FWIW I have measured a 15" (not a coax, however) driver that kept a very nice pattern to about 1.5kHz. So it is possible.
Hi nc535,
Thanks for sharing you simulations models. I "learned" ABEC last winter but haven't played with it for awhile. I bought a pair of 15OB350's to experiment and correlate simulation/measurements, but this is on-hold.
Here is the ABEC input deck for the 15OB350:
dD =316mm // Diameter of cone based (Cutout-2*ws)
//dD =330mm // Diameter of cone based on Sd=359.7cm2
tD1=59mm // Inner depth of cone to base of dust cap
dD1=126mm // Diameter of dust cap
hD1=17mm // Height of dust cap
hD2=86mm // Height of total cone on the outside
dVC=65.5mm // Diameter of voice coil
hVC=0mm // Height of voice coil
dM =180mm // Diameter of magnet
hM =56mm // Height of magnet (up to visible cone)
Ws =21mm // width of suspension BaffleCutout= 242 = dD + 2*Ws
The voice coil is not exposed so I put the height as 0. The cone attaches to the spider about 41mm forward of the back of the magnet. The spider OD is about the same diameter as the magnet, so I just fudged a longer magnet to reach the spider. Below I just replaced the lower woofer from your ABEC simulation to show the difference. I agree it's unlikely to change your results. Obviously one thing we are missing is the effect of the basket on the rear - intuitively that would rank high on my list to explain simulation vs. measurements discrepancies, but that's part of the learning curve!
Thanks for that!
However, I'm going to stick with the cone diameter based on Sd because the SPL level is derived from that. I'm also going to stick with the suspension with that sums with that diameter to match the data sheet baffle cutout. Finally, I'm going to stay with the data sheet VC height because it is apparently what spaces the magnet behind the voice coil. How far back the magnet starts is probably the most critical dimension. Ultimately, I will tweak it in with the VC height as I can scale that dimension from the data sheet dwg.
The height of the magnet needs a judgment call because of the way it is tapered. The back side of the magnet is shaded by the front and thus doesn't get reflected from but does participate in diffraction around the magnet. This really calls for a CAD based mesh file to be imported for the woofer frame and magnet; at least it would if this were not a woofer.
I think the frequency range where these details matter is about 800 Hz ++. In my 5208C modelling, I'm seeing peaks and valleys in the rear wave, not matched by any in the front wave, that I have to ascribe to interactions between the magnet structure and the rear of the baffle. I really hope someone will do a similar post for the 5208C.
However, I'm going to stick with the cone diameter based on Sd because the SPL level is derived from that. I'm also going to stick with the suspension with that sums with that diameter to match the data sheet baffle cutout. Finally, I'm going to stay with the data sheet VC height because it is apparently what spaces the magnet behind the voice coil. How far back the magnet starts is probably the most critical dimension. Ultimately, I will tweak it in with the VC height as I can scale that dimension from the data sheet dwg.
The height of the magnet needs a judgment call because of the way it is tapered. The back side of the magnet is shaded by the front and thus doesn't get reflected from but does participate in diffraction around the magnet. This really calls for a CAD based mesh file to be imported for the woofer frame and magnet; at least it would if this were not a woofer.
I think the frequency range where these details matter is about 800 Hz ++. In my 5208C modelling, I'm seeing peaks and valleys in the rear wave, not matched by any in the front wave, that I have to ascribe to interactions between the magnet structure and the rear of the baffle. I really hope someone will do a similar post for the 5208C.
I would stick with the dD based on Sd too to run simulations. I put it there to show how it would look like sticking to ABEC’s definition.
As for the magnet position I think we’ve reached the limits of ABEC’s diaphragm definition. The position where the VC meets the spider and thus the cone definition is correct. My dimensions reflect the fact the voice coil is not visible so air has to go around the spider starting from this plane, but the surround itself is a moving surface, etc.. Many questions to which I don’t have the answers now so I appreciate a different perspective.
Ultimately one has to make a full-blown CAD-based model at least once to correlate with measurements. Then introduce the “reduced-order” ABEC model’s limitations on one by one to appreciate the modeling errors and learn how to make the most of the “reduced-order” model and its limitations.
Before going full blown there some learning is to be gathered making a CircSym model with a simple round baffle or baffle-less (thanks Charlie). This is where I was when I put simulation on pause.
As for the magnet position I think we’ve reached the limits of ABEC’s diaphragm definition. The position where the VC meets the spider and thus the cone definition is correct. My dimensions reflect the fact the voice coil is not visible so air has to go around the spider starting from this plane, but the surround itself is a moving surface, etc.. Many questions to which I don’t have the answers now so I appreciate a different perspective.
Ultimately one has to make a full-blown CAD-based model at least once to correlate with measurements. Then introduce the “reduced-order” ABEC model’s limitations on one by one to appreciate the modeling errors and learn how to make the most of the “reduced-order” model and its limitations.
Before going full blown there some learning is to be gathered making a CircSym model with a simple round baffle or baffle-less (thanks Charlie). This is where I was when I put simulation on pause.
I will definitely rerun the woofer sims to see what difference the diaphragm model changes make but right now I'm working on the COAX response.
The design now is so different than when I started that I need to retrace my steps in order to tell the story.
When I first modelled a tall tapered baffle with a Coax mid on top, I got a vertical response that was tilted downwards. This is 1221 Hz:
this is obviously unacceptable. I saw two alternatives. One was to mount the coax on a small baffle separated from the woofer baffle sufficiently to allow cancellation to occur below it. The other was to cut a hole in the baffle between the woofer and coax driver. I tried both choices and the hole in baffle worked best. Here are the 1239 Hz fields:
Unfortunately, the hole in the baffle above the woofer unbalanced the woofer's vertical response:
the tilt isn't so extreme but it added to the vertical dip at XO between the woofer and the coax mid circa 400 Hz so I felt I had to do something about it. My goal is a speaker that can be listened to both seated and standing, albeit at a greater distance for standing.
Thus I ended up with two holes in the baffle, a nearly level V pattern,
and a horizontal pattern with side nulls remaining past XO to the coax mid, although not as deep as at lower frequencies
This is what the baffle model looked like at that stage:
I had added cross braces above and below the woofer because I didn't want the HinBs to be acoustic short circuits for the woofer. I ended up in a tug of war between the coax mid and woofer for the upper cross brace. The woofer response was best with a deep brace; the coax response was best with a shallow brace.
I got a really nice system response in Vituix exporting directivities for a 100mm deep brace. The olive dashed trace below is the in-room response with wall and floor reflections enabled, 30 deg toe in, 1m from each of front and side walls.
The design now is so different than when I started that I need to retrace my steps in order to tell the story.
When I first modelled a tall tapered baffle with a Coax mid on top, I got a vertical response that was tilted downwards. This is 1221 Hz:
this is obviously unacceptable. I saw two alternatives. One was to mount the coax on a small baffle separated from the woofer baffle sufficiently to allow cancellation to occur below it. The other was to cut a hole in the baffle between the woofer and coax driver. I tried both choices and the hole in baffle worked best. Here are the 1239 Hz fields:
Unfortunately, the hole in the baffle above the woofer unbalanced the woofer's vertical response:
the tilt isn't so extreme but it added to the vertical dip at XO between the woofer and the coax mid circa 400 Hz so I felt I had to do something about it. My goal is a speaker that can be listened to both seated and standing, albeit at a greater distance for standing.
Thus I ended up with two holes in the baffle, a nearly level V pattern,
and a horizontal pattern with side nulls remaining past XO to the coax mid, although not as deep as at lower frequencies
This is what the baffle model looked like at that stage:
I had added cross braces above and below the woofer because I didn't want the HinBs to be acoustic short circuits for the woofer. I ended up in a tug of war between the coax mid and woofer for the upper cross brace. The woofer response was best with a deep brace; the coax response was best with a shallow brace.
I got a really nice system response in Vituix exporting directivities for a 100mm deep brace. The olive dashed trace below is the in-room response with wall and floor reflections enabled, 30 deg toe in, 1m from each of front and side walls.
The holes in the baffle presented me with an aesthetic problem
the best I could do is hide them behind grills:
to my eye, this works fairly well for rosewood veneer and dark rose grill cloth, assuming I could find some, but falls short when the grill needs to be brown to coordinate with the wood:
I believe that form should follow function, but perhaps not quite that far.
Pondering this dilemma, I belatedly had the idea to extend the baffle above the coax driver to straighten its vertical response, instead of cutting a hole in the baffle below it.
That brings me to this morning's work, still a lot to do but this is the concept
the best I could do is hide them behind grills:
to my eye, this works fairly well for rosewood veneer and dark rose grill cloth, assuming I could find some, but falls short when the grill needs to be brown to coordinate with the wood:
I believe that form should follow function, but perhaps not quite that far.
Pondering this dilemma, I belatedly had the idea to extend the baffle above the coax driver to straighten its vertical response, instead of cutting a hole in the baffle below it.
That brings me to this morning's work, still a lot to do but this is the concept
The fields images for the woofer with single hole in baffle fell out of the long post prior to the one above. Here ithey are
for 417 Hz and 96 Hz
I've likely lost track of your design intent along the way here, so apologies in advance if you covered this earlier. I was wondering why the bass driver is that far off the floor, and whether the tug of war between the coax and the bass you described could be handled simply by having the bass more closely coupled to the floor (aka, larger air gap between coax and bass). Would having the bass driver lower also be better for the first floor reflection?
Anyway, then I spotted a bit of driver through the hole at the bottom. Is that an upward-facing sealed sub? Or otherwise, what are the plans for that lower volume there?
Separately, I've read a few other threads where a benefit was gained by separating the panels to reduce vibration transmission. I'm not sure how much worse the simulation was between the coax with the air gap vs the coax with the hole--and I'd wonder whether that simulation holds true at other angles--but the uncoupling of the upper baffle might convey other sonic advantages.
Unfortunately I don't have my own ears and experience to draw on here--only web-reading eyeballs 😵--but most of the time isolating/removing vibration seems to be a good thing in audio.
That is a rear facing sub in back and the dipole woofer is elevated just enough to make room for it. The dipole woofer is still close enough to the floor to have its SPL effectively doubled by the floor image but not so far as to have a floor reflection null within its passband, although there is one just outside it. Its also still close enough to the floor to take advantage of the Allison effect to reduce the floor null. This is confirmed by the overlay trace in post #125.
I'm going back to the version with the sloping rear baffle , as shown in the abec fields drawing in the post just before yours as opposed to the flat, up facing sub shown in the renderings. Believe it not, I was getting reflections from the coax midwoofer off the flat sub. I believe they will be dispersed by the sloping rear.
The sealed 12" 14mm Xmax rear sub crosses to the dipoles at 80 Hz, making this a full range speaker without subs, not to say that additional subs wouldn't help in room. Since the 15OB350 only needs to go down to 80 Hz instead of 30 or 40 Hz, it needs 1/6 less displacement and is in fact power limited rather than excursion limited. Since the sub is there, the 15OB350 doesn't really need any help from the floor.
The predicted H polars look like what others have shown for the combination of a sealed and dipole woofer: There is/will be a rear facing tweeter that isn't modelled here. There is an off axis null circa 1 khz that I'm investigating.
You definitely don't want that big baffle to vibrate. The sub will be in a separate cabinet that just sits behind the baffle without touching it. The baffle will have side wings 10-12" high to prop it up. The drivers on the baffle will be decoupled to some extent by gaskets and rubber grommets on their mounting screws. I've done this before and it seems to work but I haven't done any science to confirm that it is really effective. I'm open to suggestions on this aspect because with a massive 36 mm thick baffle and low mms drivers it might not be worth obsessing about.
I'm going back to the version with the sloping rear baffle , as shown in the abec fields drawing in the post just before yours as opposed to the flat, up facing sub shown in the renderings. Believe it not, I was getting reflections from the coax midwoofer off the flat sub. I believe they will be dispersed by the sloping rear.
The sealed 12" 14mm Xmax rear sub crosses to the dipoles at 80 Hz, making this a full range speaker without subs, not to say that additional subs wouldn't help in room. Since the 15OB350 only needs to go down to 80 Hz instead of 30 or 40 Hz, it needs 1/6 less displacement and is in fact power limited rather than excursion limited. Since the sub is there, the 15OB350 doesn't really need any help from the floor.
The predicted H polars look like what others have shown for the combination of a sealed and dipole woofer: There is/will be a rear facing tweeter that isn't modelled here. There is an off axis null circa 1 khz that I'm investigating.
You definitely don't want that big baffle to vibrate. The sub will be in a separate cabinet that just sits behind the baffle without touching it. The baffle will have side wings 10-12" high to prop it up. The drivers on the baffle will be decoupled to some extent by gaskets and rubber grommets on their mounting screws. I've done this before and it seems to work but I haven't done any science to confirm that it is really effective. I'm open to suggestions on this aspect because with a massive 36 mm thick baffle and low mms drivers it might not be worth obsessing about.
This is a really cool project! Or at least I think it's cool since it aligns very much with what I'm considering building for my next system. 😎
In my living/listening room I have this large planar surface called the "floor". Just kidding 😉, but I'm not sure the flat top of the sub is much different? Unless you put absorptive material there, even at an angle wouldn't those waves still bounce around and come back off other surfaces anyway as (hopefully) soundstage depth and air?
The co-located but physically separate subwoofer makes a lot of sense for isolating the direct vibrations. In my room I'll unfortunately need to be able to regularly move the speakers out from the wall for more critical listening, so I've toyed with integrating a dual-opposed sub that force-cancels. Specifically, I've looked at rebuilding the Rythmik FM8 which has dual 8" paper cone drivers driven by their their servo plate amp--pairing the drivers side to side, front to back, or even separated by a 4-5' long vertical tube that puts monopole bass into the vertical axis of the room (possibly generating semi-multipoint bass??) while also holding the coax baffle and rear tweeter. [note--I'm not sure those drivers are ok in the horizontal orientation for the tube, would have to check]
The FM8s (or their guts if available separately) would likely cost a good bit more than your solution. That said, FWIW at least one person on a different forum prefers the FM8s to the GR OB dual servo 12s and they'll take you down to 20 Hz plus XO to the 15OB350 at 80 Hz/ 12 db quite easily (some people run them up over 200 Hz).
I'm trying to change your design, of course, just sharing ideas driven by different listening room constraints. The level of rigor here is amazing, way past what I'd every attempt myself!
I originally thought of the flat top sub as just a bump in the floor but when I saw artifacts of it in a simulation of the coax mid I had second thoughts. Its somewhat inconsistent to include a bump in the floor in a simulation but not the floor itself. If the rear sloping sub gets rid of those artifacts by directing reflections away from the observation point, then I can move ahead without any hand-waving arguments as to why I can ignore the artifacts. That remains to be seen.
We all have to work within our constraints. I'm keeping the weight down and dividing the load but once I get them in place, I may well tie them together through "damping bushings" from McMaster-Carr. The last time I had OB speakers, somehow I had to keep pulling them out from the wall. Something there was that kept pushing them back. I think it was my wife🙂
Use as big and as many subs as you can. I'm well aware that my single 12" per side is a compromise but I no longer listen as loud as I did in my youth so it will probably work for me. If not, I can build more thin sub boxes and hide them around the room... My goal was to have sub that allowed me to keep the 15OB350 no higher than where I show it now which is a compromise between CTC for the XO to the midwoofer and being close to the floor so the floor null is out of its passband.
We all have to work within our constraints. I'm keeping the weight down and dividing the load but once I get them in place, I may well tie them together through "damping bushings" from McMaster-Carr. The last time I had OB speakers, somehow I had to keep pulling them out from the wall. Something there was that kept pushing them back. I think it was my wife🙂
Use as big and as many subs as you can. I'm well aware that my single 12" per side is a compromise but I no longer listen as loud as I did in my youth so it will probably work for me. If not, I can build more thin sub boxes and hide them around the room... My goal was to have sub that allowed me to keep the 15OB350 no higher than where I show it now which is a compromise between CTC for the XO to the midwoofer and being close to the floor so the floor null is out of its passband.
I've pondered whether my speakers have a strange gravitational attraction for the wall ("speaker-wall coupling force"?), or perhaps that I have an infestation of living room gnomes who push them back to make space for their gnomeball matches.
It doesn't seem to be my wife, who maintains innocence and who, I might add, has never been actually observed in the act. 🤔 🙂
I don't know about the impact the floor might have in your simulations, but I see your point about trying to isolate certain aspects to better understand their individual contributions.
What would happen without the coax baffle? I keep seeing nude/no baffle systems and wonder what the relative pros and cons are.
If you haven't already, you should read Charlie Laub's recent article on baffle-less systems.
Basically, its a tradeoff between low end SPL and upper operating range limit. Also, when you get above the woofer, you don't want to have to contend with baffle step. Baffle step is eliminated without a baffle.
With a coax driver, baffle step for the tweeter is eliminated because the midwoofer cone acts as a waveguide. For the midwoofer, I have minimized the baffle width to little more than needed to support the driver, so it operates below the baffle step frequency. With separate midrange and tweeter, the same considerations drive you away from a baffle.
Were I to mount the coax on a separate small baffle or suspend it without a baffle, it's response would still include reflections from the woofer and woofer baffle below it. These would be a 2nd order effect but nevertheless visible in measurements, I believe. With a single baffle, there is no surface discontinuity between the two drivers, so I expect it to be cleaner in that regard.
Counter-intuitively, I have seen that at least in simulation, the coax midwoofer "likes" thicker baffles. This is helpful in that it makes it easier to mount the driver. The baffle thickness also contributes to the front to back path length for incremental low end support, but it also brings up the question as to whether the baffle hole for the coax driver should be relieved or beveled.
Basically, its a tradeoff between low end SPL and upper operating range limit. Also, when you get above the woofer, you don't want to have to contend with baffle step. Baffle step is eliminated without a baffle.
With a coax driver, baffle step for the tweeter is eliminated because the midwoofer cone acts as a waveguide. For the midwoofer, I have minimized the baffle width to little more than needed to support the driver, so it operates below the baffle step frequency. With separate midrange and tweeter, the same considerations drive you away from a baffle.
Were I to mount the coax on a separate small baffle or suspend it without a baffle, it's response would still include reflections from the woofer and woofer baffle below it. These would be a 2nd order effect but nevertheless visible in measurements, I believe. With a single baffle, there is no surface discontinuity between the two drivers, so I expect it to be cleaner in that regard.
Counter-intuitively, I have seen that at least in simulation, the coax midwoofer "likes" thicker baffles. This is helpful in that it makes it easier to mount the driver. The baffle thickness also contributes to the front to back path length for incremental low end support, but it also brings up the question as to whether the baffle hole for the coax driver should be relieved or beveled.
Regarding the baffle thickness, what did you mean by "thick" above when talking about the baffle holding the coax driver?
At one time there was some effort by other members of this forum to investigate the effect of a large roundover or thick tapering (getting thicker away from the driver) baffles with rounded ends on the off-axis pattern, but I am not sure that produced any results. It might be something you can model if you are interested.
At one time there was some effort by other members of this forum to investigate the effect of a large roundover or thick tapering (getting thicker away from the driver) baffles with rounded ends on the off-axis pattern, but I am not sure that produced any results. It might be something you can model if you are interested.
I'm modelling baffles as a planar surface of constant thickness, no round overs or bevels. I get best results with 36mm thickness but in truth there isn't a lot of difference at 18 or 24 mm
I do want to investigate a tapered thickness baffle, at least the taper that results from a bevel on the baffle hole for the driver. More than that comes up against the limits of the nodes based modeling methodology; would require switching to CAD generated baffle mesh which is another bag of worms.
I do want to investigate a tapered thickness baffle, at least the taper that results from a bevel on the baffle hole for the driver. More than that comes up against the limits of the nodes based modeling methodology; would require switching to CAD generated baffle mesh which is another bag of worms.
Last edited:
I've read quite a few of Charlie's invariably interesting posts, but haven't found the article you're mentioning. @CharlieLaub, could you point me where to go?
AudioXpress issues September and October of this year (2023). The October issue also has an article on the Bitches Brew open baffle by @perrymarshall so you get a twofer.
I'm interested to know more about why you're choosing an 80Hz crossover point.
80Hz is high enough that, depending on your SPL requirements, you could skip the 15OB350s entirely, and also skip the u-frame wings, and just use the 15CXN88 on a 18" wide slab. You'd have to DSP boost it in the 80-150Hz range but I imagine it has enough excursion to handle that.
I have a similar pair of subs in my listening room. I used them with the Faital Dipoles (12" coax in Open Baffle) and crossed over at 70Hz. I had no problem cranking out high SPLs with minimal excursion on the part of the Faital coax. The (polar pattern?) mismatch between the sub and the OBs was slight, but noticeable to the trained ear.
Now with the Bitches Brews I have a 30Hz crossover (actually there's no additional xover on the Bitches Brews, I use the stock EQ & the sub simply comes in below 30 as they're rolling off) and that setup works great down to 10Hz. 11 octaves.
It seems like if you're going to the trouble to make a speaker as big as the Bitches Brews (= size of small refrigerator) you might as well get all the polar pattern and bass you're paying for, especially if it's going to have wings.
Others like @Balthazarp have made non-Uframe, non-wings versions of the Bitches Brews and I could see crossing those over to your sub at 40-50Hz and I think that would work quite well.
Also I think dipoles sound more natural when the entire range is dipole and not just the bass and mids.
Last edited:
I appreciate your perspective on the general subject of the bass. I'm leery of being bass shy as I find bookshelf speakers, even ones with subs to be somehow lacking. OTOH, when there is enough volume displacement, and perhaps also treatment, that the low bass is undistorted and articulate, SQL rises to another level. My reference there is a floor to ceiling line arrays in a 25' square treated garage augmented by 4 15" subs plus two 10" subs. Another reference was a pair of corner synergies sitting on a sealed 15" corner subwoofers. Prior to my DIY excesses, I had Carver Amazing line arrays that had 4 12" dipole woofers per side.
My driver lineup is 5208C, 150B350 dipole (not U-frame), BM-11 shallow mount sealed sub. When the design settles down, I expect it to be almost a foot shorter than the Bitches Brew, half the width at top, and to have depth only at the lowest 10". I hope that is not as big as small refrigerator. Its more like a speedskater compared to a football lineman. Both have a solid base, but their upper bodies are completely different.
I'm being conservative in setting the XO points. My estimate says the 15OB350 dipole can go down at least to 80 Hz. It depends on how loud I want to play and how it sounds and how linear the driver is near its Xmax. I'm also being conservative in setting the high pass on the 5208C at around 360 Hz, but I'm also choosing it so that there is overlap with the woofer around the floor reflection null.
My double 15OB350 U-frame simulation predicts 100 db SPL at the dipole peak, 2.83V 1m. That is 10 db higher than my single 15OB350 dipole. How much does your 30 Hz crossover have to be raised to compensate for having half the volume displacement, and starting its rolloff at 400 Hz? Raising the XO just to 60 Hz means that 1/8 the volume displacement is needed. A single woofer then uses 1/4 the excursion at the same drive level but produces 6 db less output?
Perhaps 60 Hz XO is feasible. I will have to try it and see.
There is another aspect to this. Depending on how the sub is phased relative to the dipole and on whether or not there is any overlap between the two drivers, you can get either a pure dipole response down to an XO to the sub or a cardioid-like response around the crossover or through an overlap region. There is an opportunity there to tune the directivity there reduce boundary interference. The Vituix simulation does show a boundary null near 100 Hz that has been reduced to a 3 db dip this way.
My driver lineup is 5208C, 150B350 dipole (not U-frame), BM-11 shallow mount sealed sub. When the design settles down, I expect it to be almost a foot shorter than the Bitches Brew, half the width at top, and to have depth only at the lowest 10". I hope that is not as big as small refrigerator. Its more like a speedskater compared to a football lineman. Both have a solid base, but their upper bodies are completely different.
I'm being conservative in setting the XO points. My estimate says the 15OB350 dipole can go down at least to 80 Hz. It depends on how loud I want to play and how it sounds and how linear the driver is near its Xmax. I'm also being conservative in setting the high pass on the 5208C at around 360 Hz, but I'm also choosing it so that there is overlap with the woofer around the floor reflection null.
My double 15OB350 U-frame simulation predicts 100 db SPL at the dipole peak, 2.83V 1m. That is 10 db higher than my single 15OB350 dipole. How much does your 30 Hz crossover have to be raised to compensate for having half the volume displacement, and starting its rolloff at 400 Hz? Raising the XO just to 60 Hz means that 1/8 the volume displacement is needed. A single woofer then uses 1/4 the excursion at the same drive level but produces 6 db less output?
Perhaps 60 Hz XO is feasible. I will have to try it and see.
There is another aspect to this. Depending on how the sub is phased relative to the dipole and on whether or not there is any overlap between the two drivers, you can get either a pure dipole response down to an XO to the sub or a cardioid-like response around the crossover or through an overlap region. There is an opportunity there to tune the directivity there reduce boundary interference. The Vituix simulation does show a boundary null near 100 Hz that has been reduced to a 3 db dip this way.
I also think that 80 Hz is a good crossover point to a subwoofer, or perhaps a tad lower. The basis for this is a paper by Sean Olive that I mention in my aX article:
Included in the research that Olive presents are the preferred in-room (e.g. steady-stage) response curves for a variety of listener types and loudspeakers, all on one plot. This paper is not easy to get, so instead I suggest that anyone interested download this other paper by Toole in which he cites that work and presents the same data along with some other related info in Figure 14:
Please download the article, see Figure 14, and read the accompanying text in section 4.1. The preferred responses all show an increase around ~100Hz of between 3dB and 6dB. This "bass bump" is what the listeners preferred in the steady-state in room response.
I relate the data of Figure 14 to OB/dipole systems in the following way:
A true dipole system has constant directivity, and therefore also has a constant power response.
The in-room steady-state response reported by Olive and Toole will be very similar to the power response of the loudspeaker and is not flat!
This is the same result you see time and time again from Klippel testing of loudspeaker e.g. see Erins Audio Corner, reported as the estimated in-room response.
Therefore some equalization/correction of the dipole should be made to better mimic this response. Above 100Hz I do this via EQ of the input signal to "tilt" it down.
Below 100Hz we can obtain the "bass bump" for free just by crossing over to a monopole. This is because a monopole radiates 4.75dB more power into the room than a dipole for the same on-axis SPL.
So I just cross over around 80Hz to a sealed sub. The power response will naturally rise through the crossover point by almost 5dB, exactly like the preferred listening curves from Olive. The monopole sub can also deliver a better low bass response and room pressurization at very low frequency.
This also greatly reduces the demand on the dipole woofer, which now only needs to work well down to 80Hz. By reducing the LF requirements the dipole loudspeaker can be made physically smaller, or use a smaller woofer(s), etc.
Last edited: