An Active Two-way Omnidirectional Main Channel Monitor

After the completion of the SP15 for surround and center use, work on a model for the main channels and behind screen center slowly began. The goal was to design a two-way system with wide linear dispersion, full bandwidth, neutral, open sound, low distortion at realistic volume levels, and at a moderate cost, yet quality beyond its price point.

I

Application

The intended application is for main and behind screen center channel speakers for 2-channel stereo & multichannel music or home theater.

II

Design Parameters

The design goal : An active, two-way, omnidirectional, full range (+/-3 dB 50 Hz - 20k) monitor, capable of 95 dB continuous SPL, low distortion, high resolution audio reproduction primarily for HT.

III

Research & Development

The thought of designing the SP15 with an added tweeter to make it suitable for front channel mains that compliment the surrounds had occurred to me before finishing the SP15 (fall of 2006). There were obstacles and challenges that needed to be solved before moving on with the design. The major hurdles to overcome were, crossover design, tweeter selection, and how to mount the tweeter to the SP15 enclosure. A search for a suitable tweeter was the first consideration. Many fine tweeters are available—usually very expensive. Most are monopole. How do you incorporate a monopole tweeter with an omnidirectional woofer? In reality, all direct firing woofers radiate omnidirectional in their lower range. Consequently, conventional direct radiating box speakers are omnipole in the low frequencies and become progressively more directional through the midrange and up. An up-firing tweeter, like the single driver SP15, will not do the job because high frequencies become directional even with small diameter drivers. That would have been an obvious and easy solution, but the laws of physics prevent it. Some manufacturers have solved this issue with wave guides to reflect and distribute the tweeter's output 360° horizontally, but not vertically. A simpler, more elegant solution, though not fully omnidirectional, is a dipole or bipole configuration, which would require either two tweeters positioned back to back, or an open back dipole tweeter. To my knowledge, there are no open back dome tweeters that would provide dipole operation. Because of the size of a tweeter's dome, it is impractical to leave the back open; the magnet and voice coil would interfere too much with the back-wave. Two tweeters back to back would complicate baffle construction and increase the cost. At first I thought to look for a suitable dipole ribbon or planar. Most ribbon/planar drivers are larger line source and not dipole. The large line-source creates interference patterns (comb filtering) and very limited vertical dispersion. The choice was limited to only one dipole driver small enough to provide reasonably wide dispersion, minimal comb filtering, the necessary bandwidth for a two-way system, a flat frequency response, and low distortion. This planar is also the lowest distortion planar, according to independent tests, rivaling the best and beating many of the dome tweeters in its price range, and has a very good cumulative spectral decay (CSD). Amazingly, it is also one of the most reasonably priced tweeters. There is, however, a concern with using a dipole configuration. The front wave is in phase with the woofer, the back wave is out of phase. This causes cancellation of the signal at the crossover, and partial cancellation for about an octave above and below the crossover point on the backside. There is not much sense in having an omnidirectional speaker when one side is linear and the other very non-linear. It defeats the purpose. Two dome tweeters wired in phase (bipole) would be theoretically the proper choice.

Note: The bipole configuration of the tweeters does not create a perfect omnidirectional dispersion pattern. It is omnidirectional up to the mid-treble. Above 5 kHz the polar/power response is very wide front and back, but falling off to the sides. The advantage of using the bipole configuration over using a waveguide or a reflector is for allowing better vertical dispersion; important for matching the vertical dispersion of the woofer. See: http://www.moultonlabs.com/more/panoramic_power_response/

Next issue was the crossover. At first I considered a basic 1st order passive crossover, believing it would be the simplest, most economical means. However, it's clear that few drivers are capable of operating well with a 1st order crossover and that at least a 2nd order circuit would be needed, especially for the tweeter. A 4th order would be best, but the added complication would be most likely beyond my capability. Upon further investigation, I found a number of sources for active crossovers, both custom and ready made. Active crossovers solve a mountain of problems, and in this case actually makes for not only a better means of separating the bandwidth to each driver, it also makes it easier. Of course, it requires four amplifiers instead of two, and a couple more cables, but once set up, a far better system. Best of all the crossover could be a 4th order Linkwitz/Riley configuration. . See Active vs Passive page.

I tossed around several options for mounting the tweeter baffle to the SP15. This was a challenge from both a physical and an aesthetic standpoint. The final design was a bit more complicated than I had hoped, yet visually the cleanest, and the most functional acoustically. The tweeter would be mounted 4.5" directly above the woofer, slightly closer vertically than would be possible if mounted on a single flat baffle, and about a half wavelength distance if crossed over at 1.5k. This is important for keeping interference patterns to a minimum through the crossover region where both drivers are producing the same frequencies. It also means that the drivers will act as a near point source.

It was assumed that the SP15 would be the base for the design. The one area I was not completely satisfied with on the SP15 was the upper bass/lower midrange region, a very critical part of the spectrum and one that for main channel use had to be solved. The crossover would take care of most of the rising upper midrange, but it still seemed prudent to look at other options. Reviewing possible drivers, the next size up from the same line of woofers, a 7” rather than the 5.5”, looked promising. It has more than double the the magnet strength and air volume displacement, and a lower resonance frequency. It is also well suited for small sealed enclosures, the recommended being 7L. With an adjustment to the height of the enclosure, in order to place the tweeter at ear level when seated, the resultant volume came to 14L. This volume coincidentally has a Qtc of .56, which is excellent for damping and transient response. It enables a lower f3 than the 7L, and in fact, essentially the same f3 (65.88 Hz vs 64.68) as the 5.5” in the larger (16L) enclosure of the SP15. All indications suggested that it would prove better in the upper bass/lower mid. The 7” is the largest driver that will fit the same enclosure used for the SP15, requiring only a redesign of the woofer mount. At this point I was quite pleased with the prospects of the new two-way main channel speakers.

Now with the drivers tentatively selected, the crossover chosen, the drawings worked out, and most of the technical issues settled, it was time to build and test the first prototype.

While running into several delays with getting the wood parts cut for the tweeter baffle, I found a very informative website, Zaph Audio. Zaph Audio does independent testing of drivers. I was very pleased to discover that the Bohlender Graebener Neo 3 planar tweeter tested well. So well, in fact, he rated it the best of the ribbon/planar drivers tested, with the lowest distortion, best cumulative spectral decay, and one of the smoothest frequency responses. Despite the earlier decision to go with two dome tweeters, I decided to order a couple of the BG Neo to test.

It is interesting to note that Zaph is of the opinion there is no need to buy expensive drivers. Excellent drivers can be had for under $100, and he rarely considers using anything over that price for his designs. Unless a driver is really spectacular, higher priced units are rarely worth the cost, only caveat is that low priced units usually have lower power handling capability. Yet, mid-priced drivers are often as good as the high-priced with regards to distortion and power handling. (Usually the high price reflects more expensive materials such as, aluminum, titanium, or other special coatings, and not necessarily an improvement in the magnet, voice coil, or other structures.) The measurements prove his point. When looking over the test results, it became clear there were other good choices for the woofer, ones costing considerably less. The annoying delays were proving beneficial.

With the new possibilities, I ran the numbers and found a couple of other 7” drivers quite suitable that cost less than half the original choice, and with a less serious 4k spike in the response. Even without the 4k spike, a low crossover point would be better than a higher one, necessitating a superior low end response from the tweeter. The BG Neo3 was also a good choice for this. As mentioned previously, I had already determined that a dipole tweeter would cause problems with the backside frequency response, but I was hoping to eliminate the need for two tweeters. To test the BGs, they were mounted on a cardboard baffle cut to the dimensions of the design. The upper frequencies were beautiful, as expected. However, below 2k there are highly audible artifacts. The problem is most likely edge or cavity resonances. It reminded me of the Newform Research quasi-ribbon drivers I had owned some years ago. The Neo3 exhibited a similar resonance between 1-2k that is completely unacceptable. How/why was this never mentioned, measured, or heard in the Zaph testing? (Actually, he does, sort of, by saying none of the ribbons tested should be used under 2.5 kHz.) There was no way this driver was going into one of my speakers. I couldn't live with the Newforms, and I won't live with the BG. This problem is common of ribbon/planar drivers, and so rarely mentioned. Also rarely mentioned is their delicate nature. Ribbons are easily overdriven and damaged. The BG's maximum power handling is only 20 watts. To avoid the distortion in the 1-2k range they would need to be crossed over at least at 3-4k, which is not an option for the intended design—it would spoil the near point source dispersion between the tweeter and woofer at crossover and create interference patterns in the crossover region. The distortion would be audible not only with sine waves, but at times with music. The only ribbon tweeter I've heard that doesn't exhibit this nasty characteristic is the proprietary Piega ribbons. Piegas are not available as a raw driver, and if they were, they'd be extremely expensive, certainly several hundred each, maybe more. No more diversions with ribbons or planars. This mistake cost me, but ultimately proved to be an indispensable lesson. Clearly, a standard fabric dome tweeter is the best choice for clean, smooth treble, and wide dispersion. Again, Zaph has test results for many dome tweeters. It's amazing to see how well a modestly priced tweeter compares to some of the very costly ones, a couple just as good. The need for a low Fs (resonance frequency) and crossover point helped reduce the choices. The cost savings on the woofers would more than make up for the extra cost of two more tweeters. The baffle design, despite my initial thoughts, actually turned out to be simpler, leading to a further small cost reduction.

More comments on ribbon/planar tweeters and why I cringe every time I see one used in a high-end speaker : http://www.zaphaudio.com/nondomes/.

Note: If the center is used behind an acoustically transparent projection screen and place closer than 24” to the front wall, then it would be advisable to eliminate the rear tweeter on the center. Having the rear firing tweeter too close to the front wall will create ripples in the frequency response and other confusions because the reflected signal will not be sufficiently delayed.

(II) UPDATE : Since I was using the Behringer CX2310 crossover, I also decided, based on a good review, on using their A500 amplifier. Within a month, one of the amps started making noise and microphoning through the volume knobs. It was replaced by the dealer. A few weeks later another amp went bad with the same problem. The dealer was kind enough to replace it even though it was beyond their 30 day replacement policy; Behringer support only offered warranty service. The next amp was defective straight out of the box, and the next also! Something is terribly wrong with the quality control on these amps. The last one was returned for a refund. No more A500s. Obviously, at some point, an amp cannot be made at any lower cost and still be of reasonable quality and dependability. The A500 can't deliver on its promise. There's a saying, "Spend too much and all that's lost is a little money; spend too little and you've lost everything."

The QSC RMX850 was purchased to replace the defective A500. Solid build, good finishing, and quality workmanship inside and out. The RMX850 weighs almost twice as much as the A500, delivers the full rated output, is capable of driving 2 ohm loads, has a 3 year warranty, and sounds excellent. Costs more, yet this amp is a real bargain. Not all Behringer products are of poor quality. The DCX2496 digital crossover is excellent and superior to the analog CX2310—clearly better sound quality and audibly lower distortion.

IV

Results

Finally after getting the parts cut, the speakers were assembled. Countless hours were spent testing, getting the settings adjusted, levels balanced, trying various crossover points, and retesting before the prototype was ready for listening tests. The results were decisively positive. Expectations have even been exceeded. The SP18M is fully capable of <50—20k Hz, with a respectable +/- 3 dB. Distortion levels are acceptably low and power handling very good.

I followed up the initial listening tests with A/B switching between a pair of Piega C8 LTD, a Stereophile Class A rated speaker. Unfair comparison? It should be. Yet the goal was to build something that might, expect in bass extension, rival them.

V

Observations & Comparisons

• The C8 LTD has exceptional high end extension, exceeding the SP18 and human hearing.
• The C8 has better low end extension, flat to 30 Hz; the SP 18 is flat to 50, -6 dB at 40 Hz.
• The C8 has the edge at very high volume; the SP18 sounds better at mid to low volume.
• The C8's imaging and soundstage is top notch; the SP18's is actually wider, deeper, more spacious.
• The C8 seems to have less dynamic compression at higher volume; the SP18 performs well with less impact.
• The C8 has an elevated mid-treble; the SP18 is flat throughout the treble.
• The C8's treble emphasis adds some artificial detail; the SP18 has more clarity in the midrange, with a balanced treble.
• The C8's lower midrange is fuller sounding; the SP18's is more lean.

A $2100 dollar per pair speaker is standing up to a pair that retailed for over eight times as much? But, is the C8 better? In some ways, yes. Better extension on both ends, fuller sounding lower midrange, better high volume level performance. The C8's treble emphasis is often quite enticing, and never fatiguing. Something that can only be attributed to the quality of the Piega ribbon and its low distortion. (An assumption based on listening only. I have not found any test results. It is the only ribbon tweeter I've heard that is tolerable.) Although, there are recordings where the treble becomes obviously unnatural. Brass, organ, and strings tend to get over etched. This never happens with the SP18. Soundstage on the C8 is excellent, yet more restricted horizontally. The SP18's soundstage is superior to any conventional box speaker I've heard, most likely due to the polar response and wide dispersion. This may also explain why it sounds better at low to mid volume levels. The sound appears to float in the air rather than coming out of a box. For the five-figure price, the C8 surely better offer more. The C8 seems to have greater power handling, more punch. But, the price difference is hard to justify. There are times when the C8 sounds better to me because of the fullness in the lower midrange/upper bass, and conversely, there are times when SP18 sounds more open and transparent for the lack of that thickening. Is what I'm hearing something the C8 is adding, resonance from the ported enclosure, or perhaps is it something missing in the SP18, the dip in the 100-200 Hz range? It may have to do with off-axis response. The omnidirectional nature of the SP18 creates a linear reverberant soundfield that preserves the midrange openness and 'air.' Highly reverberant recordings are more clearly represented by the SP18, whereas the C8 'dries out' some of the reverberation. I'll admit, the midrange thickening and reduced reverb imparts a pleasing smoothness. Yet, is it true to the recording? I suspect it may be a result of a higher Q, port resonance, less damping, and the passive crossover, along with a flatter upper bass response. I wish I had the test equipment to determine all that's going on in this area. That extra fullness is probably not in the recording, yet it does produce smoothness at the cost of resolution.

Once the subwoofer was added to the SP18M to bring in the bottom octave (a single powered sub made by a big name manufacturer), the sound became more complete. On many recordings, there is a definite need for the sub, on others the difference is minor. As I listen to a wider range of recordings with the SP18, I am finding its presentation and capabilities impressive for a simple two-way speaker.

More realistic comparisons can be made to other well known speakers in the $2-4k range. Without mentioning brands names, the first one that comes to mind is a small floorstanding two-way that retails for $2200, made by a well respected US company. It's their smallest floorstander. A very good little monitor. Almost bought a pair a couple of years ago. Two things stopped me. They're only 35.5” tall and the tweeter is several inches down from the top, which puts the tweeter well below ear level. The soundstage is likewise too low. They need stands to elevate them (but elevating them might mess with the time alignment of the sloped baffle). Second reason was, as good as they sounded, I left the audition with listening fatigue. Distortion perhaps? Their specs claim a low end response of -3 dB at 48 with a ported enclosure. The SP18M does as well with a sealed enclosure, which means the lowest frequencies are being produced by the driver, not by the enclosure resonance through the port.

The next comparison that comes to mind is with another well respected US maker. This time to a mid-line model that retails around $3500. These have great bass definition and extension (sealed acoustic suspension). Excellent soundstage and imaging. Beautiful smooth quality, very easy on the ears, but a decisively veiled sound. I kept feeling like there was a blanket between me and the speakers keeping clarity and detail at bay. I listened to the next model up costing nearly $8k. Better, more open, same great bass, but still veiled and lacking real presence in the midrange and up.

Third comparison is to a speaker of unique visual design, raved about by reviewers and owners. Retails for $3k. Another example of a small floorstander with the tweeter too far below ear level. Short soundstage. The vertical dispersion is limited, midrange mediocre, and the bass, though it has good extension, muddy. It was extremely disappointing considering all the raves, and unquestionably not worth anywhere near the asking price. Recently, I've listened to them again at a different dealer. The setup was better, which helped them sound more acceptable than the first audition, but not by much, the midrange was still constricted and the bass mushy.

Could the SP18 be improved? Certainly, but by how much, and at what cost? The modestly priced drivers used are excellent. Their performance confirms the test results from Zaph Audio, proving that mid-priced drivers are capable of delivering high quality. If top-of-the-line drivers were used instead, then there might be some improvement. Considering the price of premium drivers, it would more than double the cost. This may be hard to justify if the results are only marginal, though I suspect the major benefit might be lower distortion at very high volume levels and a more extended treble beyond hearing range. If there were anything that could improve them as they are, it would be leveling out the dip in the upper bass, and the rise in the 1k-2k range even though both are within the standard +/-3 dB tolerance. (See updates below.) As I listened to a wider variety of recordings, the only time I found fault is when I noticed one of those two weaknesses. The more I listened, the more noticeable they became. It finally came to a head when listening to a CD I had just purchased, a recording of Cherubini's Symphony in D, and not just for the reasons noted, but also for another major attribute I listen for, resolution/clarity. This recording sounded stuffy, boxy, congested, muddled, or what ever adjective you might choose to describe a general lack of transparency. When listening to the new CD on the Orion+, it was noticeably more open and transparent. Still, even with the Orion+, it was obvious that the recording itself was on the stuffy side, yet the difference between the Orion+ and the SP18M was obvious. This forced me to reconsider the crossover. Lowering the crossover setting to 1.2k opened up the midrange to a surprising degree. Previously I had settled on the 1.5k setting assuming that any lower would put too much stress on the tweeter. Yet, there was such an improvement with the sound that I had to give the lower setting further testing and extended listening time. With the crossover changed, retesting obviously showed a need to readjust the woofer/tweeter output balance; a boost to the low and trim on the high corrected the balance. This helped level out the unevenness noted above, both the upper bass dip and the elevation around the crossover point. Careful attention to the tweeter amp output and listening to recordings that would put extra stress in the upper midrange, didn't push the tweeters past their limits. The tweeters are taking the extra load all in stride. Lowering the crossover not only opened up the sound, it also smoothed the frequency response making a notable improvement, far more than I ever would have guessed.

(II) UPDATE : Since I decided to move on with building a subwoofer for the SP18M, it was necessary to rethink the crossover because a panel amp with its own crossover for the sub was ruled out. (See PX30 Design.) A digital 3-way crossover was ordered while waiting on parts for the subs. With EQ applied to level out the upper bass dip, an improvement was immediately audible. It filled out that missing oomph in the upper bass that was nagging me. Then I changed the crossover from a 24 dB per octave slope to a 48 dB and raised it slightly to 1.29 kHz. This will insure there is no over stressing the tweeter. The frequency response smoothed out to +1/-3 dB throughout its bandwidth. Now, I am really impressed. The Piegas, in the meantime, were sold, so I could no longer make a direct A/B comparison. Nevertheless, with the flattened upper bass, the reduction in the driver overlap, and resulting reduction in distortion, there's no doubt that the SP18M is better than the Piega in every way except bass extension. Until the PX30 was added, by which the design goals were totally surpassed.

The frequency response of a speaker appears to reveal far more than is given credit. It's rare for a speaker manufacturer to publish frequency response charts, but it would tell us a lot about its performance without knowing anything else. A simple frequency range without a +/- dB figure essentially tells us nothing. It must indicated a tolerance range to be relevant. (And even this alone, without seeing the full chart, can be deceptive. The standard +/-3 dB is a 4x, 6 dB deviation, over a broad range this can make for a very poorly balanced response. Some manufacturers also dance around the plus or minus figures by specifying only the frequency extremes, i.e., the response may be +/- 3 AT 30 AND 25k, but in between 30 & 25k the response goes well outside of the 3 dB tolerance.) Next most revealing test would be distortion, including the Cumulative Spectral Decay; a measure of how long the system keeps resonating across its bandwidth, which clouds the signal and smears the sound. Even if the frequency response is broad and smooth, high distortion and/or a poor CSD can spoil everything.

VI

Conclusions

There are three major attributes of the SP18M that explain the distinguishing features of, and the reasons for its outstanding performance.

First, is the active crossover/digital signal processor. Each driver is controlled by its own amplifier. No loss of power or damping caused by a crossover getting in the way. Each bandwidth is amplified separately, so the demands on the bass amp will not effect the treble amp or the tweeters, making an easier load for both. It cannot be over emphasized how critical it is to eliminate the passive crossover. An active crossover is the single greatest problem solver in speaker design. Add to this the high-order, low crossover point that gives more of the upper midrange to the tweeter. This is beneficial because the tweeter has a far cleaner cumulative spectral decay, due to lower mass, than the woofer driver.

Second, is the sealed acoustic suspension woofer in a rigid cylindrical enclosure with a large air volume. No stored energy from a port or passive radiator, no panel resonances, highly damped and low Q to deliver better transients and less group delay.

Third, is the primary design principle : point source omnidirectional, wide dispersion, linear frequency and polar/power response along with proper height of the tweeter. All these factors help preserve clarity, resolution, soundstage/imaging, and a reverberant soundfield that corresponds to the spectral balance of the direct sound. This reduces coloration in the reverberant field, which makes up a significant portion of the sum-total acoustic energy in the room, resulting in a more realistic sound.

The SP18M presents a real contest for every high-end speaker I've auditioned. I'm finding it more and more difficult to understand why speaker manufacturers don't design and build better loudspeakers at a more reasonable price point. It may be where they start with their design goals. It may be the assumptions they make about market demands. Assuming the market won't go for active crossovers, or relying on ported enclosures for bass extension, or more insidious, latching on to a single design parameter of lesser importance to artificially contrive a "unique" marketing angle, such as 1st order crossovers, or single driver systems, or exotic configurations/drivers, or massive and extremely expensive cabinets. These may be good pursuits, but not without their own compromises that are difficult or sometimes impossible to overcome, and usually at the risk of sidestepping more important priorities for audio reproduction. Conventional direct radiating box speakers are replete with unnecessary compromises. There's no more that can be done with that design paradigm.

The SP18M will hold its own against any other speaker in the $2-5k range, and with dual subs, I'd push that figure much higher. Fortunately, I'm not beholden to a marketing department or other commercial considerations. I've built these speakers to satisfy myself. They are designed to deliver high performance at low cost, with a minimum of compromises, strict adherence to priorities, and form-follows-function aesthetics.

For the latest conclusions and observations : Omnidirectional Loudspeakers : Soundstage, Imaging, Placement.

back to the Development page

continue to Subwoofer design

(|| ) Audio Home

© 2008-2024 Parallel Audio all rights reserved