So I've been planning a large Synergy horn build for almost 2 years now. My plan is to go elliptical (yes I know a SH should be square, I know the reasons why, but I'm going to try it anyways). The SH will be about 3' wide with a 60x50 pattern. I got the idea for this construction method from one of Tom Danley's posts on Audio Asylum. He basically said to cut out your mouth shape, lay a piece of stretchy material over it, press a rod over the material to stretch it to the horn shape, then hit it with epoxy or fiberglass resin. This is for making a conical horn.
I figured I better see if this will work with something smaller, so I thought I'd try an 18" wg to use with my BMS 4552nd cds. I built the fixture with a foam swimming noodle, particle board, and duct tape. The foam noodle is about 3.375" dia. I cut it to the required length, epoxied it together, then taped it to the elliptical shape I cut out of particle board. I drew the elliptical shape and determined the position of the foam in cad. I wanted to use an OS throat to match my cds, or at least a close approximation of one, so I drew that in cad as well. I turned my bench top drill press into a wood lathe and turned the throat plug out of wood. This wg is going to be 90x80, so I turned the plug with the 80° angle. My hope here was that the act of stretching the material would take care of the elliptical to round transition. In reality it didn't and I have a pinched section of my wg in the vertical plane.
Here's the jig:
Here's the wg in progress being 'glassed:
Reinforced the back with fiberglass mat and added the cd mount:
I also tried another version, forgoing my throat plug and using a 1.125" rod. My goal here is to avoid the pinching and just build the throat up with filler. I Smoothed over the first version with filler so the diffraction area is smooth.
I figured I better see if this will work with something smaller, so I thought I'd try an 18" wg to use with my BMS 4552nd cds. I built the fixture with a foam swimming noodle, particle board, and duct tape. The foam noodle is about 3.375" dia. I cut it to the required length, epoxied it together, then taped it to the elliptical shape I cut out of particle board. I drew the elliptical shape and determined the position of the foam in cad. I wanted to use an OS throat to match my cds, or at least a close approximation of one, so I drew that in cad as well. I turned my bench top drill press into a wood lathe and turned the throat plug out of wood. This wg is going to be 90x80, so I turned the plug with the 80° angle. My hope here was that the act of stretching the material would take care of the elliptical to round transition. In reality it didn't and I have a pinched section of my wg in the vertical plane.
Here's the jig:
Here's the wg in progress being 'glassed:
Reinforced the back with fiberglass mat and added the cd mount:
I also tried another version, forgoing my throat plug and using a 1.125" rod. My goal here is to avoid the pinching and just build the throat up with filler. I Smoothed over the first version with filler so the diffraction area is smooth.
And now some data. I measured the first good one I got which was the one with the smoothed over pinch points in the throat. This one needs more work to shape and smooth the throat but the results are encouraging. Measurements are with the mic about 4' off the ground, 4' from the wg and gated a little over 3ms. 0-20° in 5° increments and 30-90° in 10° increments. No filtering other than a 50uF protection cap.
Horizontal:
Vertical:
Impulse:
Some widening at 3-5khz but I've got some work to do yet in the throat, specially 3-4" from the cd entrance which I think corresponds to the widening pattern. Overall pretty consistent off axis. Thoughts/criticism most welcome!
Horizontal:
Vertical:
Impulse:
Some widening at 3-5khz but I've got some work to do yet in the throat, specially 3-4" from the cd entrance which I think corresponds to the widening pattern. Overall pretty consistent off axis. Thoughts/criticism most welcome!
Awesome !
I wonder, what material do you use for the stretchy bit? And from the pic#2, it seems not straight-sided, but a slight curve. I suppose the resilience of the film might affect the actual contour.
Nevertheless, your result looks excellent, in both real stuff and the measurements. Well done!
I wonder, what material do you use for the stretchy bit? And from the pic#2, it seems not straight-sided, but a slight curve. I suppose the resilience of the film might affect the actual contour.
Nevertheless, your result looks excellent, in both real stuff and the measurements. Well done!
CLS - the material I think is some kind of polyester used for swimwear. I just went into the fabric store and said "I need something stretchy, and not super expensive." They gave me a weird look when I told them what I was doing......
Also, good eye! The walls are in fact slightly curved. Not as much so as a traditional horn profile and obviously mine doesn't beam as such. I think you can see evidence of the curvature in the higher spl at the low end compared to say a SEOS and also the narrower directivity at hf. Slightly better loading but maybe not as constant directivity at hf.
norman - thanks!
My plan is to use these (once I'm settled on it's performance) over a TD15M in a 2.5 way setup. There will be a second TD sitting on the floor to the inside of the main speaker which I'll roll in around 2-300hz. I'll probably look into a cardioid enclosure for the mid TD as well once I get these waveguides sorted.
Also, good eye! The walls are in fact slightly curved. Not as much so as a traditional horn profile and obviously mine doesn't beam as such. I think you can see evidence of the curvature in the higher spl at the low end compared to say a SEOS and also the narrower directivity at hf. Slightly better loading but maybe not as constant directivity at hf.
norman - thanks!
My plan is to use these (once I'm settled on it's performance) over a TD15M in a 2.5 way setup. There will be a second TD sitting on the floor to the inside of the main speaker which I'll roll in around 2-300hz. I'll probably look into a cardioid enclosure for the mid TD as well once I get these waveguides sorted.
I did some more work to the throat, and also figured out how to use the directivity plots in ARTA. The sonogram is raw data because I can't figure out how to make it work with a gated impulse. I think it needs the peak to be at 0ms but it's at more like 600ms.
The widening directivity in the 3-4khz region has improved some. There's still work to do but this is coming along pretty well. The pattern narrows a bit more than I would like over 5khz. I think some of the hf "wobblies" over 5khz will be improved with further throat smoothing.....it's still not where I want it.
Here's 0-90° no eq:
Here's with eq/xo. 6dB low pass at 6.5khz and a few notch filters:
And the sonogram with eq/xo. 1/6 octave filtering because with 1/12 octave the raw data makes it difficult to read:
I went ahead and made another one, and I'll continue to refine this one until I'm happy/tired of sanding.
The widening directivity in the 3-4khz region has improved some. There's still work to do but this is coming along pretty well. The pattern narrows a bit more than I would like over 5khz. I think some of the hf "wobblies" over 5khz will be improved with further throat smoothing.....it's still not where I want it.
Here's 0-90° no eq:
Here's with eq/xo. 6dB low pass at 6.5khz and a few notch filters:
And the sonogram with eq/xo. 1/6 octave filtering because with 1/12 octave the raw data makes it difficult to read:
I went ahead and made another one, and I'll continue to refine this one until I'm happy/tired of sanding.
Thanks Earl...like I said in the op I got the idea from Danley.
The weirdness at 3.5khz I'm sure is because the sonogram is raw, ungated data. I couldn't make the plot with my gated measurements. What you see in the freq measurements is gated at 3ms.
That would be cool if you could plot it in your sw. Let me know what you need and I can get it done after this round of sanding and filling.
I also wouldn't mind some criticism/comments of my work if there's enough data for you to go on.
The weirdness at 3.5khz I'm sure is because the sonogram is raw, ungated data. I couldn't make the plot with my gated measurements. What you see in the freq measurements is gated at 3ms.
That would be cool if you could plot it in your sw. Let me know what you need and I can get it done after this round of sanding and filling.
I also wouldn't mind some criticism/comments of my work if there's enough data for you to go on.
I'm still hard at it..........sanding and filling, sanding and filling, sanding and filling......but improvements are being made. In this set of measurements the entire guide has been smoothed and I paid particular attention to the throat. Nice and smooth. The only thing bugging me right now is the exit angle on my wg is about 2-3° or so steeper than on my cd and I'm at a point where I can't sand anymore without sanding into the fiberglass and making the entrance bigger. I might get a belt sander and take about .200" off my mounting plate/throat. This will allow me to build the throat into the correct angle with filler, but I'm not sure if it will be worth the effort.........it's fairly close as is. I measured both left and right sides of the wg and they match up quite well so my symmetry is better than I thought it would be.
I'm liking the improvements though. Over 5khz flattened out quite nicely, and the peaks that you see are mostly driver issues seen in the manufacturer's plots. I'm still not happy about the "hot spot" at 3-4khz but I'm not sure how to approach that at this time. I was thinking it might be the on axis dip due to mouth diffraction that's spread out due to the elliptical shape, but it seems to be happening too low in frequency. Any ideas?
I'm liking the improvements though. Over 5khz flattened out quite nicely, and the peaks that you see are mostly driver issues seen in the manufacturer's plots. I'm still not happy about the "hot spot" at 3-4khz but I'm not sure how to approach that at this time. I was thinking it might be the on axis dip due to mouth diffraction that's spread out due to the elliptical shape, but it seems to be happening too low in frequency. Any ideas?
I simply do not see a "hot-spot at 3-4 kHz. Are you plotting this with a normalized axial response? That screws things up so that you can't tell what is going on. This looks more like an axial dip - very minor, which could be mouth diffraction in an 18 " wide waveguide.
PS> If you want to take data for me then use 0, 5, 10, 15, 20, 30, 40, 50, 60, 80, 100, 120, 150, 180. The data beyond 100 is not critical but it does add some accuracy. I need raw impulse responses. I will window the data myself and there can (should) be a propagation delay, but no more than 4800 points in total. They must all be on the same time base so that the phase is accurate.
PS> If you want to take data for me then use 0, 5, 10, 15, 20, 30, 40, 50, 60, 80, 100, 120, 150, 180. The data beyond 100 is not critical but it does add some accuracy. I need raw impulse responses. I will window the data myself and there can (should) be a propagation delay, but no more than 4800 points in total. They must all be on the same time base so that the phase is accurate.
Last edited:
As a car audio guy, a lot of guys use fleece since it is a little thicker. If you are going to be reinforcing from the back side, you can find Lycra that is usually pretty smooth on the surface.
I've used everything from old tee shirts, unbacked trunk carpet, grill cloth...the thin stuff only when you are worried about warping from shrinkage.
I've used everything from old tee shirts, unbacked trunk carpet, grill cloth...the thin stuff only when you are worried about warping from shrinkage.
Earl - I'm thinking you're right........ data is not normalized. Would you mind if I sent you the data for the wg, then later for the completed speaker?
Winslow - the first thing they pointed me to at the fabric store was Lycra, but it was almost $7-8/yard more than the polyester. I have a friend who builds subs (horribly tuned and badly braced) for cars with fiberglass and he mentioned fleece as well. I'll have to check it out.
Winslow - the first thing they pointed me to at the fabric store was Lycra, but it was almost $7-8/yard more than the polyester. I have a friend who builds subs (horribly tuned and badly braced) for cars with fiberglass and he mentioned fleece as well. I'll have to check it out.
That hot spot with very wide directivity in the middle of its range seems to appear on many other horns and waveguides, for example :
Eminence H290 (probably an axial accident)
JBL H9800 (this one is very pronounced)
Last edited:
Thanks for the graphs pos. I don't think I'm going to worry about that mouth diffraction at this point. I'm still busy.......I took a Dremel to the throat of my waveguides to open them up so I can form it to match my cd exit angle better. Hopefully I'll have some more measurements from that tomorrow. I'll try to get an impedance jig together and see how that looks as well.
In the meantime, no new data, just a different look at some old data. I figured out how to use gated measurements for my directivity sonongram in ARTA. It was super simple actually and I can't believe I didn't see it. Anyways, here's the data from Sunday (I think it was....). 3ms gate, not normalized, and 1/12 smoothing:
In the meantime, no new data, just a different look at some old data. I figured out how to use gated measurements for my directivity sonongram in ARTA. It was super simple actually and I can't believe I didn't see it. Anyways, here's the data from Sunday (I think it was....). 3ms gate, not normalized, and 1/12 smoothing:
Well, it lives! Both speakers are setup and I'm listening now. I'm pretty happy, but there are some issues that are gonna bug me enough to fix them. I've got some ideas on how to make these better, so I'm gonna make another "really" good one. I'll make a mold from the good one and try to lay up some new ones in 'glass and make them really nice. The reason I want to do that is.........each wg I made is different. I knew it would be tough getting them both the same doing all this work by hand and each one is one-off. You can see the differences in the polars. That and these look pretty hagard 
. Totally prototypes with primer on the business end, my crappy 'glass reinforcement on the back, and I cut the cd mounting plate out of 3/4" particle board. I'll machine the cd mount out of aluminum for the new ones. I'm a machinist and a friend just got a lathe for his garage.
I'll start with wg1, the one I've been posting data on. Here's out to 90° with a 6th order LR at 950hz. This lower cross has a better directivity match with the 15". Distortion and listening tests tell me these cds have no problem going that low. I spent some time optimizing the xo too. With a .1ms delay on the woofer I get nice phase tracking for nearly an octave on both sides of the xo. Anyways here's the graph:
And the sonogram from the same data set, 3ms gate:
That 2.5khz dip is mitigated pretty well. Looks to me like 80-90° beamwidth out to 5khz then narrowing to about 60°. I think these effects can be attributed to the slightly curved walls of the wg making it not quite OS but something else. Prolate spheroid?
Here's why I should have made a mold to make them exactly the same, wg2:
As you can see a more pronounced on axis dip, and wider dispersion north of 5khz with a real gradual narrowing. This to me looks closer to a true os profile. I went ahead and got both speakers going and integrated into the system despite the differences and I'm loving life.
The main TD15 mid woofer has a 2nd order hpf at 60hz and is about 32" from the front wall. A second TD sits on the floor to the inside of the main speaker and is pushed up against the wall. All 4 are 3 cu ft sealed cabs and well lined with Bonded Logic. The flanking woofer has a hpf at 40hz and has a 2nd order roll-off at 300 hz......this gives me pretty good midbass smoothing but still needs work. These are -6dB at about 65hz but room modes give me some output below that so I take it. A couple 4 cu ft sealed Dayton Ref 15's round out the system.
. Totally prototypes with primer on the business end, my crappy 'glass reinforcement on the back, and I cut the cd mounting plate out of 3/4" particle board. I'll machine the cd mount out of aluminum for the new ones. I'm a machinist and a friend just got a lathe for his garage.I'll start with wg1, the one I've been posting data on. Here's out to 90° with a 6th order LR at 950hz. This lower cross has a better directivity match with the 15". Distortion and listening tests tell me these cds have no problem going that low. I spent some time optimizing the xo too. With a .1ms delay on the woofer I get nice phase tracking for nearly an octave on both sides of the xo. Anyways here's the graph:
And the sonogram from the same data set, 3ms gate:
That 2.5khz dip is mitigated pretty well. Looks to me like 80-90° beamwidth out to 5khz then narrowing to about 60°. I think these effects can be attributed to the slightly curved walls of the wg making it not quite OS but something else. Prolate spheroid?
Here's why I should have made a mold to make them exactly the same, wg2:
As you can see a more pronounced on axis dip, and wider dispersion north of 5khz with a real gradual narrowing. This to me looks closer to a true os profile. I went ahead and got both speakers going and integrated into the system despite the differences and I'm loving life
. The main TD15 mid woofer has a 2nd order hpf at 60hz and is about 32" from the front wall. A second TD sits on the floor to the inside of the main speaker and is pushed up against the wall. All 4 are 3 cu ft sealed cabs and well lined with Bonded Logic. The flanking woofer has a hpf at 40hz and has a 2nd order roll-off at 300 hz......this gives me pretty good midbass smoothing but still needs work. These are -6dB at about 65hz but room modes give me some output below that so I take it. A couple 4 cu ft sealed Dayton Ref 15's round out the system.
Here's a pic of the "finished" system. Notice the foam I cut for the throat? You can see how goofy the shape of the wg on the left is from the foam. That's the one that has less of an on axis dip.
I had a friend over yesterday afternoon to watch a movie. She said the waveguides look like toilet seats. Guess I won't be painting them white..........
I had a friend over yesterday afternoon to watch a movie. She said the waveguides look like toilet seats
. Guess I won't be painting them white..........- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.