I didn’t know about this thread until Jim sent me an email to let me know. This post will be long, but hopefully it will accomplish several things – give a history of the design, explain how it all works (something that may be of more interest to Dennis than anyone else), offer some specifications, and answer the questions that have come up in this thread.
My version of the speaker looks almost identical to the one Jim made, except for the cool CNC dado joints on Jim’s enclosure. There was a comment about the look with the screws showing. I know this won’t appeal to everyone, but for what it’s worth my speakers do have grills. I can add that this method of clamping the driver offers much more mechanical coupling than the conventional method because the woofer is clamped rather than screwed down. The baffle fits against the battens in a manner that heavily reinforces it too, and measured cabinet Q is very high revealing very low leakage and high mechanical damping of resonances. This speaker is very solid and heavy for its size compared to many similar speakers.
I originally took mine to the Indiana DIY speaker gathering in 2008, where Jim and I first introduced the Archos speaker before a large group. My little speaker generated quite a bit of buzz and was picked as best small monitor by several people. The following year, due to a lot of discussion on the speaker, I brought it to the 2009 gathering, and again it was picked as best small speaker. This time there was a lot of internet discussion on the design and several people asked me if they could build a pair. After about ten sets were being built by individuals I asked Jim if he would be interested in producing a finished version. He had listened to mine and expressed some interest. So here they are.
They aren’t really true LS3/5a clones, even though they started off to be. I compiled all of the data I could round up on the little BBC monitors, including the original BBC design papers, so I could understand what the BBC was really referring to when they talked about “acoustic scaling” for small monitors. I have always been fascinated with this speaker and how balanced it sounded for the time period (granted, it was not perfect), and wanted to make something similar for my own personal use. Once I understood what was really going on in the design of the speaker I decided to build my own. However, we have learned a lot about loudspeaker design since the mid 70’s and I saw no point in going backwards, so I implemented what I felt were the best techniques that I could use based on today’s knowledge and technology. ( I will say that reading the BBC literature is fascinating considering the time period. They usee real design techniques, measured responses, etc. which were quite advanced for the time, and a lot like we do today over thirty years later). I was inspired by the look of a particular British monitor so I decided to base mine on that look, but that’s where most of the similarity ends. My goal was to build myself the best small monitor I could. Period.
My cabinet is a little larger than the LS3/5a by about ½ in. in height and width, and about two inches in depth – mine are 8”W x 12.5” tall, by 8.5” deep. My driver selection changed the crossover design a lot from the BBC design too. I tried to stay true to the woofer by selecting a 5” driver. I actually built three different versions of this speaker before I settled on the drivers in the current version.
The most expensive drivers available are not always the best drivers for a given design, and the drivers in this speaker I feel are the best available for this particular design. I went with the Aurum Cantus AC130-F1, which is a very nice little driver – mica/carbon filled poly cone, rubber surround, cast frame, large motor, and a 5mm one-way Xmax. It’s a very beefy little 5”, and better in my opinion than most of the others I have used in this range. It also has an excellent midrange reproduction too, much better than I could get from any of the other drivers I worked with before it. I could never get the midrange that made me shout, “Yes! There it is!” with those. I got it with this driver.
The tweeter is an Usher made aluminum dome tweeter. I don’t normally select metal domes, but this is really an exceptional tweeter. It is identical to their proprietary top of the line Be dome tweeter, except this one has a black Aluminum dome. It does have a copper sleeve in the motor and two shorting rings and has some of the lowest measured distortion of any dome tweeters available today. It sounds different too. I have a lot of tweeters here, and this one sounds more like a ribbon on the top than it does any of the domes I have. It comes across as very clean and detailed with no “splashiness” or grain on cymbals. I also considered that nice screen protection it has over the dome since this design may be used as a pro monitor, and most of them have some kind of protection here.
Because this driver had such a low Fs (530Hz) and very low distortion, especially in its lower range, I decided to go with a second order acoustic crossover at 1.8kHz. The woofer’s roll-off is almost a perfect 2nd order response. The tweeter’s roll-off is a little steeper than that in order to accommodate the offset between drivers, but it still sums as an in-phase crossover with excellent phase tracking and a very deep reverse null (for those who understand in-phase crossovers). A second order acoustic crossover offers the smoothest power response through the crossover region for an in-phase crossover. The use of the low crossover point yields very smooth off-axis polar response to, so the speaker’s power response is very smooth, and the off-axis response has no significant peaks or dips. Top-end dispersion may be more limited than with a 3/4" tweeter, but that is less of an issue to me, since I don't listen much off-axis, than what is going on near the crossover region and how the overall tonal balance is impacted from the power response. I simply preferred what I could do with the lower crossover point and shallow slopes. I believe that this off-axis behavior in the crossover region coupled to the speaker’s exceptionally flat on-axis response is part of what makes it sound so balanced and open, and image so well. I have yet to hear a small speaker that sounds better and more natural to my ear than this combo.
Where I did remain true to the original LS3/5a spec was in giving the speaker the slight hump in the upper bass at around 100-125 Hz, which is done in the crossover and not in the bass tuning, which is a sealed system with a Qtc of about .75. Then this hump is followed by a small depression above this region at around 350Hz of about 1-1.5 dB. This is part of the BBC “acoustic scaling” that allows a small speaker to be perceived as more extended than it actually is, while not sounding thick or tubby in the lower mids. I think it works very well and everyone seems to rave regarding the midrange reproduction of this speaker, so they must agree ;- ) Above this region the response is almost ruler flat the rest of the way up. The speaker is a nice, solid 8 ohm nominal design, with a very easy load to drive. And, has a sensitivity of 84dB/2.83V/M, which is pretty typical for a speaker this size, if not a little better than most.
You often hear mention of a BBC dip. Some larger BBC monitors possessed a small dip in the response at 3kHz to soften the proximity effect from microphones of the day. The LS3/5a does not possess this dip though. In fact, it is actually very linear through this range. Jim took a measurement of my speaker which showed some ripples in this range, but these ripples are simply edge diffraction effects from the sharp, square cabinet edges near the tweeter correlating to wavelengths in this frequency range. The thing is, they average out over a listening angle and tend to not be audible at all. Dennis’ comment about the BBC dip is correct. This may be what Jim was referring to, or he may have been referring to the small dip in the 300-400 Hz range, which is typical of this BBC design.
I close mic’ed the speaker and it was -3dB right at 60Hz, which isn’t bad for a small sealed speaker, especially since it is rolling off second order below that. I stayed with the sealed enclosure since that was a key element to the BBC monitor and makes it a little more versatile in placement. Porting this woofer in this small box would have made it peak even more though.
What the BBC did was something no one thought much of in 1975, but we take for granted in a design now. They included full baffle step compensation in their speaker in the form of a broad, deep notch filter on the Kef B110. The little Bextrene woofer had a big peak in its response at around 1kHz. They filtered the peak, but broadened the notch so it began its dip at about 150Hz. The Aurum Cantus AC130F-1 solves most of the problems of the B110 – with better bass extension capability, much higher output capability, very low motor distortion, and a stunningly beautiful midrange. Trust me - I have used a lot of drivers over the years. This is truly a gem that gets overlooked way too much. In 1975 Baffle Step Compensation was not understood well, and seldom ever implemented. This is one of the things that made the LS3/5a stand out from the crowd. The LS3/5a actually has a pretty linear response except for a shallow trough in the lower midrange and peak in the upper bass.
About that response curve, I think some explanation is in order – In my spliced response you will see this hump and dip in the response, as you will with all LS3/5a speakers and in all small BBC-type speakers that descended from it over the years from various manufacturers. This was an important part of the BBC acoustic scaling and a large part of the reason that vocals sound so perfectly balanced in this speaker (at least according to many people).
I doubt the BBC engineers (of 1977) even understood what they had found - they simply voiced the speaker by ear for the truest vocal reproduction. I don't know if they even understood baffle step compensation back then, but I figured out what was going on here. If you use conventional baffle step compensation methods the result will go one of two ways: Either you will get the compensation where the bass sounds nice and solid, but the lower vocals sound thick or over-emphasized, or you will get the vocals sounding very good but the bass now sounding a bit weak. This seems to be the trade-off people make when implementing BSC, and we have all heard it. At most gatherings or shows there will be speakers that make good examples of both types of balance.
However, if you target the response curve of the BBC monitor, and all its descendents, which includes my speaker, then you will get the best of both. The vocal balance in the room will be extremely natural with no thickness or chestiness being added to the lower vocal range, and the hump at 100-120Hz will give the ear/brain the illusion of more bass extension than there really is. This works together with the dip though, because since the vocals do not have any added emphasis in their lower range the bass gets interpreted as extension rather than a hump. The BBC called it "acoustic scaling" or how to make a small speaker sound similar, but just not as loud as a larger speaker. Also, if you move the speaker closer to the wall or a desk it tends to fill in the trough rather than boost the midrange into muddiness.
My personal impression is that any small speaker that implements this response curve will actually sound more balanced and "flat" to our brain than a speaker that actually measures very flat does. That may sound counterintuitive, but it is about perception here. I don't think very many designers understand this, so I would be glad to put my speaker up against any flat measuring similar speaker out there. I am convinced that people will more often than not pick mine as the one that sounds "correct". Now, this design approach doesn’t apply to all small speakers……
These speakers are sealed, so there is a second order roll-off of low frequencies. This also helps to blend with the room better too. If a speaker is larger and can extend deeper on own, say it is reasonably flat into the 50Hz range, then the correct sounding response will gravitate toward flatter to sound more natural. So, the response curve we are working with here really only benefits a speaker that is rolling off below 100Hz or maybe a little lower. If the speaker extends to 50Hz then the hump may not be perceived the same way, it may make the speaker sound boomy or muddy in the bass.
Finally, there was a question of mounting the woofer from the front, flush in a more conventional manner. As Dennis pointed out this changes the offset and arrival time, more specifically the relative phase, between the two drivers. To mount the woofer in this manner moves it 12mm forward. That is possible, but it does require changes to the crossover. I have an alternate schematic that results in the exact same frequency response to within +/- 0.2dB in the crossover region with the driver front/flush mounted. However, there are a couple of other issues. First, this driver has frame that can be a real pain to recess, although Jim can probably handle it OK, and second, in the current design there is a small overlap between the tweeter and woofer in their spacing, but this could probably be adjusted too. I do believe, however, that the mechanical coupling of the baffle, the driver, and use of the solid wood small enclosure contribute to its overall sound character, so these changes may change that slightly as well.
Now, for those of you who made it this far, all two of you, you get a special offer – another name that speaker contest. I have been calling this speaker the “Continuums” because it continued the design idea of the LS3/5a to the next generation. However, I really don’t like that name much for a commercial offering. Jim and I both feel that the name of the speaker should speak to its heritage, but we don’t want to infringe on any copyrighted names either. Any ideas?
Jeff Bagby
Here's my response (spliced and one Meter) and impedance:
Here it is compared to the original LS3/5a