Music and the search for better sound reproduction have always been my passion. I realized, early on, that bass was a big part of what I liked about sound and critical for the music to elicit in me an emotional response. But to achieve ultimate bass—the deepest, most solid, most seamless bass—requires getting many things right: quality subwoofers, ideal subwoofer locations with the best frequency response, speaker and subwoofer time-alignment, acoustic treatments, and equalization. While each of these factors is worthy of its own article, the focus of this article is on examining three different ways one can connect a subwoofer within a two-channel system. While there may be other approaches, I’ve included two standard ways and one less popular one with more flexibility than the other two.  

I don’t use tools that automatically equalize and time-align speakers, such as DIRAC, Anthem Room Correction, or Audyssey. Instead, I use OmniMic acoustic measurement software and my ears. For the purposes of explaining my methods, I’ll use my own sound system as a reference point. It includes a pair of dipole electrostatics, a pair of 15” self-powered subwoofers, a tube amplifier and preamplifier, and an 8-channel DAC connected to a laptop as the music source. One of the subwoofers is centered between the main speakers while the second subwoofer is positioned in the left rear corner, both raised 18″ off the floor. The room is treated with bass traps and absorption and diffusion panels.

3 Subwoofer Connection Approaches

I’ll introduce three approaches in connecting a subwoofer, with the last approach adding digital time alignment into the mix. The first two approaches are typical of an average setup that uses a subwoofer’s phase control to blend its sound with the main speakers at a crossover point. For all three approaches, the subwoofers’ positioning remained the same; only their crossover frequency and slope, phase, and polarity settings changed between them.

I relied on the OmniMic frequency response chart to examine several frequency/slope combinations to try to achieve the smoothest curve without nulls, or at least with the fewest nulls, then used my hearing to validate the chosen crossover frequency and slope. The smoothest curve doesn’t always equal the best sound, but it can get us very close.

A crossover is a way of dividing the frequencies of the music between two or more speaker drivers (e.g. tweeter and woofer) and/or between a speaker and a subwoofer to avoid that they play beyond their limits or excessively overlap. A crossover comprises a “low-pass” and “high-pass” filter that a speaker designer uses to prevent a particular driver from playing certain high or low frequencies. A subwoofer’s low-pass filter prevents it from playing higher pitched instruments and vocals, while its high-pass filter prevents the main speakers from playing the lowest bass notes. A filter is set at a crossover frequency and slope whose steepness determines how quickly the filtered frequencies drop off (e.g. 24dB filter drops off fast while a 6dB filter does not).

Approach 1 (good): Removing the Bass from the Main Speakers Using the Subwoofer.

The signal chain was as follows:

DAC > Left preamp output to left subwoofer input and left subwoofer high-pass output to left amp input (and similar connections with the right channel).

_{Note: To simplify the first two diagrams, only the left channel is shown.}

Since my tube amp had a hard time handling low frequencies and wasn’t overly powerful to begin with, I decided to try the high-pass outputs on my subwoofers’ plate amps to feed the tube amp with frequencies at 80Hz and higher. Both subwoofers’ high-pass filters were set at 80Hz with a 24dB slope for the amp and the speakers.

Approach 2 (Better): An Active Crossover Splits the Signal Between Main Amp and Subwoofers

The signal chain was as follows:

DAC > Preamp > Active Crossover > Left and right high-pass outputs to tube amp and left and right low-pass outputs to both subwoofers.

Adding an external crossover, such as a Bryston 10B STD, allowed for more flexibility in crossover frequency and slope settings for speakers and subwoofers (after the subwoofer’s internal 24dB/octave setting was bypassed). The available crossover frequency settings were 70/100/140Hz etc., while the slope choices consisted of 6/12/18dB/octave. There’s a switch on the Bryston that allows setting its bass to mono, which I used. Most bass is recorded in mono, so playing test tones in mono seemed like an acceptable option, plus it ensures that both subs are receiving the same signal, which makes it easier to find the right crossover point. The best curve I achieved was a high-pass of 140Hz at 6dB slope for the main speakers with both subwoofers low-passed at 100Hz with a 18dB slope.

Approach 3 (Best): Multichannel DAC Improves Flexibility and Control Over All Speakers

The signal chain was as follows:

DAC’s outputs 1 and 2 > left/right preamp channels > left-right amplifier channels, while the preamp outputs 3 and 4 went directly to the front and rear subwoofers respectively.

A stereo DAC can only implement frequency or time correction to both channels on a macro-level basis. What I wanted was micro-level flexibility and functions to give me the best ability to address acoustic anomalies in my room. To this end, I bought an 8-channel DAC from Toronto-based manufacturer exaSound. Equipped with digital music management services JRiver Media Center and ROON, the DAC became the foundation on which I made channel-specific adjustments. JRiver’s Digital Signal Processing functions were used to: create mono bass for both subwoofers, high-pass the main speakers at 67Hz with a slope of 24dB, low-pass the front subwoofer at 74Hz using a 12db slope, and low-pass the rear subwoofer at 80Hz with a 12dB slope. I also performed a 3ms delay and polarity reversal to the rear subwoofer. That meant that the crossover points, slopes and delays were different for the main speakers and subwoofers, even between the subwoofers. Since each speaker/subwoofer was in a unique position within the room, it stood to reason that crossover settings (and equalization afterwards) would be customized.

Analyzing the Various Approaches

I wanted to run an experiment that compared the three approaches of bass integration discussed in Part 1. No parametric equalization (EQ) was used; rather, I used the crossover settings, subwoofer phase, and volume adjustments, as well as added delay and polarity for the most sophisticated approach: #3.  The best settings for each approach were compared. 

Below is a frequency response graph showing the best curve I could generate for each approach. Also shown is the desired target curve that is the goal for each curve to try to match. The chart’s X-axis represents a frequency range of 30 – 300Hz, the Y-axis represents the loudness in 3dB increments, while a 1/24th octave smoothing was used to show more granular results. Too much room gain below 30Hz made the measurements questionable and all three curves tracked identically above 300Hz.

BLACK line: my target curve: a 10dB slope from 25 to 200Hz

GREEN line: Approach 1: Mains High-Pass 80Hz @ 24dB slope, both subwoofers Low-Pass 80Hz @ 24dB slope

BLUE line: Approach 2: Mains High-Pass 140Hz @ 6dB slope, both subwoofers Low-Pass 100Hz @ 18dB slope

RED line: Approach 3: Mains High-Pass 67Hz @ 24dB slope, front subwoofer Low-Pass 74Hz @ 12dB slope and rear subwoofer Low-Pass 80Hz @ 12dB slope with a 3-millisecond delay and reversed polarity.

Bass and Math: Quantifying Each Curve’s Variance

To act as a second set of eyes for the curves, I called upon my data scientist friend who used three different mathematical methods called “Entropy,” “Correlation Coefficient,” and “Average Absolute Weighted Error” to calculate each curve’s variance from the target curve. Each method is a different approach on quantifying the error rate. 

The variance comparisons:

The table supports the idea that the sample curves come closer to approximating the target curve when there’s more control flexibility for each speaker and subwoofer. For each math method above, Approach 3 scores closest to the target curve. This isn’t a big surprise because when “local” controls exist for every sound source you have, it provides the maximum flexibility to tailor the sound to your taste, which is what Approach 3 offers most over the other approaches. Global controls can work for left or right channels and/or for left-high or left-low frequencies (same with right), but their effectiveness and precision can only go so far.

Listening Notes

Approach 1 was my first foray into using subwoofers, so it was exciting and entertaining. I typically used too much volume due to the sub’s novelty factor and me wanting to hear what it could do. With too much bass, the response wasn’t well controlled and, since I didn’t use parametric EQ filters, could be boomy at certain frequencies. There was no digital way to time-align the subwoofers to each other and to the main speakers, so I used the phase control on the subwoofer’s plate amplifier to get the best response at the crossover point and below, as seen on the real-time frequency response charts. This, however, meant the bass from the subwoofer was delayed from that of the main speakers by one or more cycles, causing it to arrive a bit behind the beat of the music. This “slow bass” effect was unenjoyable to listen to.

Approach 2 drew me in with its less boomy, more articulate, and overall better bass. The added flexibility offered by this setup allowed for much experimentation, which was fun, and helped move the frequency response curve in the right direction. Despite having the same time alignment difficulties as in Approach 1, the smoother frequency response sounded very good.

Approach 3 allowed more of the leading transient attack of a note to be heard, such as from the thwack of a bass drum or the pluck of an upright bass. When a bass guitar was played very fast (e.g. by Brian Bromberg or Jaco Pastorious), it was easier to hear individual bass notes as separate sonic events than through the other approaches. This improvement was due to time alignment of the subwoofers to each other and to the main speakers. There was no boominess and all bass notes sounded equally loud and articulate. Compared to Approach 1, there seemed to be less bass (as Approach 1 was typically too present and boomy), which may suit an audiophile but not necessarily a bass-nut or lover of action movies. Approach 3’s bass bettered that of Approach 2 in smoothness, integration, and articulation.

To further improve the sound achieved by any of the three approaches, I suggest taking the following actions: 1) apply EQ to the bass region to better fit the actual frequency response to the target curve, 2) time align all subwoofers and speakers, and 3) reduce long bass decay times to bring them more in line with midrange decay times (e.g. average bass decay times are about 2X the average midrange decay times). These important steps in bass optimization are beyond the scope of this article.

Practical Tips and Interesting Findings

Through my bass-optimizing journey, a few things leapt out as interesting findings:

• The issue of overlap between the front subwoofer’s and the main speakers’ crossover frequency. Most of the articles I read suggest having an identical crossover frequency between the main speakers and subwoofers so there is a smooth transition between them. Maybe if different types of slopes were used, my results would have been different. To put things into perspective, the front subwoofer’s 74Hz and the main speakers’ 67Hz crossover points are only two semitones/notes apart, so they’re very close to each other. Any further overlap may muddy the high bass and low midrange.

• Placing subwoofers in ways that complement each other’s output curves. The front subwoofer had good output above 50Hz and little below it due to its room position away from boundaries. The rear subwoofer near a corner had very good low-end output (20-50Hz), which compensated for the front subwoofer’s weaker bass. Together, they did a very good job at approximating the target curve.

• Rooms are too unpredictable and require endless tinkering of subwoofer placement and settings. Relying only on one’s ears to find the ideal placement for a subwoofer, without the help of measurement software, is a bad strategy, in my opinion. Measurements are needed to note the effect of a change to the system or settings so you can determine if you’re moving in the right direction. Ears should be used for the final subjective testing.

• A curve that measures well may not actually sound great. This surprised me—I assumed that if curve A measures better than curve B, then A ought to sound better. I feel that phase artifacts introduced via parametric EQ filters are the culprit. At the end of the day, a target curve, and any desirable deviations from it, is very much a matter of personal taste. This is why I offer clients several target curves so they can find the one that sounds best to their ears.

Integrating one or more subwoofers into a stereo setup can be complex and time-consuming, requiring careful consideration of various factors. Acoustic measurements are vital for extracting the best performance from a subwoofer, showing in real-time whether a crossover frequency and slope or a volume or phase change has a positive or negative effect on sound quality. When the bass is time-aligned with the main speakers and the crossover frequencies have allowed you to reach your target curve, you will hear and feel the difference. Your musical enjoyment and emotional satisfaction will escalate, and you may find yourself searching your music library for bass-heavy music to experience what you’ve been missing. In my case, integrating two subwoofers into a 2-channel stereo system involved first finding the optimal position for the subwoofers, for which Todd Welti’s research on subwoofer placement within a rectangular room was invaluable. In my first two articles in this series, I explored three different methods of connecting a subwoofer to the system. In this final part, I’ll address subwoofer time alignment with the main speakers and compare five popular crossover strategies.

Subwoofer Time Alignment

The goal of a well setup stereo is to have, as much as possible, the direct sound from all speakers and subwoofers arrive at the listener’s ears simultaneously to reduce phase inconsistencies. The most accurate way to achieve this is to make sure the main speakers are the same measured distance to a microphone positioned where the listener’s head would normally be. Using an Impulse Response chart and test tones, you should see a single large impulse peak at time zero, which represents the listener’s location. If you see two impulse peaks, then one speaker will have to be moved closer or farther from the microphone so that the two impulse peaks become a single one, indicating successful time alignment.

Consider the picture below where the front two marching soldiers are the main speakers and the fellow behind them is a subwoofer. The front two soldiers are “time aligned”, as they are next to each other and “in phase” as their right legs are raised in unison. The fellow behind is also in phase with his right leg raised but is 1 cycle behind the front soldiers and therefore not time aligned with them. Positioning a sub behind main speakers without using digital signal processing to delay the main speakers results in slow or sluggish bass that is slightly behind the beat of the music despite the subwoofer being in phase with the speakers.

Placing a subwoofer next to and on the same plane as a main speaker doesn’t guarantee that it will be time aligned due to the subwoofer’s “Group Delay”—a delay in the signal emitted by the subwoofer’s lower frequencies, driver excursion, crossover, enclosure or signal processing. The best way to achieve proper time alignment is to use digital room correction/digital signal processing (DRC/DSP) incorporated in the playback chain.

When DSP is available, position the subwoofer where its frequency response measures best (i.e. loudest at lowest frequencies, flat or downward tilted frequency response, nulls at a frequency higher than the expected crossover point). Then, measure the closest speaker’s delay time to the microphone of the Impulse Response tone and do the same with the subwoofer. Use the difference between both figures to delay the fastest sound source. For example, if the speaker’s delay is 2ms and the subwoofer’s delay is 5ms, then DSP would delay the speaker by 3ms (5ms – 2ms = 3ms) to ensure time alignment.

When DSP is not available in the stereo chain, the subwoofer will have to be physically, rather than digitally, time aligned.* Again, using an Impulse Response test tone, move each subwoofer and measure its delay time until it matches that of the speakers. Don’t be alarmed if a subwoofer ends up against the side wall ahead of the main speaker plane. Figure 2 shows subwoofer spot ‘A’ as offering the best measured frequency response while spot ‘B’ offers the best time alignment. A good strategy is to measure the distance from the sub to the microphone, then using that distance from the microphone to draw an imaginary arc consisting of other areas that offer proper time alignment. Finally, choose the spot on the arc that delivers the best frequency response.

Crossover Settings Between the Main Speakers and Subwoofers

In the search for better stereo sound reproduction, I investigated crossover options between my main speakers and two subwoofers and compared 5 popular options commonly used: the home theatre standard 80Hz/24dB; staggered frequencies; overlapped frequencies; the highly regarded 6dB/octave slope; and the run-the-main-speakers-full-range approach. I’ve used my listening room as a “lab” to seek out the best crossover settings for each option for comparison purposes, and encourage you to likewise experiment with different crossover settings in your own room. Room effects are too unpredictable for me to recommend my findings to you with absolute certainty. Through experimentation, you should be able to discover insights into your situation and, at the very least, be entertained.

When I got my 8-channel DAC from exaSound, I paired it with JRiver Media, a popular software for organizing and streaming music that comes with rich DSP functionality. Compared to a 2-channel stereo DAC, the options for integrating the main speakers and pair of subwoofers increased exponentially with the exaSound’s 8 channels. The ability to control crossover frequency and slope, delay, polarity, volume, and parametric equalization at the individual speaker/subwoofer level was new to me as I’d never been into home theatre or multi-channel AV receivers. I was strictly a stereo guy. And while DSP tools like MiniDSP or DIRAC intrigued me, my multi-channel DAC, JRiver software, and the Dayton Audio OmniMic offer all the tools I need to optimize subwoofer and speaker placement in practically any room.

5 Popular Crossover Settings

Using the JRiver software, I compared several popular crossover options between my speakers and subwoofers. Here they are, with their descriptions:

A. “Home Theatre Favourite” filter option: borrowed from the home theatre world, main speakers and subwoofers are crossed over at 80Hz using 24dB/octave high- and low-pass filters, respectively.

B. “Staggered” filters option: the speakers are crossed over at a higher frequency than the subwoofers such that their ‘tails’—their tapering-off response below their crossover frequency setting—fill in the gap between the two crossover frequencies. That means that the speaker’s filter set to 80Hz will continue to work at 70Hz and lower to allow lower frequencies through but at quieter loudness levels; similarly, the sub’s 50Hz filter will continue to work at 60Hz and above. The stagger distance and slopes are up to the user to optimize.

C. “Overlapped” filters option: the speakers are crossed over at a lower frequency than the subwoofers with the amount of the overlap and slopes to be determined by the user.

D. “6dB-per-octave” filter option: proponents of this option espouse that shallow slopes are best for seamless integration and with little impact on phase. It’s essentially an overlapped filter with a long tail, as far as I can tell. I started with both speakers and subwoofers using 6dB filters but couldn’t get a great frequency response, so I decided to cross the speakers at 6dB with the subwoofers at a higher slope.

E. “Run The Speakers Full Range” filter option: Like Dr. Earl Geddes preaches for best bass integration, the speakers are run full range with no high-pass filter. Because I couldn’t find out what slopes his approach recommends for the subwoofers, I borrowed from the ‘D’ option and chose to cross the front subwoofer nearest the speakers at 6dB for integration while using quite a low crossover frequency. The subwoofer at the rear of the room behind the listening chair was using a 18dB slope and a higher frequency.

Testing used no equalization and focused on the bass region from 20 – 500Hz. Raw measurements were used.

_{Figure 3. Frequency response comparisons of five popular crossover settings.  The 1^st and 2^nd place winners were ‘E’ and ‘C’ respectively based on how closely their actual curves matched my target curve.}

The “Run The Speakers Full Range”, with the front subwoofer at a 6dB/octave slope, was the clear winner! Running your speakers full range provides more sources of bass to help smoothen the frequency response; using a low slope filter on a 15″ subwoofer requires a very low crossover point which, in this case, was 32Hz, but 27Hz and 20Hz worked nearly as well. Listening confirmed that E and C sounded best. I’d speculate that the overlapping has something to do with it. With option E, the front sub at a 6dB slope starting at 32Hz had a long tail (e.g. only down 12dB at 128Hz) and with option C, there was already a +1-octave overlap (e.g. 48Hz to 98Hz) before considering the tails of the speakers and subs. In both cases, the sound was richer, more engaging, and less thin-sounding than with options B and D. Option A was just okay-sounding.

The process of proper subwoofer integration with a stereo system involves many steps, some of which depend on whether analogue or digital music sources are used. Integration involves choices—subwoofer connection, time alignment, crossover settings, DRC/DSP—that can have a wide effect on the frequency response and perceived sound quality. For example, timbre and tonality are greatly impacted by whether a subwoofer’s and speaker’s crossovers are staggered or overlapped, by how much, and where within the frequency spectrum it occurs. As personal tastes prevail in sound reproduction quality, I encourage you to experiment. You may find that your imagination (and patience) to test, measure, and assess what speaker / subwoofer configuration works best for you has paid off with the best sound quality you’ve ever heard in your room. Good luck!

* If you don’t have DSP in your signal chain to time align your speakers with your subwoofers, the following table provides physical distances (in inches) between a subwoofer and the listening position. To apply it to your circumstances, you will need to calculate the speaker-to-chair distance and maybe contact your subwoofer manufacturer to obtain the subwoofer’s Group Delay value. Where those two values intersect in the table shows the time aligned distance of the subwoofer from the listening position.