My two cents on a couple things….
Here’s a simplistic explanation of what a horn does: A correctly designed horn does two things. First it focuses the sound into a tighter beam. Second the compression chamber better matches the acoustic impedance of the diaphragm to the air around it (down to the cutoff frequency of the horn – where it begins to “unload”). BOTH things increase the load on the drivers diaphragm (thus requiring stronger motors to move the diaphragm a given distance – or more electrical energy to do the same) – and increase the SPL inside of the coverage pattern of the device.
To the first, a horn acts as a kind of a wave guide and does the same thing to the sound that a flashlight’s reflector does - it directs and focuses the sound coming from the transducer. Let’s say you have a light – that measures 10 foot-lambert’s in open air at 1 meter, with the light radiating in every direction spherically 360 dgs around it (equivalent to an Anechoic environment). Now take that same light, and set it down on a perfect mirror (half space). At 1 meter distance, you’ll read double the brightness or 200 foot lamberts (double free space). Add another wall and set the light at the juncture of the two walls - (quarter space) you double its output again (four times free space ). 400 foot lamberts. Add another wall to create a corner with a point and put the light down at the apex of that corner – (8 times the intensity of free space) 800 foot lamberts. And so on..
The tighter you focus that light the brighter that single light source is inside of its coverage pattern, but less area the light covers and the higher the intensity is in front of the beam. Similar to using a Mag Light and spreading the beam super wide vs. its tightest focal point. The tighter the acoustic beam – the larger the increase in SPL measured inside of its pattern.
The next thing a horn does is operate as an acoustic impedance transformer. Ordinarily – the radiation resistance of the driver is insignificant relative to the air it is pushing around (the air – a gaseous medium with quasi viscous properties) it is difficult for a little tiny cone to “grab” that air and control very well. Think of a Fan, or the impeller of a boat, or propeller to an airplane. As the fan blades spin, turbulent air peels off the outside edges and the blades efficacy is compromised. Same with an impeller or propeller. If you wrap a duct around those blades (any group of them) – the air or water cannot so easily shear and tumble off the edges – the now focused fluid or gas moves at a much higher velocity out the back side of the blades – while they turn at the same speed, though the load on the driving motor has increased. Decrease the size of the orifice the fluid/gas is exiting through (compress it), and again you increase the velocity of that medium on exit (and load on the drive unit)..
The duct has increased the mechanical impedance of the fan/impeller/propeller. To spin the same speed – because the device is moving more fluid / gas per sweep – the ‘Load’ on the driving motor is significantly increased. Decreasing the size of the ducts exhaust – also increases the impedance / load on the driving device and velocity of the exiting medium.
This is the same type of thing the compression chamber does in a Horn. It increases the acoustic impedance of the driving element – bringing it to a much closer match to the air around it. This increased “grip” on the air – improves the acoustic coupling of the diaphragm – and increase the SPL it generates for a given movement. While at the same time – increasing the force required to move the diaphragm a given distance.
This was most easily demonstrated to me one day – when I grabbed the huge exponential 400hz midrange horn at Klipsch and tried talking through it. My voice was Massively increased in SPL – but the force required to make said voice was incredible. It was physically hard for me to make a sound through the horn. Till that point – I couldn’t fully comprehend exactly what it was that a horn does.
Two reasons bass horns can sound better than point source drivers:
1. The Spl increase brought about by the above lead to a significant reduction in distortion at a given SPL, provided the driving element has the strength to handle the increased load. The reason is simple – motors are one of the two components that generate distortion in a transducer – to quote Paul W. Klipsh’s words to me directly “If the cone Moves – the speaker Distorts, period. Increase its movement – increase its distortion. Anything you can do to reduce its travel for a given SPL will yield an immediate reduction in distortion components.”.
2. Dynamic Range. This is a distortion (dynamic compression) component RARELY talked about by the audio industry to the end consumer – and to me it is arguably the most important distortion component to address. Ever go to a fantastic concert – with exceptional sound – and drive home wishing you could have the dynamics of that system in your living room? Ever go to a HiFi store and hear a system that is purportedly ultra flat in f/r – but come away thinking the system sounded dull/lifeless/un-impressive? There is a Reason for that.
Something I have found in my own research into LF that may be almost counter intuitive to a consumer – but is true. The human ear responds more directly to increases in dynamic range – than to high Q deviations in frequency response. High dynamics will lend more of a sense of “realism” to sounds being reproduced than an uber flat on axis f/r, provided the object with the rough f/r follows the general trend of the flatter speaker.
The human ear is relatively insensitive to minor high Q deviations in amplitude – especially troughs, it roughly correlates to f/r smoothing of between 1/6’th and 1/3d octave (this is part of the reason most mfrs published material shows 1/3 oct smoothing).. If the highly dynamic systems 1/3 smoothed response is akin to that uber flat speakers response – a 20dB headroom system will sound tremendously more realistic than that Uber Flat 5dB or 10dB system. ( an XXdB headroom means the enumerated SPL capability broad band above the reference listening level. IOW if I choose 100dB for my “reference” level – my system will require 120dB broadband capability to be called a “20dB system.)
Regarding the “time delay” involved to propagate the signal from a Bass horn.
The speed of sound is 340m/s so to change it into yards the conversion rate of a meter to a yard is 1.0936133 so you would get 371.829 yrds/s and then you would use the conversion 3 ft=1 yard.
So sound travels 1115.486ft/s
So then you would use a basic physics equation of V=X/T
V=Velocity
X=Distance
T=Time
You're trying to find time so rearrange the equation T=X/V
T=3ft/1115.486ft/s
T=2.69 milliseconds
The length of this horn is approximately 6.096 meters – so a sound exiting the horn is delayed approximately 16.3982milliseconds. This does not account for the phase lead generated by the horns increased acoustic impedance match to air relative to a free air point source transducer at the corner frequency. In the 2 decades I’ve been studying and researching acoustics – including literally hundreds of AB and ABX test of various things, I’ve come to the conclusion – that regarding low frequency time delay – the human ear is as insensitive to errors in time as it is to the actual frequency.
Remember that at 4khz – an un damaged ear can hear 1dB spl – or literally molecules oscilating back and forth the distance of that molecules electron orbit (!) IOW colliding into each other. Correspondingly at those frequencies it is also the most sensitive to timing errors – between 1 and 4khz, the human ear can detect time errors as low as 2ms (1khz) and 1.4milliseconds at 4khz. 500hz the ear can detect 3.2ms deviations. At 20hz, it takes the average human ear over 70dB spl to cross the threshold of hearing. And at 16, about 100dB spl. Your ears are increasingly less sensitive as you drop (and increase) in frequency.
I subscribe to the theory my friend and a figure in our industry John Murphy does with regards to timing error's perceptibility at LF. Extrapolating Blauert and Law’s threshold of detection for “hearing” timing error’s to cycles at a given frequency – (their data covers 500hz/1khz/2khz/4khz and 8khz) it looks like we become sensitive to changes in timing of around 1 full cycle under the most controlled listening environment and using the most challenging material (clicks and pops).
Here is John’s opinion on the audibility of delay in the bass range. “My" current opinion would go something like this: "One cycle or so of delay in the bass range is probably just audible under controlled listening conditions (headphones) and with the most challenging program material (clicks and pops). Under less well controlled listening conditions and with less difficult program material the audibility thresholds are somewhat higher."
So – let’s look at 10hz, 20hz, and 40hz (the range my sub will be used under.. even though Lil-Mike has pulled out the almost impossible and designed an almost 3 octave tapped bass horn in theHouseWrecker!) and see how long one cycle is.
1 period at 10hz is 113 feet. (!!!) Expressed in milliseconds – that would be analogous to 101 milliseconds (!!!)
1 period at 20hz is 56.5 feet. (!!) Expressed in milliseconds – that would be analogous to 50.6 milliseconds (!!)
1 period at 40hz is 28.25 feet. (!) Expressed in milliseconds – that would be analogous to 25.3 milliseconds (!)
Keep in mind – these models use an infinitely ridged wall for the horn – the actual deviations will be smoothed. My experience also tells me (from ABX testing) that Most Consumers under normal listening conditions cannot tell there is a significant timing error with as much as 3 times those numbers below ~~ 60/80hz. So.. looking at theHouseWrecker’s model we have……
101ms delay at 11hz. (Just under what the potential barely detectable threshold of a trained ear under controlled conditions and with program material specifically designed to highlight timing errors.) My experience is that the human ear can’t detect even 4x that much delay at this frequency – it is very insensitive to this deep LF.
22ms delay at 20hz. Less than half of the best possible threshold.
23ms at 40. Right under the best likely potential threshold for detection.
I can use double those numbers and be quite happy.
Regarding “distortion increasing as the line length increases":100% False. Increase the compression to a degree the velocity is so high at the reproduced levels in the throat – distortion will increase. Use too small a mouth, or folded turns with tight restrictions in the high velocity area (end of line), distortion will increase. Etc..
The designs Lil-Mike and I work up have a TON of energy put into them to reduce the acoustic reactance as much as possible, and match the load to the transducers being used. (for instance theHouseWrecker has significantly less acoustic reactance than the DTS10. It presents a very easy load to the driver face) – and ensure adequate expansion and mouth size relative to the driving element and expected F/R and SPL to be reproduced.
