owner, thank you for your patience and please let us know if you feel that we should move this discussion to another thread.
 
If you decide to go ahead - thank you BesPav, it means a lot to us, it is not common that someone builds an unverified design, especially not one requiring this much effort and such an expensive type of driver.

He-he.
You know first hand, mr. martinsson, that all seems to be accurate to the contrary.
Members are often spoiled with approved designs, but where are a lots of backgroung estimating, calculating, simulation, modeling, labour samples and working prototypes, understanding in the end?
And even with huge support, mounting instructions and cutting sheets from time to time we have found gifted peoples doing things at unpredictedly wrong ways...
:eek:

Regarding the RCF LF18X401 in a rear mounting application, given the shape of the interfacing basket edge I have concerns that they might not seal properly towards the panel, which is very important that it does, do you share these concerns?

Doesn't worry, RCF's datasheet keeps undisclosed hard polyethilene foam sealing ring at the inside bracket of the driver basket.
Your design with this driver could resist and dissipate around of 3 kWt power without risk of touching magnet system pole. Driver are well vented and symmetrically well loaded through the passband so cone stays inside Xmech even at rated program power.
I want to ask you about another concern, rear cover stiffness. I would prefer to keep two hardening brackets at the whole 928 mm length/height with a lot of confirment screws.
 
Last edited:
I want to ask you about another concern, rear cover stiffness. I would prefer to keep two hardening brackets at the whole 928 mm length/height with a lot of confirment screws.
The rear panel is 924x564 if built with 18mm plywood, and the bracing of the rear panel, and its connection to the driver baffle is one of the most difficult ones to get right, it is also one of the most exposed to pressure so bracing it is very important to get it right.

Interesting idea regarding the bracing (your concern) and your idea is not bad at all, but it means taking away the brace going perpendicular across the driver cut out, this brace was added since we believed it would act as a form of primitive divider, marking the start of the symmetrizing tapping sections.

Also, bare in mind that it might give rise to a local asymmetrical loading of the cone if you replace the four bracing pieces with two longer ones extending all the way along the rear panel and past the driver, in the THAM18 we suggested making holes in the two parallel braces to relive the potential pressure difference that might act on the cone (see THAM18 example picture below).

THAM18_600x800x650_bracing_parts_and_positions_1_updated.jpg
 
Maybe put in a 45 degree triangle against the backwall to strengthen the back wall across driver cone. Does it need cone compensation?

I think I can see what you mean, to integrate a cone compensation into the bracing is not a bad idea, I looked into it briefly, my idea was to make a really rough compensator doubling as a connector well, a box fitted to the inside of the rear panel at the center of the cut out for the driver, with a cutout in the rear panel slightly smaller, in which one could mount the speakons and prevent damage to the same should the slide of something.

The triangular or pyramid shape is not a bad idea, but it might be tricky get stable, but it should be possible to utilize it in pretty much the same way should one wish to do so.

I have not yet decided if cone compensators are necessary, so we left it up to the DIY'er to implement if he/she should feel it necessary, same as with the THAM designs, it would bring the design closer to the simulations but what concerns me is that I feel there is no harm with more air cone side, granted we already have very low compression, but still, it is also kind of tricky to make one static designs for all driver cones.
 
I have not yet decided if cone compensators are necessary
Hi Andersson,

If you don't sim them, how can you 'decide' if they are necessary? ;-)

My two cents, use the cone volume and the volume of the driver baffle cut-out in the Vtc/Atc function of Hornresp during the design process or you will need cone correction afterwards.
Secondly, if you use a "stub" in a symmetric design you also create an S1-S2 segment. In the case of the Roar series that will probably show up as a small peak within the dip. Third, the length value of S3-S4 needs to be set to zero.

Regards,
Djim
 
Last edited:
May this ROAR arrangment work out well!

Gentlemen,

This ROAR design looks interesting, like a new way to fold a Series-Tuned 6th order cabinet , or much like a split-path Stepped Tapped Horn with just one big step, or a sort of Transflex with a single abrupt expansion, very smart:)

With the high aspect ratio of the symmetrical duct sections i suspect it might tune slightly lower than Hornresp predicts ....

The simulations show promise and I am looking forward to hearing/reading more feedback about performance from builders regarding the ROAR cabinets in the real world .....

Please post whatever results people report , good or bad ..

Great to see that a ROAR15 and ROAR18 are in the works ...

I found a few budget drivers which look reasonably decent in some ROAR Hornresp models on my end, and i thought i should let you all know about these .....

They are sold by Thomann , and here are the measured T/S parameters provided by our good friend Sebastian Schlager in Denmark ...... (both drivers currently seem to ship with the same motor) ..

An externally hosted image should be here but it was not working when we last tested it.




An externally hosted image should be here but it was not working when we last tested it.


I greatly appreciate the innovative spirit of those involved in the developing of this idea :) :up: I hope that it will meet or exceed all expectations!:yes: :xfingers:

Best of luck to all of those bold and brave explorers who to dare venture into unknown acoustic territory!
 
I found a few budget drivers which look reasonably decent in some ROAR Hornresp models on my end, and i thought i should let you all know about these .....

You can't accurately simulate these with hornresp, this point has been mentioned at least 3 times now.

I greatly appreciate the innovative spirit of those involved in the developing of this idea I hope that it will meet or exceed all expectations!

The innovative spirit involved in designing these was a misconception that the Danley BC sub was similar to this. It's not similar, the frequency response is not similar and it's not even a tapped horn, it's a front loaded horn.

As the woefully incomplete sim suggests, it will have bad frequency response, frequency response will get worse with power compression and there will be a big dip right at the edge of the passband. If that's what you mean by meeting expectations then it will, but none of these are good things.
 
ROAR chat

You can't accurately simulate these with hornresp, this point has been mentioned at least 3 times now.

There is always Akabak :) ... In fact JAG i think you were one of the folks who urged me to learn Akabak, and i am thankful for that :cheers:......

Someone should definitely try modeling the ROAR in Akabak to see how similar the results compare to both the real world measurements and also to see if it looks remotely similar to the Hornresp sims:magnify: ....

In Hornresp i tried both with the flare and without the flare (meaning just a very abrupt change in CSA), and the latter of course causes Hornesp to complain but still calculates and produces results which aren't much different than the flared results (i figured the sim for a 90cm cabinet depth).....



ANYWAY , not all cabinets can be fully modeled in Hornresp, but that doesn't mean they aren't worth building ... They could potentially be great cabinets (or not) but that isn't determined by the limitations of our software.. ...



The innovative spirit involved in designing these was a misconception that the Danley BC sub was similar to this. It's not similar, the frequency response is not similar and it's not even a tapped horn, it's a front loaded horn.

Ok , so someone may have missed the mark completely when trying to clone:Pirate: a Danley sub and they ended up developing some other design/alignment instead.:blush: Hehe ... A marvelous story if you ask me , especially if they came up with something useful :yes:...

As the woefully incomplete sim suggests, it will have bad frequency response, frequency response will get worse with power compression and there will be a big dip right at the edge of the passband. If that's what you mean by meeting expectations then it will, but none of these are good things.

This ROAR design isn't a wideband design to be sure but most users don't require wide bandwidth in a subwoofer............

The majority of folks will typically just run their subwoofers up to the common 80hz to 100hz range LPF or crossover point (which is well under the response hole located further up on the ROAR) ..

Power compression degrades performance with any cabinet, it is to be expected ..... Your example given in post #12 is a fantastic demonstration showing how power compression can happen gracefully in a well designed system … Impressive work on that JAG, good stuff :)...

Your given example is a very convincing argument for using drivers with stronger motors and cabinets that are large enough to show a somewhat underdamped saddle response in simulation (small signal levels model).... This however also suggests that everyone who has been shooting for flat response in standard Hornresp simulation has been doing it the wrong way this whole time ;) Including all of our DIY friends who have been designing and building Tapped Horns and enjoying their results....... So is it possible that your testing method could be on the extreme side? ........................... You may be right though, after all i suppose we should be designing the cabinets for the higher power levels which they will be run at instead of optimizing them for tiny signals....I cannot argue with that.....

Accommodating for the expected shift in parameters that occur when the voicecoil heats up ... Totally logical ...

However, most of the simulations posted on DIYaudio over the years (not just the ROAR) would fall apart when doubling the "Re"in Hornresp..... Are they all flawed? Why single out this one particular design?

It has been stated in this discussion that the response which these cabinets produce is more smooth than what is shown in the Hornresp simulation ... Johannes from post#14 believes that this is related to frictional losses ....

If we take a look at the graph from post#20 we can see that there is a purple 1m trace of the ROAR12 .... We would expect to recognize the 3 peaks which form the W shaped curve shown in the Hornresp model but for some reason those peaks are absent in the measured curve .... The real world response is indeed more smooth than what we see in Hornresp, just as the gentlemen in Sweden have observed ...................

In the simulation it is those dips in the W shaped curve that become exaggerated when the model's "Re" value is raised to emulate thermal power compression right?

Ok .... So here is the question we should be asking then:

If in the real world we are not starting off with any W shaped curve at all, will we still end up with the exaggerated W shape in the curve that Hornresp suggests when the driver is pushed hard enough to heat up the voicecoil??? I would assume not, considering the smooth curve that we are starting off with in the measurements, but who knows(?) ..... I imagine that the result would be less peaky and less exaggerated than what Hornresp is suggesting with power compression ...

In order to answer this question it would be great to see someone take measurements at very low power (1w-ish) and then overlay another measurement at very high power with the microphone at least 1 meter away for both measurements ....... This way we can see what sort of changes take place in the curve at the different power levels ..

Circlomanen, Martinsson, BesPav, do you guys think it might be possible to get someone to take such a measurement? It would be very helpful ....

 
Last edited:
The "power compression" that you're referring to is what's expected to happen when the system is continuously run at such a significant high power that the voice coil's resistance rises due to temperature change and affects the driver's operating characteristics. Note that this temperature change also depends on the design of the system as well as how much power you've run through it, so THs that are usually designed with the driver's magnet sitting right in the spot of greatest air movement probably suffer from it a lot less than say FLHs. For my POC3 TH, the driver's magnet only gets slightly warmer after running it at my usual playback levels, so I consider this form of power compression a non-issue. I'm about to replace the driver with a 3012LF which will allow me to run it at higher levels - I might do a few power compression tests then.

As for designing with an underdamped response in mind, my POC2 design had that in mind, my POC3 design with the same driver did not. POC2 also had a slightly deeper midbass response dip than POC3. POC3 sounds so much better than POC2 did that I will likely never deliberately design or build another system with an underdamped response in mind, and I will also avoid introducing a midbass dip as much as possible.
 


There is always Akabak :) ... In fact JAG i think you were one of the folks who urged me to learn Akabak, and i am thankful for that :cheers:......

Someone should definitely try modeling the ROAR in Akabak to see how similar the results compare to both the real world measurements and also to see if it looks remotely similar to the Hornresp sims:magnify: ....

In Hornresp i tried both with the flare and without the flare (meaning just a very abrupt change in CSA), and the latter of course causes Hornesp to complain but still calculates and produces results which aren't much different than the flared results (i figured the sim for a 90cm cabinet depth).....



ANYWAY , not all cabinets can be fully modeled in Hornresp, but that doesn't mean they aren't worth building ... They could potentially be great cabinets (or not) but that isn't determined by the limitations of our software.. ...

Yes, this needs to be simulated in Akabak. Whether you sim this with or without the flare in Hornresp the driver tap near the mouth won't be in the right spot. The difference may be big or it may be small but it should be simulated properly, especially if the design is promoted but hasn't been built and tested yet.

Ok , so someone may have missed the mark completely when trying to clone:Pirate: a Danley sub and they ended up developing some other design/alignment instead.:blush: Hehe ... A marvelous story if you ask me , especially if they came up with something useful :yes:...

Sure, if a mistake leads to something useful it's fortuitous. My issue with this is that the attempt to guess what was in the BC sub was so far off AND that the resulting product has serious frequency response problems. This was pointed out 2 years ago.

This ROAR design isn't a wideband design to be sure but most users don't require wide bandwidth in a subwoofer............

The majority of folks will typically just run their subwoofers up to the common 80hz to 100hz range LPF or crossover point (which is well under the response hole located further up on the ROAR) ..

The recommendation has ALWAYS been to have smooth response as far outside the passband as possible to ensure easy integration through the crossover region. With modern dsp you can fix a lot of things but you can't fix deep dips. The huge dip in this design is too close to the passband for comfort - unless using very steep filters it will be very difficult (if not impossible) to have smooth integration through the crossover region.

Power compression degrades performance with any cabinet, it is to be expected ..... Your example given in post #12 is a fantastic demonstration showing how power compression can happen gracefully in a well designed system … Impressive work on that JAG, good stuff :)...

Your given example is a very convincing argument for using drivers with stronger motors and cabinets that are large enough to show a somewhat underdamped saddle response in simulation (small signal levels model).... This however also suggests that everyone who has been shooting for flat response in standard Hornresp simulation has been doing it the wrong way this whole time ;) Including all of our DIY friends who have been designing and building Tapped Horns and enjoying their results....... So is it possible that your testing method could be on the extreme side? ........................... You may be right though, after all i suppose we should be designing the cabinets for the higher power levels which they will be run at instead of optimizing them for tiny signals....I cannot argue with that.....

Accommodating for the expected shift in parameters that occur when the voicecoil heats up ... Totally logical ...

However, most of the simulations posted on DIYaudio over the years (not just the ROAR) would fall apart when doubling the "Re"in Hornresp..... Are they all flawed? Why single out this one particular design?


There's a lot of reasons why I single out this particular design. The history of it's evolution (which I discussed previously) is one reason. The implication earlier in this thread that Hornresp is "too theoretical" is another, especially when no one even bothered to simulate it correctly in the first place. A third reason is that it was specifically stated that this is designed to be used and excel at high power levels. This last reason is mainly why I speak out about the power compression issue with this design but rarely with other designs - this is a rare claim.

All this was pointed out years ago, typically when people don't seem to be learning and keep repeating the same mistakes I get more vigilant in pointing out the mistakes.

It has been stated in this discussion that the response which these cabinets produce is more smooth than what is shown in the Hornresp simulation ... Johannes from post#14 believes that this is related to frictional losses ....

If we take a look at the graph from post#20 we can see that there is a purple 1m trace of the ROAR12 .... We would expect to recognize the 3 peaks which form the W shaped curve shown in the Hornresp model but for some reason those peaks are absent in the measured curve .... The real world response is indeed more smooth than what we see in Hornresp, just as the gentlemen in Sweden have observed ...................

In the simulation it is those dips in the W shaped curve that become exaggerated when the model's "Re" value is raised to emulate thermal power compression right?

Ok .... So here is the question we should be asking then:

If in the real world we are not starting off with any W shaped curve at all, will we still end up with the exaggerated W shape in the curve that Hornresp suggests when the driver is pushed hard enough to heat up the voicecoil??? I would assume not, considering the smooth curve that we are starting off with in the measurements, but who knows(?) ..... I imagine that the result would be less peaky and less exaggerated than what Hornresp is suggesting with power compression ...

In order to answer this question it would be great to see someone take measurements at very low power (1w-ish) and then overlay another measurement at very high power with the microphone at least 1 meter away for both measurements ....... This way we can see what sort of changes take place in the curve at the different power levels ..

Circlomanen, Martinsson, BesPav, do you guys think it might be possible to get someone to take such a measurement? It would be very helpful ....

This cab has NEVER been measured. The 12 inch version that was measured was not simulated properly AND not measured properly. I would consider a proper measurement to be an outside measurement at a distance of at least 2 meters. So any discussion of the frequency response or the measurements or the sims of either this cab or the 12 inch version are completely moot. They will continue to be so until the cab is properly simulated and measured.

I'm not sure that simply adding Re to the sim is a completely accurate way to estimate power compression - but it's the only tool we have available. When you add Re the same thing ALWAYS happens - the frequency response "melts" away leaving spikes where the impedance peaks are. If you add enough Re the frequency response will look a lot like the impedance curve.

It's not enough to have a low compression ratio and claim that a design is going to excel at high power levels. There's a lot more to it than that.

Brian is correct that due to the fanlike cooling of the motor in tapped horns that power compression won't be as much of an issue as with other designs. But Danley designs for an underdamped response for a reason - it WILL stand up to abuse much better than an overdamped design, and he is in the business of making products that are routinely abused. If you don't intend to abuse the design you don't need it to be all that underdamped at all. BUT in this case the designers are making the claim specifically that their cab is meant to be abused at high power levels and claiming it will excel. It won't.
 
I'm not sure that simply adding Re to the sim is a completely accurate way to estimate power compression - but it's the only tool we have available.

It's not completely accurate (I suspect compliance is also affected as the soft parts like the spider heat up), but it does get the sim a bit closer to what response would be like when power compression is happening, so it's better than NOT doing it. The trick here is determining how much additional Re needs to be added to sim what you're tryin to sim, which would mean knowing what the voice coil temperature would be like under the "high power" conditions you expect the system to see.

BTW, there's also another stated reason why some builders look at underdamped designs in HornResp - it's an attempt to compensate for HornResp not taking box losses into effect in its simulations.
 
the length value of S3-S4 needs to be set to zero

I see L34 as a "rubbery/flexible" section. The sim does not change much with a shorter or longer L34. I use the distance from the end of the tapped pipe section to the center of the cone/voicecoil as a guesstimate length. It is not an issue in real life. You don't create a great design by creating a perfect match between a theoretical and greatly simplified mathematical simulation and a measured response.
I judge my designs based on real life performance.

In Hornresp i tried both with the flare and without the flare (meaning just a very abrupt change in CSA), and the latter of course causes Hornesp to complain but still calculates and produces results which aren't much different than the flared results

Thanks for your interest in our ROAR-design.
I have come to the same conclusion based on Horn-resp sims and real life tests.
The start/bottom of the front quarter-wave resonator is a very interesting area. There seems to be a lof of different things going on there that Hornresp does not simulate properly. The L34 is just another crude simplified mathematical approximation of a much more complex real life series of interdependent events.

@Matthew Morgan J:

The ROAR is a series-tuned 6th order bandpass design. It exchanges Helmholtz-resonators for quarter-wave resonators to increase radiating surface and the bandwidth of the resonators. I have been following your thread "New sub design? Constricted Transflex, simple build (series tuned 6th order)" for some time, and I must thank you for a lot if interesting ideas and simulations.

Cheers,
Johannes
 
and I will also avoid introducing a midbass dip as much as possible

The front-resonator is sized and tuned to give a small peak in the midbass response (70 - 100 Hz). I really don't like the dip many tapped horns have in this region. This is partly why we (me and Anders Martinsson) tuned the THAM series of tapped horns slightly higher then normal.

This ROAR design isn't a wideband design to be sure but most users don't require wide bandwidth in a subwoofer............

The majority of folks will typically just run their subwoofers up to the common 80hz to 100hz range LPF or crossover point (which is well under the response hole located further up on the ROAR)

When used outdoors the ROAR12 sound very nice crossed over 24 dB/octave at 120 Hz. It is a quite typical user scenario for tapped horns.

Cheers,
Johannes
 
I see L34 as a "rubbery/flexible" section. The sim does not change much with a shorter or longer L34. I use the distance from the end of the tapped pipe section to the center of the cone/voicecoil as a guesstimate length. It is not an issue in real life. You don't create a great design by creating a perfect match between a theoretical and greatly simplified mathematical simulation and a measured response.
I judge my designs based on real life performance.

Here we go again. If you simulate and then accurately build what you simulated you will find that Hornresp is NOT greatly simplified. The theory is there because it works and it works well. If you accurately build what you simulated the correlation between sim and measurement will be very close. The simulation doesn't include internal losses but those don't amount to much and won't change things much.

Again, I find it ludicrous that you claim Hornresp isn't good enough when you didn't ever bother to sim the enclosure properly.

Thanks for your interest in our ROAR-design.
I have come to the same conclusion based on Horn-resp sims and real life tests.
The start/bottom of the front quarter-wave resonator is a very interesting area. There seems to be a lof of different things going on there that Hornresp does not simulate properly. The L34 is just another crude simplified mathematical approximation of a much more complex real life series of interdependent events.

There are not a lot of things going on ANYWHERE that Hornresp does not simulate properly. If you bothered to sim your design correctly you might find out how accurate it actually is.

There's no "complex real life interdependent events". We're dealing with air flow and resonances, all of this is very well understood and Hornresp is more than capable of simulating it all. The issue is that Hornresp doesn't have enough segments to sim your design so you need to step up to Akabak.

This is the same kind of bunk you were spouting in the BC sub thread with your claims that the path length was longer because there was a virtual segment at the mouth created by air vortexes. Absolutely ridiculous. There are real explanations for all the phenomenon that can happen, in that case it was cab frontal area creating a diffraction boost, in this case you simply don't understand that you need to simulate accurately.

@Matthew Morgan J:

The ROAR is a series-tuned 6th order bandpass design. It exchanges Helmholtz-resonators for quarter-wave resonators to increase radiating surface and the bandwidth of the resonators. I have been following your thread "New sub design? Constricted Transflex, simple build (series tuned 6th order)" for some time, and I must thank you for a lot if interesting ideas and simulations.

Cheers,
Johannes

The front-resonator is sized and tuned to give a small peak in the midbass response (70 - 100 Hz). I really don't like the dip many tapped horns have in this region. This is partly why we (me and Anders Martinsson) tuned the THAM series of tapped horns slightly higher then normal.



When used outdoors the ROAR12 sound very nice crossed over 24 dB/octave at 120 Hz. It is a quite typical user scenario for tapped horns.

Cheers,
Johannes

As i showed in the BC sub thread, you can easily make a normally shaped tapped horn with the same frequency response as your stepped tapped horn, but without the big problematic dip that your design has.

You simply don't have enough skill with Hornresp to create the frequency response you want with a normal enclosure so you resort to this stepped tapped horn alignment that has frequency response issues (the big dip).

There's nothing wrong with stepped horns, they have been done before and they work well. The problem is that you think that's the only way to get the frequency response you want, the frequency response you want has serious issues, and you've created a new name for your creation and claimed it was something totally new. None of this is new, your design has problems and clearly you haven't leaned much in the last couple of years since this was all pointed out to you.