In Hornresp, need explanation of Atc and general guidance

[letsbangout]

Hi all. I'm trying to build an exponential horn that will look alot like the JBL 4550a (hopefully), but with three drivers. I suppose i'll stack the drivers (JBL 2226J's) one on top of the other. Drivers will run in parallel.

I have a two-fold problem:

1) I fail to understand what Atc means, as I have a problem with nomenclature like "normal to axis" in the help file.

2) I have a really sweet design in terms of the way the curves look.. i'll call its 3db down points 25hz and 1200hz. But the only way I've found to get the kind of response I want (usable above 500hz) is to set the Atc at 1600, Vtc at 14500, and S1 at 4600. Is it possible to build such a thing? I'm envisioning that the throat chamber would be shaped like an Isoceles Trapezoid, i.e. the long side of the trapezoid is the "mouth" (S1) where the transducers sit, being rather wide, and then it would slowly taper inward to produce a vertical slit at the end of the throat chamber.

Does this make any sense at all? Can anyone give me pointers in order to maintain that good high end response up to 1200 (or more) hertz?

Hornresp Parameters for reference:
Ang = 1xPI
Eg = 180 (yes, I intend to some day run that much power)
Rg = 0.02
S1 = 4600
S2 = 7000
Exp = 70
Sd = 692
Cms = 3.15E-03
Mmd = 98
Re = 10
Bl = 27.10
Rms = 0.13
Le = 1.75
Vrc = 700
Ap = 300
Vtc = 14500
Lrc = 15
Lpt = 0
Atc = 1600
I've omitted everything that's greyed out by the program, but the one relevant thing of course is the driver config, being Nd = 3P as I mentioned above.

Any help or guidance is appreciated!

 

[David McBean]

I fail to understand what Atc means, as I have a problem with nomenclature like "normal to axis" in the help file.

Hi letsbangout,

"Normal to" is a mathematical / engineering term meaning "perpendicular to" or "at right angles to". As a general rule, most Hornresp users simply set Atc = Sd. For your proposed 3-driver design, that would make Atc = 3 x Sd.

BTW, some of your JBL 2226J driver parameter values appear to be rather strange . By my reckoning, according to the JBL specification sheet:

Sd = 880
Bl = 27.1
Cms = 1.58E-04
Rms = 5.04
Mmd = 85.19
Le = 3.5
Re = 10

Kind regards,

David

 

[letsbangout]

I guess what I fail to understand is how Atc is any different than S1 in my application. I'm doing an exponential horn, a la JBL 4550a. Can you explain that a bit?

Also, I've found that unless I use a really high compression ratio (between S1 and Sd * n number of drivers), a la 2.5 or so, I can't get good frequency response up to 1khz. Per my original post, do you have any pointers regarding getting good response up to 1khz?

Also, any idea why I'm (apparently) getting some kind of strange resonance at 680hz? With the same JBL 2226J's, my parameters are now:

Ang: 1xPI
Eg: 180
Rg: 0.02
Cir: 0.44
S1: 1000
S2: 2800
Exp: 70
Sd: 880
Cms: 1.58E-04
Mmd: 85.18
Re: 10
Bl: 27.1
Rms: 5.04
Le: 3.5
Nd: 3P
Lrc: 25
Lpt: 0
Atc: 1000

Thx!

 

[Rademakers]

diyAudio.com Wiki - projects by fanatics, for fanatics for general guidance on how to use HR.

For extended frequency response higher up, minimize Vtc (volume). Also you're using a high Le driver, you'll find mid and high range drivers generally have low Le. Use the directivity tool to see how the frequency response is, taking directivity into account.

Best regards Johan

 

[davygrvy]

To be honest, good response up to 1k is going to be tough as drivers with that size a coil like being beamy. It's one thing to see a behavior in a simulator.

What's the adage? All wave guides are horns, but not all horns are waveguides. Depends how you shape the horn progression that hornresp gives so you don't screw-up the horizontal pattern.

Here's my advice based on my experience having made a mid-bass cab before that turned-out rather beamy. After you get your hornresp data for the path expansion, start with flat side walls expanding at not less than 50° from the center if you are going for a 90°H horn and use the top and bottom parts to fill the cross sectional area. This way, your horizontal pattern won't be too mucked up, but unfortunately my advice is mostly experimental experience based on prior failures.

With three vertically aligned drivers in a single horn, I don't know how you are going to get a good horizontal pattern.

 

[davygrvy]

This looks OK. Use Ctrl+O to see combined response of horn+port. This shows a tractrix expansion rather than exponential which some people feel is more appropriate for use as a mid-bass horn.

 

[j.michael droke]

Hi there: I've been using Hornresp for tapped horns (TH) and have experienced resonances around 90-95 hz, untill I removed the back chamber (by setting s1 to 0.01 and the corresponding length c=0.01) and removing the loading chamber (Vtc=0 and Atc=0). These chambers seem to be modifing the response (SPL curve) by additional resonances which sometimes show up as hard to remove spikes or dips, requiring extream modifications to the basic dimensions to cancel. ....regards, Michael

 

[letsbangout]

diyAudio.com Wiki - projects by fanatics, for fanatics for general guidance on how to use HR.

Thanks.. my reading of Atc now is that it's the cross-sectional area of the throat chamber, whereas throat chamber is defined as the portion of the throat prior to reaching the beginning of the horn segment (although I didn't see that broken down anywhere in the wiki, that's my interpolation..)

For extended frequency response higher up, minimize Vtc (volume).

I've tried that before, it seemed to have minimal effect. I'll give it another whirl now that I have a better handle on the parameters.

Also you're using a high Le driver, you'll find mid and high range drivers generally have low Le. Use the directivity tool to see how the frequency response is, taking directivity into account.

While that may or may not be true (I wouldn't know), in the case of this particular driver, the Le is high(er) because the motor structure is literally double that of the 8 ohm 2226H. Double the wire, double the impedance, double the Le.

 

[letsbangout]

This looks OK. Use Ctrl+O to see combined response of horn+port. This shows a tractrix expansion rather than exponential which some people feel is more appropriate for use as a mid-bass horn.

I plugged that in and found that your bass response is much worse than mine (my 3db down is like 27hz on the low side, assuming a baseline of 135db, yours is something like 32hz). Also the phase response from your design is _horrendous_. And there's still the same strange resonance at 600hz. But thanks for the contribution. I'll be sure to post my final parameters back to the forum, once I've finished redesigning.

Also in relation to what you said about vertical pattern control, that's a very valid point. One thing I could do is simply not have a vertical flare - essentially repeating the Altec 817 (which, by the way, I'm thinking about modeling since their plans are available online). Also, I don't suppose there's any advantage to flaring the vertical down more narrow, right? Would probably mess up the phasing tremendously, not to mention, hornresp might not be able to account for any of that...

 

[davygrvy]

The shape of the horn that HR assumes is a cone, IINM. Better ask for a confirm. I, myself, wouldn't believe any of the polar results once I deviate from a cone shape. Hornresp is for HORNS, not wave guides. HR knows nothing about the shape of your progression.

I was assuming you we're using this for mid-bass not caring about 70 and below. The phase response is funny because of the port to control displacement. If you don't want a port, you may do so. Also note the port placement is a relative thing for the sim and not stored in the data. Placement can have a wide change in response shape as you'll discover. If you really do want to push it to 130Vrms across each driver, you'd better respect the displacement graph. Your data said Nd=3P, not Nd=3S.

Oh, and I said HORIZONTAL pattern, not vertical. Is this for a home or a PA? Personally, it doesn't make sense to me to push the range as far as you are for a PA for numerous reasons I won't begin to list.

Oh... And this is really fun... A horn (wave guide) that has a well controlled coverage pattern over it's frequency range is never flat without outside assistance I love that one! Just look at those EV constant directivity horns, they ain't flat.

Please test your horizontal pattern with a cheapy plywood test build before committing to the nice furniture grade wood you'll probably be using. I've been down this road.

Sims are just a tool out of a wide range of design tools you may want to use. Don't be surprised at how correct HR is in some things, and how wrong it may be in others due to assumptions it is making that don't hold true with what gets built. Make some sawdust and verify, please. Flaring only the sides, doesn't mean the top-end is going to stick to it and follow. For example, look at the old late 70s 120° horns. They all have flat sides with the flare on the top and bottom. Also, if you are off-axis at 45° and looking at the cone, and can't see it, you certainly aren't going to hear 1kHz from it.

I look forward to pictures

 

[letsbangout]

These are all very good points that you make.. the implications of which I'm probably not yet fully aware. I'm definitely going to test with cheap plywood for sure. As for home/PA, this is mostly for home.. but I can also see myself using them for sound reinforcement gigs on the side (and building more for the same purpose, albeit with other drivers which are easier to get ahold of, probably 2226H's). So one of the design constraints is that the system must not be deeper than the size of most front doors/sliding glass doors/etc. (and for that matter, a constraint of any home I'd choose to live in, in the future).

Yes, 130v is realistic, although my data said 180. I keep it jacked up a bit for modeling purposes. At this point I'm pretty much resolved to go with 2 of those drivers. I figure that 2 is much more realistic than 3 - both for these J's and for any future cabinets that I build and equip with the 8 ohm 2226H's. The only advantage of 3 drivers is being able to extend both the low end and the high end response out a little more (because of the wider electrical Q), but it doesn't seem viable long-term to keep building speakers with J's. Not to mention, I have 4 drivers already, so I won't need to wait on any good deals in order to get more.

Two things that are relevant to what you said, which I have questions on:

1) You mentioned that the full-range nature that I'm trying to design for is unideal for PA. I'm aware of the monotonicity and power-compression aspects (of using such a wide-range midbass system down low and with mucho power). Is there something else I'm missing with regard to the PA aspect?

2) I'm employing an extremely high compression ratio in my current re-design, specifically on the order of 3.52:1 ((Sd * n drivers) / S1). I've never seen a ratio this high being used by conventional low-frequency horns - only in midrange/tweeters. Is there some aspect I should be aware of in this regard? Will the narrow "slit" in the throat that the high compression ratio requires - probably 5 by 40 inches - adversely affect my dispersion, distortion, or anything else relevant that I'm not aware of?

THX!

 

[davygrvy]

1) You mentioned that the full-range nature that I'm trying to design for is unideal for PA. I'm aware of the monotonicity and power-compression aspects (of using such a wide-range midbass system down low and with mucho power). Is there something else I'm missing with regard to the PA aspect?

This is mainly a personal thing, but I look at a well controlled horizontal pattern as being top priority. Maybe it's my scars from not having had good results? I have yet to be happy with any of the mid-bass horns I've built trying to stay with a three-way system to cover 90-5k. A 4-way with a mid horn (2" compression driver) covering 800-5k I've had great success with but the cost is astronomical for the good stuff. Generally, above 500Hz is where beamy starts taking over without special control for it.

Therefore, the perfect mid-bass cab should not be used above 500Hz unless some wave guide principles are used to control the horizontal dispersion (?)
Maybe 1k is a fine upper limit and I'm just pushing the envelope too far and taking it out on you (?)

Your ability to get deep bass is compromised by the horn length, so you'll end up with it being rear chamber loaded which is a great loss in efficiency compared to a long path bass horn covering 35-90. I like a midbass to run deep even when not required just cause the chamber volume is easily had for near free from the horn sides. It's just a little wider on the back, but doesn't need to be deeper. What you end up with is a loudspeaker that is displacement limited below a certain point. IOW, if your highpass is out so you're running it deep, you'll need to limit power due to a displacement peak at some freq around 57Hz or whatever. When not running it deep with a high-pass of 110Hz, displacement isn't an issue, so you can run the loudspeaker at its thermal limit.

2) I'm employing an extremely high compression ratio in my current re-design, specifically on the order of 3.52:1 ((Sd * n drivers) / S1). I've never seen a ratio this high being used by conventional low-frequency horns - only in midrange/tweeters. Is there some aspect I should be aware of in this regard? Will the narrow "slit" in the throat that the high compression ratio requires - probably 5 by 40 inches - adversely affect my dispersion, distortion, or anything else relevant that I'm not aware of?

High compression is good. It's a mark of how efficient the horn can be at turning pressure at the throat into volumetric displacement at the mouth. The realistic upper limit is found when the cone tears from the pressure. More pressure means the cone moves less. Less motion is lower distortion by virtue of the non-linearities of the motor. I'm sure there is a better description somewhere. That's just my take on it.

Jim Bell, in another thread here, is experimenting with an MCM 55-2421 8" LF driver and running it at 8.3:1. He says it doesn't sound good below Fc of 40Hz which might be due to an acoustical impedance limit being too high which wreaks the Q (just a guess). IOW, if the chamber is too small, wind resistance takes over. The bigger the driver, the less CR possible due to the strength of the cone diminishes (generally speaking).

I say go with 3.52:1 until, by experiment, it is proven to be bad.

This is good for me. Now I'm really thinking about my midbass horns and where I want to go next to improve them:
- One driver per cab this time for ease of moving, but design them for running as a group.
- Flat sides with a top and bottom flare to produce the progression.
- Consider experimenting with a phase plug and try ideas for wave guide control above 500Hz.
- Consider fiberglass over chicken wire instead of laminating to achieve compound curves if needed.

 

[letsbangout]

Likewise, you've got me thinking about this again. First of all, displacement (I presume you're referring to excursion of the diaphram) is a non-issue for me. My crossover of choice (for the moment) is a Behringer CX3400 with a 25hz switchable highpass filter on the input stage - presumably this is 24db/octave just like everything else within the device. If in doubt, I can also run the first eq band down 3db - those two measures surely take care of excessive diaphram jumping.

Regarding one driver per cab - I mentioned previously the Altec 817.. I was also looking at the 816 to achieve that goal - one driver per cab, can stack and run in a group, etc. If I ran two or three in a stack per side, I could also extend the entire apparatus using a field-configurable "horn extension" which would make the entire horn longer for that deep bass that you speak of. I'm not knowledgeable enough to know whether that extra 2 or 3db at 30hz makes that much of a difference. But the horns would still fit through a door, would be easier to transport than 250 pounds of plywood a la the JBL 4550, etc.

Ok the glass over chicken wire - please elaborate. You mean lay out the chicken wire, then put fiberglass mat over top of it, laminated with resin? I presume you've tried (in my opinion the simplest of all,) laminated posterboard?

 

[davygrvy]

I think you'd better look at diaphragm displacement in HR. I see 26mm at 45Hz as per the data you posted with 135L for Vrc. You can get that under control with a port or duct. Don't make the rear chamber too small to control excursion as it increases the potential for overheating.

Here's something just for fun. Nice and flat, but looking at the schematic and how focused it is aside from being a large monster. I think that'll bomb for good horizontal control. But fun to look at just the same.

Laminated poster board -- that's a neet idea. I used single directional grain thin sheets, which were surprisingly stiff on my last. It was the best available to me at the time. Even with poster board, you can't do compound curves where the bending is happening in multiple planes.

 

[David McBean]

The shape of the horn that HR assumes is a cone.

Not true .

 

[davygrvy]

don't stop there, David Is HR assuming assuming a horn that is square shaped progression where the wall widths are equal?

 

[letsbangout]

I think you'd better look at diaphragm displacement in HR. I see 26mm at 45Hz as per the data you posted with 135L for Vrc. You can get that under control with a port or duct. Don't make the rear chamber too small to control excursion as it increases the potential for overheating.

Yes, even in my current design I'm seeing 3cm at about 35hz. This is indeed "pushing it a bit" within the confines of the Maximum Excursion Before Damage portion of the 2226J spec sheet. Yet, while that is 1.6 inches peak-to-peak, the Xmax is only 0.3 inches. What's the difference? Is Xmax a measured parameter based on the enclosure they're using to model some of those parameters (which is a vented box)?

Laminated poster board -- that's a neet idea. I used single directional grain thin sheets, which were surprisingly stiff on my last. It was the best available to me at the time. Even with poster board, you can't do compound curves where the bending is happening in multiple planes.

Perhaps construction paper then. Make your compound curves using multiple pieces. Just remember that you're (probably) going to need something to back-fill these compound curves from the rear - maybe a two-part liquid foam and then sand/laminate/mat it to make the entire structure rigid. On the horn face, you could try the construction paper and chicken wire, laminate with epoxy, then start laying up more chicken wire and at some point switch over to epoxy with microbaloons (to make it easy to sand). Another limitation you'll have with these complex curves, is that in order to make a (somewhat durable) laminate, you'll need to resinate and lay up a few laters of glass mat. My worry would be at the joining of these curves - you'd probably have to cut the mat into sections just like you would the construction paper...it could make the joining of the curves the weakest part of your apparatus. For that, maybe you can tolerate some chopper?

 

[letsbangout]

don't stop there, David Is HR assuming assuming a horn that is square shaped progression where the wall widths are equal?

The question is - does it even matter for the purpose of modeling the horn, given the graphs produced by HornResp? Based on my (extremely rudimentary) understanding of horn theory, the info given by HR (SPL, excursion, etc) is based on cone theory.. but isn't it true that the science (being based on the areas of the throat, exit, segment length, etc.) does not change when you mangle the cone a little bit?

Looking forward to David's answer on this as well

 

[davygrvy]

Xmax is a T/S parameter of the driver. Off hand, it represent the distance where the coil is 30% out of the flux field. It is a general point of interest to consider where distortion begins. If 'Maximum Excursion Before Damage' is 1.6 inches p-p, that puts Xmech (also called Xlim) at 1.6*25.4/2 = 20mm. 0.3" Xmax is 7.4mm. Xmax and Xmech are distances in one direction oddly enough.

Epoxy with micro balloons? Thanks for the tip. I'll look into it. For experimenting, I really wish I could do it in clay on a test cabinet sitting on turntable. From that, make a mold like it's done for boats -- à la turbosound but hopefully not as ugly.

 

[letsbangout]

Thanks for the tip, I did figure out that Xmax is the one-way parameter after posting that. Good to know about the distortion figures. Along the same lines, where do you suppose power compression comes into play? I'm thinking that the measured power compression (i.e. in the 2226) is based on a vented box and not on the driver itself. So if the driver [presumably in a vented box] compresses at 400 watts, how then shall we determine power compression in a horn? Is it reasonable to assume that we could scale up that 400 watt figure by a factor of (Exit size / (Sd * n drivers)) which is the new effective diaphram size of the motors as determined by the exit?

Regarding fiberglassing, that's _perfect_ actually. Shape your horn on the turntable using clay, fire it, and glaze it. What you have is now the plug for a mold. Put parafin wax on it and spray gelcoat on top.. then start spraying layers of resin and chopper. What you come up with will be a mold, although you could just as easily skip the mold and simply use the plug to create the final product - probably a lot easier.