James
Lord of the Erg\o/s
This piece was originally posted on the Naim Forum, but for the benefit of those who dont visit the other place, here it is in its entirety.
The Ergo IIIs are now complete and I think there may be a few people interested in how this project went. I welcome all comments, and will attempt to answer any questions that this post might generate. Hopefully, words and pictures will be enough to start telling the whole story.
Background
Loudspeakers are inherently the most variable component in the hi-fi chain. Thats because the way they sound depends on the room in which they are housed. Besides the acoustic signature of the room (the only signature free one is an anechoic chamber), loudspeakers are voiced differently. By this, I mean the tonal attributes (shading, texture, colour, vibrancy etc) can be quite different, and not always to everyones liking. This is why seasoned audio maniacs demand nothing less than a comprehensive home demonstration of loudspeakers they intend to buy. I take this one step further by designing and building a pair to suit my taste and circumstance. Besides, I have found nothing that I like at a price I consider fair.
Design Brief
I want a pair of loudspeakers that connect me to the music viscerally, emotionally and truthfully. By visceral, I want scale, dynamics and vitality. By emotion, I want to understand meaning, context and experience what the singer or player is feeling. By truth, I want clarity, timbral and temporal accuracy, and even-handedness. Not an insufferable wish list, but what a challenge nonetheless.
To do scale and dynamics, the loudspeaker needs to move lots of air and be able to move it and stop it quickly. Vitality is harder to determine, but there seem to be a direct connection between efficiency and a sense of liveliness. So this means big drivers, with fast and efficient numbers. Emotional connectivity is, I think, an outcome of coherent design with properly designed crossovers for a specified complement of drivers. The scope for getting this wrong is considerably broader than it is for getting it right, even with CAD assistance. Finally, truth may be found by operating carefully chosen and good quality drivers well within their design parameters with minimal filter intervention or correction. Lastly, I wanted a 3-way system simply because properly designed, they offer far fewer compromises than their 2 or 2.5-way brethren do.
Driver Selection
There is a huge range of drivers available to the DIYer, ranging from el cheapo Taiwanese copies to stratospherically priced hand-made Skaaning Flex Units. Incidentally, manufacturers also use most of the drivers available to DIYers with few exceptions; Dynaudio being one of them, but thats no loss in my opinion. I chose the upper middle ground across three brands.
Bass Driver Tight and fast, and ported boxes are oxymoronic to a large extent. Sealed boxes are much better in this regard. Regrettably, there are very few large drivers that work optimally in a sealed box these days. I also wanted paper-pulp cone, high mechanical damping (Qms) for a crisp bass, and a low reach. That makes the Scan-speak 25W/8565 an optimum choice. The question was which variant? The 00 goes lower but requires a 100-litre box, whilst the 01 sacrifices about 5Hz and needs only 65 litres. Domestic acceptability dictated the 01 variant.
Midrange The Scan-speak breaks up badly above 2kHz, and my preference for 2nd order slopes meant that I needed a midrange driver that can go down cleanly to 200Hz to pick up where the 10-incher leaves off. Few true midrange drivers will go that low, so the next best thing was a smallish mid-bass driver. Scan-speak makes an ultra-desirable 15W/4531 that would do perfectly, but alas it was a 4-ohm design and a bit too expensive for what I was looking for. Then the Seas CA15RLY made its appearance, and it seemed a perfect partner for the big Scan-speak.
Tweeter My perennial favourite is the Hiquphon OW1. To my ears, these have the most natural sound and impeccable credentials. Since I already had two pairs on hand, it was a no-brainer choice once I knew the Seas would go high enough to cross cleanly with it.
Design Methodology
There are two parts to loudspeaker design. The physical aspect is about building a suitable box to house the drivers and is, in my opinion, the easier aspect of loudspeaker design. Considerations for the box design include volume calculations, cabinet materials and damping, assembly methods, aesthetics, and most importantly, driver alignment.
The other aspect is crossover design, and this is far more challenging than first meets the eye. First, there is the acoustic element, which is determined by the native response of the drivers in question, which is never flat. More complication is brought about by the varying roll-off rates at either end of their spectrum, and the effect the baffle has on the radiation of high vs low frequencies. Then, you have the electrical elements determined by a brace of capacitors, resistors and inductors. How these are brought together to tailor the native response of a driver to your intended acoustic target is not something that can be easily done without taking actual measurements and modelling that on a computer. Those who successfully design by ear from the ground up without CAD assistance or acoustic measurements are an especially talented lot. I dont have such talent.
So which comes first, the box or the crossover? Well, neither actually. Its the dummy box that comes first. Thats because in order to align the drivers correctly, one needs to know the relative offset between the drivers. Well, the purist would. Otherwise, the design CAD software can usually correct for it, but who really wants a boringly vertical slab-faced box for a loudspeaker? So I built a test-box and found the necessary offset by CAD simulation. Only then can the box be designed and built in earnest and confidence.
Enclosure Construction
I wanted a free-space design, which meant that I should aim for a total system Q of around 0.7, which gives a maximally flat response. If I erred, Id rather err on the lower side of 0.7 than the higher, so I built my boxes based on 65 litres for the bass enclosure, and allowed for light fibreglass stuffing to kill internal reflections without lowering Q too much.
The midrange and tweeter are housed in a separate sub-enclosure to avoid the intermodulation effects of the woofer. Otherwise it would defeat the purpose of a separate midrange driver. Again, this sub-enclosure is stuffed with fibreglass to kill internal reflections and minimise leakage of internal sound back through the cone.
Aesthetics is taken care of with judicious use of Sapele, which is an African hardwood with gorgeous striations when quarter-sawn. Internally, the real cabinet is constructed from 18mm MDF, which is heavily braced.
The MDF inner carcass is bonded to the external 20mm timber cladding with Liquid Nails, which allow for some natural movement in the timber, unlike aliphatic timber glues. The Liquid Nails also serve another purpose, and that is to damp the dual-composite panels. Hence, I found no need to use bituminous gunk or mass loading trickeries to achieve a fairly inert cabinet.
The piece de resistance is the contoured baffle. Sharp cabinet edges cause ripple like diffraction of high frequencies and is a key contributor to an uneven tonal balance in the upper reaches. Round the edges with any radius less than two inches apparently doesnt do much either. Hence, I glued three layers of 18mm MDF for the upper baffle and shaped it with an electric plane and plenty of sandpaper.
The baffle proper is routed to flush mount all drivers to the right depth. Both the midrange and bass drivers are secured with M5 bolts and T-Nuts for proper tension that woodscrews cannot reliably provide in MDF.
The tweeter is secured using woodscrews only because it is light and does not move with the same magnitude. Furniture-grade black leather cowhide covers the baffle and matches the Ergo IIIs nicely with the black leather lounge suite.
The Ergo IIIs are now complete and I think there may be a few people interested in how this project went. I welcome all comments, and will attempt to answer any questions that this post might generate. Hopefully, words and pictures will be enough to start telling the whole story.
Background
Loudspeakers are inherently the most variable component in the hi-fi chain. Thats because the way they sound depends on the room in which they are housed. Besides the acoustic signature of the room (the only signature free one is an anechoic chamber), loudspeakers are voiced differently. By this, I mean the tonal attributes (shading, texture, colour, vibrancy etc) can be quite different, and not always to everyones liking. This is why seasoned audio maniacs demand nothing less than a comprehensive home demonstration of loudspeakers they intend to buy. I take this one step further by designing and building a pair to suit my taste and circumstance. Besides, I have found nothing that I like at a price I consider fair.
Design Brief
I want a pair of loudspeakers that connect me to the music viscerally, emotionally and truthfully. By visceral, I want scale, dynamics and vitality. By emotion, I want to understand meaning, context and experience what the singer or player is feeling. By truth, I want clarity, timbral and temporal accuracy, and even-handedness. Not an insufferable wish list, but what a challenge nonetheless.
To do scale and dynamics, the loudspeaker needs to move lots of air and be able to move it and stop it quickly. Vitality is harder to determine, but there seem to be a direct connection between efficiency and a sense of liveliness. So this means big drivers, with fast and efficient numbers. Emotional connectivity is, I think, an outcome of coherent design with properly designed crossovers for a specified complement of drivers. The scope for getting this wrong is considerably broader than it is for getting it right, even with CAD assistance. Finally, truth may be found by operating carefully chosen and good quality drivers well within their design parameters with minimal filter intervention or correction. Lastly, I wanted a 3-way system simply because properly designed, they offer far fewer compromises than their 2 or 2.5-way brethren do.
Driver Selection
There is a huge range of drivers available to the DIYer, ranging from el cheapo Taiwanese copies to stratospherically priced hand-made Skaaning Flex Units. Incidentally, manufacturers also use most of the drivers available to DIYers with few exceptions; Dynaudio being one of them, but thats no loss in my opinion. I chose the upper middle ground across three brands.
Bass Driver Tight and fast, and ported boxes are oxymoronic to a large extent. Sealed boxes are much better in this regard. Regrettably, there are very few large drivers that work optimally in a sealed box these days. I also wanted paper-pulp cone, high mechanical damping (Qms) for a crisp bass, and a low reach. That makes the Scan-speak 25W/8565 an optimum choice. The question was which variant? The 00 goes lower but requires a 100-litre box, whilst the 01 sacrifices about 5Hz and needs only 65 litres. Domestic acceptability dictated the 01 variant.
Midrange The Scan-speak breaks up badly above 2kHz, and my preference for 2nd order slopes meant that I needed a midrange driver that can go down cleanly to 200Hz to pick up where the 10-incher leaves off. Few true midrange drivers will go that low, so the next best thing was a smallish mid-bass driver. Scan-speak makes an ultra-desirable 15W/4531 that would do perfectly, but alas it was a 4-ohm design and a bit too expensive for what I was looking for. Then the Seas CA15RLY made its appearance, and it seemed a perfect partner for the big Scan-speak.
Tweeter My perennial favourite is the Hiquphon OW1. To my ears, these have the most natural sound and impeccable credentials. Since I already had two pairs on hand, it was a no-brainer choice once I knew the Seas would go high enough to cross cleanly with it.
Design Methodology
There are two parts to loudspeaker design. The physical aspect is about building a suitable box to house the drivers and is, in my opinion, the easier aspect of loudspeaker design. Considerations for the box design include volume calculations, cabinet materials and damping, assembly methods, aesthetics, and most importantly, driver alignment.
The other aspect is crossover design, and this is far more challenging than first meets the eye. First, there is the acoustic element, which is determined by the native response of the drivers in question, which is never flat. More complication is brought about by the varying roll-off rates at either end of their spectrum, and the effect the baffle has on the radiation of high vs low frequencies. Then, you have the electrical elements determined by a brace of capacitors, resistors and inductors. How these are brought together to tailor the native response of a driver to your intended acoustic target is not something that can be easily done without taking actual measurements and modelling that on a computer. Those who successfully design by ear from the ground up without CAD assistance or acoustic measurements are an especially talented lot. I dont have such talent.
So which comes first, the box or the crossover? Well, neither actually. Its the dummy box that comes first. Thats because in order to align the drivers correctly, one needs to know the relative offset between the drivers. Well, the purist would. Otherwise, the design CAD software can usually correct for it, but who really wants a boringly vertical slab-faced box for a loudspeaker? So I built a test-box and found the necessary offset by CAD simulation. Only then can the box be designed and built in earnest and confidence.
Enclosure Construction
I wanted a free-space design, which meant that I should aim for a total system Q of around 0.7, which gives a maximally flat response. If I erred, Id rather err on the lower side of 0.7 than the higher, so I built my boxes based on 65 litres for the bass enclosure, and allowed for light fibreglass stuffing to kill internal reflections without lowering Q too much.
The midrange and tweeter are housed in a separate sub-enclosure to avoid the intermodulation effects of the woofer. Otherwise it would defeat the purpose of a separate midrange driver. Again, this sub-enclosure is stuffed with fibreglass to kill internal reflections and minimise leakage of internal sound back through the cone.
Aesthetics is taken care of with judicious use of Sapele, which is an African hardwood with gorgeous striations when quarter-sawn. Internally, the real cabinet is constructed from 18mm MDF, which is heavily braced.
The MDF inner carcass is bonded to the external 20mm timber cladding with Liquid Nails, which allow for some natural movement in the timber, unlike aliphatic timber glues. The Liquid Nails also serve another purpose, and that is to damp the dual-composite panels. Hence, I found no need to use bituminous gunk or mass loading trickeries to achieve a fairly inert cabinet.
The piece de resistance is the contoured baffle. Sharp cabinet edges cause ripple like diffraction of high frequencies and is a key contributor to an uneven tonal balance in the upper reaches. Round the edges with any radius less than two inches apparently doesnt do much either. Hence, I glued three layers of 18mm MDF for the upper baffle and shaped it with an electric plane and plenty of sandpaper.
The baffle proper is routed to flush mount all drivers to the right depth. Both the midrange and bass drivers are secured with M5 bolts and T-Nuts for proper tension that woodscrews cannot reliably provide in MDF.
The tweeter is secured using woodscrews only because it is light and does not move with the same magnitude. Furniture-grade black leather cowhide covers the baffle and matches the Ergo IIIs nicely with the black leather lounge suite.