Why do you think it's not a problem at higher frequencies?
Back in the day when I had fullrange drivers this issue was a particular problem. Moving my head as little as a couple of inches would give an entirely different frequency response so I could "eq" the response by slouching vs sitting up straight or leaning a bit to the left or right. All stereo speakers will have destructive interference issues unless you specifically design the speakers with pattern control and place the speakers so the coverage does not overlap. Or you can at least use relative location of the drivers to steer the lobes away from the listening position, but in small rooms most of the indirect sound eventually comes back to the listening position so that is a compromised situation at best.
There's a good reason why speaker design uses very specific center to center distances and it's strongly advised to keep the mids and tweeters as tight together as possible. If you use a crossover design program you can visually see the effect of lobing if the speaker allows you to pan around and view frequency response at different locations, and it's ugly, even with the smallest mids and tweeters placed as close together as physically possible.
Last edited:
Still has destructive interference issues like most arrays do, even with the tight center to center spacing of the individual drivers the drivers at opposite ends are quite far away from each other.
Benefits include a somewhat planar wave for less than 6db reduction in spl per doubling of distance (in the nearfield at least) and a huge baffle face providing some pattern control similar to barn doors, not allowing the higher frequencies in the sub's passband to radiate backward and pushing them forward instead.
Very expensive approach though, and leaves the drivers open to the elements of nature unless you put it all away indoors after use.
Comb filtering occurs through the entire audio range with multiple sources, but for stereo speakers there's very little to be done about it. It's especially problematic in the bass frequencies though, because unlike the upper frequencies, you can have entire areas with huge gaps in bass coverage. This is much less of a problem in the upper range. You aren't going to find a large area in front of the speaker with no midrange, but for multiple subs the lobing can cause massive dead spots.
And I guarantee you cannot hear the source of a 30hz tone in a test like that. What you're hearing, I'd bet, is a shift in the room modes being excited. Whichever speaker is playing the tone will excite different modes and that will certainly be audible. Perform the test again in an open field and it will not work.
Unless there's a lot of harmonics and distortion at higher frequencies that accompany the 30 hz tone, which is actually pretty likely. Even with a steep crossover at 80 - 100 hz, any higher frequency harmonics and distortion at higher frequencies are still going to get through a bit (the cone itself even has resonant frequencies well above the 30 hz tone that can be excited and produce noise), and since the ear is MUCH more sensitive to these higher frequencies there can be situations where you can detect the location based on the small but still audible higher frequency distortion being produced.
And that's just the driver portion of the sound considered. The cab itself can (and does) have resonant modes well above 30 hz that can be excited by the energy from a 30 hz tone, and the cab itself can be a sound source at the resonant frequency of the cab panels. Far down in level, but somewhat audible due to the increased sensitivity of the ear at higher frequencies.
And then when you push it hard, the whole system is a distortion machine - the surround can make extraneous noises not related to the tone frequency, items placed on or near the sub can audibly vibrate, drywall can slap on the wall studs, the whole space is energized.
It's not really surprising to me that subs can be localized even when playing low frequencies, but it's all about distortion, not the actual tone of interest. Especially cheap subs, as they are distortion machines.
1) Simulations are useful when you lack measured data, measured data is proof, simulations are not.
2) Persistence of vision using a dot on the cone along side a ruler is accurate to within 1/2mm, certainly not statistically "a huge margin for error". I could easily tell when excursion was <1mm peak to peak, and Xmax is one way, so figure less than 1mm maximum error on measurements of up to 18mm peak to peak. One does not need a laser reader for this kind of work.
3)I can assure you the the peak excursion was captured, as I also ran manual sweeps with an analog tone generator prior to the documented tests. The excursion measurements with 5 Hz increments with the BR enclosure and three different driver compliments in the Keystone was a lot of work, but provides plenty of data points for comparison. You won't find too many other designers on the planet that have provided as much data for you, which you dismiss as if I was some wet behind the ears newcomer to the processes required to make a product with superior performance aspects.
4) If you bothered to look at the low corner frequency response you would notice they are as near to identical as possible.
5) The driver in the ported box (a BC18SW115-4) was not only mentioned in the text you replied to, but in the data reprint that you obviously also ignored. The BR test also used the same driver as was used in the Keystone to avoid any possibility of driver to driver differences accounting for doubt in any reader's mind regarding test protocol.
The tests were conducted with the cabinets front baffles in the same exact position, to insure no external impedance variables were in play.
6)You continue to ignore measured results that don't agree with your simulations, which seems out of character considering your extensive research when you found "large voice coil" drivers measured in box response did not match Hornresp sims until David adjusted the program to include your input.
You say that the Keystone is not unlike any other tapped horn, when unlike others it's mouth exit can't be properly modeled in Hornresp.
If you want to understand it without building and measuring your own cabinets, you will have to use a program that is capable of it's proper simulation.
You will know you are using the proper program when your simulated frequency response and excursion matches those of the actual cabinet as built and measured.
Cheers,
Art
Last edited:
You and Danley keep saying that, but it's been proven beyond any doubt (as far as I'm concerned) that at small signal power levels accurate sims match measurements to a very high degree of accuracy.
I've shown on a few occasions that both you and Danley are not exactly accurate when simulating what was actually built.
Once you get past small signal levels the sims start to lose accuracy, but as data-bass testing has shown you have to get pretty near the limits of the driver before you start to see very major deviations from the small signal levels. Besides, we know very well what is going to happen under those circumstances so the deviations are not unexpected and can be accounted for in the sim somewhat.
I have at least a couple of issues with this.
1. You would need to have a line of vision as close to parallel to the cone as possible to measure visually with any accuracy. That would require crawling right inside the mouth of the horn, a very uncomfortable position.
2. Persistence of vision in this case is very blurry at the excursion max and min positions. I've tried to visually measure excursion before and concluded that it can't be done accurately by eye. This is like trying to measure frequency response by ear. Sure you can do it to some extent, but the proper tools are going to do a better job.
Your work is noted and appreciated. The issue is that you dismiss the accuracy of the sims, although you haven't accurately simulated what you have built, and then measured certain characteristics with tools that are not suitable for the task.
When large deviations between sims and measurements are noticed, it becomes necessary to change the sim protocol. So far the only large discrepancies I've notice so far are measurements at power levels far exceeding small signal levels and measurements of drivers with huge coils that suffer greatly from lossy inductance. Both can be at least somewhat accounted for in the sim.
I don't know of any easy and cheap way to measure excursion accurately so all I could possibly do is measure frequency response at the limits and compare the different designs that way. But even that is influenced by power and port compression differences in the different designs, so all you can truly conclude beyond any doubt is which design has more max output and how much for any given signal and power level.
This is inconsequential, the fact that no one has accurately simulated the Keystone doesn't mean it acts differently than other tapped horns. The physics are still in full force.
There are several reasons I haven't accurately simulated the Keystone so far.
1. I don't intend to build one, I design my own stuff.
2. It would take about 2 days of work to do it properly. That's how long it took to reverse engineer the F20 cab here - Horn Folding - a brief study of the centerline vs advanced centerline method - AVS Forum | Home Theater Discussions And Reviews
As you can see the work pays off - an accurate sim comes reasonably close to measured results.
3. The only reason I would have to simulate the Keystone in an accurate manner would be to prove that it can be simulated accurately and to show that for an equal volume and low knee it can be beat by ported boxes, like all tapped horns (although it would take more drivers and more power).
But since you seem convinced that it can't be accurately simulated and think it doesn't act like other tapped horns, it would also require a full build and measurement of both the Keystone and the ported box - something I am not in any position to do.
I could prove this pretty easily with a tapped horn I already have on hand - I have a few of the same spare drivers from the same production batch that I could use to make the ported box for comparison. But there seems no point, as you seem to be of the opinion that the Keystone is unique and does not operate under the same rules as other tapped horns. I also don't have time, this is my second annual busy season and the only reason I even have time for discussion is due to a few rain days lately.
There doesn't seem to be any way that you can convince me of the validity of your excursion test done by eye (or the validity of your comparison as a whole), and there's no way I can convince you that any experiment I would be willing to perform would have any validity to the nature of tapped horns in general, including the Keystone, so at this point we will have to agree to disagree. Maybe at some point, life will put us in the same place at the same time and we can come to an understanding via real world experience and measurements of an example project.
Last edited:
Post #160 - Desibel
Hi Entropy455,
Just on note: The big green horn project of which David McBean attached a picture in Post #160 is the "Desibel". Here is a link to Bjoern Kolbrek's page:
The Desibel Horn System
I think you can get a lot of information from there as to size, number of horns required to cover the full audio band, and such. Naturally, Desibel had to be portable, and you don't have the same constraints...
Regards,
Hi Entropy455,
Just on note: The big green horn project of which David McBean attached a picture in Post #160 is the "Desibel". Here is a link to Bjoern Kolbrek's page:
The Desibel Horn System
I think you can get a lot of information from there as to size, number of horns required to cover the full audio band, and such. Naturally, Desibel had to be portable, and you don't have the same constraints...
Regards,
Jag,
A) My professional use of equipment is seldom done at small signal power levels, as that use would require hauling an order of several magnitude more speaker enclosures than required. I have no argument with your pointing out that Hornresp simulations don't accurately reflect what I built in the case of the Keystone sub.
Many of my cabinets designs, such as the L4 (both the FLH and later ported truncated horn version) "C horn", WS 2xLAB12", all are accurately represented by Hornresp sims.
1. The Keystone driver position does not "require crawling right inside the mouth of the horn" to have a line of vision quite close to parallel to the cone. I take the parallax error caused by off axis viewing into account when recording excursion. The "Keystone" name is in fact derived from the effect of parallax error in a visual projection.
2. My vision is not "very blurry" at the excursion maxima and minima. I am nearsighted, my vision is good at close range, and does not get blurry until a bit over 130 dB at certain frequencies that resonate the ocular cavity, which literally makes your eyes flop around in their sockets. The Keystone excursion tests were done at a level where my eyes behaved normally.
3. Thanks for the appreciation, but it sounds like you are confusing me with a fellow from Toronto, unlike he, I am quite positive anything can be accurately simulated given enough time and resources.
4. Understood.
5. I am convinced that thus far, nobody has presented an accurate simulation of the Keystone sub as built. It is certainly possible that somebody has done a proper simulation, and has used the information gathered for their own use, but they have not shared the results with me or this forum.
I understand that given the extensive measurement testing of the actual response, there would be little to be gained by anybody spending the time to enter the build parameters into Akabak or any other program that may produce a simulation that accurately reflects the measured response.
I know from using Hornresp that adding 5" height to the Keystone design will drop the Fb to 29 Hz, but I know from the exit size and shape experiments I conducted before arriving at the exact dimensions that they will likely need adjustment for the "B-Low" version.
If I were to spend dozens of hours learning and becoming proficient in Akabak program, I may be able to accurately predict the optimum exit dimensions for the two different drivers I will use in the new design, but would still feel compelled to confirm whether the prediction was optimal by trying physically different exit size and shapes to confirm that my simulations were in fact optimal.
Since I will be measuring the results of different exit size and shapes, spending time to become proficient in another far more difficult simulation program does not seem to be as fun as finishing several speaker designs in the works, going to the theater, sailing, creating new music, or any number of other things that make up the limited 24 hours a day we have to play with.
Art
Last edited:
More OT 
Hi Art,
In Hornresp: adding 15" (you are only adding 5") to the height of the Keystone, and only modifiying the upper portion of the cabinet (no modifications to anything in the lower portion) I get roughly: Fb=25.7Hz F-10dB=20Hz F-6dB=22Hz F-3db=24Hz. I agree, it would have to be built, and measured.
Regards,
Hi Art,
In Hornresp: adding 15" (you are only adding 5") to the height of the Keystone, and only modifiying the upper portion of the cabinet (no modifications to anything in the lower portion) I get roughly: Fb=25.7Hz F-10dB=20Hz F-6dB=22Hz F-3db=24Hz. I agree, it would have to be built, and measured.
Regards,
Akabak isn't as hard to learn as you might think. I learned to use all the features that I needed in about 2 hours, and I was also watching a movie at the time so it's not like I was working hard on it. There's still a lot I don't know how to do - stuffing, diffraction, acoumass, acouresistance, acoucompliance for example, but I don't really need that stuff.
Within 2 hours I was able to read scripts and understand what is being simulated and how, and I was able to write simple scripts (although that isn't usually required because there's a wizard for virtually every input and you can import a Hornresp file so Hornresp does most if not all of the heavy lifting for you).
Simply follow this thread and find out how easy it actually is - AkAbak for Dummies
- AVS Forum | Home Theater Discussions And ReviewsOn the other hand I understand if you don't want to - learning a new skill is rarely any fun - the fun doesn't start until you have a working proficiency at it.
As to the other stuff, I am noting all you've said and I'll try to keep an open mind until such time as I can do my own experiments.
Oliver,
Using the "advanced centerline" approach, the 5" height increase does not result in a full 15" horn path increase, hence my Hornresp estimate of 29 Hz Fb.
After testing, I may revise the name accordingly, it may be an "A" or "G"-Low rather than a "B-Low", but neither sounds as cool
.That said, adjustment of the Keystone exit area gives an adjustment window of several Hz for the low corner, I will shoot for a solid 30 Hz response, as I'd prefer more output at 30 Hz and up than extended response into the mid 20 Hz range, as neither pipe organists nor EDM LF freaks have ever been on the list of clients I seek to work with, as loud LF much below 30 Hz tends to make me feel ill
.Art
Last edited:
Everything I try to do lately requires like four times the time I expect it to. Remembering the days I spent getting used to Hornresp (and Windoze...), with it's relatively easy interface, would mean if I started learning Akabak now, it would displace approximately the time I expect it to take to build six Keystone "B-Low" cabinets. Their pre-cut plywood has been taunting me, sitting in the shop since just before Hurricane Matthew de-railed my to-do list.
Since you have already learned to use Akabak (although diffraction, acoumass, acouresistance, acoucompliance may also be required in addition to the features you learned) it is possible you could figure out the correct exit size and shape for a solid 30 Hz response in the taller Keystone enclosures before I do it empirically, were you interested, and not in your second busy season.
Art
Last edited:
XRK is the resident Akabak expert and has done considerable work already with simulating the Karlson exit shape - and he's friendly and helpful. If he can sim a Karlson exit the Keystone shouldn't be a problem. It probably wouldn't take him more than a few minutes to an hour to sim it up.
Unfortunately my Akabak computer needs a power supply (virtual machines don't work well on my regular computer for some reason that I haven't figured out, so no Akabak on the main computer). The weather report suggests this will be our last rain day so I can't be much help, although it does seem like an interesting challenge.
Unfortunately my Akabak computer needs a power supply (virtual machines don't work well on my regular computer for some reason that I haven't figured out, so no Akabak on the main computer). The weather report suggests this will be our last rain day so I can't be much help, although it does seem like an interesting challenge.
More pondering on the horn subwoofer
Hi Entropy455,
Post #202: "...mouth is under 12 feet square. These are dinky horns compared to what I was originally planning to build (i.e. full-space horns). I’m a happy about this...
Question: Should I drop the tune down to 18 Hz... or should I just stick with my original 20 Hz goal?"
I'm just messing around in Hornresp w/ a fixed length (40ft long Hyp) subwoofer horn w/ different throat/mouth/chamber sizes. From this, I would stay w/ the single horn, and just increase the mouth to the point I'm comfortable constructing. The larger the mouth (for a given chamber/throat/length) the lower the ripple.
Somewhere around 25m^2 I hit the point of diminishing returns w/ a 4:1 compression ration (I think that's a lot of compression.)
I recommend you go w/ a single horn, and determine the horn flare, etc. from the construction viewpoint. The radius for a 50m^2 horn mouth is 5.64m[18.51ft], and for a 10m^2 mouth it would be 2.52m[8.28ft].
Regards,
Hi Entropy455,
Post #202: "...mouth is under 12 feet square. These are dinky horns compared to what I was originally planning to build (i.e. full-space horns). I’m a happy about this...
Question: Should I drop the tune down to 18 Hz... or should I just stick with my original 20 Hz goal?"
I'm just messing around in Hornresp w/ a fixed length (40ft long Hyp) subwoofer horn w/ different throat/mouth/chamber sizes. From this, I would stay w/ the single horn, and just increase the mouth to the point I'm comfortable constructing. The larger the mouth (for a given chamber/throat/length) the lower the ripple.
Somewhere around 25m^2 I hit the point of diminishing returns w/ a 4:1 compression ration (I think that's a lot of compression.)
I recommend you go w/ a single horn, and determine the horn flare, etc. from the construction viewpoint. The radius for a 50m^2 horn mouth is 5.64m[18.51ft], and for a 10m^2 mouth it would be 2.52m[8.28ft].
Regards,
Attachments
Last edited:
JAG,
xrk971 does do very good simulation work with Akabak, his simulations of my SynTripP virtual single source point, two section, conic/exponential horn design were fairly close to the response of the cabinet as built.
The porting scheme I devised for that project proved to be a bit of a challenge to simulate, resulting in a slightly smoother measured response due to the port resonance not being as rough or large as predicted, and the woofer injection ports not having quite the predicted peaks.
As they say on the carnival midway, "close, but no cigar"
.I may hire Mr. X to make some printed plastic parts for some upcoming SynTripP models in the works.
Art
Oliver,
That would explain it, the difference of a few inches that the advanced centerline measurement on a 5" increase in the height of the cabinet would be a lot less than adding 15", which would really not work well for 4x8 plywood, but would be a natural for 60" x 60" Baltic Birch.
60" would be too tall for my trailer pack though, even if I wanted the cabinet response to go that low.
That response would be in Entropy Eric's wheelhouse, if he were to consider multiple portable enclosures for his desired sonic potential, rather than tons of poured concrete permanently placed in a potentially compromised placement.
TOPCPPIAPCP.
Cheers,
Art
Question: if I go with a single mono bass horn, how exactly do I go about getting my mono signal? I primarily listen to CD recordings.
I was hoping to keep the audio signal(s) in the digital domain, as long as possible. Specifically, I’d like to find a high-end (or otherwise capable) computer sound card that has an all-in-one adjustable multi-channel active crossover (digital), as well as a multi-channel time-delay feature (digital). Some of the newer amplifiers have D/A converters built right into them – and I’d be willing to pick up some of these amps if necessary.
I’ve currently got a really nice analog crossover that sounds great. However the only adjustable time-delay devices I've found (to properly time my horns) are digital. It has been my experience that whenever analog signals are converted to digital for processing, then converted back to analog – the system fidelity suffers (compounding errors).
I was hoping to keep the audio signal(s) in the digital domain, as long as possible. Specifically, I’d like to find a high-end (or otherwise capable) computer sound card that has an all-in-one adjustable multi-channel active crossover (digital), as well as a multi-channel time-delay feature (digital). Some of the newer amplifiers have D/A converters built right into them – and I’d be willing to pick up some of these amps if necessary.
I’ve currently got a really nice analog crossover that sounds great. However the only adjustable time-delay devices I've found (to properly time my horns) are digital. It has been my experience that whenever analog signals are converted to digital for processing, then converted back to analog – the system fidelity suffers (compounding errors).
Ummm, while some folks is running off playing with their sims late at night to three decimal places, you're still at Stage One in system design.... as you should be.
In short, you get a Behringer DCX2496, like a whole lot of those on DiYaudio. It does the rest beautifully. My system is likewise mostly digital. You can feed the Behringer with an AES coax digital signal from an inexpensive USB device ("Breeze" on eBay is one such) and thereby avoid slipping into analog between the computer and the Behringer downstream digital stage.
A mixed bass is simply L and R added together, not that there are many sources where they differ in the low bass and if they do, probably was simple-minded of the engineer to do so. And if they differ, certainly just cooked up on the recording console, not authentically spatial. (There are some people who say it is better bass to have distinctive sub signals, but a lot of reasons not to such as building two horns and two amps.)
As I said before, you need to sort out system design (including considering your room) before fussing over horn mouth size.
Ben
In short, you get a Behringer DCX2496, like a whole lot of those on DiYaudio. It does the rest beautifully. My system is likewise mostly digital. You can feed the Behringer with an AES coax digital signal from an inexpensive USB device ("Breeze" on eBay is one such) and thereby avoid slipping into analog between the computer and the Behringer downstream digital stage.
A mixed bass is simply L and R added together, not that there are many sources where they differ in the low bass and if they do, probably was simple-minded of the engineer to do so. And if they differ, certainly just cooked up on the recording console, not authentically spatial. (There are some people who say it is better bass to have distinctive sub signals, but a lot of reasons not to such as building two horns and two amps.)
As I said before, you need to sort out system design (including considering your room) before fussing over horn mouth size.
Ben
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.