| A bass reflex, TMM, 2.5-way, floor standing speaker. | |||||||||
| My goals for this project were to build a speaker using high quality drivers that would provide a smooth, detailed
midrange, good highs, and sufficient natural bass. I chose the TMM/2.5-way configuration for a couple of reasons,
both of which may be questionable. I wanted to attempt to achieve some degree of time alignment of the tweeter
and mid-woofer and used a sloped baffle to physically align the drivers. This wouldn't be possible with an MTM.
I also believe that the imaging and dispersion characteristics of the TMM/2.5-way can be more natural than an MTM. Designing a TMM/2.5-way speaker is definitely more difficult than doing an MTM 2-way. The crossover is more complex due to the need for a low frequency low pass for the 0.5 woofer. Matching mid-woofer level to tweeter level is harder -- many MTM's will require little or no attenuation for the tweeter. Component values, inductors in particular, are larger, not just because of the 0.5 low pass, but because each driver provides an 8 ohm load to the crossover versus 4 for an MTM. The project follows below. If you want to skip to the final crossover design, click here. |
|||||||||
|
|||||||||
|
Drivers |
|||||||||
|
|||||||||
|
Enclosure |
|||||||||
| I used BassBox 6.0 and Woofer Tester measured Thiele-Small parameters to simulate the box design. Since my goal was to produce a full range speaker for primarily music listening, a vented enclosure was called for. BassBox Pro provides an optimum extended bass enclosure of around 2.0 cu.ft. I would have constructed the enclosure to this size, but it would have been slightly larger than I wanted. The BassBox standard alignment called for 1.2 cu.ft. I decided to go ahead and build the box to 1.66 cu.ft -- about halfway in between and tune it to 38Hz which provided an F3 point only slightly higher than the 2.0 cu.ft. alignment and resulted in a symmetrical dual mode impedance. | |||||||||
| The baffle is a departure from my usual designs which generally employ the narrowest practical width to fit the drivers. This time, the baffle would have a second beveled doubler. This meant that the baffle width would be around 9 5/8" which is 2 1/2" wider than the driver's 7 1/8". | |||||||||
| The enclosure construction is the same as I've used on several other speakers but it is complicated by the fact that the sides aren't rectangular and all box walls are doubled with a second layer of 3/4" material. There are five internal braces. The doublers for the sides are added inside, between the braces while the second layer for the front and rear is solid wood. The braces and top/bottom pieces are set into 1/8" deep dadoes in the sides. This make assembly much easier and about as square as you can get. I also used the overlap method described in my Links and Tips page. | |||||||||
| Here are the drawings. For the detail drawings, clicking on the drawing links to a Word document with the drawing which you can print and zoom in. | |||||||||
|
|
|||||||||
|
|
|||||||||
|
|
|||||||||
| A couple of pics showing the dadoed sides and the braces which slide into the dado slots. | |||||||||
|
|||||||||
|
The Lefty-Waldron Base |
|||||||||
| The base is used to add stability -- important if you have kids or large pets. On a carpeted floor, it helps to use spikes. The method shown here is called the "Lefty-Waldron Base". Lefty came up with the idea for using the disks/rounded corners and the truck bed liner paint and it was Barry Waldron's idea for the spacer between the speaker and the base. | |||||||||
| Made from 3/4" MDF. Before attaching the 5 3/4" x 12 1/8" piece to the 8 3/4" x 15 1/8" piece, glue on the four 2 1/2" disks. Then you can use the disks as a guide to router the rounded corners. Paint and mount to the speaker using 2 1/2" drywall screws. | |||||||||
|
|
|||||||||
| The pics show the base painted with Dupli-Color Truck Bed Coating. $7.95 for a can from Krangen's auto parts store. | |||||||||
|
|||||||||
|
Driver Measurements |
|||||||||
| These are BassBox predictions of the 1.66 cu.ft. enclosure tuned to 38.6Hz. | |||||||||
|
|
|||||||||
| Prdicted impedance -- upper curve is one woofer, lower curve is two woofers. Both at 38.6Hz Fb. Minimum Z is 2.98 ohms at 260Hz. | |||||||||
|
|
|||||||||
| Phase prediction with cursor at 260Hz. Note that phase rate of change is small. | |||||||||
|
|
|||||||||
| Here are Woofer Tester impedance plots for each driver in the enclosure: | |||||||||
| Tweeter -- Fs is close to the published 750Hz. | |||||||||
|
|
|||||||||
| Upper Mid-Woofer -- Note port tuning frequency at approximately 37 Hz. Glitch at 160Hz is probably an enclosure resonance; hopefully this will be go away after the baffle doublers and some acoustic foam is placed in the enclosure. | |||||||||
|
|
|||||||||
| Lower Woofer -- the 160Hz glitch is less prominent in this trace -- possibly confirming that it is enclosure-related. | |||||||||
|
|
|||||||||
| Both woofers -- driven in parallel. Tuning frequency 37Hz. Minimum impedance around 3 ohms at 250Hz -- both woofers will operate in this range. | |||||||||
|
|
|||||||||
| Here are unfiltered driver SPL response curves taken using Speaker Workshop MLS at 18" on an axis 2" below the center of the tweeter: | |||||||||
| Tweeter: | |||||||||
|
|
|||||||||
| The lower 0.5 woofer: | |||||||||
|
|
|||||||||
| The upper 1.0 woofer: | |||||||||
|
|
|||||||||
|
Crossover |
|||||||||
| I used measured SPL data and Woofer Tester impedance data to define the drivers for Speaker Workshop crossover development. My goals for the initial filter design were 1) to provide a crossover point around 250Hz for the 0.5 woofer and 2) to achieve an initial 2200Hz crossover for the 1.0 mid-woofer to tweeter. I also planned to accept a slightly raised bass response which I believe may be appropriate for larger sized rooms and rooms with high ceilings -- but I was prepared to reduce the level of this based on listening. My third goal was to produce a time coherent design -- more on that later. | |||||||||
|
|
|||||||||
| Here is the response. The sum is in blue. The 0.5 woofer is fairly well down at 250Hz and the HP/LP cross point is right at 2200Hz. A 2 to 3 dB bass elevation appears to exist, however, I suspect the data may not be totally valid in the 50 to 100Hz range due to room interactions in my relatively small test room. You'll also note a 2dB dip through the HP/LP crossover region. My goal here was to prempt any tendancy for glare in this range -- a technique promulgated by Paul Holsopple. Tweeter polarity is inverted in the Speaker Workshop crossover circuit however, this may end up being changed back to normal polarity -- I've experienced this with SW previously, that is, the reversed polarity isn't valid even though SW requires it to properly sum the drivers. | |||||||||
|
|
|||||||||
| One other component of the crossover is a notch filter centered at 1000Hz. I noticed my measurements showed a peak at this point which was persistent and made creation of a smooth rolloff of the 1.0 woofer difficult. I believe this to be caused by a resonance in the driver, possibly from the speaker's frame/motor structure. You can see this peak in the manufacturer's response data too, so, I believe it is actually there and not a measurement artifact. I used the spreadsheet I developed to determine the values for the notch filter. It works well if you know the frequency you need and you can play around with it to change the steepness (Q) and magnitude of the filter. Here's a link to it: http://users.d-web.com/dbrown/speakers/NotchCalc.xls | |||||||||
| Here is an adjusted crossover after taking measurements and doing some listening. R1 and Rn were decreased to 1.5 ohms and L2 was increased to 0.82 mH. | |||||||||
|
|
|||||||||
| Here is the SPL response. It should be noted that with the 13 degrees of baffle tilt back, the tweeter is that much off axis. The roll off is noticeable in this and the previous measurements. | |||||||||
|
|
|||||||||
| Third version of the crossover. C1 has been adjusted to raise the tweeter's high pass point and L2 returned to 0.7mH. Cnb, Czb, and C3b were added as bypass caps on the large value electrolytics. | |||||||||
|
|
|||||||||
| Here is the SpeakerWorkshop simulation. | |||||||||
|
|
|||||||||
| And here is the SPL response. | |||||||||
|
|
|||||||||
| At this point, the speaker sounds well balanced. Imaging and detail are good. Bass is extended with no tendancy to boominess. I'm almost ready to call this final, but with the goal of opening up the mid-range a bit more, I will try some different values for L2 to raise Woofer 1.0's low pass cutoff a bit. | |||||||||
| One more change -- decreasing the value of L2 to 0.5 mH to flatten the response in the mid-woofer/tweeter crossover range. That made a major difference in the level of detail and openess in the midrange. (Also note that the values of the bypass caps have changed. These will be the 0.47 uF Dayton film and foil caps). | |||||||||
|
|
|||||||||
| And the SpeakerWorkshop measured response. In this measurement, quasi ground plane data is spliced below 200Hz. This is a quasi-ground plane measurement because ground plane measurements should be taken in a large open area on a hard surface, such as a parking lot. Since I couldn't do that, I tried it with the speaker laying on its side on my living room floor. The lower frequency depiction seems somewhat better but probably still not totally accurate. | |||||||||
|
|
|||||||||
| And the impedance: | |||||||||
|
|
|||||||||
| Everything was fine with the preceding version, except that I noticed an excessive sibilance on female voice and an overall brightness that needed to be dealt with. It was apparent that this tweeter wasn't responding well to the second order crossover. You can see a rise in the 7000 to 9000Hz range which I believed to be the problem. I considered switching tweeters, but first I decided to try a third order alignment and it seemed to work. The third order seems to allow more of a plateau effect in the tweeter's response, thus minimizing the effect of the peak. Now, there is slightly less in terms of absolute detail recognition, but the sibilance is normal and imaging and impact are excellent. This is the final version I'm went with for my customer (my daughter). | |||||||||
| Here's the SpeakerWorkshop simulation: | |||||||||
|
|
|||||||||
| And here's the SPL response plot for this version. Notice the absence of the 7000 to 9000 Hz peak which was the culprit for the sibilance. Overall, treble response is shelved down by a couple of dB, but it results in confortable listening. Builders can try reduced L-pad attenuation to tweek this up a bit... | |||||||||
|
|
|||||||||
| The result with this crossover is a warm presentation with a smooth, neutral midrange. Overall the speaker's sound is non-analytical and pleasant to listen to. Soundstaging and imaging are excellent. |
