BEST 18" Sub for a folded horn enclosure!

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
HA! Thats not me! i want something with headroom! Id rather have a set of subs at half volume than at full with the same SPL ! :)

This is it! I need to speak to someone that has first hand knowledge ! Its all well looking at graphs, but id like to know that what im building is going to impress!
Simple.
Build these.

http://www.diyaudio.com/forums/subwoofers/185588-keystone-sub-using-18-15-12-inch-speakers.html

"Half volume" is about a 10 dB difference in level. The Keystone is about 6 dB more efficient than a bass reflex nearly the same size.
It is an easy to build cabinet, and works well with a variety of speakers, though best with the B&C 18SW115-4 or B&C 18TBW100-4.

Having designed subs to beat any other designs I have heard for over 30 years, the Keystone, for it's size, does that.

Art
 
The Keystone is about 6 dB more efficient than a bass reflex nearly the same size.

I found this statement to be extremely compelling so I did a quick sim using tb46's Hornresp inputs for the Keystone in post 96 of the Keystone thread compared to a ported box. Both sims use the B&C 18tbw100, both shown at xmax. The ported box is half the size of the Keystone. The ported box is obviously the light grey line.

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


Clearly the graphs don't tell the whole story, the ported box will suffer more from both power and port compression than the tapped horn but the answer doesn't look quite so simple to me. I know you've mentioned that you get different measured results than simulations predict but this simulation shows that either you are exaggerating a bit or Hornresp has no correlation to reality at all.
 
Last edited:
Both sims use the B&C 18tbw100, both shown at xmax. The ported box is half the size of the Keystone. The ported box is obviously the light grey line.

I know you've mentioned that you get different measured results than simulations predict but this simulation shows that either you are exaggerating a bit or Hornresp has no correlation to reality at all.
As David McBean has mentioned, he is a programmer, not a cabinet builder.
He relies on builders to convey differences in the actual results to predicted.
Many builders have noted that Hornresp often exaggerates peaks and dips.
There is also the basic problem of Hornresp using a circular expansion, while very few bass horns are ever built in that fashion.

I find his program quite useful, though seldom find exact correlation to the actual builds.
The actual measurements and performance of a cabinet are more useful to me.

Of interest, the ported cabinet does suffer from "port compression" at high power levels, having less LF compared to upper output, while the tapped horn is the inverse, LF gains linearly, while upper output reduces.

The results below are the ported cabinet and the Keystone, both loaded with the BC18SW115-4, and a pair of Lab 12s, each measured in the same place outdoors, same amp, filters (25 &125Hz BW) same pink noise output.

I have no reason to lie about any test results, the only way to learn how (and how well) a design works is to measure, listen and compare. If the results are not good, change the box (or driver) until they are.
The Keystone design encompassed more than 100 separate cabinet changes and tests to determine the results of those changes.

Art
 

Attachments

  • Keystone, BR.png
    Keystone, BR.png
    142.7 KB · Views: 369
Hi just a guy,

I should not have posted any Hornresp simulations for the Keystone sub (even though I noted that they do not take the Keystone mouth into consideration) as these simulations only lead to confusion. Because of the Keystone exit (mouth) Hornresp will not arrive at a correct simulation. It really does take AkAbak to get close on this one. Djim tried different methods, e.g.: see Page 13: http://www.diyaudio.com/forums/subw...ng-18-15-12-inch-speakers-13.html#post2933411 . Even Tom Danley states that measurements beat simulations.

PASC build a Keystone sub and posted pictures and measurements in the Keystone thread. He seems to be very impressed with the results.

Regards,
 
I have no reason to lie about any test results...

I understand, and please don't think I'm claiming you are doing so.

As David McBean has mentioned, he is a programmer, not a cabinet builder.
He relies on builders to convey differences in the actual results to predicted.
Many builders have noted that Hornresp often exaggerates peaks and dips.

There are a couple of good reasons why peaks and dips (especially very narrow band, high amplitude peaks and dips) might not agree completely with the simulation. First, the physical wooden cab walls are not infinitely rigid so there are some internal losses that the simulation cannot accurately account for. This is mainly going to affect the really narrow peaks you often see in Hornresp tapped horn sims, if the peak is narrow enough it might not show up in the measurement at all. Second, when there are solid boundaries in the physical build you can create nulls, just like the nulls that are created by the solid walls reflections in a room. There is no way for Hornresp to account for this because Hornresp does not concern itself with the physical boundaries inside an enclosure, especially a folded horn.

There is also the basic problem of Hornresp using a circular expansion, while very few bass horns are ever built in that fashion.

From my own experience and what I've read this isn't really a concern unless your horn mouth gets big enough to cause problems with dispersion - and it needs to be pretty big to affect the frequencies a sub is playing.

I find his program quite useful, though seldom find exact correlation to the actual builds.
The actual measurements and performance of a cabinet are more useful to me.

You might (or might not) remember a member here named soho54. (He posted here a bit but was much more active at avsforum and hometheatreshack.) He did dozens of accurate simulations and over and over again, he showed that measurements actually do match simulations very very well. He used Akabak and broke the horn into 50+ sections and his simulations regularly matched measured results. Several times people complained that Hornresp was not accurate and he simulated their physical build and showed that they in fact did not build what they simulated and that's why the sims and measurements did not match. He showed that if they simulated what they built accurately the sims match the measurements very well.

This matches my experience, my sims overlay my measurements almost exactly. I've had problems a couple of times with nulls in the 80 - 90 hz region which is most likely due to having multiple segments of the physical build having the same length and being long enough to cause nulls based on their physical dimensions.

I agree with your assessment that response can change with extremely high power level but I'm not seeing the large discrepancies that you are talking about.

The results below are the ported cabinet and the Keystone, both loaded with the BC18SW115-4, and a pair of Lab 12s, each measured in the same place outdoors, same amp, filters (25 &125Hz BW) same pink noise output.

In the simulation I provided the ported box needs quite a bit more power to hit xmax than the tapped horn, so I'm not sure this is a fair comparison. A more fair comparison would be to apply enough voltage to each box to get it to the point of onset of clearly audible distortion. And to be fair, to accurately compare there should be more details about the ported box. All I know about it is the driver used. No info about tuning frequency, volume, port size, etc etc etc.

Please understand that I'm not trying to give you a hard time but I'm not seeing the gross discrepancies that you are describing. I can see power and port compression accounting for maybe 3 db difference but not 6. And to be honest, I'm much more interested in the flh vs tapped horn discrepancies you mentioned earlier but that discussion didn't go anywhere. You didn't provide a simulation for the front loaded design.
 
Last edited:
Hi just a guy,

I should not have posted any Hornresp simulations for the Keystone sub (even though I noted that they do not take the Keystone mouth into consideration) as these simulations only lead to confusion. Because of the Keystone exit (mouth) Hornresp will not arrive at a correct simulation. It really does take AkAbak to get close on this one. Djim tried different methods, e.g.: see Page 13: http://www.diyaudio.com/forums/subw...ng-18-15-12-inch-speakers-13.html#post2933411 . Even Tom Danley states that measurements beat simulations.

PASC build a Keystone sub and posted pictures and measurements in the Keystone thread. He seems to be very impressed with the results.

Regards,

Understood, but soho54 has shown over and over that accurate simulations match measurements very well, at least at small signal levels. Weltersys has claimed over and over that accurate simulations don't match measurements very well at all. I'm just trying to get to the bottom of this, I think it's in everyone's best interests to determine how much trust to place in our simulations.

I'm not trying to say the Keystone is a bad design, I just don't see it being 6db ahead of a similarly sized and tuned well designed ported box with the same driver when both are provided enough power to get them to the onset of audible distortion (not the same power applied to each).
 
Last edited:
And to be honest, I'm much more interested in the flh vs tapped horn discrepancies you mentioned earlier but that discussion didn't go anywhere. You didn't provide a simulation for the front loaded design.
You provided none of your simulations or measurements...

I used a Hornresp FLH simulation that did not represent the actual results.
It would be of no use to post up a simulation that bears so little resemblance to the finished cabinet.
It is not possible to model a DFLH in Hornresp, the ducts did bring it closer to the original FLH Hornresp model.
I was not greatly pleased by the design, other than it is adequate for providing bass for a small system that can be towed behind my Mustang.
As it sits, I spent far more time designing, building and trying to get the DFLH to work decently than I have using them. I posted them more as a curiosity, as the design is quite unusual.

If you want to take the time to simulate the DFLH design in Akaback and find you get the same results as the cabinet as built, good on you.
 
You provided none of your simulations or measurements...

I used a Hornresp FLH simulation that did not represent the actual results.
It would be of no use to post up a simulation that bears so little resemblance to the finished cabinet.
It is not possible to model a DFLH in Hornresp, the ducts did bring it closer to the original FLH Hornresp model.
I was not greatly pleased by the design, other than it is adequate for providing bass for a small system that can be towed behind my Mustang.
As it sits, I spent far more time designing, building and trying to get the DFLH to work decently than I have using them. I posted them more as a curiosity, as the design is quite unusual.

If you want to take the time to simulate the DFLH design in Akaback and find you get the same results as the cabinet as built, good on you.

So to be clear, you didn't even try to make an accurate simulation of the front loaded horn at all but still claim that the simulation doesn't match the measurement? I think that's all I need to know.

I did show you the sims and measurements I could find online from several years ago, which were first run measurements with no care taken to get an accurate result, only to make sure there were no major problems like air leaks. Even after changing the design in Akabak, raising the tuning, the sim still almost exactly overlays the measurement except for an unexpected null at 90 hz which I've already explained twice. I already explained that my old computer died and I lost my more accurate measurements hat I never posted online (taken outside at 10m) which look just like the measurements I did show. I went out of my way to show you what I have. I haven't built or simmed anything new in the last 2 or 3 years so I don't have anything new to show you.

As far as I can tell you are the only one claiming Hornresp is not accurate at small signal levels. Clearly there are differences at high power levels but if you can't get your response curve shape to match your sims at 1 watt you are doing something wrong. Lilmike is a member here (but more active at avsforum) and he's posted sims and measurements of ALL of his SEVERAL designs and they ALL overlay within about 1 db across their entire bandwidth. He's posted both front loaded and tapped horn sims and measurements overlaid on the same graph of each design. His experiences mirror mine and dozens of others.

I understand that you are a trial and error kind of guy and there's nothing wrong with that, but from what I can tell you never bothered to try to get an accurate simulation on either of your designs.

If you want to take the time to simulate the DFLH design in Akaback and find you get the same results as the cabinet as built, good on you.

I have no desire to take the time to simulate your design just to prove that it can be simulated accurately and that you could have done so yourself if you had taken the time.
 
Last edited:
Understood, but soho54 has shown over and over that accurate simulations match measurements very well, at least at small signal levels.

To attain accurate measurements one would need to measure the TS Parameters of the driver to confirm if indeed the TS Parameters are close to the manufactures stated TS Parameters. The majority of designers don't measure the TS Parameters of the actual driver. This is why many attain measured results differing from the simulation.
 

Attachments

  • LF corner.png
    LF corner.png
    236.8 KB · Views: 328
Looking at them it appears the low corner as measured is about 4 Hz higher than simmed, nearly 1/3 octave higher than the sim predicted.

To me, that is a fairly big difference, but close enough for rock and roll ;).

Art

In both this thread and that one I stated that I imported this design into Akabak to refine and also shorted the horn a bit to raise the tuning at that time for more max spl vs the shown lower tuning, but kept the response curve shape the same (only the lowest impedance peaks shifted up a bit as a result). Unfortunately I don't have the final Akabak sim anymore but the higher tuning was not a mistake, I did that on purpose in Akabak after this Hornresp data was captured but unfortunately this is all I have left to show.

I'd be more than happy to show you the final Akabak sim and the measurements that I took outside at 10m and well away from any boundaries but I don't have them anymore, I never posted them online anywhere and my old computer died.

In the absense of more accurate data on my part, I suggest we can look at Lilmike's data instead (if you are interested), he's got at least a few examples of very well documented sims and measurements done properly outdoors.
 
Last edited:
To attain accurate measurements one would need to measure the TS Parameters of the driver to confirm if indeed the TS Parameters are close to the manufactures stated TS Parameters. The majority of designers don't measure the TS Parameters of the actual driver. This is why many attain measured results differing from the simulation.

Further to OMNIFEX's comments -

In Hornresp, fs is taken to be the driver free-air resonance frequency. Manufacturers often don't state under what conditions their fs values are measured. If a given fs figure happens to be for a driver mounted in an infinite baffle, then the Hornresp predictions will be inaccurate. This is one reason why I prefer wherever possible to use the fundamental electro-mechanical parameters as primary inputs to specify the driver, rather being required to derive the values from perhaps questionable Thiele-Small equivalents.

Kind regards,

David
 
Hey guys.
Hope that opening this old thread isn´t a problem. I only need to "exploit" one. Or, I hope so. After (still to do) reading the 100+page topic about keystone, I might be tempted to pop even there.

After having 18SW115 in compact 135l bassreflex box, which works awesome, I´m looking further. I considered 21" 170l bassreflex, and also a tapped horn design.
But I did my homework, and I quite don´t buy that TH superiority yet.

Firstly, when I set some strict box volume, I never see compact TH outperform BR where I need it - under 50 or 40Hz. Then there is all that sensitivity and power talk, and efficiency, but NOONE looks at it from the impedance and real loading and power level angle of view. Therefore Efficiency is not what is measured. Just VOLTAGE SENSITIVITY, which is not really good way to look at it to be fair.
High BL speakers have close to 200Ohm peak impedance, and still over 60Ohm in the upper peak when enclosed in BR box. That´s not a lot power going into the speaker. How is TH?
 
Firstly, when I set some strict box volume, I never see compact TH outperform BR where I need it ...

Acoustic gain (and therefore ultimately efficiency) is largely a product of enclosure size (vs low knee frequency). If you keep the box sizes the same your efficiency is going to be roughly the same no matter what kind of box style you use.

If you look around at tapped horn designs they are usually about 4x larger than a normal sized ported box for the same driver and low knee. That's where the efficiency comes from.

There are also advantages that can't be seen so easily like the lack of port compression from any type of horn. Also horns (which are usually larger) hit their excursion limits with less power than ported boxes (which are usually smaller) and this might show some small gains in lack of power compression issues as well.

There's not much to gain from a compact TH, the whole point is that you can make them large and get some benefits over a ported box.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.