top of page

Kimchi Pot Speakers

Getting high-end sound on a modest budget takes a little imagination. And earthenware.

The last time I was in Vancouver to sift through some stored worldly possessions, I came across some high-end speaker components I'd bought ages ago for an unrealized project. They'd languished in boxes because I'd been completely satisfied with the largish speakers I'd built several years before. Those were 12” Tannoy Monitor Golds (circa 1968) in mahogany bass reflex cabinets I designed with the help of an early edition of Vance Dickason's indispensable Loudspeaker Design Cookbook. For the record, they were asymmetrical pentagonal prisms with angled baffles. This approach minimized internal and external reflections and put the straight port in a favorable spot near the driver. It was also hard to go wrong with the highly engaging dual concentric Monitor Golds and their smooth top end. They sounded wonderful: big and musical, even if they weren’t particularly detailed. I was very happy with those speakers - at least until a tenant's boyfriend fried them. But by then I was moving around a lot, so I sold them and joined the masses whose music came primarily from headphones or earbuds.


But when those forgotten drivers resurfaced, I brought them back to my temporary home in Seoul, Korea, knowing they stood a better chance of finding an application if they were close at hand. I started a speaker project folder where I collected all the relevant data I could find, including spec sheets, designs incorporating similar drivers, observations from other builders, etc. I didn't have a specific design approach in mind yet – I was just scratching around for inspiration.

Canadians and Americans are never more than a close relative away from a decent woodworking shop. I had cut the mahogany pieces for my aforementioned pentagonal tower speakers in my brother's basement, then bought some power tools of my own when I moved out of borrowing range. Most weekends involved at least one ‘necessary’ trip to an immense home improvement center. Until my older sister married an Englishman, I believed that all men everywhere grew up hammering pieces of wood together, and spent the rest of their lives and disposable incomes doing much the same thing. But in South Korea, there were people who called an electrician just to change a light bulb. Seriously. Few households there had screwdrivers, let alone a table saw. This meant that my long-delayed speaker project would have to be completed with a modified cabinet kit, or through a clumsy, sketch-intensive dialogue with a Korean furniture builder, or by some other means.

Most speakers - and especially homebuilt ones - are in the form of wooden boxes for reasons of convenience. However, the ideal shape for a speaker cabinet would be a sphere: no parallel surfaces, hence no acoustic reflections to muddy the sound. A sphere also has uniform rigidity, so it doesn't affect acoustic directionality the way a box can. The next best shape would be a cylinder for similar reasons. Neither shape is likely to emerge from anyone's brother's table saw, so we try to make the inevitable wooden boxes stiff, dense, and proportioned to minimize unwanted resonances. Or consider alternatives, like kimchi pots.

Kimchi is a Korean staple food, served with almost every meal. It's made from fermented cabbage spiced with red pepper paste, garlic, anchovy paste, etc. Koreans take great pride in their own family's kimchi recipe, making batches at home in earthenware vessels too large to keep indoors. The mere smell of some interpretations can knock some Westerners flat. Others love the stuff. Thanks to kimchi's ubiquity, the pots it's made in are readily available in many sizes. They resemble outdoor plant pots, but their insides are glazed, and there are no drain holes.

There was a shop near my Seoul neighborhood's major crossroads that had evidently been selling kimchi pots for decades. Rows of the dark brown vessels lined the street for perhaps a hundred feet extending from the tiny storefront. The pots were out there day and night, uncovered, unlocked. No one ever stole any. That's Korea. The unguarded inventory was actually an area landmark, with “go left at the kimchi pots” appearing in the directions to countless expat rooftop barbecues and going-away parties.

Once I hit upon the idea of using kimchi pots as speaker cabinets, I had some numbers to crunch. With a fairly high Qts of 0.61 and a middle-of-the-road EBP of 72, the Morel MW166 woofers I had were reasonable candidates for acoustic suspension - what I like to think of as the forgotten alignment. Acoustic suspension speakers are sealed boxes whose enclosed air mass acts as a spring, providing a restoring force for the woofer cone's excursions. The concept dates from the 1930's but didn't hit its stride until the 1950's when woofers with a loose enough suspension were created. Compared to the room-filling exponential horn designs that preceded them, acoustic suspension speakers produced a surprising amount of clean, controlled bass from a simple, home-friendly cabinet. The gentle 3dB per octave bass rolloff starts a bit higher than in modern ported designs, so acoustic suspension is not known for bringing the thunder. That said, there's a lot of nice, taut, clean, realistic bass to be had from this now largely neglected design. I was keen to try it for my new kimchi pot speakers.

For an optimum Q of 0.707, the MW166 was going to need a 1.48 cu ft (41.9L) enclosure. According to Morel's published specs, it would be down 3dB at 53Hz - not too shabby for a 5” woofer. The usual practice of stuffing the enclosure with polyester pillow filling would increase the effective volume by about ten percent, so I was in the market for two kimchi pots of about 38 litres each. That's roughly twice the volume of a water cooler bottle. Lots of kimchi pots were in that size range. So far, so good.

I'd also need a pot with a mouth wide enough to house the woofer and the excellent ScanSpeak D2905-970000 textile dome tweeters I had. I'd have to mount the drivers on some sort of baffle – I hadn't quite figured that out yet. The pots shouldn't be too tall and narrow, or there'd be pipe resonances to deal with. It seemed sensible to look for pots with a golden ratio relationship between height and diameter, for both resonance and aesthetic reasons. I plugged the target volume of 38 litres into the cylindrical volume formula V=πr²h and solved for the golden ratio where h=1.618d, and found I'd be looking for kimchi pots about 50cm high with an average inner diameter of about 31cm. The pots are not perfectly cylindrical, of course; they're all hand-made, so I'd be lucky to find two that were roughly the same. I headed down the hill to the kimchi pot store, tape measure in hand.

The biggest obstacle to the kimchi pot shopping trip was the traffic along that busy narrow street. A regular string of pedestrians hiked alongside the row of pots as cars and buses whizzed by inches away. Not much allowance was made for lunatic foreigners fumbling with their tape measures and smartphone calculators, let alone the extra real estate required to dismantle the wall of giant pots to reach a potential candidate. After twenty minutes of borderline suicide, I finally exposed two pots that completely fit the bill: their heights, average diameters, and opening sizes were all perfect, and the two were as close to identical as I was ever going to find. There was a wide flat rim to support the speaker baffle. They even had a nice simple decoration their maker had etched into the wet clay – possibly with a stick or a finger – making them more home-friendly than your average speaker enclosure.

The tiny ancient woman who'd been running the kimchi pot store since just after the Korean War vigorously scrubbed the layers of road grime from my finds while I trotted to the ATM. The pots were too heavy to carry, so when I got back to the shop I asked her in my halting mock Korean if she used a delivery service. She suddenly threw up a hand and a passing cab driver instantly stopped, popped open his trunk, and tried to assist her in loading the two pots. Korean taxi trunks universally contained an LPG tank, dramatically limiting their luggage capacity. Some also contained a hodge-podge of the driver's personal stuff: gloves, umbrellas, changes of clothes, boots, newspapers, etc. The kimchi pot lady clearly had a long history of dealing with taxi trunks, and ably arranged the contents to provide protective support for the two large ceramic pots. On the short ride to my place the driver asked me how much I'd paid for the pots. I gave him an insanely low figure to avoid hearing a painful recounting of how I could've gotten a better deal in some far-flung corner of Seoul. I didn't care about finding a bargain - I wanted to help the little old lady whose offspring were most likely waiting to close the shop the moment she expired. I'd tipped her. It felt great.

Next, I had to find round baffles to fit the pots. Remember, I had no jigsaw or nearby home improvement center, so the baffles would have to be repurposed from something at hand. The round wooden kitchen cutting board my mother used for years would've been perfect, so on a restaurateur friend's recommendation I took a bus to a big kitchenware market to look through the stalls. With the help of some red-clad tourism volunteers, I found a stall jam-packed with stacks of semi-professional kitchenware. Normally, in Korean markets you’re not supposed to dig through the goods to find something yourself; you’re supposed to ask the proprietor, who can reach into a congested pile of wares and quickly extract what you’re looking for. Thankfully, the shopkeeper's adult son understood English well enough for me to convey what I wanted and why. After failing to find an appropriate cutting board, we eventually unearthed what seemed to be a sort of traditional wooden hotplate or trivet for some kind of kitchen vessel. It was too small, but otherwise perfect. It had a rounded edge, was nicely finished, and even had some knots that would complement the kimchi pots' rusticity. The son said he'd try to get some larger ones if I could come back in a few days. I told him anything with a diameter between 30 and 32cm would work.

Three days later, the shopkeeper called me. “This is Lee! Wood circles! I have! You remember?” And he was right. They measured a perfect 31cm across, cost me very little, and looked like they'd been made to sit on top of my kimchi pots. They even made a satisfyingly deep “toom” sound when I plopped them onto the empty pots. Things were looking good, even though I had no idea what those wooden discs actually were. I still don't. I cut the holes for the drivers by hand with a small coping saw I found at an art supply shop near Seoul’s largest arts university. The wood was thankfully soft enough that I could “drill” the starting hole with a small blade screwdriver and some elbow grease.

Next came the crossover design. Second order Linkwitz-Riley seemed to be the way to go for an even response at the crossover point, as long as 12dB per octave was steep enough to protect the tweeter. The Scanspeak D2905-970000 has a low-ish Fs of 500Hz, so I knew I'd want to cross over between one and two octaves above that, or 1-2kHz. I also knew it would be ideal to space the drivers’ centers at 7/12 of the crossover wavelength if possible, although the baffle size allowed very little flexibility. The drivers pretty much had to be directly adjacent, making their centers 132mm apart, which is 7/12 of the wavelength of 1522Hz. Fortuitously, this is only 2Hz from the midpoint between F#6 and G6, meaning that the crossover point would fall smack between two notes on the musical scale. Of course, tolerances in the driver and crossover specs make this sort of precision a questionable pursuit, but I'd always rather design for it anyway and adjust what can be adjusted.

Crossover calculations always yield component values that don't exist in the real world. You can always wind your own coils (which I did), but the capacitors can be tricky. The values I needed were 8.71uF and 6.53uF. My solution was to buy a bunch of 6.8uF and 2.0uF mylar capacitors with 10% tolerances, and measure them all to find the closest matches to the target values (using a 6.8 and a 2.0 in parallel to get close to the 8.71). This approach was possible because I was only a short bus ride away from what I called “The Warren”, a seemingly endless maze of tiny electronics parts shops in an older part of Seoul that has since been razed and redeveloped. I went there a lot. I was the only foreigner I'd ever seen in there. Good-quality parts were astoundingly cheap even in small quantities, and I could find almost anything with a little work. My terrible Korean language skills were less of a hindrance in that techno-Brigadoon thanks to the near-universality of electronics terminology, and I could always fall back on a scribbled schematic symbol. Anyway, with capacitors, a 10% tolerance doesn't mean that there will be significant drift in normal use: you can pretty much expect the cap to remain close to its measured value. I lucked out, needing only 10 pieces of each to find combinations well within 1% of the target values.

Winding the coils was quite a bit trickier. The formulas I had were all for AWG, but wire gauges are different in Korea. I bought 0.80mm diameter magnet wire, whereas 20AWG is 0.83mm. My solution was to overwind by a few turns, then tune as needed. One inductor needed a form of exactly the same diameter as the cinnamon jar in my kitchen cupboard. The other needed a form slightly larger than one of narrow soft drink cans there, which I built up to the correct diameter with a few layers of thin duct tape. I put a single layer of very thin double-sided tape between each layer of windings to keep the coils together. The magnet wire spool was suspended on a pencil across the mouth of a small bowl to keep the winding process tidy and kink-free. Counting the turns and keeping the windings neat was slow work, but I was fairly happy with how the four coils turned out. I left about a foot of extra wire at each end, and bound the coils with nylon tie wraps.

I didn't have an inductance meter on hand and didn't want to buy one there just for the speaker project, so I measured the coils by wiring up the crossovers' high-pass and low-pass sections separately into dummy loads, feeding them with an oscillator app on my smartphone through a small transformer, and using a DVM to measure the AC voltages across each component. The initial accuracy was surprisingly good, and I only needed to add or remove two windings at most to get each crossover section to -6.0dB at 1524Hz. Both finished crossovers were matched within 0.05dB. Not bad for a first attempt!

I'm a big fan of genuine point-to-point wiring - a term that gets commonly misapplied to eyelet board and turret board construction. Real point-to-point wiring lacks the strict orthogonal aesthetic many builders cling to, but it’s very organic, potentially compact, and can minimize unwanted signal interactions. I built my crossovers onto the backs of the baffles, using terminal strips and distance to minimize inductive coupling between the coils. A 9x13 baking pan allowed me to work on the backs of the baffles with the drivers' fronts safely suspended below. A pair of power resistors formed a 3dB pad to compensate for the tweeters' higher sensitivity. Finely-stranded 12ga OFHC speaker wire fit into a notch I cut into the edge of each baffle with the same little handsaw and chiseled out with a jeweler's screwdriver. It was soldered directly to the drivers' crossover components for purity and simplicity.

For a first test, I just threw some pillow stuffing into each kimchi pot and plopped a finished baffle assembly on top. The result was clearly not airtight, but would give me a rough idea of what the final sound would be like. The test setup included a beast of a JVC receiver (their flagship at the time), an Oppo DVD player, and a little Sansa MP3 player loaded with some FLAC files. I placed the speakers about six feet apart in the middle of the room so I could hear them from any angle.

Initially the speakers sounded very bass-shy with both classical and contemporary music, but then I realized that the receiver had been configured for a subwoofer and was rolling off everything below 120Hz. Correcting that problem resulted in a tight, full sound that left me speechless. The tweeters were very fast and detailed, and the bass, though not room-shaking, was natural, controlled and all there. Best of all, I couldn't hear any artifacts around the crossover point. It was like listening to a single driver with a very full range. Most surprisingly, the frequency balance was insensitive to placement. The speakers had a consistent balance from every angle in my (admittedly lively) living room, even with horizontal baffles. I had every reason to expect them to sound superb once the pots were properly filled and sealed.

The final step was to completely fill the pots with polyester stuffing (courtesy of some inherited sofa cushions) and silicone the baffles onto the pots, relying on the slight air gap at the cable entrance for pressure equalization. Considering that all this project's tools and materials were either already on hand or painlessly acquired, the biggest challenge was waiting 24 hours for the silicone to set.

The final audio test was highly satisfying. The bass had become even more precise with the additional stuffing and sealing. The tweeters continued to impress me with their speed and clarity, allowing me to hear all kinds of new details in recordings I've been listening to for years. There was less bass compared to my old Tannoys, but the bass from the kimchi pot speakers was much more even from note to note. And the speakers just vanished, filling the whole room with wide, even, natural sound. Recent symphonic recordings were especially realistic, like 3-D sound on a black velvet background.

To try to get a bit more lower midrange chestiness, I tilted the speakers forward, resting them on some folded towels until their drivers were roughly time-aligned. The result was unexpected. The lower mid filled out only slightly, but at the expense of all that 3-D glory. Localization improved, but they sounded too monitor-like in this more conventional orientation. Without stopping the audio source, I tilted the pots upright again, and all the music came back. It was actually a relief to discover that the most satisfying sound came from the most home-friendly positioning.

Despite the fairly gentle rolloff of 12db per octave and very clean bass notes (down 3dB at 53Hz) I was craving one more basement octave. Because I also had a pair of Morel MW-267 8” woofers in the same driver cache, I opened a new file to design a subwoofer. Reaching once again for The Loudspeaker Design Cookbook, I settled on a fourth-order clamshell isobaric bandpass configuration. I'll make this brief because anyone who knows about fourth-order clamshell isobaric bandpass subwoofers doesn't need to hear my description, and no one else cares. But simply, it's two identical woofers embedded face-to-face deep inside an enclosure, with the top chamber connected to the outside world only by a port - in this case, a pipe or three. The lower chamber is sealed. The volumes of the chambers and the pipe's dimensions are quite critical. It’s an alignment that was often chosen for boom-car installations due to its ability to deliver gobs of thunder, thus earning it an undeserved reputation for peaky, one-note bass. But it doesn’t have to be that way; a wider passband and less excursion can result in clean, musical deep bass. 

With a move to Budapest on the horizon, I set the subwoofer project aside hoping that more materials and resources would be available to me there, although kimchi pots wouldn’t be among them. I still wanted to stick with the ceramic vessel theme though for its freedom from resonances and aesthetic harmony with the existing two-ways. I figured I’d be able to find a local garden center in Budapest that would have glazed plant pots of the right volume. The subwoofer’s compound driver arrangement allows for a smaller-than-normal enclosure (31 litres in this case) and reduces distortion significantly. The first Hungarian garden center I checked out had just what I was looking for.

I was determined to spend as little money as possible on the rest of the materials, so what I did was admittedly kind of makeshift. Let’s face it: it was going to be somewhat improvised anyway owing to my being in yet another country where I didn’t speak the language. The baffles ended up being made from large laminated wooden pizza spatulas, which happened to be very close to the planter's irregular internal diameter. Because the planter's mouth was narrower than its middle, I had to cut each baffle in half after it was shaped in order to get them into the planter, then somehow fuse them into a one-piece baffle again. This is where it got tricky.

The two 8” drivers had to be mounted on either side of a baffle that had to be assembled and glued inside the planter so that it sat at a specific height. I siliconed machine bolts up through the mounting holes of the bottom driver, then siliconed a carefully-measured perforated cardboard tube to the bottom of the planter. The heavy speaker wires from each driver were passed through a hole in the side of this tube and out through another hole in a plastic insert I siliconed to the bottom of the planter. I used some polyester filling from a fabric store to lightly fill the two chambers. The drivers themselves ended up providing the clamping force for the wood glue holding the lower baffle pieces together. Once that baffle was whole and the glue fully dried, I sealed the baffle’s perimeter with more silicone.

The upper chamber was a little more challenging because there would be no drivers to compress the top baffle’s glue joint. Instead, I glued an arrangement of wooden dowels to the bottom surfaces of both halves of the upper baffle. These would sit around the upper driver, holding the baffle pieces at the right height. But before the upper baffle could be assembled and glued, I had to attach the port tubing, which was made from three pieces of copper-colored plastic downspout pipe from a hardware store. These were friction-fit into the baffle holes I’d made with a hole saw. Not gluing or sealing them right gave me the option to slide them in and out of the baffle to fine tune the frequency response once the mechanical assembly was done. I used small bags of reptile sand from a pet shop to weight the top surface of the baffle while the wood glue was drying. The next day, I put silicone around the baffle perimeter as before to seal it.


Lacking access to an anechoic chamber, it seemed sensible to do the final tuning in my listening area. Cables from both drivers had been brought out through the bottom of the planter, making it possible to use separate amps down the road if I needed more oomph. For now, I wired them anti-phase (the drivers were in a face-to-face, push-pull configuration) and drove the resulting 4-ohm load with a bridged 400W linear MOSFET amp with a DSP-based 4th order 80Hz low-pass filter. Sweeping the oscillator manually revealed a deep, clean, room-shaking bass that was devoid of any peaks that couldn’t be corrected by tweaking the placement. The ports were fine where they were, so I siliconed them in place to provide the final airtight seal.


The reptile sand was for filling the area between the top baffle and the rim of the planter. I arranged some seashells I’d collected over the years in the sand. Why? The random angles of the vent pipes made them look like seashore pilings - a familiar sight from my westcoast youth. It worked: most subwoofers are an eyesore in a living space, but this one really fits the style of our home. It has also survived my daughters’ many impromptu high-volume dance parties and every Telarc bass drum I could throw at it. More importantly, it doesn’t sound like a separate system element. The sub blends perfectly with the kimchi pot two-ways, simply filling the main part of the house with music that sounds like music - at any volume. 


It would be great to see a renaissance of home listening on systems capable of delivering big, clean sound. Their only drawback is that the shortcomings of many contemporary sources (e.g. streaming services like Spotify) and connection methods (e.g. Bluetooth) become very apparent once you start listening to uncompressed music via clean signal paths and transducers. Even people who can’t pinpoint the difference vote with their (dancing) feet. Music exists to move us in one way or another, and creatively eliminating the obstacles can be highly satisfying.

So… what’s your next audio mission?   


5 views0 comments

Recent Posts

See All


bottom of page