Arpeggio Music & Sound


Shopping Basket Home Text Menu Glossary Contact

Speaker Design System Tutorial

Following is a step-by-step example of how to use this site using an actual speaker system that we designed and build here. It is a high-quality home theater 5.1 (to 7.1) system based upon the popular VIFA P17/D25 pairing for the surround and stereo channels, a 3-way center channel from the VIFA MG-series, and a Peerless 12" push-pull subwoofer. It will be easiest if you print this web page and follow the printed instructions (rather than attempt to alternate between windows on the computer screen).

We don't pretend that this is the "best" combination of drivers to use, or the "best" possible crossovers. There are as many opinions on such matters as their are speaker designers and builders. However, the tutorial demonstrates the capabilities of this design tool for assisting in the design and construction of simple to complex speakers, regardless of the crossovers or drivers selected.

NOTE: This site is under constant improvement/development. It may behave slightly differently than described in the tutorial.

General Project Structure

A 'project' consists of one or more speaker enclosures. Each enclosure can contain one or more drivers, plus the circuitry to support the drivers. To design a speaker project:

  1. Initiate the Project / Add an enclosure
  2. Add drivers to the enclosure
  3. Design the supporting circuitry
  4. Adjust the enclosure dimensions and style to meet the drivers' parameters and desired output
  5. Repeat the above steps for each enclosure in the project
  6. Save the project to your computer at appropriate intervals
  7. Order the parts
  8. Construct and test the crossovers
  9. Construct and finish the enclosures
  10. Install the drivers and electronics
  11. Connect to your amplifier and enjoy

For a simple, single speaker project, the entire hands-on time might be a few hours. A complex project may take a full week or more. Our project will consist of four different kinds of enclosures, some of which will be duplicated. There will be five different kinds of drivers, and a passive crossover in every enclosure except for the subwoofer, which will have its own amplifier.


Enclosure 1
Stereo L/R (x2)

Driver 1
Vifa P17 (x2)

Driver 2
Vifa D25

Enclosure 2

Driver 1
Peerless 315SWR

Enclosure 3
Surrounds (x2)

Driver 1
Vifa P17

Driver 2
Vifa D25

Enclosure 4

Driver 1
Vifa MG18 (x2)

Driver 2
Vifa MG10 (x2)

Driver 3
Vifa D25

You'll learn the most about this system by manually going through each of the steps, below. However, you can load the finished project by clicking here.

As the design progresses, menus for each of your enclosures and drivers will appear in the upper left of any screen (they will appear as they are created and added). Additional commands affecting the entire project -- save, upload, generate parts list, etc. -- appear only in the Speaker Design menu.

Adding an Enclosure / Initiate the Project

You can start a project or add to an existing one by selecting 'Add Enclosure' from the Speaker Design menu. The new enclosure's data screen will appear. Give the enclosure a name -- for example, 'Stereo L/R'. You can enter a brief text description, if you wish. We will make two of these, so enter 2 for the quantity. Enter the power that the amplifier will supply to this channel; ours supplies 85 W per channel. It is important to enter the channel power. It will be used to suggest the power rating of resistors in your circuitry. Underestimating the power may result in damaged components. For example, if your amplifier supplied up to 200 or 300 Watts per channel (and the drivers could handle it), the crossover circuit will require higher power resistors than the one we are designing for an 85 W amplifier.

There are many different kinds of speaker boxes. This system can assist with designs of the two most common models; sealed or ported. More complex boxes are (presently) beyond this site's capabilities. If you already know what kind of enclosure you wish to make, sealed or ported, you can select that now, as well as the enclosure's dimensions. But we will add the drivers first and see what the system suggests. You can leave all other figures blank or at their defaults for now.

Click the Save Changes button. A new menu box appears at the top of the screen with the name of your enclosure. You'll use that menu to edit the enclosure or add drivers from this point forward.

IMPORTANT NOTE: None of your changes will ever be saved unless you also click the save changes button. Clicking any other link will discard any changes you've made and go to the new link-address without warning.

Adding a Driver

Select the Add Driver option from your new enclosure's menu at the top of the screen. The driver data screen will appear. Enter a name for the driver -- usually its part number or something similar -- and the driver's parameters; not every parameter is critical, especially if you will not be wiring a crossover, but the following are quite important as initial entries for this system:

The more variables you fill in, the more useful and reliable the system's recommendations will be. Failing to fill in critical data or mis-entering information may result in less-than-useful results, or even damaged components. Where do you get these values? The manufacturers provide the data on their web sites, sometimes with the individual drivers. The variables can (usually) also be provided by the reseller from whom you purchased your drivers. Or, they can be measured directly if you have the proper equipment. We directly measure most of the variables for our own entries, and supplement our measurements with the data from the manufacturer. We entered:

Driver NameP17SJ-00-08
Nominal Impedance8
Voice Coil Resistance5.76
Voice Coil Inductance0.64
Nominal Power20
Short Term Max Power350
Long Term Max Power150
Operating Power8
Free Air Resonance41
Low end F335
High end F35000
Diaphragm Area136
Moving Mass12
Force Factor6
Total Q0.35
Effective Displacement.43
Face Diameter17
Quantity in this Enclosure2
Wiring Method (if multiple)Parallel

The remainder can be left blank for now. Click the Save Changes button. If you look at the enclosure's pull-down menu, you'll see that your new driver now appears in the list.

Add the Tweeter

Add another (single) driver to this using the same method as before. Call it D25ASG. Enter the parameters for the driver. We entered:

Driver NameD25ASG05-06
Nominal Impedance6
Voice Coil Resistance4.9
Free Air Resonance1500
High end F320000
Effective Displacement.1
Face Diameter10.4
Quantity in this Enclosure1

The rest can be left blank or at their default values. Save your entries.

Tweak the Enclosure

To return to an enclosure's data screen select the Edit Enclosure option from the enclosure's menu. Depending on how many driver parameters were entered, the screen may now make general recommendations about enclosure style (ported or sealed), dimensions, etc.

Baffle Diffraction

Baffle diffraction is a complete topic in itself, and is one of several phrases referring to a reduction in low frequency sound pressure being projected towards the listener is a function of the speaker face (or baffle) size. It is much more complex than it seems at first, and attempted solutions raise as many questions as are answered. We will discuss it only insofar as it is used in this system, and avoid the more subtle theories for now. For the sake of simplicity, the baffle effect frequency range is determined solely by the minimum face dimension of the enclosure. To determine the approximate components to use, enter the nominal impedance of the drivers to which compensation circuit will be connected, and enter the decibels of correction desired. For free standing speakers several feet from any nearby wall, this might be as much as 6. For speakers nearer a wall, 2, 3 or 4 is a more common entry. If you enter 0 for either of these figures, the baffle compensation calculations will be skipped and no related components will be included in your projects parts list. If you choose to include a baffle compensation circuit, there are other important matters to address.

First, you must determine which drivers will be connected to the baffle circuit, and which will connect directly to the signal line. Any drivers with an output range that ventures into the baffle step range (and that will not be cut off by a filter) usually connect to the baffle circuit. If multiple drivers are connecting in parallel, the Zb will need to be calculated using parallel impedance calculations (two 8 ohm drivers connecting in parallel will have 4, not 8 or 16, ohms impedance). It may also be necessary to add a zobel circuit to every diffraction-compensated driver, even if it is not needed for other reasons. Next, when you enter the driver data, the DC resistance of the baffle coil should be added to the DC resistance entry for series coils; a small amount of iteration may ensue. Finally, the power rating of the baffle circuit resistor Rb is determined by integrating from Fb3 to 20000Hz to determine the power seen by Rb. It then distributes that power between Rb and Zb, and rounds up to the nearest typical power rating. In some circumstances you may need a higher power resistor. If there is not a stock inductor available at the projected value, the general consensus is to choose the next larger one. For example, if the system suggests a 2.4mH inductor, and the nearby values are 2.2 and 2.7, most builders appear to recommend using the 2.7.

The formulas we use to determine the baffle diffraction compensation components are a mix of 'status quo' formulas and our own developments. The baffle step midpoint is the common Fbs = 4560 / Baffle Width [in]. The resistor value is derived from n = 20 log10((Rbs + Zn) / Zn), the result being Rbs = Zn x (10^(n/20) - 1), where n is the number of decibels correction (as also mentioned in 'Simple Sizing of the Components in a Baffle Step Correction Circuit' by Martin King). The advantage of this over other 'stock' formulas is that you can adjust the desired compensation amount, where the more common Rbs = Zn (or Re) assumes you are attempting a 6dB correction. Our derivation for L is much more complex -- and we think more accurate -- than we have seen elsewhere:

Lb = (1/(2*pi*fb3))/((1/(Zn * (10^(n/40) - 1))) - (1/Rb))

The formula assumes, and this is an assumption with a capital A, that the step midpoint is taking place at the generally accepted frequency (4560/W). This, of course, may not actually be the case. Barring the availability of equalization, it may be prudent to insert a 5W potentiometer in the place of R, and some inductor switching may be called for.

Try changing some of the variables to see how they affect the enclosure projections. Width, height and depth refer to the inside dimensions when looking at the FRONT of the enclosure. If you enter box dimensions and material width, a rough drawing of the box will be displayed below the data entry area. In addition, information about sound diffraction will appear showing the low, mid and top frequencies it is likely to affect.

We'll be placing some bracing in the enclosure, so let's assume for now that it will take up a liter of space, too. Enter 1 in the "Other internal displacement" box. The system should be recommending a ported enclosure with about 48-49 liters of free air and a 3" diameter port, a little over 5 1/2 inches long. It will also recommend some internal dimensions, and states that the box's resonant frequency will be about 44Hz. There is nothing necessarily "magic" about the suggested enclosure dimensions. They are merely a product of taking the air volume recommended by the driver parameters and arranging the width, height and depth according to the 'golden rectangle,' which is a mathematical ratio popular in artistic and architectural circles. It is intended as a starting point. Go ahead and copy the dimensional information exactly as it appears to the equivalent entry fields:

Interior X9
Interior Y14.5
Interior Z23.5
Wall Thickness.75
Port X-Y Location4.5,3.5
Port Inner Diameter3
Port Outer Diameter3.5
Port Length5.68

The system now shows, at the top of the projections/recommendations window, a lower projected box resonance. This is due to differences in the formulas that the site uses to predict ported enclosure resonances. In 'real life,' the differences may be negligible (relative to other issues), but the top prediction is physically, acoustically (mathematically) more rigorous. The good news is that this means we can probably reduce the box's dimensions and still get a good low-end response. For our project we settled on 7 x 24 x 10.5 as the interior dimensions, and a 6 inch long port.

The site handles two basic assembly materials, and adjusts the drafting accordingly; simple MDF and mid-difficulty hardwoods. If you select MDF as the material, a simple construction layout will display. MDF box construction requires little more than good glue and clamps, though screws or nails can help, too. If you can walk and chew gum at the same time, you can probably make speakers with MDF. Add some veneer and stain or quality paint and polish, and you can have a nice looking speaker. Hardwood is a little more difficult. The hardwood design assumes that you have the tooling and skills necessary to cut dados and rabbets, and that the box will be made with solid hardwood casing and inset plywood panels (inset roughly 1/4 inch). Whichever construction material you choose, the box design is, like the box dimensions, merely a suggestion. It is easy to adapt the hardwood box design, for example, for dovetail joints, or even simple butt-joints.


Advanced speaker designers may want to estimate in advance the spacial sound coverage ("lobing"). To do this with any hope of accuracy, one has to enter the driver location, phase shift, and other related variables. An important variable in this is the z-offset. This variable attempts to quantify the difference in the effective source of sound generation in different speakers. A larger, deep speaker may 'generate' its sound from a point an inch or more in from the speaker surround. A smaller speaker may appear to generate its sound very near its surface. At low frequencies this does not matter, but let us imagine an unlikely scenario where we have a crossover at 6750 Hz between a woofer with a 1" offset and a tweeter with a 0" offset. For the sake of simplicity, let's also assume no phase shift between the two speakers. At 6750 Hz, the wavelength is ~2" long. The wave produced by the woofer will be reaching the plane of the tweeter just as the second half of the wave is being produced. Both will combine some feet away to cancel each other. The result will be an audible loss of volume around 6750 (and multiples thereof). It is better to know this in advance and adjust, if possible, than to design and build the whole speaker only to find that it sounds awful. It takes specialized equipment to accurately arrive at these figures, but they can drastically affect sound lobing and cancellation (and this is the only calculation affected by the entry). Barring the purchase of the necessary equipment, your drivers could be sent to a specialist to make the measurements, or some people estimate the offset by measuring the distance from the speaker's surface/mounting plane back to where the cone connects to the spider. This isn't precise, but may be better than omitting the measurement altogether and relying on blind luck to get it right. A small cancellation at a high frequency might not be noticed or matter, but the z-offset and phase differences can combine such that, the greater the difference, the lower the frequency at which there may be an audible problem.

The entry should be made relative to a common reference for all drivers in the same enclosure. For example, we are treating our D25 tweeter as the zero point for z (assuming that the sound is primarily generated at the surface of the dome at the face of the tweeter), and estimating that the sound generation is originating about one inch in from the speaker face. So for the D25 we enter a z-offset of 0, and for the P17 a z-offset of -1. It can be entered on the driver data screen along with x and y in the format x,y,z, but it is usually best made from the enclosure data screen. If sound lobing and band-cancellation are not concerns, you can simply leave the z-offset (and the phase, for that matter) at 0. It is worth noting, however, that with 3rd order filters, almost any relative offset up to 2" total is possible and, with a properly-aligned MTM configuration, you'll still get good sound coverage -- in other words, it takes some effort to mess up an MTM design.

Enter the x-y location for each of the drivers and the port. We entered:

P17 (1)4.520.5
P17 (2)4.59.5

Adjust the enclosure's dimensions to achieve a projected box resonance of 44 or fewer Hz. To summarize, we decided to make this a ported enclosure, using a 5 inch length of 3 inch inner diameter ABS pipe. The walls of our enclosure are .75 inches thick, and the enclosure's inner dimensions are 9 x 24 x 10.5 inches. We are guessing that our interior bracing will take up about one liter of space. This enclosure is now roughed out. (In the actual construction of this project, we made this enclosure a little narrower, 7 x 24 x 10.5, and increased the port length to 6 inches. Doing so increased the box resonance to about 49Hz, which is still okay, since we are adding a subwoofer, anyway).

Save any changes.

Design the Crossover Circuitry

Go back to the data screens for each of the drivers, beginning with the P17. In the recommendations box it displays information about the relationship between the displayed driver and any others in the enclosure, specifically, the ranges over which their frequencies overlap. It shows the midpoint for that range, and, if the driver face sizes have been entered, shows the maximum frequency at which the drivers can be crossed and still spaced one wavelength apart (what some consider to be an ideal situation). Pick a point in that range -- usually near the middle -- as the crossover frequency. We started out by selecting 2500 because we wanted this to be an MTM box with the drivers spaced one wavelength apart. Select 3 for the order, Low Pass as the type, and CPC as the style.

At the bottom of the page it will already show a suggested crossover circuit with 'ideal' components. But we have some issues to iron out. The D25 is a bit "hotter" -- more power-efficient -- than the P17; it has an SPL value 91, while the P17 has 87dB; at the same power input level, a single D25 will sound significantly louder than a single P17. Under simpler circumstances, with just one D25 and one P17 in a box, we would add a resistor to the D25 circuit to absorb some of its power (we will actually do just this with the surround units, later). But we have not one, but two P17s in this enclosure. Combined, they have an effective SPL of 93dB, 2 more than the D25, and we need to reduce it so that it doesn't overpower the D25. Enter 91dB as the target SPL at the bottom of the P17 screen. This will prompt the system to add an attenuation resistor to the P17 crossover circuit, Rs, of a little over 1 ohm. The addition of this resistor causes the system to change the calculated values of the inductors and capacitor, too. Those values were originally computed in relation to the resistance that they 'see' downline. Adding the attenuation resistor increased that resistance, so the other components change, as well.

What happens if we do not match SPLs between the drivers in an enclosure? Some frequency bands may be annoyingly louder than others. A great example of this is a television (like the $600 one we recently purchased) with multiple, apparently-unmatched SPL speakers. Higher frequencies are much louder than lower ones. As a consequence, it is often difficult (or tiring, or just plain annoying) to try to hear conversations in shows and movies while there are other background noises on the track. The volume seems to get so loud we have to turn it down, even though we can't hear what the actors are saying, and later so quiet we have to turn it back up not because of a real change in the volume of the soundtrack, but because the primary frequencies being produced changed. If mismatched the other way, the sound will seem dead, distant, and have a lack of detail. Some makers use a specially-designed adjustable resistor (potentiometer) so that they can manually fine-tune the balance after construction.

Another method of SPL matching is an L-pad; a pair of two resistors, one in series and one in parallel with the drivers. The effect of an L-pad is to keep the resistance as seen by the filter the same as Zn while attenuating SPL. This allows a designer to reduce the SPL without having to recalculate inductor and capacitor values. Because inductors and capacitors are generally more expensive than resistors, and manual calculations are error-prone, L-pad designs could save some money and time. But this is usually not necessary, now, for two reasons. First, the computer automatically and easily recalculates the component values before construction. Second, if mid- and lower-end amplifiers need anything, it is usually increased impedance in the crossover circuit. By using a single resistor and recalculating the inductors and capacitors, we achieve the desired SPL-matching result, but also keep the overal circuit impedance higher than if we used an L-pad, which is especially desirable for some situations (and we eliminate an unnecessary component from the circuit).

But there is still a related and often-overlooked issue. On the P17 circuit there are two inductors in series with the speakers. In addition to their frequency-based impedance, they have a DC resistance due to the sheer amount of wire they contain. We measured ours with an ohm-meter and found a total DC resistance of .8 ohms (this data is also available direct from the manufacturers and suppliers, so you don't need to wait until you receive your inductors to enter this figure). Not a lot, but it will make a difference. This resistance will convert power to heat, reducing the SPL. The site 'understands' this and will automatically reduce the resistance of the attenuating resistor we added a couple steps ago to a .24 ohms when you enter .8 as the resistance of series inductors.

How much effect does this small a resistor really have? Stated simply, the total power entering the circuit will get divided between the various loads; if the driver is the only effective load, it will get all of the power. If there is another resistor (or resistor and inductor) in series with the driver that is, say, 10% of the total circuit resistance, it will absorb 10% of the power and the remaining 90% will go to the driver. If we clear out our entry of 91 in the Desired Net SPL blank, the estimated output SPL increases to 91.4. The .24 ohm resistor is attenuating the output of the two, parallel P17s by about half a decibel. A machine might be able to measure the difference, but it is unlikely that a typical human being could discern it (most people cannot discern audio power differences under 1dB). This is a small-enough amount (not to mention that it is entirely theoretical and will be vastly overshadowed by other factors) that one could probably just eliminate it from the design altogether, but do as you wish.

As long as we're on the resistor, this is a good moment to consider the resistor power ratings. The site considers (1) the power being supplied by the amplifier to the enclosure, (2) the percentage of that power that will be allowed through the filter, and (3) the percentage of that power that will be absorbed (converted to heat) by the resistor. It then recommends a power rating above that projected maximum power amount; usually 5, 10 or 12 watts. In almost every case, a 5 watt resistor is sufficient. The two notable exceptions are high power systems and resistors found in many woofer circuits. In general, such resistors have to dissipate more heat and so end up needing to be 10 or 12 watt resistors.

Frequency/Impedance Graph for P17SJ-00-08 from Vifa

Frequency chart for Vifa P17SJ

Impedance Problems: Our inductor/capacitor filter is designed with the assumption that the load/driver impedance will remain constant. But it does not. At Fs -- the driver's resonant frequency -- and in the upper half of most drivers' frequency ranges the impedance rises sharply, sometimes as much as three to four times the driver's nominal impedance. This has serious consequences for crossovers. Inductors and capacitors don't behave just like a resistor -- they don't absorb power and change it to heat -- but they have similar effects. Ideally, they simply reject power in proportion to frequency -- capacitors tend to reject lower frequencies, and inductors higher and allow the remainder to pass. The net effect is similar to the resistor scenario, above, but where the resistor absorbs 10% of the input power, the crossover components might simply reject 10% and allow the remaining 90% through at a specific frequency. In the typical CPC-style crossover, the crossover component values are selected so that, at a specific frequency, the crossover is rejecting and dumping about half of the input power, and allowing the remaining half through to the remainder of the circuit. Components are therefore selected so that their reactance is equal to the driver's impedance at the crossover frequency. If the driver's impedance changes significantly, it can cause our crossover frequency to shift in undesired ways. For example, the P17 has a 'nominal' impedance of 8 ohms; we therefore want our crossover components to have a similar total reactance at our crossover frequency. But at our crossover frequency, the manufacturer's specifications graph shows that the P17's effective impedance is actually twice that, 16 ohms. If we install crossover components expecting an 8 ohm impedance at 2500 Hz, the results will be less than spectacular. The crossover won't be rejecting half, but only about one third, of the power. In other words, the low pass filter will be less effective and will shift up in frequency. This could be a Very Bad Thing. We have a couple options; we can select our crossover inductors and capacitors in relation to the system impedance at the crossover point (but it is constantly changing, so that poses real problems), or we can use the nominal impedance as our benchmark and add circuitry that forces the apparent impedance to stay at that value. This second method is the most common. Click the Impedence Compensation checkbox and the system will add the calculated resistor and capacitor to the circuit, Cz of 32uF and Rz of 3.6 ohms. When placed in parallel with the driver, this will help keep the impedance constant. This is also known as a 'Zobel' circuit.

Also in the circuit area is a note about the driver's phase. This is an estimation related to the components of the selected circuit type and how they will affect the sound signal. No phase shift will actually be assigned to your driver unless you personally enter a shift (in degrees) in the driver data area. The phase shift is used when the system estimates the sound lobes at specific frequencies on the enclosure data screen. It has no effect on any other system calculations. A related matter is the speaker polarity wiring. By clicking the reverse wiring checkbox, this will automatically add 180 degrees to the phase shift for the driver, a common technique for even-order crossovers. A final, related element is the acoustic location entry. This can be entered and changed on the driver data screen, but it is easiest to enter it on the enclosure screen, where you can also see where the driver will appear on the enclosure's fase.

The P17 side of the crossover circuit is now ready to go. Save any changes. Go ahead and review the D25 side and add a high pass 3rd order CPC crossover at 2500 Hz. Save the changes.

Frequency/Impedance Graph for D25AG from Vifa

Frequency chart for Vifa D25AG

Some designers might consider adding a notch filter at Fs for the D25 because of the proximity of Fs to Fxo, especially if a 2nd or 1st order crossover were being used. Similar to how the increasing-with-frequency impedance affected the P17 crossover, the Fs resonance impedance might affect the D25 filter. Assuming Zn of 6 ohms, at 1500 Hz the D25 filter will be sloughing 93% of the power, passing 7% to the driver. Because the driver impedance rises to ~10 ohms around 1500 Hz, this reduces the relative effect of the crossover to about 89%, increasing the power to the driver at the worst possible point. At resonance, this could make a difference, as it takes very little power to get the diaphragm moving. Ideally, the crossover frequency will be far-enough away from Fs, or the attenuation steep enough, that the resonance impedance is not a factor. If you want to add a notch filter that balances the impedance at these frequencies (much like the Zobel circuit, above), just click the notch filter box and the necessary components will be inserted into the schematic (as long as Qes, Rvc, Qms and Fs have been entered). Because the manufacturer did not provide Qes and Qms specs for this driver, we would have to determine the values impirically, which is another tutorial in itself.

Reality Check: The above system assumes that you will be able to find stock parts close enough to the ideal values to keep you satisfied. But this is not always the case. Capacitors can be easily combined in parallel, and resistors in series. Inductors can be combined, too, but are problematic due to their bulk, magnetic coupling, and other factors. Consequently, the recommendations box approximates other resistance values that could be added to the system load to require stock inductor values for the same Fxo. It also suggests nearby (and some not-so-nearby) crossover frequencies that might result in stock inductor values. For 1st and 2nd order crossovers, there may be many suggestions. For 3rd order low pass, there will be fewer (or none) because it is showing only those recommendations for which there are stock inductors available for both L1 and L2. Now, none of these figures is going to be perfect. But by gradually tweaking your crossover frequency and desired SPL (decreasing the desired SPL increases the circuit loading Rs, which affects the inductor and capacitor selections), you can usually arrive at projected component values that allow a selection of stock inductors. For example, in the 'real-world' build of the the above speaker, we changed our target crossover to 2520, which resulted in stock inductor recommendations of 0.47mH and 0.15mH. This is an almost trivial change. Had we not knowingly entered the different Fxo, it almost wouldn't have mattered, because we would have had to purchase the nearest components anyway. At least this way we know that, on the P17 side, the crossover frequency probably is moving up a little. For the same reason we also increased the D25 Fxo to 2650. In theory we now have a small spread between the crossover frequencies. Having little or no spread in CPC-style crossovers usually results in a little output power peak at crossover. A slight spread reduces this peak. Too much of a spread will cause a power output drop at crossover. We have a 5% spread ((2650 - 2520) / 2584 = 5%), which should be fine. An anticipated shift of acceptable proportion and in the right direction can be managed, but the shift can be significant -- as much as 10% or more -- and might not go in the desired direction if you are simply picking the closest components while ordering. It is for this reason that the system points it out in advance. If effective crossover frequencies are shifting around by 10% or more without anticipation and advance adjustment, it can result in audible volume changes at crossover.

Advanced Issues

At this point some people will (rightfully) want to return to the enclosure screen, enter any phase shifts for the drivers and supporting circuitry, any z-offsets (this is the relative difference in acoustic centers of the drivers), and even test the predicted sound pressure for problems at key frequencies. We entered -135 as the phase shift for the P17s, 135 for the D25, and -1 z-shift for the P17s. We check the boxes under 'F-Inspect' for all drivers, enter 2600 as the test frequency, and click 'Update Image.' The drafting will now show the mathematically-predicted polar sound pressure pattern that will be produced by the speaker given the current driver locations, phase shifts, SPL, etc.

Save any changes you want to be able to restore later.

Add a Subwoofer

There is little need to add a subwoofer to a stereo system that can generate frequencies as low as 44Hz, unless one is looking for ground-shaking power... so let's do it. Add another enclosure and name it 'Subwoofer,' save the changes, then add a driver to the enclosure.

This will be a step up from a simple subwoofer, a push-pull box with two identical drivers. Enter the specs for the Peerless 315SWR. As with the P17, enter two for the quantity, but select Parallel Push-Pull as the configuration. Note the change in polarity on the schematic. The push-pull configuration reduces the effective VAS and allows strong output from a box of about half the 'normal' size. Save the entries and return to the enclosure's data screen.

This woofer is right on the border between recommending a sealed or a ported box. A sealed box could be smaller, but the larger ported box can have a lower resonance. Pick your poison -- we're using a sealed one in a 14 x 20 x 14 inner-dimensioned enclosure -- finalize your dimensions, and save the changes.

Our humble receiver/amplifier generates 85W per channel. We'd like to be able to pump more power into the subwoofer under some circumstances, so, rather than add a low pass filter to the enclosure, we will purchase a separate power amplifier just for the subwoofer that includes a low pass filter. This amp will be installed directly on the subwoofer enclosure, so no circuitry work is required for this box.

The drafting system doesn't handle push-pull drivers, so you can leave them off of the image. Indeed, our actual box design will be substantially-different from the site's generic suggestion.

Take a Breather....

You've now designed a fine speaker project with two unique enclosures and a respectable array of drivers. In fact, the above alone would be an above average stereo system for most music, movies, television and games.

Save our Progress

Select the 'Save' option from the Speaker Design menu to have the server create a text file of the project and send it to your computer. This allows you to design projects and save them for later review (the server will usually remember your current project from visit to visit, but there is no guarantee that it won't go nuts without warning). On most computers, the file will appear on your 'desktop' as spkrprj.txt, and contain what appears to be gibberish. DON'T MESS WITH THE FILE. If you make any changes to the file's contents, you will not be able to re-load it to the server, and all of your work may be lost.

To resume work on a saved file, select Upload from the Speaker Design menu. Click the browse button and locate the desired file on your computer. Then click Begin Upload. It should send the file to the server, where it will be decoded and opened.

Let's complete the project. Make two surround enclosures based upon the stereo L/R one.

Duplicating an Enclosure -- Adding Surround Channels

Return to the Stereo L/R enclosure page (you should know how to do this now). Click 'Duplicate' right under the enclosure name at the top left of the screen. A new enclosure will appear, 'Stereo L/R Copy,' containing all of the same drivers as your original enclosure. Change the name to 'Surround' and save the changes.

Go to the P17 driver in your new enclosure and decrease its quantity to 1. This will change the output SPL as well. Save the changes and then go back to the Surround enclosure data screen. As before, adjust the dimensions so that the box resonance is 44Hz or lower. We specified 7 x 15 x 10.5 inches, along with a 5 inch long 2 inch inner diameter port.

2nd Reality Check: It is not an accident that this enclosure's width is the same as the stereo l/r enclosure. This is not simply because it happens to be an integer larger than the largest driver, but to ease the future construction process. By doing this, at least four of our project's speakers will have the same width for their hardwood panels. This will allow us to rip all 16 of the casing panels without having to change the tool setups. In addition, we are going to use traditional dovetailing for the joinery, and having all of the panels the same width makes that process much easier. In other words, we are simply thinking ahead.

Note that the predicted output of the P17 is not the factory-noted 87dB, but is closer to 86 due to the inductors' resistance (under the circuit diagram on the P17 page). Attenuate the D25 to 86dB SPL. This will add a resistor of about 4.5 ohms to the D25 circuit.

After tweaking our crossover frequencies so that we could use the nearest stock inductors, we ended up with a low-pass filter at 2850Hz on the P17, and a high-pass filter at 3000Hz on the D25. In addition, the D25 is being attenuated for a target SPL of 86.7 to match the P17.

Locate the drivers and port as you wish, using the same procedure as for the Stereo L/R channels.

Add a Center Channel

The center channel enclosure has been entirely redesigned since the last revision of this tutorial, now contains five drivers, and has crossovers at 300 and 3000Hz. The results are much-improved overall performance, but considerably-more expense due mostly to the increased capacitor cost for the 240μF AudioCap capacitor sets. Load the tutorial project to view the changes.

A good home theater system needs a decent center channel. Our stereo l/r box would be fine, but we're going to do a little more to demonstrate how to make a more complicated speaker on this site. Create a new enclosure now using the above procedures, but this one will contain three different drivers; MG18 (x2), MG10 and D25. You might try importing the driver data from the database instead of adding it manually. The design process is the same as before, we are adding a band pass filter for the MG10. Our MG18 low pass filter is set at 490Hz, and the SPL out targeting 84.8 (limited by the MG10). The MG10 has two key crossover frequencies, 500 and 4375Hz -- enter "500,4375" -- and no additional attenuation as it has the lowest SPL of the set. On the D25, the high pass filter is set at 4400Hz, and it is also being attenuated to 84.8dB for SPL. These particular frequencies were all selected (1) based on the manufacturer's performance graphs, (2) the system recommended crossover frequencies, and (3) the frequencies that make it most likely for us to be able to use stock inductors. We are building 3rd order CPC crossovers for all drivers. The MG18 and MG10 will need Zobel circuits for reasons discussed earlier. The MG18 and D25 will need to be attenuated to match the lower SPL out of the MG10. Don't forget to include the resistance of the crossover inductors. (When doing bandpass filters, the site automatically accounts for interaction between the many related circuits and adjusts the component values accordingly. This helps avoid power level and impedence problems common to 3+ way speakers, and is why the system-recommended components 'add up' to key frequencies slightly different from 500 and 4375Hz -- more like 530 and 4120.)

We gave the enclosure inner dimensions of 8 x 20.5 x 20 and a 5 inch long, 3 inch inner diameter port. Unlike the other ported enclosures, we'll be mounting this port on the back. (It would be nice to match the stereo l/r and surround box widths, but acoustic performance is taking priority, and the size, orientation and eventual location of the speakers for this box requires a larger width.)

Note that the MG10 has its own little VAS property that we've basically ignored. Some designers would make a small internal chamber right behind the MG10 in the box, and then adjust the overall box volume and dimension accordingly. This is, effectively, making a speaker box within the larger speaker box, and treats the MG10, physically, as if it is not really part of another system -- it isn't "breathing the same air". All of the electrical connections remain the same. Theoretically, this is usually the right thing to do. For the sake of simplicity, however, we will 'skip' it for this tutorial.

The drafting system won't properly show the drivers because, when placed in-line as it will attempt to do, they exceed the enclosure height. But we'll be stacking the mid-range and tweeter drivers, so it isn't a problem. Although the speaker appears in the drawing as a vertical tower, when finished we'll be resting it horizontally on a shelf above the television (with plenty of clearance behind it to avoid blocking the back-firing port).

You are finished with the design stage. This is a good time to do another download of your project file.

Design Review

To avoid the cost of making an obvious mistake, we go back through each enclosure, one at a time, reviewing every variable. We perform manual "sanity checks" on some elements using paper, pencil and calculator. We purposefully check the sound lobes at crossover points, driver placement, circuit components, etc. A complete speaker system can easily cost thousands of dollars for parts alone. It is much easier to catch design problems now, before ordering parts and well-before beginning enclosure fabrication.

Order the Needed Parts

Once you are confident in your design, select 'Parts List' from the Speaker Design menu to get a list of current drivers, crossover components and other materials needed to complete the project. The system will automatically suggest components from our inventory that are as close as possible to the ideal components mathematically calculated by the server. By clicking a button on the parts list page, you can roll these items into an order and we'll ship the parts to you. Or, you can print the parts list and order the items from any supplier of your choosing.

When you components arrive, be sure to check them against your original order and vendor packing list. In our experience about one in three shipments from audio supply houses have errors ranging from small to significant. If you have the equipment, manually measure and label your components with their capacitance, inductance and resistance values at the crossover frequency. The actual values can vary enough to significantly-impact your circuits.

Assemble and Rough-Test the Crossovers

Print each of your driver pages to get the crossover schematics. Crossovers are easily made directly on plywood or perforated circuit boards. We use a hot glue gun to hold the components firmly in place (and avoid future problems). We also add nylon zip straps to heavy items like inductors. The component leads get soldered directly to each other where possible, or connected by heavy wire. Anticipate the final location for the crossovers and make sure that your boards are small enough to fit. For example, if we mount crossovers boards on the inside bottom of our our stereo l/r enclosures, the size cannot exceed the internal x-z dimensions of 7" x 10.5". Not a problem in this case, but it could be for smaller boxes. If you want to be able to remove the crossover later, you need to plan an access panel on the box, or make the crossover small and accessible-enough that it can be installed and removed through a driver hole. For our actual finished speakers, we created pedestal bases and mounted the crossovers outside the enclosure, in the pedestal. This required us to make them a little smaller.

Anyway, get your components on the board and connect them to your speakers. Place the speakers a few feet apart -- far enough that you can discern which one is producing sound. Connect a signal generator to the input of the crossover and slowly sweep from 20Hz through your crossover frequency and beyond. As the frequency approaches the crossover point, you should be able to clearly tell that the output is transferring to the tweeters. At the crossover point, the tweeter may sound a bit stronger than the woofers, depending on which way it is pointing. As you proceed beyond the crossover point the woofer output should diminish until it cannot be discerned.

If this does not work as described, then recheck your wiring, component values, etc.

A note about complex crossovers: The crossover for our center channel, when we add the baffle compensation, has 27 different electrical components; 8 inductors, 15 capacitors (many in parallel to arrive at desired values) and 4 resistors. Yes, this is a costly (but awesome) speaker. To avoid interference, especially between inductors, we will place the crossover as far as possible away from the driver coils, place the inductors always at right angles to each other and as far apart as possible on the board. Because there is no way to place more than three inductors on the same board at right angles to each other, in this unusually-complex case we will build two or more separate crossovers boards for the enclosure and place them on opposite ends of the enclosure.

Making the Enclosures

You are free to make your own enclosures from scratch, order custom-cut panels from us as a kit, or order the enclosures (or the entire project) custom made. We have a complete woodshop at our location, and use quality, solid hardwoods to make our own boxes. Some people prefer to use MDF and apply laminant or paint. Hardwoods take a little more care in construction, but when properly assembled, are superior in strength, rigidity and appearance to MDF.

Making the boxes requires more tooling, time and manual skill than probably any other part of the project (indeed just the cost of the mid-range tools for making speaker cabinets can cost as much or more than some of the best made speakers available). Print each of the enclosure pages to get the drawings for the boxes (the cut list is also on the parts list page). Cut your panels to their finished dimensions. Cut the holes for the drivers, remembering to counterbore the tweeters so they lay flush on the panel face. This is usually best done with a router and a hole cutting jig. Drill holes for the binding posts or cut a hole for the terminal cup.

Remember, the server-generated box designs are just examples. Adjust them to your liking. For example, our subwoofer casing is quite different from the server's suggestions, especially if we are doing somethign like push-pull drivers. If your crossover is too big to fit through a driver hole and you aren't adding an access panel, mount the crossover in location before assembling the box. We use four longer woodscrews with nylon standoffs and one inch thick foam. The foam goes between the crossover and the panel to which it is mounted to eliminate vibrations. The screws go through the top corners of the crossover panel, through the nylon standoffs, and into the enclosure wall.

Dry-assemble all enclosures first. This means to put them together exactly as they will be when glued, but without the glue. This practice run allows you to confirm that everything fits properly and figure out how you will have to arrange your clamps. Once you put glue on, any problems can become big problems very quickly. Much better to address and anticipate these before glue is running all over and drying on you and your speakers. You'll be amazed at home many clamps it takes (and how much the clamps cost). Even a small enclosure may need as many as six clamps. If you are gluing up 8 enclosures at once, this means you may need almost 50 clamps, which is more than even most small wood shops have available for simultaneous use.

After a good dry-run, go ahead glue your panels together using a glue with considerable 'open time'. Open time is the amount of time you can leave the joint open or make adjustments before the glue becomes too hard. Fast drying glues can set up before you are ready, which can ruin your whole day. We use either Titebond III or urethane glue. Newspaper, cardboard or waxed paper can be used to protect (and keep the boxes from adhering to) your floor or table. Apply an even bead of glue to both surfaces that will be in contact with each other. Smooth the glue with a stick, brush, roller (or even a finger, unless you are using urethane).

If you are using a good glue and clamping technique you won't need screws or nails to hold the enclosure panels together, but they can help. You'll know you have the right amount of glue and clamping pressure if a small, even amount of glue squeezes out of all seams. If no glue seeps out of an area, there might not be enough glue, enough pressure, or something might be seriously wrong with the assembly. Use a rag or paint scraper to clean up as much glue as possible; this will make later sanding and trimming much easier. A portable drill with adjustable clutch and countersink bit (or better yet, a complete pocket-hole jig setup) is good for screws. A pneumatic nail gun can make quick work of nailing and help keep things from shifting as they might under the repeated blows of a hammer.

If the enclosure has large panels, they should have at least one brace, somewhere, to help dampen wall vibration. Ideally, a pair of such braces would go from near the center of the panel they are attempting to stop, to opposite corners. The goal is to deaden the box -- to keep it from artificially attenuating or amplifying certain frequencies.

Let the glue cure per the manufacturer's instructions, then trim, scrape and thoroughly sand the box's exterior. If there is any chance that you didn't have a full glue joint, you should also caulk all inside seams. MDF takes paint and veneer fairly well. On hardwoods we use Danish Oil stains, but varnishes and polyurethanes are also popular. Once the exterior is finished, install the binding posts (or terminal cup), connect them to the crossover using spade connectors (or solder). Connect and install the speakers. Foam, caulk, or similar materials should be used to seal the binding posts, terminal cups and speaker perimiter. We use weather stripping foam with an adhesive back. If using caulk, apply a very light but continuous bead before installation. Be ready to perform clean-up immediately if any squeezes out onto your finish.

Connect the enclosure to your amplifier, and you are done!

Final Notes

We were careful to match the SPL of the drivers within enclosures by adjusting the circuitry. However, the different enclosures have significantly-different SPL. The MTM stereo enclosures have ~91. The surrounds have ~87. The center channel has ~84. We could do additional circuit-work to reduce all enclosures to the lowest SPL, but this would waste a lot of amplifier power. In this case we will simply allow our recevier/amplifier's built-in power equalization process balance out the speakers.

Enclosure resonant frequencies are guesses. Many factors not taken into consideration by this program may affect them, especially the box's actual proportions and the location of interior frames or stuffing.

The system does not presently address phase/polarity issues except in the most simplistic way. The good news is that, once your drivers are in their boxes, you can connect a signal generator to the box and roll through the crossover frequencies. If there is an audible loss of volume around the crossover point, try reversing the polarity on the higher frequency driver. This is usually an issue only on 2nd order filters, but it might pop up unexpectedly in other scenarios.

Anything Else?

If you have any questions, you are welcome to contact us.

Copyright 2005 Simple Way Limited Partnership DBA Arpeggio Music and Sound. All Rights Reserved. Terms of Use

Arpeggio Music and Sound, 805 NW Alder St, McMinnville, OR, USA. 97128
Telephone 503-472-0688; FAX 503-434-5803; Hours 10 AM - 5 PM Pacific Time, Monday through Friday
Visits by appointment only.