> After reading that I understand why everyone keeps telling me to build a kit.
A kit is safe.
You can't build a speaker from book-theory. Steen has collected a lot of info together, and some good points, but there is much more to it than that.
You CAN build a perfectly useful speaker without any book. Buy a cone speaker 10" to 4" that looks good. You can stare at curves until you go blind, or thump/scratch cones to get an idea of the sound, but "looks" seem to be about as reliable as any other selection.
To build a sealed box cabinet for any common speaker, calculate a cube with sides 1.1 times the advertised diameter, then build a non-cube with the same internal volume.
(This "1.1 rule" actually gives a Q of about 1, a nice value, for any speaker with efficiency near 1%, which is very typical.)
Example: the Fostex FE-103 is a lovely little 4-inch driver. 1.1 times 4 inches is 4.4 inches on each side of a cube. But if we really build a cube then all the internal reflections are the same length, same frequency. Also a cube does not leave much face-area for a tweeter, which will be our first "frill", at least for anything bigger than 4-inch. So build a non-cube of 4.4*4.4*4.4= 85 cubic inches. A good size might be D*1.0*1.1*1.2 or 4.0"*4.4"*4.8", or D*0.8*1.1*1.5 or 3.2"*4.4"*6". This decision depends more on available space or pre-cut boards than acoustics.
Note that this is internal dimension. You have to figure the outside dimensions and how the 6 panels fit together.
Note that the internal volume is not critical. 20% shift in volume gives about 10% shift of resonance with a complementary shift of peaking, so you don't hear any great difference with large changes in box size. The FE-103 can be happy in 50 cubic inches or 200 cubic inches.
Build the box very tight. Leaks will wheeze and whistle. Use caulk under the rim of the speaker. Always front-mount unless there is a very good reason to back-mount.
Do that with a 4-inch FE-103 and you get smooth on-axis response from 15KHz down to 200Hz. You can't get a heap of bass from a 4-inch, and if you try it will fuzz-up the midrange.
Do that with a 10-inch speaker and you get good bass down to 80Hz, and on-axis response to around 3KHz, but the off-axis response is drooping above 800Hz. A 10-inch makes a very good speaker for a large AM radio: full bass, clear on-axis, or you can sit off-axis and lose treble if reception is bad.
Yes, a cone speaker is a terrible thing to listen to. But nothing else is better. We can argue this point for years: some non-cone speakers have some really wonderful properties. Adapting a cone/dome to work with a horn is a powerful (and costly) technique. But after 80 years and thousands of odd drivers, the plain cone is still what we use almost everywhere.
No single speaker can cover the whole range 20-20KHz, or even 50-15KHz, with uniform directivity and enough power even for a living room. 8-inch to 4-inch speakers cover most of the critical midrange and are very popular. If you go to the larger sizes, 6" and 8", you get respectable bass with beamy droopy treble. A 2-inch speaker covers treble well and wide, but won't even get down to 500Hz well. But an 8" plus a 2" cover most of the audio band OK.
The key problem is how you splice the two sounds together. It is just impossible to make two speakers sound as one. If they overlap, it confuses the ear. If they have different sensitivity where they come together, that stands out.
And filter theory is only a general starting point. Real speaker responses are very complicated and don't act like ideal filters.
What seems to give best result for "reasonable effort" is to find a woofer that is fairly flat to a point like 3K or 6K and then drops off quickly. They build them this way on purpose. Letting speaker limitations be the necessary filter works quite well. This typically means your vital midrange 600-3KHz comes out of the woofer, which is a bad plan, but still perfectly acceptable. It also means you are pushing a big cone about as far as it can go, up to frequencies where it gets a little ratty. But getting a smooth splice anywhere within the vital midrange is very hard to do.
So note where your woofer runs out of steam, and pick a tweeter and filter-capapcitor to cover the rest of the audio band. Don't, like my friend, use a 80Hz-3KHz woofer with a 5KHz+ tweeter: this sounded really queer until I realized he had bought the wrong drivers.
So with a sharp-cutoff woofer, a complementary tweeter w/cap, and a 1.1X box, you can build a pretty decent speaker without any heavy thinking. We did this all the time in college and it usually worked fine.
There is no great magic involved.
In fact quite wacky ideas often work better than they should. Put 16 $2 4" speakers in a box, either 4x4 square or a 1x16 line-array. You get a lot more than $32 of sound. Accurate? No, but fun.
But..... build your own reference monitors? This is like building your own ruler. It is a Precision Tool. You have to become a Real Expert. By the time you have a hope of getting close to what you can buy, your life has been dedicated to speaker-thinking and you have no time for anything else. Get a life, or get on with the life you already picked. If you were put on earth to build monitor speakers, you would already know that and be doing it all the time.
