DIY Cheap Three-Way Speaker with 15" Woofer

Because a few people asked...




You guys already know many of the reasons behind a speaker this like. Before pressing GO, please be reminded that such speakers will likely not suit modern audiophile tastes. In fact, they are likely to offend modern audiophile sensibilities because they use non-coated paper drivers used in the mids and lows. I like paper drivers because they hit fast. Nothing is for free however. Paper drivers decay slower than plastic, metal, exotic material drivers.

The ideas behind this were low distortion/high-SPL, high-efficiency, a lively immediate fast punchy sound, the ability to put them against a wall and not in the middle of the room (to avoid wives and children thinking you are the loser kind of audiophile at age 55 and leaving you). The downsides are lack of ultimate bass extension (37Hz -3db), ain't small, and what was just said above: poor decay characteristics.

The restrictions I put upon myself: be as cheap a possible. It's easy making a good sounding speaker using very expensive parts. This is called spending money to solve a problem. It's challenging making a speaker to cost. Did I succeed with what I sought out as my vision? Absolutely and even more. This project exceeded my expectations. I was further convinced I had reached my goals when I saw Tony (from Schiit California) cranking up Tool at 105db+ and when Evan (from Schiit Texas) asking me if he could permanently borrow them.

The drivers chosen for this job are as follows:

• Eminence Kappa 15LFA 15” Woofer, retail price $130
• Eminence Beta 6A 6.5” Midrange, retail price $65*
• Dayton Audio RST28F-4 1.25” 4-Ohm Fabric Dome Tweeter, retail price $37

*A better midrange would have be the Faital 6PR122, but this part at $157 belows the "cheap" aspect out of the water. FWIW, I did initially built the speaker with this midrange.

The Dayton was the only tweeter that had the voltage sensitivity requirements required so I wouldn't have to pad down the woofer. It does this by cheating with 4-ohm impedance and 1.25" dome. The bigger dome will have poorer dispersion characteristics. Tradeoffs ya know.

The design goal for the crossover network was to use a minimum of parts. Lively sound means using a minimum of crossover parts. I've since had some ideas to improved on the xover - more later. This approach is very different from that of say GR Research, which is to make everything perfectly flat on-axis, or to appeal to the ASR objectivists. There will be imperfect frequency response. Note that is is possible to flatten the frequency response to perfection with the use of more xover parts, but this is not what I wanted to do here. I feel that more xover parts results in a deader sound. Again, tradeoffs. It's is still good enough that one doesn't need to put their head into vice.

The entire package uses x4 inductors, x4 caps, and x1 resistor per speaker. Here is a rough frequency-response taken in-room with the speaker a mere 4" from the back wall. The microphone is aligned at the level between the tweeter and midrange cone. The midrange is where is the intended listening height (think low couch or sofa). Note that this measurement was without the resistor to pad down the tweeter.



The cabinet is 7.5 cu.f ft. Ports were x2 4" diameter. Port frequency was set to highish 38Hz. I could have gotten deeper bass extension with a lower port frequency, but neither my wife nor I did liked the results. The bass just ended up sounding drier and less punchy. Sometimes it's best to follow ears than box theory or pure measurements. We can see the room I put the speaker in has room modes at 70Hz and 110Hz along with a null at 85Hz. These things cannot be avoided in a room. They exist even in fully acoustically treated mix stages with low ceilings. If you are super worried about them, these effects can be reduced a bit by moving the speaker about a foot from the back wall or moving them to a room with cathedral ceilings.

Will post xover stuff next.

 

All crossover parts were purchased from Parts Express.

x2 Jantzen 6.8mH 16 AWG P-core DCR = 0.30-ohms
x1 ERSE 1.1mH 18 AWG
x1 ERSE 0.22mH 18 AWG
x2 Audyn (Red) 33uF 400V Poly Foil caps
x1 Audyn (Red) 3.3uF 400V
x1 Audyn (Red) 12uF 400V

x1 1.0-ohm 10W wirewound resistor (could use a better GR-Research approved part here).

Code:

Woofer

-----6.8 mH--------------     (+)
                    |
                   33uF
                    |


Midrange

-----1.1mH------------------33uF----------------    (-)
                    |                     |
                   3.3uF                6.8mH
                    |                     |


Tweeter

----12uF------------------1Ω------     (+)
                    |
                   0.22mH
                    |

Notes:

• Can probably get away with slightly higher DCR part in the woofer section
• Can get away with much less expensive electrolytic cap in the woofer section
• Can use a better cap than Audyn red in the tweeter section.
• Can probably stick a zobel in the woofer section to make the xover slope more effective (it has to fight against rising impedance) to prevent even less woofer bleed in (tighter more modern sound), but this will require a redo of the other parts.
• Move tweeter and midrange closer together
• The midrange is offset behind the tweeter, thus moving the sweet spot down to where one's ears are at with a low couch. It's probably best to align midrange and tweeter on the same plane. Play with the location of mid/tweeter up and down for the correct vertical sweet spot.

 

The above won't get you totally there. The effortless bass and thunderous physicality, yes...
The Sansui E138 or JBL 4333 types use horn loaded compression drivers in the mids and highs. The Dayton tweeters can pull off that tactility, but the biggest missing piece will be the replace the 6" paper mid with a suitable 1" compression driver. You'll then get the estat speed with dynamic physicality of the E138.

Maybe I should get together with @CEE TEE at Nitsch to bring back those old school blue baffle JBLs.

 

Absolutely not!

Part of the reason was to test whether it a separate mid cab made difference since the JBL 4698 I now use as instrument cab actually does have a sealed mid cabinet. I shoved a cut up pillow behind the mid instead. The other reason is that I wanted the mid driver venting into a large of volume as possible to minimize any damping for snappy attack transients. Large cabs let us get away with less damping compared to small cabs. This was a chosen design approach / philosophy, for better and worse. Small cabs seem to require heroic bracing and massive internal damping materials - think Magico. Again, it's about tradeoffs.