About 10 years ago, I built some loudspeakers for college course credit. The speakers were based off of Professor W. Marshall Leach's 3-way speaker plans. Prof. Leach is an electrical engineering professor at Georgia Institute of Technology, my alma mater.
I needed a new project (got tired of doing the big Sunosubs) so I decided to build a a center channel speaker that had the same sonic signature of my main front speakers.
I wanted to say "Thanks" to EdP for all the help and guidance he gave through the process. He was extremely patient with my questions and pushed me to get me to this point.
I decided I wanted a maximum height of 10" for the CC speakers. This was mainly due to the height of the tweeter and midrange drivers, and factoring in 1.5" for the two 0.75" thick MDF panels. This made the selection of the woofer a little tough because I had to settle on the size of the woofer compartments without making a really huge CC speaker enclosure. That's why I settled on two 6.5" Vifa M18WO-09's (wired in parallel), and I gave each one about 14 liters of volume, and each compartment was tuned to 50Hz (2" wide port 7" long). I thought about going with 8" woofers, but I would have needed really large compartments for them, so I went back to the 6.5" woofers. The final box dimensions are 28"x10"x14".
I picked up parts for the crossover, ports, and drivers from both Audio Lab of Georgia, and Madisound. I got the MDF, glue, and screws from Home Depot. A friend of mine has a nice workshop with a table saw and helped me make all the MDF panel cuts.
As an exercise, I picked up a Vifa D27TG-45 tweeter for testing an afternate tweeter later on.
Update 8/26/01 - I revamped the crossover once again using 2nd order Linkwitz-Riley filters. Here's my "final" crossover below. It's a gaudy crossover with 12 components for 3 levels of drivers:
Here's the frequency response for the tweeter:
After 4KHz, the tweeter gets a bit "hot".
Here's the frequency response for the midrange:
Around 3KHz, the midrange gets really peaky.
Here's the frequency response for the woofer:
I'm not too concerned with bass response under 100Hz since I usually run the center channel speaker in the "small" mode, and allow the bass management from my receiver to migrate the CC's bass to my subwoofer(s). But it does put out bass all the way down into the 50Hz range when I run it as a "large" speaker.
To test the overall frequency response of the CC speaker, I took measurements from my main seated position, about 3 meters away, on axis with the CC speaker. Please bear in mind that I am using a BassZone Test CD from Stryke and a Radio Shack SPL meter, which doesn't measure too well above 12KHz, so I am basically ignoring SPL reading over 12KHz for now. (Note, as of measurements taken after 8/01, I've been able to come up with "correction values" for the SPL meter readings, so the graphs will look more "normal").
Amazingly enough, when you compare the individual driver response curves to the overall response curve, you can see each driver's response reflected in the overall frequency response of the CC speaker. I tried my best to tame the peaks around 400Hz and 3000Hz. What I've wound up doing is EQing the center channel by bumping 500hz by 1dB or 2dB, and cutting 3100Hz by 1dB. This smoothes out the dialogue quite well to my ears.
I have tons of graphs showing the incremental changes and their effect on the overall response of the CC speakers. I will put this up at a later date. This is definitely a trial-n-error crossover design process (I don't have the fancy software to do extensive modelling of speakers).
I have, on the drawing boards, another crossover network that's made up of 2nd order Linkwitz-Riley filters and zobels on the midrange and woofers. The crossover points are different from my own initial crossover design. The woofer/midrange is crossed over at 700Hz (in an attempt to get rid of the 400Hz hump caused by the midrange and utilizing the woofer's smoother low end response before 1000Hz), and the midrange and tweeter are crossed over slightly separated due to the response of each driver at 3800Hz(midrange) and 4100Hz(tweeter). I may give that crossover a try when I get the right components.
Update 4/21/01 - I finally get the crossover components I need, and install a new crossover network. I crossed the tweeter with a 2nd order L-R around 3250Hz, I cross the midrange's lowpass with a 1st order Butterworth filter around 2000Hz to take advantage of the rising response from 2000-3500Hz and remove the need for a Zobel to level out the rising impedance. The highpass of the midrange is a 2nd order L-R filter around 700Hz, and the woofer's lowpass filter is a 2nd order Butterworth around 700, which is used to gain a little back around 500Hz. The response looks a little peaky and not too hot on the high end, but it sounds more natural than the original crossover network I came up with last month. I have the intermediate attempts for this new crossover network in the Crossover Tweaking section below (after the construction picture section).
Here's the "final" frequency response curve (as of 8/26/01):
Hint: For faster navigation, when you click on your first picture link, don't close that 2nd window, but resize this window and position it so you can see both windows concurrently. I've designed it so that you can leave that 2nd (photo) window open, and you can click to your heart's content in this first window, and the images will only show up in that 2nd other window. This should also speed up your visit here. If you want separate windows, then right click of the photo links and select the "open in new window" option.
Here we go:
Day 1 (Saturday 2/17/01)
The raw materials - wood panels:
Photo 1 : Here are the scraps of MDF from my previous Sunosub projects.
Photo 2 : More MDF scraps.
Photo 3 : Here a look of the MDF panel cut to my specifications by a friend of mine with a wood shop that would put Tim "The Tool Man" Taylor to shame.
Day 2 (Sunday 2/18/01)
Photo 4 : For the front panel, I decide to use T-nuts so I can take out the drivers easily, this means glueing in a piece of plywood to give the T-nuts something to sink their teeth into because MDF is easily stripped. Here I am marking where the plywood panel should go on the backside of the front panel.
Photo 5 : I test for fit of the panel, roughly, and this give me an idea of how things will be put together.
Photo 6 : Here's the plywood panel and the glue.
Photo 7 : I proceed to squirt the glue on the back of the front panel.
Photo 8 : The plywood is glued in place.
Photo 9 : It doesn't look like it, but I'd got that front panel and the plywood underneath that stack of MDF, and this will apply the weight for a strong glue bond.
Photo 10 : Now I start placement of the tweeter and midrange, as well as the inner walls/braces for the enclosure.
Photo 11 : I mark off the walls and the drivers on the plywood for cutting purposes.
Photo 12 : Here's a close-up of the markings I made.
Photo 13 : Using a ruler and a compass, I draw the circles I need to cut into the front panel for the tweeter and midrange.
Photo 14 : I start with drilling the pilot hole for the router and the circle jig combo.
Photo 15 : The router doing its thing, cutting circles for me.
Photo 16 : Both the holes for the tweeter and midrange are now cut.
Photo 17 : I test for fit of the drivers.
Photo 18 : Next I start on the hole for the woofers. This gives me an idea of the total area the woofer will occupy on the front panel.
Photo 19 : I do the hole for the other woofer.
Photo 20 : After finding the center of the woofer hole and the inner width I need, I drill in a pilot hole for the router/jig combo.
Photo 21 : Routing out the woofer hole.
Photo 22 : Woofer hole finish and woofer waiting for placement.
Photo 23 : Test for fit of the woofer hole.
Photo 24 : I finish cutting out the hole for the other woofer.
Photo 25 : Test for fit of the other woofer hole.
Photo 26 : On the back panel, I mark the center point for the small terminal cup, this requires a 2" hole.
Photo 27 : While I drill out a pilot hole, I quickly discover my jig won't allow me to cut in a 2" hole.
Photo 28 : I use the compass to draw the circle, and then use the router free hand and cut out the cup hole.
Photo 29 : Test for fit of the terminal cup in the back panel.
Photo 30 : I drill holes for the screws to be used to firmly attach the cup to the back panel.
Photo 31 : I use 6-32 1.5" screws/nuts for the terminal cup.
Photo 32 : Screws are in.
Photo 33 : Nuts are screws on with washers.
Photo 34 : I drill the holes for the tweeter.
Photo 35 : I drill the holes for the midrange.
Photo 36 : To use T-nuts, the hole needs to be widen just a little bit, about 1/4" deep, which explains the masking tape on the drill bit.
Photo 37 : I apply a little glue around the holes for the T-nuts for extra adhesive.
Photo 38 : Then I hammer in the T-nuts.
Photo 39 : To seal the T-nuts, I apply some more glue around the T-nuts.
Photo 40 : I do the same for the midrange regarding T-nuts.
Photo 41 : I start on the woofer screw holes and the T-nuts, but then I realize I need to cut in the port holes...
Photo 42 : I place the ports for fit on the front panel.
Photo 43 : I draw in the port's outer ring and the hole needed for each port.
Photo 44 : I freehand the port with the router (since it doesn't have to be perfectly cut).
Photo 45 : Test for fit of the port and the new port hole.
Photo 46 : Both port holes have been cut.
Photo 47 : Test for fit of the ports.
Photo 48 : Next I start on the inside walls and the bracing for the back panel (which will allow me to use wood screws to screw in the back panel, and give me access later when I need to make changes for the crossover network.
Photo 49 : I use a saber saw to cut out of the corners out of the inner walls for the bracing to sit on.
Photo 50 : Test for fit of the bracing and the walls.
Photo 51 : Here's a shot of both inner walls with the required corners cut out of them.
Photo 52 : Here is where I draw in the grooves that I will use to hold piece of plywood that have the crossover network for the tweeter and midrange on one piece, and the bass network for the other piece.
Photo 53 : I use the router, and the straight piece of MDF to provide a rail to slide the router down to route in a 1/4" deep groove into the walls.
Photo 54 : The router in action, routing out the groove.
Photo 55 : One wall's grooves are done.
Photo 56 : The other wall's grooves are done, and the plywood pieces are tested for fit.
Photo 57 : This is how the crossover network will look inside the enclosure.
Day 3 (2/19/01)
Photo 58 : I finish applying glue for the rest of the mounting holes for the T-nuts.
Photo 59 : I hammer in the T-nuts for the woofer mounting holes.
Photo 60 : I seal up the T-nuts for the mounting holes.
Photo 61 : finally done with all the holes and T-nuts.
Photo 62 : Now is the time to glue the top, bottom, and side panel together.
Photo 63 : Apply the glue on the bottom panel.
Photo 64 : Apply the glue to the top panel.
Photo 65 : Apply the glue to the inside wall panels.
Photo 66 : I use the front panel (on the bottom) to line up the panels before clamping the panels together.
Photo 67 : The clamps are used to keep the panels glued together as tight and close to being a rectangular box as possible.
Photo 68 : One more look at the clamped up box. Notice the crossover panel stuck between the inner wall panel for fit.
Photo 69 : While the box is clamped, I was playing around with the bracing, and noticed that the bracing is sitting a bit too high, so I route just a little bit off the bracing so I can account for the weather-stripping later.
Photo 70 : Here's a look at the bracing.
Day 4 (2/20/01)
Photo 71 : This is the day I glue the front panel to the box, so here's all the necessary glue on the front panel.
Photo 72 : More clamp action, but this time, it's to clamp the front panel to the box.
That's it for now.
I'm waiting for some crossover parts, mainly the polypropylene capacitors from Madisound.
Day 5 (2/22/01)
Photo 73 : I test for fit on my bracing, and determine I need to sand down the inner walls just a bit, so that's why the sander is there.
Photo 74 : The long brace needed a bit routed off (like Lincoln logs), and I applied glue on the required surfaces.
Photo 75 : The glue is also applied to where the bracing will be attached. I use a pen to draw a pair of lines where the brace should be aligned.
Photo 76 : Clamps are placed to squeeze the bracing's glue for a tight bond.
Photo 77 : The cabinet is flipped over, and I apply glue for the other long bracing piece.
Photo 78 : Again, clamps are used to squeeze the bracing for a tight bond.
Photo 79 : More glue, more bracing, this time, it's the piece that goes along the side of the cabinet.
Photo 80 : Using the clamp, and the short piece, I position the side bracing between the 2 lines I previously drew for final placement (not shown). This is to ensure the fit of the "T" style bracing I'm using.
Photo 81 : I glue the other side bracing.
Photo 82 : This gives a better picture of the placement of the side brace (I also glue the bracing's bottom which butts into the inside front panel).
Photo 83 : Here's a shot of both bracings being clamped for a good couple of hours (or more). You can also see all the glue I apply to the inner walls and the long pieces of bracing. Go nuts with the glue!
Photo 84 : This is a close-up of the clamping and how the "T" style bracing is being utilized. At this point, I haven't glued that short piece of bracing yet.
Photo 85 : I just remembered to drill holes through each inner wall so I can feed the woofer drivers' wires to their terminals. As you can see it's a tight fit with the power drill. I recommend drilling the holes earlier than this, before installing them into the box.
Photo 86 : I test for proper hole size by feeding a piece of wire through the hole I just drilled, this is the second wall. I needed 2 passes to get the hole oblong enough to fit the speaker wire through it.
Photo 87 : Finally the top little piece of bracing is ready to be glued to the side bracing to form the "T".
Photo 88 : There's lots of surfaces to apply glue for these little short pieces of bracing.
Photo 89 : More clamping is involved in getting the tight bond from just glueing the bracing to the cabinet's inner walls.
Photo 90 : I quickly get both little pieces of bracing glued into place and clamped. I give the glue of the bracing the entire next day for the glue to dry.
Day 6 (2/23/01)
Photo 91 : It's now time to caulk the inner seams of the cabinet with all the bracing in place.
Photo 92 : As you can see, I've caulked the backside of the cabinet inner seams.
Photo 93 : Finally, all the inner seams are caulked.
I wait an hour for the caulk to dry. While it's drying, I start on the crossover network. This takes forever and a day to do. There's lots of places where you just have to be patient, and do all the little things to get it wired up correctly.
Photo 94 : I originally wanted to put my tweeter and midrange crossover components on 1 board, and the woofer on another board, but it just wasn't going to happen, I ran out of real estate on the board for both the tweeter and midrange, so I put the midrange on one board and the tweeter and woofer on the other board. This part took a while since I was playing around with the best way to lay everything out, so to get to this point, I must have spent 1-2 hour mucking around. There is one capacitor missing (I used an electrolytic cap in the upcoming photos just to get the crossover running while I waited for the polypropylene cap to arrive).
Photo 95 : I mark the holes for the terminals (black with screws on them) and drill them. I got the terminals and the other terminal mounts at Radio Shack.
Photo 96 : More drilling of holes for the components on this board. I do the same with the other board, of course.
Photo 97 : Here is what the boards look like with most every component tie-wrapped to the board. The terminals are attached with little machine screws and nuts (bought at Home Depot).
Photo 98 : Here's a shot of me screwing around...actually getting the components connected to one another based on the crossover I cobbled together. This board is the midrange crossover. It's always a good idea to keep your inductors oriented 90 degrees from one another.
Photo 99 : Just a shot of most of the bits and pieces I used, like male and female spade connectors, and the ring terminal connectors (I spent the next day getting all my inputs terminated with the ring connectors for a nice clean fit, it was a tedious, and yet, therapeutic exercise. I use the insulated female spade connectors when I knew it was going to be used to connect to a driver (to avoid shorting out the connection). The crimper is a godsend. Can't do good connections without it. At this point, it's around 3 a.m. and I'm tired, so I go to sleep.
Day 7 (2/25/01)
Had a wedding to go to on Saturday, and just didn't feel like working on this project, so I took the day off. Now's it's Sunday.
Photo 100 : Doesn't my cat look so impressed with me? I bring in the cabinet, and the drivers because it's time to put it all together!
Photo 101 : These are wires for the woofer drivers. For the connections to the crossover network, I crimp on the insulated female spade connectors to avoid short circuits (If you're cheap, just wrap them up with electrical tape, red for the positive lead, black for the negative lead). I always wrap the negative lead wire with black electrical tape to avoid confusion later. The female spades that will be connected to the woofer driver is not insulated because they will be a couple of inches apart when connected to the driver.
Photo 102 : A shot of me crimping these nifty ring terminal connectors. They allow be to really neaten up the crossover on the input terminals where I have make those little bitty jumpers for provide the positive and negative terminals for 3 sets of drivers (seen on the left side of the board). I had to solder 3 inductors in series to get 8 mH of inductance (FYI). And there's the electrolytic caps I am using temporarily until I get my poly cap (needed two 10 uF caps parallelled to get a 20uF cap) for the woofer crossover.
Photo 103 : Here's a close-up view of both crossover board slipped into their new home.
Photo 104 : Here's a wide view of the backside of the speaker with the crossover installed. I measured it just right to fit 2 levels of crossover components.
Photo 105 : I install the midrange driver, and then connect the wires for the tweeter. My other kitty is fascinated by the whole process.
Photo 106 : Next up is one of the woofer drivers, notice I don't need the insulated female spade connectors.
Photo 107 : Make sure you take your time screwing in the driver screws. I use my other hand to shield the head of the screwdriver from the cone, lest it slip and puncture it. This is where using T-nuts really pays off. I think I used 10-24 1.5" machine screws for the woofers and midrange. Again, kitty is mesmerized by the process.
Photo 108 : What can I say, kitty (Fred) just can't wait for me to finish up so I can play with him.
Photo 109 : The tweeter screws are 8-32 1.5" machine screws IIRC. I use the corresponding T-nuts for the tweeter driver.
Photo 110 : And here is the midrange. This is laborious since I dare not use a power drill.
Photo 111 : The other woofer's wires are connected to the driver.
Photo 112 : Finally, all the drivers have been screwed into all the front panel.
Photo 113 : Here's my testing which wood screw I want to use for attaching the back panel to the cabinet. This one is a #6 wood screw. I test for depth as well as "unscrewability" in case I need to get back inside the cabinet later.
Photo 114 : I test a #8 wood screw, and this is the one I decide to use. It was harder to strip its phillips head, so that was part of my rationale to use it over the #6 wood screw.
Photo 115 : I mark where the bracing is the strongest, so I can intersect some lines to get the screw hole placements in a moment.
Photo 116 : Just thought that it was really tough to get the back panel off if I can't push a hammer through the woofer hole (which I had been doing before I screwed in all the drivers to the front panel). I drill 2 holes, and loop some speaker wire through the holes (with 2 knots for added strength/support when I pull on it later).
Photo 117 : This is what the back panel will have hanging out of it for the interim (I'm thinking about putting in a "handle" later). But it works, and you never see it once I put it in its place in the HT.
Photo 118 : I connect the crossover leads to the terminal leads of the back panel. I use a long piece of speaker wire for flexibility when I take off the back panel and don't want to disconnect it from the crossover.
Photo 119 : This is foam weatherstripping which I'll apply the bracing to produce a good seal mainly for the woofer enclosures.
Photo 120 : I apply the weatherstripping all around the bracing. Yes, I forget to put it on the inner walls, but I'll be back inside once I finalize my woofer crossover components, so I'll fix that at a later time.
Photo 121 : It's finally time to close up the cabinet.
Photo 122 : Remember those marks I made? Now I use them to make the intersections which give me the holes to drill for the wood screws.
Photo 123 : I use an 1/8" bit to make the pilot holes for the wood screws.
Photo 124 : To countersink the wood screws, I use a thicker bit to drill a depth of 1/8" or so. The masking tape is used for the depth marker.
Photo 125 : The screws are screwed into the bracing nice and evenly. I confess, I forgot to include additional depth for the foam weatherstripping, so the back panel is not completely flush with the rest of the cabinet. I will remedy this by routing out a perimeter 1/16" deep, 1.75" wide, on the inside back panel. I got the medium compressible weatherstripping. Doh!
Photo 126 : The last thing to do is create the ports for each woofer. I need roughly 7" long ports 2" wide. I found these 2" wide, 4" long ports, so I bought 4 of them with the intention of sawing off 2 of the ports to use to extend the other 2 ports. I love my PVC pipe saw, just handy for this application. I wound up having to saw it off around to get the 3" of length for each pipe add-on.
Photo 127 : Both of the chopped down ports. They get chopped down where their fins disappear.
Photo 128 : I use black electrical tape for the adhesive on the inside and outside of the port.
Photo 129 : Here's the total new length for the port.
Photo 130 : Ta-da! Once the ports have been inserted and hammered into the port holes, the center channel speakers is finally operational.
Photo 131 : Here's a shot from the left corner.
Photo 132 : And here's a shot from the right corner (sorry, proud papa operating the camera!)
Photo 133 : Since I can't connect the CC speaker up to my current setup (have to re-arrange everything to fit this speaker over the RPTV), so just for fun, I listened to it as the left speaker just to see if I had gotten my connection all wired up correctly. Amazingly enough, I got my crossover hooked up right on the 1st try. Oh yeah, next to it is my Giant Pickle Sub.
Photo 134 : A gag shot of the CC speaker incorporated into the HT setup.
Photo 135 : Here's what my HT setup looked like before I moved everything around to accomodate this new CC speaker.
Photo 136 : This is how my HT setup looks after I moved my HT gear around. I had to relegate my Harman Kardon PA5800 5-channel amp into the closet behind the subwoofer where my subwoofer amp also resides (but that's okay, I rarely need to "see" it anyway).
Photo 137 : Here is how my current HT looks like with the new CC speaker in place. I'm tempted to forego the veneering as the front panel's color is a close match with my mains, and it's not like I'm impressing anyone with my stuff anyway.
Hopefully I'll get the veneering done by the end of April.
Well, I have just a bit of testing to do to finalize my woofer driver crossover. I did some measurements of the CC speaker with the woofers connected in series. I'll post the graph as soon as I can get the numbers all graphed out. But, from a cursory glance at the numbers, I can tell that I will need to connect the drivers in parallel to get a 4-6dB increase in the SPL output from 400Hz and down to even out the output from the rest of the drivers.
I calibrated my CC speaker with the rest of my system, and I needed to bumped up the center channel speaker output by +2dBs on my receiver. I'm thinking when I go with a paralleled woofer load, I won't have to boost the center channel output much at all.
Well, I basically spent today fiddling around with the crossover network to smooth out the response within +/- 3dB. This is not necessarily hard work, but it is time consuming, and lugging my CC speaker back and forth from the testing spot to the dining table can be construed as working out.
Given the deficiencies of my RS SPL meter in the high end of audio spectrum, I'm not going to worry about the frequencies past 12KHz for now. You'd have to be a dog to enjoy those frequencies anyhow.
Due to my concern that my original Philips tweeter was not too good (due to my not realizing the RS SPL meter's usable bandwidth being 40-12KHz), I decided to install another tweeter, the Vifa D27TG45-06, and measure/listen to how it sounded as it was integrated to my CC speaker. I had to drill some more holes through the front panel, and couldn't install T-nuts, so I just used the nuts that came with the original machine screws. That was the easy part.
I spent some time getting a preliminary tweeter high pass crossover design (2nd order with the break frequency at roughly 4650Hz) plus a tentative L-pad (around -3dB to start), bought some resistors, capacitors and inductors to give me a range to play around the crossover with. The midrange and woofer crossover networks had already been tentatively designed.
Here's 7 graphs that summarize my whole day as a process of trial-n-error to get the response smoothed out:
Graph 1: This is what my starting point is after I install my first "guess" for the tweeter crossover. As you can see, it's pretty rocky, and the high end is just too high. You can ignore the low end on this graph, I've since figured out why it was measuring so low, I had placed the CC speaker over the TV, and it was inhibiting decent results of my measurements. Placing the CC speaker on a make-shift stand around 18" tall helped the next set of graphs on the low end frequency response.
Graph 2: Thinking that my midrange's 1st order crossover wasn't providing enough slope to kill off the hump from 5000Hz to 8000Hz, I decided to install a 2nd order low pass midrange crossover. I was able to just parallel a 2uF capacitor with my original .68mH inductor to get the new 2nd order crossover. The break frequency is 4300Hz for this crossover. Well, as you can see, that just stiffened up the hump. Please ignore the low end response for this graph as well.
Graph 3: So, I thought maybe I needed to pad down the tweeter response, so I swapped out the L-pad resistors (-3.52dB of attenuation) for another set which gave me -8.24dB of attenuation. As you can see, it didn't really help. Notice now that my CC speaker is on the make-shift stand, its low response got a lot healthier.
Graph 4: The next thing I did was to push the tweeter high pass break frequency out from 4650Hz to around 6800Hz with a 2nd order Bessel crossover. I was constrained by what components (L's, C's, and R's) I had on hand, and this is a low Q (~0.5) crossover, so the break is more gradual than most normal crossovers (like a Butterworth/Q~.63 or Chebychev/Q~.7) I though this to be a good thing given what I had on hand. As you can see, pushing the tweeter break frequency out to 6800Hz got rid of the hump. This means the midrange is doing a little extra work from 4000Hz to 6500Hz, but it should be able to handle it. The tweeter gets to slack off until 6000Hz or so (a guess on my part). But as you can tell, the rest of the tweeter response falls off fast, so this isn't quite desirable either.
Graph 5: Since I needed to boost the high frequencies, I decide to de-install the L-pad resistors that gave my -8.24dB of attenuation. Then I re-install the L-pad resistors that gave me my original -3.52dB of attenuation from my initial design. As you can see, this worked out pretty well. I'm not too concerned about the low end because I'll be running this CC speaker as a SMALL speaker, so the low end output isn't a major concern.
Graph 6: Not convinced that I need the midrange 2nd order low pass crossover, I decide to see what happens when I take out the capacitor on the midrange low pass crossover that makes it a 2nd order LP crossover with a break frequency around 4300Hz. With only the inductor in place, the LP crossover is now a 1st order with break frequency around 3500Hz. As you can see, my 5000Hz-8000Hz hump comes roaring back and creates a tumultuous and mountainous response curve once again.
Graph 7: Knowing that I need that midrange LP 2nd order crossover to keep my frequency response smoother, I re-install the capacitor and decide that this will be my final crossover design for these speaker components.
Granted, there's some pockets of lowness in the 125Hz-500Hz range, it's something I can live with. I'm afraid that if I re-do my L-pad for the midrange anymore (or rip it out altogether - it's only providing -1.5dB of attenuation as it is anyhow), I'll screw with my high end tweeter response. The same goes with the 4000Hz range, it's a bit low, but not terribly so. I may reverse back the midrange polarity connection since I went to the 2nd order LP crossover, to see if it gives me back a little more SPLs in the midrange.
You'll notice that on graphs 4-6, there are 2 set of measurements, one is from my HK PA5800 amp, and the other is from my Sony DB930 receiver. I can't explain why the Sony's SPL output track the HK well on one graph, but then lags by 2-4dB on the other 2 graphs. It may have been things moving around since I had to lug the speaker to and from the measuring spot after each change to the crossover network. Also, the overall volume level would go up and down a few dB from different measurement sessions. I tried keeping the receiver's volume level the same, but I'm guessing there were minute variation of the actual placement and some room interaction was going on. But at least the general shape of the curves were consistent.
I'm so glad I borrowed the company cordless power drill to screw and unscrew the back panel off and on.
Next up: After that I document the crossover network for the Vifa tweeter, I'll go back to my original Philip tweeter, and see how far I need to push my tweeter's break frequency to smooth out the frequency response with the Philips tweeter in place. The fun never ends.
Think how much quicker I'd be done if I had some speaker modelling software! :)
Day 9 (3/8/01)
Here's what the response looks like when the CC is placed over the TV per normal use (still using the Vifa D27TG45 tweeter):
Graph 8: As you can see, it's a little rocky all the way through, but it's +/- 3dB through the important part of the audio spectrum for a CC speaker. The 3m response (measured from where I usually sit) is pretty decent given the RS SPL meter's high end metering deficiencies. This is my final design using the Vifa D27TG45 tweeter along with the rest of the drivers.
Now I embark on getting my original tweeter, the Philips AD11600, integrated into the CC speaker to match my main speakers (I'll be re-doing their crossovers as soon as I'm done with fixing the response with the Philips tweeter).
I use a starting point from the previous design with the Vifa tweeter.
Graph 9: It's obvious that the Philips tweeter has a higher sensitivity (I'm guessing around 94dB/1W/1m) and can see the high end hump from it.
Graph 10: I put in a L-pad that lowers the tweeter output by almost 8.5dB using resistors of common values (I needed 5 ohms//5 ohm L-pad combo, but had to settle for 5.1 ohm//5.1 ohm combo since I had to series a 1.1 ohm with a 4 ohm resistor to get close to 5 ohms. This definitely helped get rid of the high end hump, but it's pretty rocky, so I try to get the crossover slopes to line up better.
The next couple of graphs show me going the wrong way on smoothing out the response.
Graph 11: I revert back to a 1st order midrange lowpass, and it creates more ups-and-downs. This wasn't a good thing.
Graph 12: Keeping the 1st order midrange lowpass, I move the woofer and midrange crossover point near 700Hz to try and close of the valleys. Well this just raised the high end on a gradual, but upward sloping response.
At this point, I decide I'm not getting anywhere going down this road, so I revert back to my design from Graph 10.
Graph 13: I ripped out all the components from the design in Graph 11 and get back to the design in Graph 10. I was able to obtain actual 5 ohm resistors, so the high response is smoothed out just a tad more than in Graph 10, though it now has that hump around 800-2000Hz, so that's what I try to smooth out next.
Graph 14: Well, I tried moving the midrange lowpass frequency to almost 800Hz, so that produce a nice fat dip from 250-800Hz.
Graph 15: So, I go in the opposite direction and move the midrange lowpass frequency further lower to 400Hz. Now, it's a little rocky, but within +/- 3dB throughout the audio spectrum of concern, so this is where I decide to stop the design process.
I plop the CC speaker on top of the TV and proceed to take measurements.
Graph 16: Again, the low end frequency response suffers with the CC speaker on top of the TV, but I'm not concerned as most of that gets filled in when I run the CC speaker as a "SMALL" speaker, and let my subwoofer do most of the heavy lifting. The 1m response is also within +/- 3dB throughout the audio spectrum of concern, as well as the 3m response, though the room gives the high bass region a little hump from 200-500Hz, but I can live with this. The high end response rolls off as expected due to the RS SPL meter.
So, at this point, I'll do some listening with the Philips tweeter in place and evaluate it qualitatively.
Day 10 (3/11/01)
Okay, so I listened to my CC speaker, and I found it a little dull, so looking back at graph 16, I noticed my high response was dying quickly after 8KHz at my seated position nearly 3 meters away, so I decided to work getting more high end output at 3m away.
Graph 17: I change the L-pad which gives me -8.4dB of attenuation, and used the original -3.75dB L-pad. This gave back my high end, but it was little too bright.
Graph 18: To bump the high end down, I went with a L-pad which gives me -6dB of attentuation. This was getting me close. I notice a slight trough in my midrange, so I will raise the midrange next.
Graph 19: I changed up midrange L-pad, just leaving a 1 ohm resistor in series with the lowpass network. I raised the midrange output by 1dB from 2KHz-4KHz. Now the midrange and the high end are more on axis with their SPL output.
I think now I'm done with the Philips tweeter in the CC speaker. Unless I want to do something about the baffle diffraction that reduces my bass output on the low end (worst case is by 6dB). I may leave it alone unless the male vocals are too thin. If I go with a parallel woofer connection, I could gain back the loss from the baffle diffraction, maybe I'll try that when I have some more time.
Next, if I can make the Vifa D25AG05 tweeter fit into the tweeter hole (and mounting screw holes) easily, I'll see use the Vifa D27 crossover design as a starting point, and get the response smoothed out and see if I like how it sounds compared to the other 2 tweeters.
Day 11 (3/15/01) So I'm not done with the Philips tweeter. Here are my final final final graphs/crossover networks for this tweeter. I have graphed the response for the center channel speaker operating in the Small and Large speaker mode, this will be important when I wire up the woofers in parallel later on.
Graph 20: I rip out the L-pad on the midrange crossover, and change the midrange's 2nd order high pass into a 1st order high pass by taking out the capacitor. This doesn't produce good high end results (but it was expected, thought I wanted to see what would happen when I took out the midrange L-pad).
Graph 21: I change the midrange high pass back to a 2nd order to smooth out the high end. The low to midrange looks a little too humpy now without the pad resistor.
Graph 22: I put back the 1 ohm resistor pad in the midrange to smooth out the low to midrange. This is about as much as I can do unless I totally redesign the crossover. So I'm calling this my final crossover for when the woofers are connected in series, even though my low end response is lacking.
Now, I wire up the woofers in parallel to get some low end back (up to 6dB if I'm lucky).
Graph 23: Using my last crossover as a starting point, I connect up the woofers in parallel, and swap out the woofer crossover components. Using a low pass crossover point of 400Hz for the woofer, I get this big hump/spike around 400Hz, so I decide to fiddle with the high pass of the midrange to smooth out the response in the 300-700Hz range. If you'll notice, my low end has dramatically come back to life since I gained woofer output with each woofer connected in parallel. It causes a drain on the amp, but I should be okay.
Graph 24: I change the midrange's high pass crossover point from 460Hz to 640Hz to spread out the hump. It appears I was successful in taming that 400Hz peak, and notice when the speaker is set to Small, the response is relatively flat +/-3dB from 80Hz to 12KHz (I'm discounting the upper high end due to my RS SPL meter's high end deficiency). I can live with this crossover setup with the Small speaker setting for the center channel. So I'll give this setup some listening time and see if it's a keeper.
Day 12 (3/17/01)
Minor confession: I've been using an 11 year old midrange that I salvaged a couple of years ago, so I decided last week to order a younger one (same vintage as my main speakers). I got the new driver yesterday (thanks, Madisound!) Here is a graph showing slight improvement in the upper midrange after replacing the old midrange driver. Now all the drivers are pretty much "new".
Graph 25: I got a little smoother response from 800Hz-4000Hz with the new midrange. This graph also shows the difference between using an amp capable of a 4 ohm load (my HK PA5800) vs. an amp that limits the power (my Sony DB930 receiver), the difference is basically 3 dB across the board. I've included my raw numbers (yellow) for the HK response, it's got a decent response from 80Hz to 10KHz before my RS SPL meter dies on me.
I've thought about putting in a Zobel across the woofers (would need a 5ohm/18uF Zobel) but I don't think it would clean up the woofer impedance too much at the 400Hz range. If I have the parts, I may try it any how (I think I have to use a 20uF cap though - all I have in that range).
Today I listened to a lot of mono-music (mainly the left channel of many CDs in the large speaker mode with no sub hooked up) just to listen for the vocals and whether or not that sounded right to me. I noticed I have a slight dip at the woofer crossover point near 500Hz, but I've found that I can EQ it by giving a 1 dB or 2 dB bump at 500Hz on my receiver. It gives back the manliness to male vocals, and they now don't sound thin at all. Female vocal were tamed a bit. When I listen to the CC speaker with normal TV programming in Dolby Pro-Logic and DVDs in DD/dts, the vocals sound right to me now. Sometimes I will have to fiddle with the Pad of the CC speaker on the receiver because I need to fix my crossovers in my main speakers (pull up the midrange, pad down the tweeters). Overall, it's getting to a point where I'm almost ready to try the other Vifa tweeter and then veneer the box.
Day 13 (4/21/01)
I start the revamp of the new crossover network with Linkwitz-Riley crossovers (all 2nd order filters).
Graph 26: Although you can't tell from this graph (which includes a Zobel on the midrange driver in an attempt to normalize the impedance at its high end response level of 3500Hz), there was some really nasty midrange cone breakup (wailing sounds) because I crossed the midrange over too high (around 3250Hz). This meant I had to re-do the midrange section. This just goes to show you that you get something to "measure well" and sound like poop. The tweeter has a L-pad with 6dB of attenuation, but it's still too "bright". But with the midrange wailing, I can't discern how much more padding I need on the tweeter just yet.
Graph 27: I remove the Zobel from the midrange, and decide to lower its lowpass crossover point to around 2000Hz to take advantage of its rising response until it gets to the tweeter crossover frequency around 3250Hz. I keep a 2nd order L-R filter for that midrange lowpass, but this cuts out a little too much of the speaker's vocal quality, so it's something that needs to be addressed in the next crossover. I also push the woofer's lowpass crossover point to 1000Hz in an effort to get some response around 500Hz, but it doesn't help as much due to the L-R filter 6dB/octave slope. This leads me to introduce some Butterworth filters in my next set of changes in the crossover network. Also, the tweeter output is still too high, around 2-3dB too high.
Graph 28: I use a Butterworth on the woofer and cross it over at 700hz. Then I use a 1st order Butterworth filter for the lowpass on the midrange at 2000hz to allow the midrange to be "active" until I cross the tweeter over at 3250Hz, where the midrange's output starts to fall off. I then replaced the tweeter L-pad with one that put the brakes on its output with a -8.5dB L-pad. While the new overall response looks peaky in spots, it sounds more natural, especially voices, which is a major consideration for a center channel speaker. I will use this crossover network for the time being unless I think I can improve it some more.
I listened to a few DVDs with 5.1 audio, and I was pleasantly surprised by the consistent soundstage I was getting with the new crossover in place. It was just a more coherent sound field that I was hearing now. I then listened to some Dolby Pro-Logic sources (taped shows from this past week), and the vocal quality has improved as well. I'm finally getting the front 3 speakers plus the subwoofer integrated better than before. I can finally call this an improvement over my original setup now.
So, after I veneer the enclosure, I will be done with this project. I have already picked up some red oak veneer, so it's just a matter of time to get it completed soon.
Day 14: August 5, 2001
Now that I've gotten my SunOnes and SunTwos squared away (in my mind, at least) this past Friday and Saturday (until I get another opinion on their sound later), I figured I'd take another stab at my Center channel speaker beast I created about 4-6 months ago.
I asked one of my friends what he thought of the sound quality of it about 3 months ago, and he thought it sounded a little funny, but once you got used to it, it started to sound okay. Well, that just wasn't music to my ears, so I ordered some capacitors a few weeks ago so that I could be flexible in my CC tweaking/tune-up.
I had nothing to do Sunday, so I decided to give it a shot using the lessons I've learned since, plus the perspective I've gained by tweaking the other 2 sets of speakers. I was sort of amazed at the speed of my tweaking this time around. Then again, I didn't "document" every tweak/change in an orderly fashion, but I knew where I was, and where I wanted to be. I did make notes of each change, but didn't produce graphs upon graphs with each change this time around.
I was bummed that my plans to replace the tweeter with the Morel MDT30s that I used in my other speakers were thwarted by the existing tweeter hole being too big! Doh! So I decided to press on with the original tweeter (from my old DIY speaker project). (after tweaking, I can't really find any reason to feel too bummed about not being able to use the Morel tweeter - I'm using an old Phillips AD11600/T8 fabric tweeter).
I used a more systematic approach this time (the previous efforts had me changing something and then taking measurements of the whole frequency response and then deciding if the change was beneficial - I had no idea if the individual filter slope were actually working as designed). I broke the tuning down into 3 phases: measuring the current individual drivers with their current filters in place, graphing their response, and then making changes to get the filter slopes to cross at the desired frequencies.
So once I was happy with the woofer low pass filter, I moved on to the midrange high pass filter, got the integration of it with the woofer filter where I wanted, and then moved onto the integration of the midrange low pass filter and the tweeter high pass filter. Sounds simple, eh?
I can't believe I didn't do this the 1st time around, and I was tweaking pretty darn blindly.
I can't show the "before" situation yet (left it at home - will post it later tonight), but I can show the "after" frequency response down below. I started with the woofer response. I knew I wanted to slope off the parallelled woofers' response around 500-700Hz (to try and get that baffle step loss minimized), and then use my midrange from around 500-3000Hz, and then let the tweeter take care of the rest.
Let just be nice and say that my initial attempt at the crossover was pretty marginal.
(Editor's note: I was able to graph my initial attempt, and it's shown below to show where I was before I started tweaking this weekend:
Once I plotted the individual response of the driver/filters, I knew why my friend said the CC sounded weird: the tweeter was padded too much, and the midrange was carrying too much of the load on the high end, and IIRC, the woofers were playing too high up the road (around 1200Hz).
After about 5 hours of tinkering with the filters, seeing if the textbook filters were doing their things (or if I noticed that the filters were "delayed" in when the slope started to kick in, I would calculate a new filter with the break frequency that was lower by 1/3 or 2/3 of an octave to get the filter to cross over as desired), I got something that started to sound "normal" to me, with slightly more tweeter output, and was able to shape the midrange output to covered the region I wanted it to cover.
I think it took 2 changes to the woofer filter, 5 changes for the midrange (for the high pass and low pass filtering), and another 3 for the tweeter filter.
What also helped was when I got my other speakers measured on the LAud setup, I was able to produce a table of RS SPL meter measurement correction values for my particular meter that gets me in the ballpark. I put these corrections values (what I needed to add or subtract from any given frequency's SPL reading - since I use 1/3 octave test tones, it's a simple matter to standardize on the corrections) in an Excel spreadsheet, and it gave me a "better" idea of the frequency response, especially in the high frequencies after 3KHz. Believe me, it's a lot closer than the correction values I was using before (which led me to pad my tweeters too much - usually 3dB too much).
Just keep in mind that for the low end (60-150Hz) I have this weird room mode that always accentuates the bass response, it's not as bad as it sounds/looks once the speaker is put in its place above the TV. The extended ridge in the 1KHz range is actually beneficial once the speaker is in place as well, so that's why I did not take steps to pad down the midrange. It's not all that "flat-line" pretty, but without serious measuring gear and filter design software, it gets most of the job done. So without further ado...
August 12, 2001
The previous crossover sounded okay, but there was too much growl in male vocals, still a little boxy, and some yucky glare that I'm sensitive from the midrange in the 3KH-5KHz region (more apparent when music is played through it).
So just for more kicks, I revamped the filters once more (all 2nd order LR), the tweeter filter (originally crossed over around 3.2KHz) is now crossed even lower using a LR - 1.9KHz, against good judgment given the old tweeter I'm using (with a high resonance frequency around 1.6KHz). I factored in the cross-over for some of the benefits in crossing it over so early. The current L-pad on the tweeter is -5.5dB.
Then I changed the filters on the midrange. The highpass is a LR crossed over around 575Hz. I used the LR for a more damped response in the crossover region with the woofers (crossed over around 500Hz) to smooth out the minor summation hump in the 500Hz region (originally it was a Butterworth filter, so it was a little strong at the XO region). The lowpass is rolled off around 1.9KHz as well using a LR. This is done to take out as much midrange glare as possible. I also popped in a -2dB L-pad to equal out the output level. This appeared to work well enough for me and my ears. Originally it was crossed higher, around 3KHz, I think, and was probably a Butterworth as well.
The woofer filter was left alone as it since it appeared to be working as planned.
I was a little lazy because I wasn't really sure if I would like what I heard with this crossover melting pot (using textbook filters, but adjusted based on observed behavior - i.e. to get the roll-off I wanted on the woofer, when I wanted to roll it off around 500Hz, I had to ask for a filter that broke around 250Hz). Strange but true.
Now, the tweeter is carrying more of the upper frequency load. The midrange is carrying less, but it an important portion that gives depth to mainly male vocals without over-powering the rest of the audio content coming through the center channel.
I had entertained bypassing the midrange all-together, and just running a MTM configuration, but due to a dip in the woofer response around 1.5KHz, I decided to give it one more try at the current W(T/M)W configuration.
I re-listened to DPL material with strong center channel content (like TV programming on tape, and DVDs) that showed me the growling male vocal problems from the last crossover design, and now it's gone, plus the voices sound more natural now.
So while I was sitting on my couch I decided to at least take a set of measurements with the center channel installed in its normal position, right on top of the TV, about 3 meters away.
Here's just the overall frequency response of the current crossover incarnation:
Now things look pretty good, and actually sound good too. I plan to leave things the way they are for now unless I hear a need to revisit the crossover. So for now, I'm done!
August 26, 2001
I finally found some free time/energy to take individual driver frequency response readings, and have graphed them below. I took these readings at 1 meter, but they are pretty close to the 3 meter reading and response from the last graph:
My home theater web page was a featured
home theater for October 1999!
Thanks for stopping by! Now go watch some DVDs and listen to some music!