Whetstones


Photo credit J.K. Andreasen - DIY Iowa 2009

"Single driver speakers for people who hate single-driver speakers."

"These might just be the very best single-driver speakers I've ever heard."

"They image at least well as my Maggies."

"I'm continuously amazed at how good these speakers sound...with everything."

"The bass is surprising for such a small driver."

1. Introduction

Note: This section is long; if you want to skip it entirely, you won’t be missing any information that you need to build the ‘Stones…there’s no penalty for invoking tl;dr here. It just describes where I was coming from when I designed the Whetstones. To me, they’re much more than a “driver in a box”.

Prior to this project, I’d experimented with full-range drivers on and off ever since getting into DIY loudspeakers for the first time. When Parts Express had the 4” Pioneer full-range drivers (269-570, the venerable “NSB”) for sale, I picked up four of them and experimented with them in a couple of enclosures, but was never impressed enough with them to make a serious design effort with them. I did use them in a pair of experimental transmission-line cabinets loaded at either end, which didn’t work as expected, and I generally found the drivers listenable, but underwhelming. I built another single-driver design as rear surrounds for a friend’s home theater system using the Tang Band W4-616S, a 4” paper-cone driver with a stamped frame, an aluminum phase plug, an enormous shielded ferrite magnet, and a very respectable 3.4mm of linear excursion each way. When a crossover component had come unglued inside the main L/R speakers (another of my early designs), I swapped the W4-616S-based surrounds in place of the mains while performing repairs on the mains, and was surprised at the three-dimensionality of the sound field, though of course not nearly as resolving of treble detail as the multi-way design of the mains. I chalked the imaging up as a novelty and replaced the main speakers, and didn’t think much of the experience afterward.

As I continued onward in my DIY audio journey, I continued to be intrigued by the fanatical level of devotion that single-driver speaker fans had for their one-way, one-driver, filter-less loudspeakers. This kept me from putting the single-driver concept firmly to bed in my mind, because even though I generally thought of filterless full-range designs as inferior (shouty, screechy, unruly, lacking low end power or definition), there was something that continued to fuel the single-driver fringe element in the hi-fi world, and I believe it might be the startling imaging ability owed to the minimum-phase behavior of the driver. A loudspeaker driver can be thought of as a ‘minimum phase’ device, though not a ‘zero phase’ device, because some phase rotations are noticeable in the measured phase plot of any driver that is encountering heavy cone breakup. Still, the minimum-phase designation means that a single driver with a single voice coil is as close to an acoustic singularity as you can get. The entire passband emerges from a single point at the apex of the cone, mostly in-phase with itself. Crossovers introduce phase rotations between the passbands, so care must be taken in designing a multi-way speaker to keep the phase plot as linear as possible.

The Whetstones (often referred to simply as the ‘Stones) are a filtered single-driver monitor design that began life during my senior year at Purdue as a thought experiment into an idealized single-driver loudspeaker design. While single-driver filterless designs remain something of a fringe element in hi-fi, they are still highly sought-after for their imaging characteristics, and for tonal warmth that seems to work well with low-powered tube amplification. I’ve long believed that the allure of tube amplification is due to euphonic distortion – that is, the amplifier itself adds a tiny amount of distortion that is pleasing to the ear. I carried this line of thought into the speaker itself and realized that many single-driver loudspeakers are also adding euphonic distortion to the system output, due to cone breakup and another phenomenon called ‘cone cry’ by which dry-fiber paper cones add additional self-noise to the audible output as their fibers rub against each other, trying to lock together to transmit the stimulus from the apex of the cone to its edge, with no matrix in the composite material to prevent this relative movement of the fibers. This ‘cone cry’ would be perceived as additional harmonic richness. However, I have always considered the addition of any sort of distortion to the sound system’s output (even purposefully) to be antithetical to the goal of hi-fidelity music reproduction, though it may increase the listener’s enjoyment if that happens to be the listener’s particular taste.

This realization about the nature of cone cry and the importance of material selection in wideband driver and single-driver loudspeaker design drove me to search for as ‘quiet’ of a driver as possible for use with the ‘Stones design concept. The quietest possible material would have a homogeneous composition and a high level of self-damping; polypropylene fit the bill, and Tang Band happened to offer just such a driver, the W4-1052S. The W4-1052S has a very thin poly cone with ridges molded into the diaphragm that kill ringing modes (by dissipating them chaotically through the diaphragm instead of letting them travel to the cone edge and reflect directly back to the center of the cone) and also add stiffness to the cone. It also has a Santoprene cone surround, a stamped frame and a classical long-coil motor design using Neodymium magnetic material. The Stones load this driver to a vented enclosure for bass reinforcement.

I also took note of the success of full-range driver designs by John “Zaph” Krutke, Ben “Wolf” Shaffer, and others which incorporated filtering networks to apply contouring to the driver’s response, notch filtering of the driver’s ringing modes, and baffle step compensation to infuse life back into the bass. These designs include Zaph’s design using the Hi-Vi B3S, and Wolf’s “Dragon Foal” design using the Hi-Vi M3N, and his “3CR” design using the Tang Band W4-1337S. In particular, I drew inspiration from the “3CR” design, which uses a driver similar to the one I had chosen, except with a cast frame and a titanium cone for an extended top end. However, one area of loudspeaker performance that can have a profound effect on the speaker’s character but is often neglected is how the driver behaves in the time domain. In comparing Zaph’s spectral decay measurements of the W4-1337S and the W4-1052S, the W4-1337S had a very long ring-out time at its breakup frequency in the treble range, which (after having the change to audition Wolf’s 3CR design) sounds like treble hiss trying to masquerade as extension. The W4-1052SD doesn’t have the W4-1337’s ultra-extended treble range, but it was much better-behaved in the time domain, and I think that makes the driver more conducive to frequency response correction through active or passive EQ because once the magnitude response is flat, there is less need to worry whether a ringing frequency is going to jump out at the listener due to having a longer decay time.

2. Enclosure Design

One issue that's peculiar to single-driver loudspeaker designs is the need to control internal reflections. The driver cone is radiating material across the entire frequency range, not just midrange and bass frequencies, and the energy that's radiated from the back of the cone into the enclosure needs to be dealt with in an orderly fashion. Careful application of damping material is crucial to prevent this stray energy from reflecting around inside the enclosure and then being re-radiated through the thin material of the cone, which can easily pass an incident wave from behind into the forward-facing output. I lined the entire upper portion of the cabinet with 1/4" thick SAE F-15 wool/poly felt from Southeastern Felt, attempting to attenuate the back wave of the driver after only a few reflections. I feel like this was exactly the right amount of damping to achieve enough back wave attenuation without making the enclosure sound over-damped.

Because the rear radiation from the cone includes upper-midrange and treble information, it's important to treat the interior of the enclosure not only as a pressure vessel for the purposes of bass loading, but also as an acoustic environment in its own right, affected by internal standing waves at the longer wavelengths that no amount of damping material can control (or 'too much' damping material would be required to control, stifling openness and musicality). The enclosure's shape is designed using an acoustic ratio chosen from a list published in Koonce and Weems' Great Sound Stereo Speaker Manual, which is a ratio of internal dimensions that spreads acoustic modes within the cabinet so the standing waves inside the enclosure cannot interact constructively at some frequencies and destructively at others - this provides for the smoothest possible upper-midrange reproduction. In addition, the location of the driver, port, and internal brace were chosen according to the Golden Ratio (1.6:1:0.6), with each being placed a distance of 0.6*cabinet height from the top or bottom of the enclosure. Placing the window-style brace according to the golden ratio also spreads the transverse resonant modes in the material of the enclosure panels, so that the enclosure itself is more harmonically expressive instead of seeming to 'ring'. It’s important to consider the contribution of panel resonances in a full-range design, since completely eliminating panel resonances would require carving the enclosure out of a chunk of granite, which would be thoroughly impractical.

For a speaker with a small monitor-style form factor that is expected to have any sort of bass output at all, a bass-reflex design is the only game in town. In an enclosure this size, any type of transmission line or horn loading is out of the question, which is fine by me – though transmission line and horn design have been mostly demystified in recent years, they are still more difficult to construct than a reflex enclosure. It seems to me that most horn or transmission line enclosures are attempting to wring additional bass extension from a driver that is reluctant to give it without adding significant distortion, which is then amplified by the resonant modes of the horn or transmission line. In this case, the driver appeared to be capable of acceptable bass output on its own, so no acoustic trickery was required to extract another half-octave of extension. Also, I wasn’t expecting to power this speaker with a flea-powered amplifier; the two amps in my stable that I had in mind for these speakers were the Sonic Impact T-amp and the NuForce Icon-1, each capable of 12-15W of output into the drivers’ 4-ohm load. So a bass-reflex design was settled upon, tuned to 65 Hz – the Fs of the driver.

With a full-range driver in a bass-reflex design, the problem of re-radiation of internally-reflected energy also affects the port. The port itself can pass midrange energy quite easily, behaving like a plane wave tube above the enclosure tuning frequency. In addition, the port has its own quarter-wave resonance, which must either be damped or accounted for. For the vent in the Whetstones’ enclosure, the calculated pipe mode for the vent is around 400 Hz, so in simulating the baffle step filter for the enclosure, I expected an additional 2dB of output from the vent near this frequency, and that led me to choose a slightly lower corner frequency for the baffle step filter than the textbook equation would predict.

I took every measure to simplify enclosure construction since I myself am a woodworking novice, and I wanted this project to be easy to build. I decided to standardize on ¾” MDF, since I have it in abundance, and for an enclosure this size, it results in fairly rigid panels. Butt joinery is used throughout, and only two screws were used. I incorporated a removable bottom panel in the enclosure design, to allow access to the filter components, which in my prototypes are mounted to the bottom panel using zipties mounted with screws, and additional dabs of hot glue to prevent buzzing. Because I wanted to be able to demonstrate the effect of the contour filter, I installed a Tyco Alcoswitch toggle with gold-flash-over-silver contacts into the terminal cup to bypass the filter network. The outer edge of the vent is rounded over, and all the outside edges of the cabinet received an ornamental roundover. I used a Jasper circle jig and my router to recess the driver frame.

3. Filter Design

The Whetstones are a filtered full-range design - you omit the filter at your peril. The filter is integral to the design. If the sight of a capacitor or an inductor makes you cringe, find another project and have fun listening to your vocal recordings. To me, what makes the Whetstones enjoyable with a wide variety of music is the very fact that they use a contouring network to flatten their response, so they neither sound shouty nor bass-starved. The Tang Band W4-1052S is virtually unlistenable without a contouring filter, in my opinion. With it, the driver discovers the rest of its range.

(Schematic from Audua Speaker Workshop)

ERRATUM: The above schematic is incorrect. The "2.700" resistor should be a 10 ohm resistor. The "2.400" resistor should be a 3.3 ohm resistor.

Right off the bat, anticipating a FAQ on the subject: You'll probably need to combine capacitors to get the 8.8 microfarad capacitor. This is fine to do. Capacitance values sum when in parallel with each other. The BOM provides a 6.8 microfarad and a 2.0 microfarad capacitor for just this purpose.

The 'Stones contour filter uses eight components, and has four discrete sections. The first section, comprised of the 1.8mH inductor and 5.6 ohm resistor, provides nearly 4dB baffle step compensation. This is just right for an application where the speakers sit on stands about two feet away from the back wall, but if you use these at your desk where they're much closer to the back wall, I'd cut the R1 value in half or else the speaker will sound quite thick, as I discovered through experimentation. I made some allowance for the pipe mode of the reflex vent, which occurs near 300 Hz, so the actual baffle step frequency is near 250 Hz. The .20 mH inductor and 10 ohm resistor provide the 'tilt' which counteracts the driver's rising response, while the .05 mH inductor, 3.3 ohm resistor, and 8.8 microfarad capacitor notch the cone breakup peak that occurs near 8kHz. I arrived at the initial component values using textbook formulas, then tweaked the values according to simulation results and available values in Audua Speaker Workshop. I added the 1.5 microfarad 'air' capacitor as an experiment in the simulation and I liked the results - above 15kHz, where the driver naturally begins rolling off, it 'gets back' some of what is being cut through the midrange to bolster the treble range of the driver and try to eke out a bit more top-end extension.

You'll notice that all the stages of the filter except for the 1.5 uF 'air' capacitor are in series. They represent a measured series resistance of 1 ohm.

(Prediction from Speaker Workshop; green trace is unfiltered, blue trace is filtered; click to enlarge)


The effect of the contouring filter on the W4-1052S response is readily apparent. The contour filter shaves up to 9dB off the W4-1052's rising response. Without the filter, the driver sounds screechy, shouty, honky, and virtually unlistenable without a great deal of reinforcement from cabinet resonance, or maybe reflected midbass output from a front-firing vent. (Ben "Wolf" Shaffer commented that the Stones' contour filter makes some fairly aggressive cuts, but I wanted to optimize the driver response as much as possible. The goal is not necessarily a high-sensitivity design but a music monitor with flat response and minimum-phase behavior.) With the filter in place, the predicted response is within +/-3dB across the board, neglecting the predicted rise from the baffle step compensation filter from 100-300 Hz. This graph also doesn't include the response characteristics of the enclosure. The FR curve of the unfiltered driver for simulation purposes was taken from the Tang Band literature.

(Plot of Measured Response, Close-Miked - vent output not included; click to enlarge)


Here's an Excel plot of the frequency response dataset for the 'Stones, close-miked on-axis in Brian Powers' basement using ARTA. (The notch near 350 Hz is floor bounce, and the output of the vent is neglected.) As you can see, the results follow the simulation very closely, though not much has been done to this data to clean it up. Through the midrange, the 'Stones measure flat - within +2/-3dB. I don't think the top end of this trace describes what's going on in the real world, since it appears to be falling off far too rapidly. I'm going with the ears on this one. Room reflections may be playing a role - this measurement was taken in a basement (far from anechoic), and a longer gate was used to try and pick up some of the midbass behavior, possibly at the expense of top end accuracy. These speakers have a respectable amount of treble output when measured on-axis, but when listened to from more than 15 degrees off axis, the treble response falls off rapidly.



4. Bill of Materials

This is what you'll need to build the Whetstones, minus the cabinetry, finish, and mounting hardware:

PE Part Number   
 Item Description
 Qty Unit Price
 264-828 Tang Band W4-1052S 4" Driver (rev. may vary)
 1 per
 $27.50
 255-264 Jantzen 1.8mH 18AWG Air Core Inductor
 1 per
 $10.44
 255-200 Jantzen 0.05mH 18 AWG Air Core Inductor
 1 per
 $2.81
 255-210 Jantzen 0.20mH 18 AWG Air Core Inductor
 1 per
 $4.00
 004-3.3
 Dayton DNR-3.3 3.3 Ohm 10W Resistor
 1 per
 $1.25
 004-5.6 Dayton DNR-5.6 5.6 Ohm 10W Resistor
 1 per
 $1.25
 004-10 Dayton DNR-10 10 Ohm 10W Resistor
 1 per
 $1.25
 027-412 Dayton DMPC-1.5 1.5 uF 250V Polypropylene Cap
 1 per
 $1.12
 027-414 Dayton DMPC-2.0 2.0 uF 250V Polypropylene Cap
 1 per
 $1.48
 027-424 Dayton DMPC-6.8 6.8 uF 250V Polypropylene Cap
 1 per
 $2.79
 260-313 Recessed 5-Way Binding Posts
 1 per
 $3.49
    
  Total Per Speaker
  $57.38
  Pair Total
  $114.76

All of these items were purchased at Parts Express for this project. For my prototype, a Tyco Alcoswitch DPDT toggle with gold flash over silver contacts was used to bypass the contouring network for demonstration purposes. You can use a similar technique to add the ability to switch the baffle step compensation in/out, but I don't recommend bypassing the other stages of the contour filter. The 3/4" MDF and 1.5" Schedule 40 PVC was purchased at Lowe's, but you can typically find these at any plumbing/lumber/home improvement store in your area. The speaker wire, spade terminals, etc. need not be fancy.

Prices are listed for reference only. Still, your $120 takes you a long way.

5. Listening Impressions

Punchy low end, smooth midrange, slightly subdued treble and upper midrange that lends itself to a darker coloration, though satisfying. Overall, a nice speaker for music use in a small bedroom sound system. They are not the most sensitive, and they have a narrow sweet spot - best used at your PC or somewhere that you'll be seated while listening to them. They do well with a Sonic Impact T-amp, which is what they were designed for, but in a larger room they really open up dynamically when fed by my Dayton APA150. I'll fill this in more as I feel the need to wax poetic about my achievement. If you build these, tell me what you think and I'll post your thoughts here alongside mine, even if they're less than complementary. One criticism is that their midrange can (rather counterintuitively) sound slightly 'closed-in', and while their treble reproduction is good for a single-driver design, they sometimes leave me wishing for a supertweeter. In the end, the Whetstones do what they were designed to do very well, and I'm satisfied enough with the design that I'm publishing it.

The Whetstones' design won third place in the 2009 Dayton DIY competition in the under-$200 category, upstaged by a two-way mini-monitor and a three-way standmount. It was the only single-driver speaker to place in the competition among the three that were presented.

(Update - and a bit of bloggery: It seems to me like the Stones have only gotten better with time. The treble seems more extended, and the midrange sounds smoother to my ears, as though the already very clean time domain performance of the driver has improved with break-in. These are honestly the best single-driver speakers that I have ever heard, blending smooth response with spectacular imaging, and I've heard several popular designs using drivers by CSS and MarkAudio, though someday I would still like to hear a design using Fostex drivers. I have become more enthusiastic about these speakers' performance over time, rather than less enthusiastic, so I consider these a successful design. When the price is considered, they are nothing short of stunning.)

6. Downloads

Enclosure Plans (.jpg)

Enclosure Plans (.pdf, printable, ANSI B-Size)

Filter Schematic (.jpg)

7. FAQ's

Here I'll try to anticipate some of the questions I expect you'll have about the Whetstones design.

  1. What is the sensitivity of the Whetstones? The Whetstones are about 82 dB @ 2.83v/1m sensitive. Really, the sensitivity of the W4-1052S driver is overstated in the manufacturer's specifications, since it is an average across-the-board sensitivity rating that includes the driver's rising response (shoutiness) and peakiness (breakup/hash). After alleviating the above two conditions, applying sufficient BSC, and taking note that the polypropylene cone has high self-damping, 82dB is closer to 'right' for this driver. You'll get back about 4 additional dB of sensitivity if you eliminate the baffle step compensation, which I don't recommend unless you plan to EQ the Whetstones' lower range in your application. You can also recover some sensitivity by cutting the BSC resistor value in half if you use them close to a wall. Still, these are for use with smaller solid-state amplifiers, not flea-powered tube amps.
  2. I want to use the Whetstones for _____. Would you recommend this? A speaker's suitability for a particular application is dependent on more than just the speaker's size, frequency response, and power handling. The Whetstones have a fairly narrow sweet spot as is inherent in the majority of single-driver designs; 15 degrees off-axis is the furthest I'd sit from an on-axis position. Any further off-axis than that, and you begin losing treble extension. I've found the Whetstones work well at the PC (which is where I originally intended for the design to be used) or placed on stands in the living room of my apartment for music listening. Their narrow dispersion may make them unsuitable for use as surrounds in a home theater system. I strongly suggest using them in an application where you sit down to listen to the system rather than moving around the room.
  3. Can the Whetstones outperform a properly-implemented two-way system? I've received quite a bit of positive commentary on the Stones from virtually everyone that's listened to them, and while they come within spitting distance of multi-ways, I don't believe they can outperform a good two-way loudspeaker, but neither can any other single driver speaker, in my opinion. The aim in designing the Stones was to idealize the single-driver loudspeaker concept in an unconventional way, and I feel I've succeeded. The result is a minimum-phase reproducer with flat response in-room across eight octaves and acceptable bass performance. If I discover a tweak that really supercharges the Stones' performance beyond what I've already published, I'll mention it in this article.
  4. You're a monster for using a filter on a full-range driver! I know, but give them a listen and tell me that the concept doesn't work. I think you'll find that it really does. I know it's a little 'Bose'-y, but Henry Kloss was known for similar passive filtering of small wideband drivers (I own a specimen), and filtering for flatter response enhances the usefulness and broadens the appeal of the speaker. If you're going to be a purist and demand a filterless design, there are other designs out there for you.
  5. Why on earth did you pick the name “Whetstones”? When I developed the design for these speakers, I had been reading audio reviews by David Kan at 6moons.com, and for a while even considered naming these speakers the “Kans” in his honor, since they seemed like the type of product he often reviewed (small single-driver loudspeakers for use with smaller amplifiers) but Linn beat me to the name. I had chosen the dark-gray color in advance to complement the black cone and silver phase plug of the drivers, so I eventually settled on the name ‘Whetstones’ (“Stones” for short), since the enclosures have a similar shape and color to a whetstone, and if I do say so myself, they look pretty sharp. As a side note, in the unlikely event that David Kan reads this, and he happens to be interested in evaluating a pair, I’d be happy to build a set in his choice of color.
Happy building, and if you have any questions, PM me via the Parts Express Tech Talk message board.

Comments