Unity horn with 2 x Eminence Alpha 6 drivers

I finally decided to throw my Unity horn hat into the ring! I hope to add a bit to the knowledge base by documenting with measurements what happens with each sequential change to the horn. My design targets are -3db at around 200 to 220 hz for high pass, and -3db at >1000hz for low pass. 115 db+ in 4pi space. As you can see, lots of "oopsies"... nothing bondo can't handle! Since these things will hide behind speaker cloth, I really don't care. I will use a minidsp for crossover/eq. More to come...

Attached is the google sketchup for the horn, as well as the Hornresp model. I had to learn sketchup from scratch, but I figured out what I needed in about an hour. I used the program's measuring tools (including the protractor- key for the angles) to figure out the cuts. Unfortunately, I don't have CNC machinery to cut the panels- only a bandsaw and Festool gear. Therefore, some amount of error got built into the horn.

Here's a link to an inspirational project at diyaudio:

http://www.diyaudio.com/forums/multi...ve-x-over.html

 

alpha6-1.txt 0.39453125k . file

 

rectangle unity with markup.zip 15.9287109375k . file

All measurements taken with a radioshack SPL meter through room eq wizard. Levels not calibrated for volume, only relative performance.

The first pic shows a ground plane measurement of the horn with 1 vs 2 drivers connected. I wanted to see if the FR trend would change when the second driver was added.... the passband of interest didn't- it only got louder.

The second pic shows the response of the raw Eminence Alpha 6 divers, measured while held in my hand, close mic'ed at about 1 inch. At first, I couldn't figure out that dip at 630hz in the first pic- I thought it was related to some diffraction or high order mode cancellation within the horn. Turns out it was just the drivers... whew! Hopefully, I can tame that with little EQ.

The third pic shows how the top end of the FR got pushed out to around 900hz after routing the frustrums into the ports. I used a plunge router with a chamfer bit after drilling 5/8" holes. This is compared to the first ground plane measurement with no frustrums.

The fourth pic shows a bump in the low end of the FR after sealing the back of both drivers with blue masking tape- It acutally worked! However, the volume is too low for what I want, making the boost too high with too much roll off below. I'll need to seal the drivers in PVC or wood boxes. Interestingly, the top end of the FR got pushed out a bit as well- didn't count on that one.

Also, note that plugging the <1 inch hole back of the horn where the CD will mount really didn't affect measurements much- can't explain that one.

I've also orded some BMS 4550 compression drivers, recommended by the man himself, Tom Danley, when discussing these horns on the diyaudio thread linked above. Once I have them, I'll fab up a mount and see what happens.

Hey Bill,

I made the assumption that the port area is fixed at the entry into the horn and that the effective length of the port decreases when the frustrum, or cone, is added. I believe that adding frustums also increases the volume of the compression chamber a bit.

I may rout wider frustrums- I haven't decided yet. Since this isn't a fine woodworking project for me, I can always go back and add bondo, then try something else. I'll be sure to take measurements along the way.

John

edit- Bill, I forgot to directly answer your first question. I first drilled 5/8 holes. I then used a plunge router with a chamfer bit to create the frustrums.

I sealed the backs of both Alpha 6s with thick cardboard and HVAC metal tape. I took groundplane measurements today, ranging from on axis to 40-50 degrees. I also took a 2.83v groundplane measurement at 1 meter... not too bad at 103.5 db.

After playing w/ hornresp, adjusting driver and chamber parameters to match my FR, I don't think I can get much more out of a 2 segment horn with these drivers without significantly changing the overall horn size. Looks like I'm getting a fairly flat response from around 300hz to 1.1khz off axis, which is not too bad. I would have preferred 200-1000hz, but I think I can make up for it with a well designed bass horn that covers around 60-350hz... we'll see.

Unless you folks give me some other ideas, I think I'll press on with this current horn design and mount up the BMS 4550 when it gets here.

John

The W box?

I need some advice. Please look at this graph and tell me if this dip at 630 hz will go away with speaker burn in. As you can see, it's in the raw driver response. How do you guys burn in your drivers?

John

Thanks for the time, guys. I've included another graph, showing the raw response of both Alpha 6 drivers superimposed (blue/red), as well as the horn response (purple). Note that the Alpha 6s are not baffle mounted for these raw response graphs, only hand held and close mic'd at about 1 inch from cone center. Those dips are in the raw response, and I don't see how they can be horn related.

You're right, mtg90, that dip isn't present in the Alpha 6 published graph, shown here:

http://www.eminence.com/pdf/Alpha_6A.pdf (I posted a close up of the FR)

Thanks,

John

Here is a comparison of the raw Alpha 6 response to that of an Elemental designs 160v.2 15" sub. Again, neither driver was baffle mounted, and both were close mic'd at about 1 inch from cone center. Phase response also shown.

My money's on cone breakup, as both drivers show a similar dip: alpha 6 at 625 hz, ED sub at 418 hz. Those peak/dip waveforms resemble one another quite a bit to my eye, although why a midrange speaker (Alpha 6) would have cone breakup smack dab in the midrange doesn't make much sense...

Here's yet another plot. This one shows, from top to bottom:

1) horn response with 2 alpha 6 drivers, ground plane

2) multiple handheld sweeps with sealed back single alpha 6 driver, measured outdoors at about 2.5 feet.

3) original responses of raw unsealed drivers, handheld, close mic'd at about 1 inch.

4) impedance plot of sealed back driver, not horn mounted.

Again, that dip in the mid 600 range is there, in the original raw sweeps, the sealed sweeps, and represented in the impedance plot. At this point, I'll have a VERY hard time believing it is caused by the horn response. I still don't understand why a midrange driver would have that kind of breakup (or whatever it is) in the middle of the mid range, and on 2 units... very odd.

John

John

That impedance plot is from the sealed driver sitting next to my computer, not horn mounted.

That's a good point, Bill, and I had looked at the sealed back alpha 6 graph before you posted. If the blip is related to the homebrew seal job that I did, I can live with that, because I need the low end boost that it provides. I can just EQ the blip in the raw resonse with the minidsp, if I even notice it during listening. Either way, moving on...

I found out that the BMS 4550s will be here this Wednesday. My dad has a shopsmith w/ lathe capability, and hopefully I'll get to use it this weekend to make the CD mounts w/ contoured throats, ala Paul Spencer. Hopefully, more measurements to come soon!

John

Just a heads up: since I'm getting close to a minidsp based crossover, I thought I'd link some good threads related to phase and crossover design.

http://redspade-audio.blogspot.com/2...alignment.html

http://www.diyaudio.com/forums/multi...-tracking.html

My BMS 4550s came in today... YES! Man, these things are heavy! I routed a hole through my horn so that I could hand mount the driver and get some quick measurements. No bondo shaping in the corners of the horn... yet.

I posted two graphs:

1) The first shows ground plane measurements of the driver sitting on its side on the ground (green) vs handheld against the back of the horn (red). Things look really good until about 8 khz, where the FR rolls off pretty quickly, faster than it does for the raw compression driver measurement. Also, I had hoped for lower extension from the compression driver, to make my crossover easier... Oh well. Hopefully, bondo contour shaping will help things.

2) The second graph was lifted from jmorken's unity horn project, documented here:

http://www.diyaudio.com/forums/multi...ve-x-over.html

He made this measurement before he had done final contour shaping of the horn throat. I figured this was the best benchmark for my project at this time. I'll need to fab up a mount for the compression driver, likely out of 1/2 inch wood, just to support its weight. This will probably increase the effective distance from the midrange ports to the mouth, and I expect the midrange to rolloff a bit earlier than it does now.