The impetus behind this project was my sister's fiancé's need for a high output, low distortion, loudspeaker with excellent low bass extension. The caveat was to achieve these goals on a budget under $ 500.00 per pair. It is always a challenge to design a good loudspeaker, regardless of price. But it is even more difficult when cost is an issue. Throughout the design, I tried to maximize sonic performance for every dollar spent, and searched for the highest performance to price ratio components on the market today.
The Tonegen/NHT 1259 woofer is an extremely well made, high quality, heavy duty woofer. It has gained considerable admiration on the Internet, and has a growing history of successful amateur designs, not to mention its critical acclaim as used in the NHT 3.3 loudspeaker system. It is distributed through Just Speakers, has a thick textured polypropylene cone and dust cap, heavy duty cast basket, large half roll rubber surround, extremely long excursion, and vented voice coil that can absorb 300 watts of power. There is a considerable amount of excellent information pertaining to this driver on the Internet, and I encourage you to research this impressive driver. And if you mention the Internet to Just Speakers when ordering, you can save quite a bit of money off their regular price for the 1259.
The 1259 is optimized for a sealed enclosure. This design uses a 2.8 cubic foot enclosure. As recommended, the enclosure was stuffed with approximately 48 ounces (approx.1 pound per cubic foot) of acoustic grade polyester fill to increase the enclosure's "effective volume"; this lowers the systems Qtc, box resonance frequency, and bass cutoff (F3). Depending on room interaction, this woofer is said to be capable of producing bass into the low 20 Hz. region with significant output, low distortion, and minimum phase disturbance.
Many designers believe that one driver should cover the critical frequency range of approximately 300 to 3000 Hz. This keeps crossover disturbances and other problems out of the bandwidth where the ear is the most sensitive. However, there are relatively few drivers that can reproduce this wide range well. The Tempest utilizes the Vifa P13MH midrange. The P13MH is the midrange version of the Vifa P13WH woofer. The P13WH and P13MH are both excellent sounding drivers, and considering their relatively low cost, I believe they are among the best polypropylene drivers available in terms of performance to price ratio. Vifa's P13 drivers have not gone unnoticed by several "high-end" loudspeaker manufactures. Notable features include a cast magnesium frame, rubber surround and dust cap, and a robust, vented magnet structure.
The midrange's performance is almost identical to its woofer sibling, albeit with a bit more high frequency extension, at the expense of a slightly less precise high end rolloff symmetry. Its response 30 degrees off-axis is on par with its response on-axis to about 3.0 kHz.
The Tempest uses Vifa's 1.85 liter isolation chamber to subenclose the P13MH and prevent woofer interaction. The chamber (a thick, tapered, plastic cylinder) was filled with acoustic grade polyester to lower the resonance and impedance peak, and attenuate rear cone output so as not to interfere the driver's primary output.
Incidently, the fact that both woofer and midrange cones are both fashioned from polypropylene plastic helps insure they have similar "sonic signature" or "voicing". In addition, the two drivers look good together on the baffle. This is sonically insignificant, but adds to the design's aesthetic appeal.
This Vifa aluminum dome is highly regarded, and is one of the best tweeters available today. Its rigid aluminum dome allows piston like behavior to the limit of human hearing. It has a relatively low, well controlled, fundamental resonance provided by an aperiodic subenclosure and ferrofluid damping fluid. It has a very high power handling capability if crossed over properly, and has excellent of-axis response. Its off-axis dispersion stands out among tweeters, as its performance 30 degrees off-axis is almost identical to its on- axis response to 20 kHz. A state of the art tweeter in all regards, and is used in commercial loudspeakers costing thousands of dollars. And at around US $26.00, I believe it represents an unparalleled value among tweeters.
The crossover was designed with the help of the excellent CALSOD computer model. At the outset, it was determined to keep the crossover as simple and uncomplicated as possible. This was possible because of the excellent characteristics of the drivers, which do not require substantial crossover manipulation to compensate for shortcomings intrinsic to the drivers. The crossover presented here is the one that "sounded" best to us after listening and tweaking. Second order filters are used to keep driver distortion low and preserve power handling. A schematic of the circuit is included with this article along with a CALSOD plot of theoretical response, impedance, and phase response.
The tweeter crossover occurs at 3200 Hz. using a 2nd order electrical filter to provide the required acoustic transfer function. A bypass capacitor bypasses the tweeter's series capacitor to reduce hystersis and improve transient response.
The midrange crossover consists of a bandpass circuit that rolls off above and below 3200 Hz. and 300 Hz. respectively. The bandpass circuit employs 2nd order electrical filters to yield the required acoustic transfer function at each end. The bandpass filter works in conjunction with the midrange's acoustic rolloffs (determined by the subenclosure's resonance characteristics and the driver's high frequency rolloff). The bandpass circuit uses a transposed component layout instead of cascaded sections because it yielded better phase and frequency response. The midrange polarity is reverse phase relative to the tweeter and woofer to place the drivers in phase around the crossover frequencies. The midrange series capacitor is bypassed by a 0.47 polypropylene capacitor to minimize hysteris loss and improve transient response.
The woofer is crossed over at 300 Hz. using a second order electrical filter. The woofer is well attenuated at the point where its upper frequency response suffers from cone breakup and directionality beaming.
The crossover uses Bennic and Solen polyproylene capacitors throughout, except for the large
value capacitor in the woofer filter, which is a 100 volt electrolytic. All inductors are air core except the large value in series with the woofer; it is iron laminate to keep dcr low and cost down. The inductor's dcr and its effect on woofer efficiency and Qes/Qtc are considered in the CALSOD model. The large 70 mfd. Solen polypropylene capacitor in series with the midrange was on sale for $14.00 (about ½ price). If you are really on a budget, you could gang together several mylarTM and/or polypropylene capacitors in parallel to achieve large capacitance values. Often you can buy "surplus" mylarTM and polypropylene capacitors at bargain prices from Madisound, Speaker City, and other distributors. I do not recommend electrolytics in series with the midrange and tweeter. There is a limit to reducing cost, and at some point sound quality will suffer disproportionately to monetary savings.
A photograph and conceptual drawing of the enclosure are provided with this article to show what the finished loudspeaker looks like. The enclosure is veneered in cherry wood. The front grill is removed to show the relative position of the drivers. The L-pads and input terminal are located on the rear of the enclosure.
Loudspeaker enclosures can take many forms, and some shapes exhibit superior acoustical properties when compared to others. Although a rectangular box is not the best shape acoustically, most builders inevitably choose this shape. Often aesthetics will override acoustical considerations and you may have to compromise by building a "less than ideal" enclosure shape that will fit physically and aesthetically into your living space. Believe it or not, not everyone wants a 4.0 foot diameter sphere made of Corian TM magnetically levitated off the floor (to prevent vibration transfer) in their living room.
The materials used are also up to the builder. We used a good grade of 3/4 particle board and braced the enclosure extensively. MDF is used by many builders, and is acoustically superior. But it is expensive and often hard to find. Again, it is up to you, and how much money and time you want to invest in the eradication of enclosure vibration.
In all seriousness, I encourage you to experiment with foam and felt damping pads, rounded edges to eliminate edge diffraction, DeflexTM Panels, sand-filled panels, MDF, etc. Surely there is room for improvement. But for those not so inclined to experiment, take heart, the simple, well built, inexpensive enclosure presented here will provide very good results.
A final note, I recommend that you build or buy a sturdy stand that is tilted/angled back. A stand should improve bass response, and if it is angled, will contribute to aligning the driver's acoustic centers. This may further improve imaging by synchronizing the arrival time of the sound of all three drivers.
Upon completion, and after listening tests, an audio generator, frequency counter, and Radio Shack SPL meter were used to generate the "nearfield" amplitude response curve. The graph was compensated for the SPL meter's high frequency rolloff beyond 15 kHz. A graph of the frequency response curve is included with the article.
The measurements were performed with the Tempest positioned on the floor (no stand), 8.0 inches from the rear wall, no furniture within 3 feet, and the grill removed. The listening room was a 36 x 24 x 7.5 foot finished cellar, with a carpeted concrete floor, drywall walls, and acoustic ceiling tile. I consider the room to be absorptive ("dead"), as opposed to reflective ("live"). The measurements were taken at low acoustic output, with the microphone close to the drivers (nearfield), to minimize the effect of the room on frequency response. I believe this curve gives a good idea of how this loudspeaker will preform, with respect to frequency response, in a similar room under similar conditions. One amplifier watt produced an 89 dB sound pressure level, 1.0 meter away, on axis, in this room. In a real listening environment (not anechoic chamber) the Tempest's overall response is within +/- 3 dB of the 89 dB reference mark, from 28 to 20 kHz., with two exceptions (explained below).
Please note that loudspeaker response below approximately 300Hz. is largely dependent on the listening room. Also, Vifa's response graph shows that tweeter will respond to 30 kHz. However, I can not confirm this because that is well beyond the range of the Radio Shack SPL meter's microphone.
One of the two exceptions to the +/- 3 dB tolerance is a peak in low bass centered around 36 Hz. When I first saw the acoustic response curve for the NHT 3.3 posted on the Internet, I noticed it had a low bass peak also. I am not sure whether this phenomenon is related to room interaction, my measurement techniques, the loudspeaker itself, or all of the above. In any event, we heard no audible evidence of this peak during music.
I believe the second anomaly is a manifestation of Roy Allison's "Wall Dip". I have read Allison's published work on this phenomenon and it seems to explain this dip in response. In fact, years ago, Acoustic Research and Allison Acoustics used a "side-firing woofer" loudspeaker configuration to compensate for this "dip" that plagues almost all conventional loudspeakers. Probably not coincidentally, the NHT's 3.3 also uses the "side-firing" configuration. I did manipulate the 300 Hz. crossover to minimize the "wall dip" effect. Fortunately, in the Tempest, the dip is represented by a modest trough in response between 200 and 300 Hz., and would likely change depending on the loudspeakers position in the room. We did not find it audible.
In addition to a frequency response sweep, we used a compact disc test recording to evaluate the Tempest's imaging, phase, presence of crossover "suckout", dynamic range, etc. The disc contains pink and white noise, individual instruments, etc., and does a fairly rigorous job at evaluating different aspects of a loudspeaker system. The Tempest faired extremely well on all tests.
Upon initial listening tests two things were apparent. First, in comparison with Craig's 2-way (we will call them Brand X so I am not sued) commercial loudspeakers (8.0 inch woofer and 8.0 inch passive radiator) the Tempests sounded "bass shy" when listening to rock and roll and heavy metal music. I surmised that the X's had an emphasized mid bass at the expense of low bass extension. In fact, many people have grown accustomed to overemphasized midbass, often using their receiver's loudness control to further boost this range. Many listeners associate speakers that have emphasized midbass as having good bass; and that is fine; who am I to judge. Indeed, most music does not contain much bass below 40 Hz., and extended low bass response is generally less noticeable than emphasized mid bass. Loudspeaker manufactures have known this for years, and often design high "Q" acoustic suspension, or nonflat (peaked response) vented enclosures to make their speakers "saleable". They know that "the general public" will buy a mid-bass heavy loudspeaker over a loudspeaker with taut extended bass response.
So, I pulled out my sound pressure level (SPL) meter and sine wave generator and did some measuring. Sure enough, the X's had a pronounced bass peak at around 55 Hz., and rolled off rapidly below that frequency. I measured the Tempest, and found the NHT 1259 woofer to be flat in response to around 30 Hz. It rolled of gradually below that point, but still had output into the 20 Hz. region. Not leaving good enough alone, I fed a 30 Hz. sine wave into the Tempest to achieve an output level of over 105 dB to demonstrate its low bass capabilities. The NHT's exhibited no audible harmonic distortion, just a pure and VERY powerful fundamental 30 Hz. This driver has an amazing throw ! In fact, the 1259 actually modulated my voice at this output, an effect similar to talking into a fan. In comparison, the X's cones slapped at the air violently at 30 Hz., but could only produce harmonic overtones (sounding like a motorboat "put put"). This is not to criticize Brand X loudspeakers, which are very popular, and own a "lion's share" of the commercial loudspeaker market. However, they were clearly outmatched in low bass capability.
This led to a tweak. We removed the 3.0 lbs of polyester stuffing to "plump up" the midbass by decreasing the "apparent box volume". Although stuffing lowers Qtc and box resonance frequency, we discovered the tradeoff was a significant loss in overall efficiency because of frictional losses associated with the woofer working (pushing air) against the surface area of all that polyester fill. There are techniques to minimize such losses, such as not putting the damping too close to the rear of the woofer basket or using a buffer of loose damping directly behind the woofer, grading into more compressed damping in the rest of the enclosure. All I can say is experiment for yourself by changing the box volume with solid fill material and/or experiment with different quantities of acoustic damping material. In this design you can subjectively alter the bass quality, to some degree, in order to make it "sound" the way you like it. The bass sounds "right" now, but going "cold turkey" and kicking the "peaked midbass" habit in favor of flat extended bass can take some getting used to.
First listening also revealed a slightly forward sound in this listening room, and the loudspeakers sounded better with the midrange and treble slightly attenuated (hence the addition of the tweeter and midrange l-pads). The attenuation yielded a softer, laid back sound. Indeed, the sound of these loudspeakers is very dependent on the source. These loudspeakers can be merciless on bad recordings, and reveal everything you "never wanted to hear". On a positive note, they can also reveal "good" sounds you never heard before. I realized I could not build a loudspeaker that will sound "the same" in any conceivable listening environment. Therefore, the addition of the l-pads gives some control over uncooperative room acoustics.
We believe this loudspeaker sounds very good! Bandwidth is near flat and borders both subsonic and ultrasonic. Its wide dispersion presents an excellent and very large sound stage that everyone can enjoy, regardless of seat. The Vifa midrange and tweeter are detailed and dynamic, and of course, the NHT 1259 produces a tremendous amount of low-distortion, low frequency bass. And for you Rock and Rollers and home theater enthusiasts, these loudspeakers can absorb a tremendous amount of power, and play EXTREMELY loud (without distortion) while doing so. They are not some fragile, "don't play me loud ... I only image" loudspeaker. They will blow you house down man ! And if you are interested in the home theater/surround sound thing, these nicely compliment the Esquires. So if you want to build a loudspeaker that does not cost a fortune, and is just plain fun to listen to, perhaps a pair of Tempests is the ticket for you. We think the finished product is commensurate with the effort. Good Luck.Scott Blaier