Midbass Conical Horn - Keele vs Hornsresp which is the better FR?

[steve71]

Background
---------------

My goal is to build a midbass conical horn to cover 80-500hz (80-800hz would be better). Below 80hz I'll be using a tapped horn to 16hz and above I'll be using an Altec 288 CD in a MR-64 horn. The 288/MR-64 horn can go as low as 500hz, but looses it's dispersion control under 800hz. However if the MR-64 horn doesn't work out, I'm willing to consider other horns for the 288 driver in the future. The speaker will be crossed actively with a 24/db/decade slope using the linkwitz riley alignment. In the long term I plan to use a digital X-over. I'm also considering a dedicated HF horn above 7Khz. Anyway, on to the matter at hand.


Keele Calulations:
-----------------------

I've spent the day reading over Keele's 1977 AES paper and doing the calculations for the Altec/GPA 515G driver.

http://www.xlrtechs.com/dbkeele.com/PDF/Keele (1976-05 AES Preprint) - Efficiency, Horns vs DR.pdf

Constants for Altec/GPA 515G 8ohm driver

Cms = 3.42 E-04 m/newton
Re = 6.2 ohms

Sd = 0.0848m^2
Le = 0.0058H

fs = 37hz
fc = 80hz

Qts = 0.269
Vas = 0.351m^3
Vfc = 0.002m^3


My calculations yield:

St = 640cm^2 (Throat area for optimal efficiency)
Vb = 49.9 liters (back chamber volume for proper reactance annulling)
Fhm = 275hz (indicates 6db/oct roll off point)
Fhvc = 340hz (indicates 12db/oct roll off point)
Fhc = 3494hz (indicates 18db/oct roll off point)


Conclusions & Questions
-------------------------------
I modeled Keeles values (rear chamber volume and throat area) in Hornsresp along with my own numbers (arrived at by trial and error). The two frequency responses are shown in the attached pic below, along with the HR parameters.

Seems to me that the black response curve gives the better result , no?

The Keele FR rolls off to early due to the very low compression ratio (read large throat area). However if I increase the compression ratio to 2.11:1, I gain a few 100 Hz on top. The lower 3db point suffers, but if I decrease the back chamber volume, I can boost the FR around 80hz to compensate. Seems like a win, win or am I missing something critical?

I'm also a little concerned about going bellow a 2:1 compression ratio. Is it still a horn with such a low compression ratio

Sorry if I'm asking dumb questions, I have zero practical experience with this.

 

[steve71]

Keele's Data

 

[steve71]

My HR data based on trial and error.

 

[norman bates]

I'll take a stab.

[ 36 hz/.269 (qts) ] x 2 = mass corner.

so you'd get horn loading (and it's boost) till 275hz.
Then it will lose spl as freq goes up till it acts as a direct radiator.
I can't tell you what freq it will act as a direct radiator.

There is a P audio 12" with a qts under .2 around $150 that'd work well.

800hz is going to be a stretch without a phase plug.

http://db.audioasylum.com/cgi/m.mpl?forum=hug&n=108371&highlight=12+horn+ev+djk&r=&session=

http://www.audioasylum.com/cgi/m.pl?forum=hug&n=135398&highlight=12+horn+ev+djk&r=&session=

http://db.audioasylum.com/cgi/m.mpl?forum=hug&n=81826&highlight=12+horn+ev+djk&r=&session=

http://db.audioasylum.com/cgi/m.mpl?forum=hug&n=72979&highlight=12+horn+ev+djk

and bass horn loading is tricky, bass goes through walls.
if .5 is free space, then 1 pi (on that program) would be floor loaded.
So floor loaded better be on concrete, not wood.
and side wall better be concrete (not drywall) also or it cannot be used.

2:1 compression may help it go a bit higher than classic formulas.

Horns tend to unload below mouth / flare, so I wouldn't add too much boost (but I am no expert).

norman

 

[SpeakerBob]

The 515 driver is meant to work well in the Edgar midbass horn, covering the 80Hz to 500Hz that you are planning. Have a search over at audioasylum for details.

Also this guys done a conical horn covering the same range:

http://www.volvotreter.de/pics_plans.htm

Rob.

 

[SpeakerBob]

Just noticed that you've already posted on AA, so my post above was useless - sorry!

Rob.

 

[steve71]

I'll take a stab.

[ 36 hz/.269 (qts) ] x 2 = mass corner.

so you'd get horn loading (and it's boost) till 275hz.
Then it will lose spl as freq goes up till it acts as a direct radiator.
I can't tell you what freq it will act as a direct radiator.

There is a P audio 12" with a qts under .2 around $150 that'd work well.

800hz is going to be a stretch without a phase plug.

http://db.audioasylum.com/cgi/m.mpl?forum=hug&n=108371&highlight=12+horn+ev+djk&r=&session=

http://www.audioasylum.com/cgi/m.pl?forum=hug&n=135398&highlight=12+horn+ev+djk&r=&session=

http://db.audioasylum.com/cgi/m.mpl?forum=hug&n=81826&highlight=12+horn+ev+djk&r=&session=

http://db.audioasylum.com/cgi/m.mpl?forum=hug&n=72979&highlight=12+horn+ev+djk

and bass horn loading is tricky, bass goes through walls.
if .5 is free space, then 1 pi (on that program) would be floor loaded.
So floor loaded better be on concrete, not wood.
and side wall better be concrete (not drywall) also or it cannot be used.

2:1 compression may help it go a bit higher than classic formulas.

Horns tend to unload below mouth / flare, so I wouldn't add too much boost (but I am no expert).

norman

Thanks for the response norman! The FR graphs I posted are for 2pi (ground plane), so according to HR I should be good to 80hz. My floors a wooden so maybe this is a bit optimistic

Thanks also for the links. I had come across those horns during my researching, but I don't think they'll go low enough for my purposes (80hz).

The 515 driver is meant to work well in the Edgar midbass horn, covering the 80Hz to 500Hz that you are planning. Have a search over at audioasylum for details.

Also this guys done a conical horn covering the same range:

http://www.volvotreter.de/pics_plans.htm

Rob.

Thanks for the response Rob. The 80Hz Edgar horn and Erik's 77hz conical horn have provided much inspiration for this project.

 

[GM]

Greets!

First things first, for those coming a little late to this 'party', here's our initial correspondence beginning here in another thread..........

Next, lacking any actual measured specs of your drivers, I'm basing my remarks on GPA 515-8G published specs (which are actually decades old Altec published specs), well proven Altec horn designs for the Altec 416, 515 series drivers and basic horn design theory.

There are basically two ways to cover a wide BW with a FLH, either combine the horn + appropriate driver's rising on axis response as Altec did or compression load it all the way and since it needs some form of phase plug to get both good HF extension and dispersion with the latter due to the much smaller throat required (not to mention the additional design problems the 515's vented DC presents WRT the high compression ratios required), the former is by far easiest to design/build along with being the most efficient, least distorted of the two.

So again I recommend using Altec's well proven ~129"^2 throat as a basis for your design as so many others have successfully done over the decades and it will allow you to use up to a ~1200 Hz XO if you so choose.

Of course you can choose an alignment between these two as you seem to think you should, but it won't be optimum, so for best performance you'll wind up using some form of EQ to smooth it out through the XO BW and you'll still have the added throat distortion, increasing with increasing CR, so Caveat Emptor.

While we're on the subject, the best performing conicals I did were of the two flare waveguide type like Tom Danley uses for his Unity (though I used a bit different design routine than he's posted), so while it will be somewhat larger, it's worth it IMO if you don't need the HF narrowing up on axis to blend to the HF horn a straight flare yields.

GM

 

[steve71]

Greets!
.... the former is by far easiest to design/build along with being the most efficient, least distorted of the two.

So again I recommend using Altec's well proven ~129"^2 throat as a basis for your design as so many others have successfully done over the decades and it will allow you to use up to a ~1200 Hz XO if you so choose.

Of course you can choose an alignment between these two as you seem to think you should, but it won't be optimum, so for best performance you'll wind up using some form of EQ to smooth it out through the XO BW and you'll still have the added throat distortion, increasing with increasing CR, so Caveat Emptor.

While we're on the subject, the best performing conicals I did were of the two flare waveguide type like Tom Danley uses for his Unity (though I used a bit different design routine than he's posted), so while it will be somewhat larger, it's worth it IMO if you don't need the HF narrowing up on axis to blend to the HF horn a straight flare yields.

GM

Thanks once again for the very informative post GM. I didn't realize the small throat would cause those problems. I guess I'm too focused on the FR and I'm ignoring other equally important issues. A little bit of knowledge is a dangerous thing as they say

I tried running some HR simulations with a 832cm^2 throat, but the FR still keeps dropping about 300hz. Obviously I'm doing something wrong. The FR plots that are supplied with the Altec 515G Pdf show a very wide BW in a 816A enclosure (just as you suggest) with none of the downward sloping response after 300-400hz.

 

[GM]

You're welcome!

The only thing you keep missing is that the horn's rolled off HF response sums with the driver's response to fill in the gap between the horn's roll off and the driver's break-up modes BW to create a nominally flat on-axis BW of a much higher efficiency lasting well into the HF, though of course the off axis rolls off such that a much lower XO point is required than it implies for a decent power response.

Since you have the capability, sim the 816's response except with a max flat sealed rear chamber as a reference to compare your design's HF roll off to.

GM

 

[steve71]

You're welcome!

The only thing you keep missing is that the horn's rolled off HF response sums with the driver's response to fill in the gap between the horn's roll off and the driver's break-up modes BW to create a nominally flat on-axis BW of a much higher efficiency lasting well into the HF, though of course the off axis rolls off such that a much lower XO point is required than it implies for a decent power response.

Since you have the capability, sim the 816's response except with a max flat sealed rear chamber as a reference to compare your design's HF roll off to.

GM

OK, that makes things a lot clearer. I didn't realise that HR's FR graph was only the horns contribution. I think you've saved me from making a few critical mistakes, so hats off to you for helping me out GM!.

Also good idea on modeling the 816's response for comparison. I'll dig up the plans for the horn and model it in HR and post up the results tomorrow night.

I just got of the phone with Bill at GPA and he's making up my 515's as I type this. Hopeful if all goes well I'll have something to listen to by the end of the month.

While we're on the subject, the best performing conicals I did were of the two flare waveguide type like Tom Danley uses for his Unity (though I used a bit different design routine than he's posted), so while it will be somewhat larger, it's worth it IMO if you don't need the HF narrowing up on axis to blend to the HF horn a straight flare yields.

That sounds interesting (did you document it on DIYA or the LH forums?) and something I'll have to revisit for version 2. For now I'll be happy if I can build a "simple" conical horn. It's turned out to be a lot more work than expected but I've learned a lot along the way and I'm sure the DIY upgrade bug will bite every few years. The next project after this is going to be a 18-80Hz tapped horn.

 

[GM]

You're welcome!

Hardly, I did the vast majority of my horn building between '64 - '75 and since they were learning exercises for me there was no need to document more than what I learned along the way. Note too that my many different designs were for 500 Hz XOs to keep them a reasonable size/cost. My <500 Hz horns were these Altec ~70 Hz expo VOTT bass horns with the vented cab removed and a sealed rear cab and corner loaded for a ~flat response to 30 Hz in room, which was low enough for vinyl/RTR when I chopped them up in '70. Of course the wings weren't used and the horns rotated with the drivers vertical (not optimum from a 500 Hz XO POV ) to keep the HF horns from being way too high.

Yeah, the 'simple' conic horn is harder to get truly right design wise than most others IMO. A simple megaphone has a point where its sound-field collapses into a prematurely rolling off response that's narrowing on axis to boot if not allowed to 'breathe', so to speak.

Sim is my old corner horns loaded with dual 515-8Gs (I used 515Bs). Not too shabby.........

GM

 

[GM]

Yeah, the 'simple' conic horn is harder to get truly right design wise than most others IMO.

Added overnight on a WG thread showing a megaphone's extremely poor wave propogation/narrowing up on axis if no end correction (some form of mouth flaring/damping) is used and why they should be severely BW limited for HIFI apps IMO. Ditto Dr. Geddes's insistence on a proper throat expansion: http://www.diyaudio.com/forums/showthread.php?postid=1590534#post1590534

GM

 

[steve71]

Thanks for the extra info GM! After finding some plans I tried to model the 515G in a 816 cabinet for comparison, but after playing with it for awhile I couldn't get anything worthwhile out of it, namely it didn't have any bass due to the massive BC volume.

I downloaded winISD and modeled the 515 8G in a sealed enclosure with the same volume as the horn's BC. I though this would give me a good indication of the direct radiator response. The 515 8G in a 37liter enclosure is down 3db at about 105hz and it's pretty flat above that. Not exactly a complementary response to the horns frequency response plot. The only way I could create a complementary direct radiator curve was to increase the BC volume to something huge - around 800liters.

I thought the 515G was supposed to work well in Bruce Edgar's 80Hz conical horn, or does the 515G only work in a horn with huge back chamber?

 

[steve71]

Added overnight on a WG thread showing a megaphone's extremely poor wave propogation/narrowing up on axis if no end correction (some form of mouth flaring/damping) is used and why they should be severely BW limited for HIFI apps IMO. Ditto Dr. Geddes's insistence on a proper throat expansion: http://www.diyaudio.com/forums/showthread.php?postid=1590534#post1590534

GM

I'm lucky enough to have my father-in-law who's a builder/carpenter here to help, but he's going back to Australia very shorty. Even if the horn has issues I'll get my moneys worth with the carpentry lesson . Unfortunately I don't have the time to redesign it with any sort of flaring. I can however have a play around with damping foam at the mouth (the wife is just going to love that!). If I can get 80-500hz out of the horn that'll be good enough for version 1.0 and MDF is cheap .

 

[SpeakerBob]

I'm pretty sure winISD is only useful for checking the bass response of the driver - like hornresp it doesn't take the driver response into account. The best thing to do would be to build a test box and measure. I'm ~pretty~ sure the manufacturers mount the driver into a wall so it is measured in 2pi.


When I was playing with prototypes I knocked up test boxes out of 15mm ply and used 'no nails' to fill any gaps in the joints up. ie just butt join without mitres and fill with goop. Worked for me!

Rob.

 

[GM]

Thanks for the extra info GM! After finding some plans I tried to model the 515G in a 816 cabinet for comparison, but after playing with it for awhile I couldn't get anything worthwhile out of it, namely it didn't have any bass due to the massive BC volume.

I downloaded winISD and modeled the 515 8G in a sealed enclosure with the same volume as the horn's BC. I though this would give me a good indication of the direct radiator response. The 515 8G in a 37liter enclosure is down 3db at about 105hz and it's pretty flat above that. Not exactly a complementary response to the horns frequency response plot. The only way I could create a complementary direct radiator curve was to increase the BC volume to something huge - around 800liters.

I thought the 515G was supposed to work well in Bruce Edgar's 80Hz conical horn, or does the 515G only work in a horn with huge back chamber?

You're welcome!

Yeah, so? As I said, you're only interested in the horn's response as a guide, not the cab's bass response. Nor are you interested in the driver's response since you're buying essentially Altec NOS replacement drivers that are optimized for Altec horns and why you should use their throat geometry as a basis for any horn + driver response width BW app IMO.

Regardless, there's no simming program I'm aware of that can accurately sim any of the Altec horn driver's responses because none follow flat piston theory throughout their usable BW, being instead tailored to suit their respective horn designs. Ditto their various vented alignment rear chamber responses in combination with its horn response.

Yes, the dims I have for his 80 Hz horn does indeed look good for use with 515s since he used Altec's ~1:1 CR.

GM

 

[SpeakerBob]

The 9" square throat in the Edgar 80Hz suggests a ~1.6:1 compression ratio, unless I've got that throat size wrong (pretty certain I haven't)

Rob.

 

[GM]

......and used 'no nails' to fill any gaps in the joints up. ie just butt join without mitres and fill with goop. Worked for me!

Me too! Liquid Nails came along just as I was getting serious about speaker building and between a lack of patience and time I was quick to take full advantage of its caulking gun application and quick setting capability figuring my time was worth far more than any hardened leftover glue.

GM

 

[SpeakerBob]

Right now the forum won't let me upload any pics but I've just put together a pic showing my JBL 2020H's response from JBL, the hornresp model and the final groundplane measurement.

It should give an idea of how to interpret the graphs. Will try and post tomorrow.

Rob.