|
|
||||
|
|||||
|
The UBT
The UBTs are a 4 ohm, TMM 2.5-way sealed box design. They produce deep bass (F3 in the 40 to 45 Hz. range) from a .67 3^ft. cabinets. The woofers used are the Dayton 6.5" woofers (part# 295-305) and tweeters are the Dayton silk tweeters (part# 275-070). ObjectiveI could try to make up something nice here, but the truth is, these speakers are the product of an obsessed, stubborn old man. I like the sound of deep sealed bass and I also wanted small speakers. Unfortunately, these two desires seem to be at odds with each other. It seemed to me that to get deep bass from a sealed box, I would need a large woofer and a large box. As I contemplated this dilemma, I conjured up the .5S (S for sub) idea. While it may not be a new radical idea, it was new to me. Getting a little Einsteinian, the thought occurred to me that all things are relative; including sound. Bass only sounds weak if all the other frequencies are louder. If I could attenuate the other frequencies, the bass would sound good. Most speakers are capable of producing far more SPL than I need, so why not trade in (throwaway) that extra SPL in the mids & highs and get relatively louder bass? The objective of this project has been to implement this concept and build a pair of speakers the produce good sealed bass from a relatively small box. Whether or not it is worth the trouble and additional expense is another question. As for me, I am happy with the outcome. Driver SelectionThere are tradeoffs to consider when trying to cover the spectrum with only 2 speakers. If I was going to do a ported design I could have chosen a woofer that wasn't as strong with bass. Though the Dayton woofers have good bass capabilities, they also have some weaknesses. Unlike the woofer response illustrated in the next section, the woofers I purchased, exhibit a dip or lull around 1.5 kHz., so they cannot be crossed high. Fortunately, the Dayton silk tweeter has a low Fs and can be crossed as low as 2 kHz. Since it is being padded down to reach the target system response of 82 dB, I was able to cross at 1.5 kHz and nullify the dip in the woofer's response. Considering other possibilities, most of the woofers I looked at in the next price category up, can be crossed higher, but just don't have the bass capability that this woofer has per the size box I wanted use. The Vifa P17J woofer could have been used without sacrificing any bass, but the box would have been nearly twice as large. One of my goals was to produce deep clean bass from a small box (under 1 cft.). The Dayton drivers are also inexpensive, which was enticing to a novice such as myself. In my opinion, they produce excellent sound for their price category and had some strengths to meet my requirements that more expensive drivers did not have.
To fully understand the problem description, it is essential to be familiar with the drivers response and their impedance. Since I did not have a measurement capability when I began this project, I had to rely on others. PE Paul graciously gave me his measurements of the Dayton 6 1/2" woofer. Paul measured the woofer in a box with the same baffle width, but larger volume. The SPL response in the bass region is not as critical to this design as the impedance for the size box used. Before I could finish this project, I had to measure the impedance in the actual box. I finished the project using a measurement taken by Dennis Murphy for me in my test box at DIY DC 2001. Paul's SPL response and my impedance are shown below.
Dennis Murphy also very kindly measured my tweeter at DIY DC 2001 in my test box. His SPL response and my impedance are shown below.
The project was difficult because of my dependence on others for measurements. Eventually, one of my silk tweeters went bad. Parts Express replaced it with a generation 3 tweeter. So I bought a second and switched to the project to the generation 3 tweeters. I got a measurement of that from PE Paul. The final outcome was close to what I modeled, but I had to interpret my model differently to get the desired results. I won't elaborate, but will state that a mic preamp for SPL measurements is now at the top of my list. Rather than show every measurement I've looked at and used, I will use the ones above and will make a note of the final results where they are posted. ProblemMy goal was to extend the bass by adding a second woofer to obtain a 6 dB gain and then throw away the extra SPL gained in the mid & high frequency range. There are two approaches I could have taken. First, I could have created a single crossover for both woofers that would have attenuated their mids & highs. This crossover would have been built around a large inductor with a resistor in parallel (something like 4 mH & 1.5 ohm). The second approach was to attenuate the lower woofer (the .5S woofer) and let the main woofer run full range to the crossover with the tweeter. This crossover would be built around an even larger inductor (8 mH). Both approaches had advantages and disadvantages. They also shared a significant challenge - the changing impedance close to the impedance peak at Fc. The first approach had a smaller inductor, but a larger cap and the mids would have gone through a resistor. It would have also required an MTM design to place the woofers equal distant CTC from the tweeter. The second approach has a larger inductor, but a smaller cap and no resistor for the mids. I chose the second approach. The problem with impedance arises from the need to begin attenuating response in the range where the impedance is changing rapidly. As can be seen in the woofer's impedance graph, the impedance is changing rapidly below 100 Hz. The degree of significance of the changing impedance was not apparent to me when I first began. My initial thought was that I would use a large inductor and live with the "little" hump that would result from the impedance. The following graph illustrates the fallacy of this thought. I modeled a 24 mH inductor on the .5 woofer and a crossover on the main woofer.
The target system response is around 82 dB. Ideally, the system response would follow the response of the main woofer down to 150 Hz. then continue at 82 dB down to 60 Hz. Two problems show up with this graph. First, the 6 dB rolloff doesn't attenuate the .5 woofer enough and it contributes to the system response all the way up. (This problem is addressed in the Crossover section.) This could be improved a bit with an 8 ohm-47 uF cap zobel, but not enough. The second problem is the .5 woofer is not attenuated enough at the impedance peak and the inductor is already too big. It is starting to attenuate the response on the lower side of the impedance peak - attenuating the bass that I am trying to get. The next step is to use a 2nd order crossover on the .5 woofer instead of a 1st order, but this makes the problem worse. The graph below shows the effect on a L-R 2nd order crossover at 100 Hz. using an impedance of 8 ohm (25.4 mH & 99.5 uF).
The 2nd order crossover does attenuate the response in the mid range, but it also changes the phase and begins to detract from the response of the main woofer. Also, for some strange reason, the peak is actually higher than it was before (87 dB instead of 85 dB). Alas, I had to admit to myself, there is no getting around the impedance peak - I had to do something about it. My first attempt to address this was to calculate an LCR circuit to flatten the impedance, but this would have been too expensive. I came up with a compromise that is a little less expensive - but still not as cheap as I would like. It required a large cap and resistor in a "zobel like" arrangement. By paralleling the .5 woofer with an 830 uF cap (yes, you read it correctly) and a 5 ohm resistor (two 10 ohm 20 watt in parallel), I was able to get the following impedance curve.
There is still some change below 100 Hz., but it is much less significant. It is smooth enough to create a crossover for the .5S woofer. The impedance dips to 3.25 at 150 Hz., but this will change when the attenuation components are added. Cabinet DesignThe suggested sealed box size for the Dayton 6.5" woofer is .23 ft.^3 which yields a box Q of .71 and an F3 around 75 Hz. I chose to use a box nearly 50% larger (.335 ft.^3) which yields a box Q of .62 and an F3 of 76 Hz. with a single woofer. The shallower roll off helps accomplish the goal of extending the bass. With two woofers, the size is .67 ft.^3. I used a 3/4" front baffle of 8 1/2" x 19" and 1/2 lb. of fill in each cabinet. A 3/4" inner baffle is glued to the front baffle to yield a 1 1/2" front baffle. The layout of the front baffle is presented below. The center for each of the drivers is shown. As you can see, the spacing is tight. A little extra space can be added between the woofers. There is already a little extra space between the main woofer and the tweeter. Of course, for the second baffle, the tweeter is offset to the other side.
At the risk of overkill, I've included list of the pieces I used and illustrations of how the cabinets are assembled. The exterior dimensions are: 8 1/2"W x 12" D x 19" H.
Below is an illustration of a cabinet viewed from the side with the sides removed:
A view from the top with the top & bottom removed is presented below:
And finally, to avoid confusion, here is an illustration of the rear baffle mount:
When I began, I did not have a crossover topography in mind. I modeled several crossover topographies in LspCAD and optimized each. The crossover that had the smoothest predicted response was a 2nd order series topography. This crossover also grabbed my interest the most. I must admit that as a novice, the lure of doing something different and unique was a factor. Being an inexperienced beginner, I didn't hesitate to try something different. I feel lucky the project turned out as well as it did, since I learned that there are a lot of unpredictable quirks that can occur with this topography. Since I'm new to crossover design, I do not have a lot of extra components lying around that would have allowed me to try several different crossovers. So I chose the 2nd order series and ordered the components. At the end of this project, and the 6 months of tweaking and learning that went with it, I now have a few components available for my next project :) My initial predicted system response is shown below. My target system response was 82 dB, but I couldn't resist the temptation to use the .5 woofer to get a little extra SPL even though it was violating my concept of obtaining relatively louder bass. This turned out to be a mistake.
When I assembled this crossover, I was not satisfied with the quality of sound in the range from 500 to 1000 Hz. The mids were muddy, so I tweaked the crossover a bit to cut the .5 woofer off earlier. At my request several fellow DIYers (Nowd, Pjay & Mike Keenan) listened to the speakers and provided feedback. Honest feedback can be a nasty pill to swallow, but I couldn't have improved without someone willing to administer it to me as gently as possible. Again the comment was that the 500 to 1000 Hz. range was gritty or muddy and again I tweaked the crossover to cut the .5 woofer off earlier. The following graph was my final predicted system response. The rest came be ear and feedback. The impedance chart shown below does reflect the final crossover. The third graph is the final measured impedance.
At DIY DC 2002, Dennis Murphy measured the speakers which helped me identify a loose connection in my crossover, and he pointed out that the speaker was a little hot. After more tweaking (it never ends) Brian Smith measured them for me. By studying his measurement and my model, I was able to determine the final tweaks before DIY Dayton. With the v3 tweeter, the response does not have the dip-peak-dip shown from 5.5 kHz to 12 kHz. Darren measured the speakers response at DIY Dayton 2002 and there was a broad, shallow dip from 3 to 3.5 kHz. I also received some feedback that the speakers were still a little bright. When I got home, I modeled a tweak to address this. This led to my last change, I replaced a 6 ohm resistor in the tweeter portion of the crossover with a 10 ohm. This lowered the response a little above 4 kHz. I also replaced a 4 ohm resistor in the woofer portion with a 5 ohm. This lowered the response from the woofer a little above 1000 Hz. The result is the mids sound cleaner and more pronounced, there is a better balance and the highs are a little softer. My tweaking began in April and now, 6 months later, I think they sound good enough to freeze the perpetual cycle of tweaking. The final crossover is shown below. NOTE: The impedance dips below 4 ohm from 70 to 80 Hz in the model. Actual measurements reported Rmin is 3.95 ohm. Also, the woofer tester is not accurate in measuring system impedance above 10 kHz., so the dip in the measurement between 10 & 20 kHz is not correct. I measure the 10 ohm resistor that comes with the woofer tester and observe a similar significant dip in this range. A more amp friendly version of the .5 woofer crossover is to replace the 830 uF cap with a 1000 uF cap, bump the inductor up to 10 mH and increase the resistor from 5 ohm to 6 ohm. The 125 cap remains the same. This version produces about the same results (the difference is probably inaudible) and it does not model a dip below 4.5 ohms. I do not used this alternative and I have no problems driving these speakers. Also, less expensive resistors could probably be used. In talking to individuals I trust and respect, they tell me the components that is most likely to produce audible differences are the resistors. I used Mills, so I kept them in the project.
After the first stage of tweaking, Mike Keenan and I compared the UBTs to several speakers he has and they compared very well. The muddiness that was present in the midrange earlier was gone and they presented a very clear image. Now after the second, third, fourth, fifth ..., rounds of tweaking, it is finished and I am very happy with the midrange. They have been improved significantly from my first crossover. I have finally padded down the highs enough and they are pleasantly sweet. The bass was good from the start. It can be felt in the floor - it is strong, clean and accurate. The sound of an acoustic standup bass is exciting. Getting an accurate measure of bass response is difficult because of room space and reflections. Using Paul's measurement of the woofer, F3 is predicted around 45 Hz. Using the LspCAD box tool, F3 is predicted closer to 42 Hz. I created a CD with test tones (sine waves) spaced at 1/6th of an octave apart from 20 Hz. to 20 kHz. Based on subjective listening tests, I believe F3 is between 40 & 45 Hz (40 Hz. is where I could detect a difference first). Paul's measurement was taken in a 4 pi setting. Since the UBTs are a sealed box design and are placed close to the wall (more like a 2 pi setting), there is more bass than what would be assumed or predicted from the measurement. When comparing the UBTs to the DIIIs, a speaker with a known F3 below 40 Hz., Mike and I found the UBTs to have bass nearly as deep, but not quite. This comparison was replicated later with Brent Langdon's DIIIs. It is also possible that the shallow rolloff is creating an illusion that there is no loss in bass beyond the actual F3. Lest it be forgotten, though the UBTs are nearly as deep as the DIIIs, the cost to get that bass from a sealed design is 6 dB of efficiency. As stated above, this is a tradeoff I can live with easily. They are as sensitive as the Dayton IIs and the BR1s. Anyone considering this project would need to decide for themselves if the sealed bass sound is worth the extra cost. The .5 woofers account for about $76 of the project. If the boxes were made a little smaller (.6 cft.) and a 1.5" D x 4" L port were used, the f3 would be a little deeper. From my perspective, I like sealed bass. It sounds natural to me. I concluded that I couldn't build a sealed sub that could blend in as well and do what the two .5 woofers do for anything close to $76, so it's a good tradeoff to me. Needless to say, I'm very pleased with the outcome of this project. Parts List
Here are some comments from people who have listened to the UBTs at DIY Dayton 2002 and at other times. Some of the remarks I obtained by posting an open invitation for comments. I promised to append them uncut, thinking they would be good or bad about the UBTs. I feel a little uncomfortable that comparisons were made to other speakers. Also, though no one took the opportunity to slam the UBTs I do know people who prefer speakers based on better drivers. "Ken Ahern's 2.5 ways are possibly the best implementation of the Dayton 6 1/2" I have heard." Dave Tenney "The Dayton designs were truly surprising....Ken A.'s TMM and the D3's were very good." Tim Kolody "Ken, your speakers were perhaps the best of the small speaker group and you are way too humble!" Bob "Audiohead" "I attended the Dayton DIY and was very impressed with your design. I'm new to speakerbuilding, but could tell that you have designed a great speaker. The UBTs sounded much better than the DIIIs. Please post your design on the PE web site. I would like to build a set for my son." Bruce "The drivers that Ken selected for this design are incredible. The Dayton silk dome has built a reputation as the best value tweeter on the market and the Dayton 6.5" woofer is a well-behaved driver that produces a tremendous amount of bass for it small size. Both of these drivers compete well against drivers with twice the price tag." "I own a pair of DIII speakers that I love. The DIII's use the same drivers as Ken's design. Ken has been able to capture many of the things that I adore about the DIII design into a package that is considerable smaller in size." Brent Langdon "Nice clean bass
The baffle is solid oak. I rounded the tops and bottoms to the sides then wrapped the veneer. After the veneer was on, I rounded the edges of the front baffle. This helped the baffle blend into the tops and sides. BTW, the speakers and stands are the same wood and veneer and stained the same.., just 9 months apart.
I used 1/2 lb. of stuffing in each box. If you look close, you can see the red mouse pad on the being hidden inside the box by the stuffing. There is another mouse pad on the other side too. I used copper foil tape to make traces for my crossover. The stuff is cheap at craft stores which carry supplies for stained glass windows. I got a 30 ft. roll for about $4. It works well, but in all my tweaking (constantly soldering in then removing and soldering again) the adhesive would melt and then it wouldn't stick where I was soldering all the time. I thought it would look cool to use clear Plexiglas for the crossover (you never know when someone is going to demand that you take the rear baffle off so they can see what you made your crossover out of :) ). The stuff is too brittle! I had to drill holed to mount the inductor and to mount the crossover to the rear baffle. A couple of holes I drilled cracked the corners off and I had to glue the corners back on with model glue. The Plexiglas was not a good idea.
Nylon spacers are easy to get at any hardware store. All my previous crossovers were glued on. With all the times I was taking it out, tweaking it and putting it back in, it was nice to have it screw on and off instead of being hot glued, or even worse, gooped to the rear baffle. Of course feeling proud of myself for this easily removed crossover, I took the UBTs to Pjay's house to measure. It was there that I realized that normal people develop their speakers with the crossovers outside the box and don't mount them till they feel like they're finished.
I'm not a master wood worker and I'm sure that no one else has ever been faced with this dilemma. When I routed to flush mount the speakers, I made the holes a little bit too big. (I don't have a rabet bit, so I route a mortise, then the speaker hole.) Nothing will ever make the problem go away, but the simple solution I found was black spline. It's cheap and you can get it at most hardware stores.
I don't mean to sound boastful, but I've gotta admit that I'm very pleased with the end results of this project. Black spline and all, they look pretty good and they sound even better. They are easily the best speakers I've ever owned. But now its time to look forward to my next speakers; a Vifa TC 3-way. The Builder This project was submitted by the venerable KenA, a frequenter of the Tech Talk board. Look for him there with questions. DISCLAIMER - Parts Express posts the "Project Showcase" as a community service. Projects displayed have not been tested or evaluated by our staff. We make no warranties whatsoever in regard to the projects' fitness for a particular purpose. Claims of performance or fitness are those of the contributor and not of Parts Express. Project re-production for commercial resale may be a violation of United States patent or trademark laws. Copyright 2002 Parts Express
|
||
|
|
|
|
|
||||||||||
|
||||||||||||||
| Parts Express • 725 Pleasant Valley Dr. • Springboro, Ohio 45066 • USA © 1996-2008 Parts Express, All Rights Reserved |
||||||||||||||
