The classic tower speaker. We all know it. It is burned into our brains. It stands well off the floor. It has impressive looking multiple woofer in a line going up. It impresses friends. It sells loudspeakers. An absolute staple in the hifi world. If you ask the average person on the street to draw you a picture of a hifi speaker I'll bet my speaker collection that they draw sketch out a tower on that piece of paper.
So lets discuss the dark art of multiple woofers. I am going to ask some questions along the way. For this we will focus on the 500hz and down end of the woofers. Just to simplify things a bit
1. Multiple woofers because the sensitivity of woofers is, in general, lower than the mids and the highs.
We have all padded down a tweeter. We have all experienced the loss of sensitivity because of those resistors. Everyone knows that more speakers equals more louder. I hear it is +3db. I have a question though. Lets take 2 woofers. Specifically, lets take two 8 ohm 6.5" woofers. Dayton RS150-8 for example. I probably wouldn't ever use these but its easy for me to sim this.
Lets look at a single woofer and lets say we are only going to use it from nill up to 500 hz.
We have about 89 db sensitivity there. Lowest impedence is 6.0 ohm.
Cool, we have a baseline. Now, lets add a second woofer. We will wire this one in paralell
Alright, now we have jumped up to about 95 db sensitivity. +6db. We are now down to 3 ohm
Let's now wire that woofer in series.
Alright, we are back where we started. 89 db except now we have a 13 ohm lowest impedence. So, what happened to all of sensitivity increase?
It went out with impedence is my guess. Then was ou6r previous +6db gain from the impedence as well? I do not know.
Do we only gain our +6 db when wire in paralell or are we getting +3db from the dual speakers and +3db from the impedence?
Lucky for us, there is a 4 ohm version of this driver. Lets compare a single 4 ohm and series 4 ohm.
91.7 db, 2.8 ohm lowest impedence Same in series, 91.7, 2.8 ohm
Now, we can decipher that adding another woofer in series doesn't make it play louder. However, two woofers with the same impedence as one woofer gives us a +3db sensitivity.
The two in parallel with a combined impedence of 4 ohm is about +3db louder than a single driver with an impedence of 4 ohm.
Is this how this works then?
2. Crossover values diminish with lower woofers
Lets take a look at inductor values
Lets cross at around 250 with an 8 ohm. 3 mh is needed. 43 uf cap. You see we are crossing 90db at 204 hz with these values
Now, same values with a 4 ohm driver. We see we would need higher values of components to cross as low as the 4 ohm driver. We are crossing 90 at around 300 hz now.
So, from this we need higher values with an 8 ohm driver. So what if we use two 8 ohm drivers in paralell.
95 db at 109 hz. Cool, this looks nicer. Look at all that low end gain and we get to use lower values for components
Alright, one more time but in series.
Booooo, well that stinks. Now we end up with a hump where we don't want it and it looks like we have actually lost high end extension
If we do try to use the low end here we would need huge components and we get no gain. Lose lose situation as far as I can tell.
Ok, so with this, I have concluded that there are senstivity benefits, low end extension benefits, and lower value component values if we use multiple woofer but ONLY if we use them in parallel. Am I on the right track here? I am not seeing why anyone would wire them in series other than to save their amplifier because it has an impedence limitation and all they have lying around are lower impedence drivers.
I am seeing how we get more low end extension with two drivers. At least in theory
3. More Cone area = More Bass
Now this I have heard all my life. However, when I start to run this through the ringer of physics, I am not seeing it play it out the way I would expect.
Lets toss a single 10" woofer, a single 12" woofer, and dual 10" woofers into WinISD.
5' box for the single 10, 6' for the single 12, 10' for the dual 10"
We have wayyyyyy more cone area with the dual 10 yet, basically no more bass extension even though we doubled the box size. Not only this but the single 12 still hits harder. I see this across the board with any driver I have ever tried to sim in whatever combination. It doesn't matter. Never does multiple smaller drivers hit as low as a single large cone. Never. I could sim drivers all night long (and I have) and not see this standard change.
So, is WINISD completely off base? I find that hard to believe because I have used it in conjunction with Hornresp and have had it spit out eerily accurate results in the world I reside.
Does more cone area just mean louder? It doesn't mean lower?
4. Floor Bounce and Diffraction
These two, in my mind, are combined. Far as I can tell, the floor and the diffraction are two metrics that combine to give you a low end response. The baffle step takes away, and, if done right, the floor can give back. Or the floor can take away. Or the floor can take away AND give back. Ahhhh, the wizardry begins.
I understand these the least. Or rather, I have done the least testing with these. I fully understand the physics of what is happening.
I am a visual person so lets do some examples. I just learned how to do this so let me know if I am doing this entirely wrong. Which I think I might be.
6" woofer at 20" heigh on a 10" baffle where you'd normally place a single woofer close to mid.
We lose about 6 db off low end to the baffle.
Then we gain some from the floor at 20".
Then we get the summed response and we end up with that stupid dip at 100 hz that I have experienced in a build and very much annoys me. Though we still gain some bass so thats cool.
Lets slap a second woofer on the floor. Right down below. I have seen some speakers designed like this and honestly do not understand how they get away with the phase issue unless they are crossing very low like 120hz or something.
Doesn't matter! We still get a stupid dip at 100 hz.
Questions:
Do I sum two woofers together in this simulator?
How do you avoid the 100 hz dip? Do you need to go up much higher from the floor? WMTMW style?
Is there any way to apply this to a simulated response just for design purposes and factor this in?
So lets discuss the dark art of multiple woofers. I am going to ask some questions along the way. For this we will focus on the 500hz and down end of the woofers. Just to simplify things a bit
1. Multiple woofers because the sensitivity of woofers is, in general, lower than the mids and the highs.
We have all padded down a tweeter. We have all experienced the loss of sensitivity because of those resistors. Everyone knows that more speakers equals more louder. I hear it is +3db. I have a question though. Lets take 2 woofers. Specifically, lets take two 8 ohm 6.5" woofers. Dayton RS150-8 for example. I probably wouldn't ever use these but its easy for me to sim this.
Lets look at a single woofer and lets say we are only going to use it from nill up to 500 hz.
We have about 89 db sensitivity there. Lowest impedence is 6.0 ohm.
Cool, we have a baseline. Now, lets add a second woofer. We will wire this one in paralell
Alright, now we have jumped up to about 95 db sensitivity. +6db. We are now down to 3 ohm
Let's now wire that woofer in series.
Alright, we are back where we started. 89 db except now we have a 13 ohm lowest impedence. So, what happened to all of sensitivity increase?
It went out with impedence is my guess. Then was ou6r previous +6db gain from the impedence as well? I do not know.
Do we only gain our +6 db when wire in paralell or are we getting +3db from the dual speakers and +3db from the impedence?
Lucky for us, there is a 4 ohm version of this driver. Lets compare a single 4 ohm and series 4 ohm.
91.7 db, 2.8 ohm lowest impedence Same in series, 91.7, 2.8 ohm
Now, we can decipher that adding another woofer in series doesn't make it play louder. However, two woofers with the same impedence as one woofer gives us a +3db sensitivity.
The two in parallel with a combined impedence of 4 ohm is about +3db louder than a single driver with an impedence of 4 ohm.
Is this how this works then?
2. Crossover values diminish with lower woofers
Lets take a look at inductor values
Lets cross at around 250 with an 8 ohm. 3 mh is needed. 43 uf cap. You see we are crossing 90db at 204 hz with these values
Now, same values with a 4 ohm driver. We see we would need higher values of components to cross as low as the 4 ohm driver. We are crossing 90 at around 300 hz now.
So, from this we need higher values with an 8 ohm driver. So what if we use two 8 ohm drivers in paralell.
95 db at 109 hz. Cool, this looks nicer. Look at all that low end gain and we get to use lower values for components
Alright, one more time but in series.
Booooo, well that stinks. Now we end up with a hump where we don't want it and it looks like we have actually lost high end extension
If we do try to use the low end here we would need huge components and we get no gain. Lose lose situation as far as I can tell.
Ok, so with this, I have concluded that there are senstivity benefits, low end extension benefits, and lower value component values if we use multiple woofer but ONLY if we use them in parallel. Am I on the right track here? I am not seeing why anyone would wire them in series other than to save their amplifier because it has an impedence limitation and all they have lying around are lower impedence drivers.
I am seeing how we get more low end extension with two drivers. At least in theory
3. More Cone area = More Bass
Now this I have heard all my life. However, when I start to run this through the ringer of physics, I am not seeing it play it out the way I would expect.
Lets toss a single 10" woofer, a single 12" woofer, and dual 10" woofers into WinISD.
5' box for the single 10, 6' for the single 12, 10' for the dual 10"
So, is WINISD completely off base? I find that hard to believe because I have used it in conjunction with Hornresp and have had it spit out eerily accurate results in the world I reside.
Does more cone area just mean louder? It doesn't mean lower?
4. Floor Bounce and Diffraction
These two, in my mind, are combined. Far as I can tell, the floor and the diffraction are two metrics that combine to give you a low end response. The baffle step takes away, and, if done right, the floor can give back. Or the floor can take away. Or the floor can take away AND give back. Ahhhh, the wizardry begins.
I understand these the least. Or rather, I have done the least testing with these. I fully understand the physics of what is happening.
I am a visual person so lets do some examples. I just learned how to do this so let me know if I am doing this entirely wrong. Which I think I might be.
6" woofer at 20" heigh on a 10" baffle where you'd normally place a single woofer close to mid.
We lose about 6 db off low end to the baffle.
Then we gain some from the floor at 20".
Then we get the summed response and we end up with that stupid dip at 100 hz that I have experienced in a build and very much annoys me. Though we still gain some bass so thats cool.
Lets slap a second woofer on the floor. Right down below. I have seen some speakers designed like this and honestly do not understand how they get away with the phase issue unless they are crossing very low like 120hz or something.
Doesn't matter! We still get a stupid dip at 100 hz.
Questions:
Do I sum two woofers together in this simulator?
How do you avoid the 100 hz dip? Do you need to go up much higher from the floor? WMTMW style?
Is there any way to apply this to a simulated response just for design purposes and factor this in?
Attachments
I can't answer all of your questions here but my last build was an attempt to remedy the "floor bounce" problem. The first step is to get the woofers closer to the floor. Yes, there are still problems and complications with this but it's very effective. My woofers are crossed to the mids at 200Hz. It works well. The calculations you're seeing here are useful, but remember you have to also calculate the walls behind and to the sides. It's not just the floor. So having the woofer be the same distance from the floor, the rear wall and/ or the side walls is the worst. You want them to be different distances from all of them.
The result in my highly reflective listening room was a much flatter low end with less drastic dips and problems. The two graphs here show a L and R measurement and the other graph is an average of the two all at the listening position. The average was quite smooth in the low end compared to any other speakers I've had in there.
I also think that the Mofi stuff haws the woofers strategically spaced vertically for just this reason and Andrew Jones did some "mathing" to get the best spacing.
The result in my highly reflective listening room was a much flatter low end with less drastic dips and problems. The two graphs here show a L and R measurement and the other graph is an average of the two all at the listening position. The average was quite smooth in the low end compared to any other speakers I've had in there.
I also think that the Mofi stuff haws the woofers strategically spaced vertically for just this reason and Andrew Jones did some "mathing" to get the best spacing.
Attachments
Rather series or parallel you have added doubled the Xmax
So less distortion, even series you should have more, so max spl is higher
Or at normal listening levels, distortion is lower.
Also forgot 1 thing.
Data Sheet Frd = Half Space
It assumes or is a infinite baffle. To be more specific is a infinite baffle with no edge diffraction
down to 100 Hz is wha IEC standard measurements use. Most data sheets will actually merge measurements 450 Hz on down.
So the measurement is considered half space. Above 100 Hz IEC baffle is considered ideal or is not difficult to get infinite baffle measurements
with smaller wavelengths or higher frequency, and infinite baffle will be a STRAIGHT line for accuracy of frequency response. No diffraction
or losses.
With infinite baffle considered forward radiation only , no diffraction losses.
In real life below 250 Hz full radiation, non directional hence = full space losses.
Not only are bass waves longer and need more energy, they also loose another 6 dB to full space radiation.
So long story short non of the sims included full space losses.
Which is up to 6 dB. Also people swear up and down that skinny baffles are magical= they are not.
So additional losses and diffraction will be included simulating on a baffle in full space.
So you loose 6 dB to full space and also create a big peak from a small baffle which all gets= removed
with the crossover.
So adding a second Woofer will gain back 3 dB of the 6 dB you loose to full space and magical skinny baffles.
Otherwise instead of insisting that speakers need to be " pulled away from the wall" ignore that
and have something closer to a ideal infinite baffle. Aka mount too the wall.
Or get even more bass if you corner load, which has less losses than Half space.
Simulate in full space, on a baffle. With the second woofer, the crossover can be set at the step in the response created by the baffle.
Hench baffle step compensation can be used. At least 3 dB of baffle step response compensation added to the 6 dB of losses.
Impedance wont be 4 ohms with 2x 8 ohm woofers, be closer to 6 ohms even 8 above the baffle step. Since the second woofer is crossed
lower
The simulation shown is Half space measurements on a IEC baffle or 1650 x 1350 with the driver offset for flatest response.
It also has merging 450 Hz down to be considered infinite. The measurement will have no edge diffraction down to 100 Hz.
Also second woofer has no Y coordinates so any phase cancellation from center to center spacing is not shown.
The sim assumes 2 woofer in infinity mounted directly on top of each other, all of which is impossible.
So less distortion, even series you should have more, so max spl is higher
Or at normal listening levels, distortion is lower.
Also forgot 1 thing.
Data Sheet Frd = Half Space
It assumes or is a infinite baffle. To be more specific is a infinite baffle with no edge diffraction
down to 100 Hz is wha IEC standard measurements use. Most data sheets will actually merge measurements 450 Hz on down.
So the measurement is considered half space. Above 100 Hz IEC baffle is considered ideal or is not difficult to get infinite baffle measurements
with smaller wavelengths or higher frequency, and infinite baffle will be a STRAIGHT line for accuracy of frequency response. No diffraction
or losses.
With infinite baffle considered forward radiation only , no diffraction losses.
In real life below 250 Hz full radiation, non directional hence = full space losses.
Not only are bass waves longer and need more energy, they also loose another 6 dB to full space radiation.
So long story short non of the sims included full space losses.
Which is up to 6 dB. Also people swear up and down that skinny baffles are magical= they are not.
So additional losses and diffraction will be included simulating on a baffle in full space.
So you loose 6 dB to full space and also create a big peak from a small baffle which all gets= removed
with the crossover.
So adding a second Woofer will gain back 3 dB of the 6 dB you loose to full space and magical skinny baffles.
Otherwise instead of insisting that speakers need to be " pulled away from the wall" ignore that
and have something closer to a ideal infinite baffle. Aka mount too the wall.
Or get even more bass if you corner load, which has less losses than Half space.
Simulate in full space, on a baffle. With the second woofer, the crossover can be set at the step in the response created by the baffle.
Hench baffle step compensation can be used. At least 3 dB of baffle step response compensation added to the 6 dB of losses.
Impedance wont be 4 ohms with 2x 8 ohm woofers, be closer to 6 ohms even 8 above the baffle step. Since the second woofer is crossed
lower
The simulation shown is Half space measurements on a IEC baffle or 1650 x 1350 with the driver offset for flatest response.
It also has merging 450 Hz down to be considered infinite. The measurement will have no edge diffraction down to 100 Hz.
Also second woofer has no Y coordinates so any phase cancellation from center to center spacing is not shown.
The sim assumes 2 woofer in infinity mounted directly on top of each other, all of which is impossible.
Last edited:
Absolutely correct. As cone size increases, its acoustic impedance match to the surround air's impedance becomes more favourable, and the efficiency of mechanical movement of the cone to acoustic output improves.
However, no home systems will ever have an acoustically 'large' radiating area.
Last edited:
You forgot a very important factor: marketing trends.
Big large boxes are out. Tall sliim towers that can fit on either side of your HUGE wall mounted TV are in.
Big large boxes probably weigh a lot. A tall slim tower can be lighter. This makes shipping less expensive.
A big large box with one large woofer can be replaced by 4 smaller (e.g. 8" woofers) and to arrange these a vertical array works well, e.g. as opposed to a 2x2 grid in the same large box.
I think these type of pressures on loudspeaker manufacturers is at least as important as the performance aspects that you have listed, all of which are correct.
IMO I would call a monkey coffin type speaker "classic" and the slim tower is a more recent development.
Big large boxes are out. Tall sliim towers that can fit on either side of your HUGE wall mounted TV are in.
Big large boxes probably weigh a lot. A tall slim tower can be lighter. This makes shipping less expensive.
A big large box with one large woofer can be replaced by 4 smaller (e.g. 8" woofers) and to arrange these a vertical array works well, e.g. as opposed to a 2x2 grid in the same large box.
I think these type of pressures on loudspeaker manufacturers is at least as important as the performance aspects that you have listed, all of which are correct.
IMO I would call a monkey coffin type speaker "classic" and the slim tower is a more recent development.
Yes, with two drivers in parallel, you get +3dB from the doubled cone area, and +3dB from the doubled power draw. So spl is 6dB up. But if you put them in series, each driver is only getting half the drive. Compared to a single, you are gaining 3dB from the doubled cone area, but losing 3dB from the reduced power draw, so same spl.
I have that. My wall is 1" plaster over wood lathe, low tech 1936. Gives me a watt free bass boost. Speakers backed to wall, datasheet - 3db to 54 hz, - 10 db to 40 hz, actually -4 db at 40 hz. So I don't feel the need of a sub even though I used to play a lot of pipe organ records.
Want more bass, lay more masonry at the speaker wall. Less elsewhere.
As for dual woofers, I had that once but have settled on 15" woofer 54-1800 hz. Pinched back case resolves the beaming issue, - 6 db over +- 45 deg from straight ahead. With books & furniture breaking up standing waves, I can walk around the room without sound changing much.
I had dual 10" woofers, Peavey 1210's. Triple Motorola piezos on top pointed in 3 directions. One 1210 had original 8 ohm impedance, other had parallel 8 ohm woofers instead of 8 ohm total. At about 10 watts, 4 ohm one burned a pair of MJ15003 transistors after a 3.5 hour choir rehearsal. Dual 8 ohm woofers, buy parallel output transistors amp, or class D.
Last edited:
It depends what you mean. The lower the frequency, the more cone displacement you need to achieve the same SPL. So more cone area = more bass capability.
And more cone area will be more efficient. But strictly speaking it gives you more output across the board, rather than just more bass.
Increasing cone area by adding multiple drivers doesn't automatically change where the LF roll-off happens.
But if you wanted, you could exploit the increased bass capability, and adjust the crossover to trade some of the extra efficiency for greater bass extension.
(PS It's possible to have very good bass extension from very small drivers - but they will be very limited in SPL capability. This is precisely what headphone drivers do.)
Last edited:
One problem I didn't mention is that having 2x 8 ohm woofers in parallel will give you problematic impedance response. Many of these woofers dip well below 8 ohms and with a passive crossover will end up getting close to 2 ohms in some areas, especially where it overlaps with a 4 ohm mid range. If I was going to do it again, I would use two 4 ohm woofers in series along with an 8 ohm mid.
Perhaps, if you consider 40 years ago as more recent. By the mid 1980's, tower shaped speakers with dual woofers were becoming more common. I remember seeing them from Infinity, A/D/S, Dali, Canton, PSB, Paradigm, and others. Here is a Boston Acoustics T-1000 from about 1987.
Tower speakers with twin woofers are common because the basic layout works so well. The tweeter is at ear level without the need for a stand. Two woofers spaced vertically puts one of them near the floor, and spreads out the woofer-to-floor spacing which helps reduce the floor bounce effect. Two woofers double the sensitivity in the bass region, which makes the baffle step easier to deal with, and it allows the mid+tweeter to operate with less padding. The net result is the speaker's overall sensitivity is higher, usually by 4 - 6 dB higher. And yes, they consume less floor space.
Just so we are all on the same page, here is a synopsis of how multiple drivers sum together in terms of impedance, voltage sensitivity, efficiency, and maximum SPL capability
j.
I understand all of these limitations of the simulation I ran for these examples.
I do not, however, know how to simulate all of these things. What program are you using? Are you willing to show me how to do this?
The important factor here I don't think is marketing
The important factor here "wife acceptance factor"
I am usually allowed a certain footprint of space in the house for speakers. The garage is my domain but our small house in Chicago has limited space. I think a lot of people have these same limitations. So yeah, if I can get extension of a 15" woofer with 3x 6.5"s in a line, I'm doing that.
Got it, so no real point to series unless you're just trying to stay within Xmax of the woofer in question.
Now this makes more sense to me. I saw that in my little experiments. It seemed, though, that ultimate low extension was off the table.
I'm not seeing a bunch of 6" woofer hitting down to 11hz flat and accurate like you get with a set of 12s in a big ported box. You could add so many that it boosts the low end but you'd have to roll off the rest of the bass to keep it accurate. Not sure if that made sense
I try to stay above 2.8 ohm minimum. I have half a mind to buy one of those big chinese clone blocks that are just fine going down to 1 ohm simply for more freedom designing my next main living room system. But yeah, I try to make designs that can be run by your average, off the shelf, class D amplifier
Per
I found Boxsim easier to figure out how to use than Vitruix.
I use Boxsim. From Visaton, but can be used with any drivers. It is easy to upload the FRD/ZMA files for a given driver. Then you can select how it was measured (off of a spec sheet/DIN baffle, or driver measured in the box). Once you model your speaker shape and driver placement on the baffle, it will give you the graphs adjusted for baffle step loss.
I found Boxsim easier to figure out how to use than Vitruix.
Just keep in mind, that when the voltage remains the same, and impedance is halved. Then the current demand is increased by 100%.
And as you touched on already, amplifier current output capability might be just as limiting as voltage swing.
And the general change over to use of switch mode power supplies has some backsides.
Just not efficiency/size/weight and price of manufacturing.
Xmax is a driver's linear excursion, which is identical whether the woofers are wired series or parallel.
Doubling excursion requires a +6dB more power to achieve +6dB SPL, disregarding voice coil heating raising it's impedance, which will happen four times faster with one driver producing the same SPL as two.
As you know, a different impedance changes the frequency and slope of a passive crossover using the same components.
Right track for the coil, but wrong track for the capacitor.
Let's consider two "8 ohm" drivers, series or parallel, same crossover slope (LR12) and frequency (~250Hz).
16 ohm uses 20mH inductor, 20uF capacitor.
4 ohm uses 5mH inductor, 80uf capacitor.
Inductor is 1/4 the value, capacitor 4x the value for the higher impedance version.
Large value coils with low DCR require so much more material that cost tends to discourage their use in high impedance low frequency passive crossovers, while the (cheaper) bypass capacitor of only 1/4 the value may save very little cost by comparison.
Well, just like when they're in parallel, a pair in series means that each driver is doing half the work - so for a given SPL you have less heating, less excursion, less distortion.