 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
Been trying to design an 80Hz M=.6 midbass folded horn extending to 500Hz+ with an EVM15L. I'm getting a conflict with the various design approaches. Edgar advocates Keele's optimum throat size, which puts the throat in the 80si range, while Leach's is in the 30-36si ballpark (just winging these numbers from memory). The 15L fh is around 344, according to Keele/Edgar. With Leach, it comes out to 700Hz+.I keep seeing discussions stating a minimum length horn of 1/4 wavelength of fL, which for 80Hz, would be about 42 inches. Leach's throat size calcalates this length. However, using Edgar's throat size, the horn is only around 32 inches.
What gives? What are the tradeoffs between the design approaches?
Also, Leach says to reactance annul the driver around 60Hz. Doubt it.
I remember reading in the "Show Horn" article that Edgar expressed doubts about Leach's formulas, but they do bring about proper length. Edgar also thought "a" (Vas/Vb) was an independently adjustable parameter, and seemed to be able to annul anyway.
It would seem the shorter horn of Edgar would cause reflections and roughness as compared to the Leach model. Also, doesn't a smaller throat help extend fh? It would seem important to extend up as high as possible to cross over to a compression driver.
Follow Ups:
HiI have used Leach's math to build dozens and dozens of horns over the last 20 years or so, some as products for our company, some for scientific R&D, some just ideas.
Having used Don’s math prior to that I can tell you that from my experience, I have never had a horn that wasn’t close to what it should be using Marshal’s approach.
Like all horns, any driver on any horn does something but Leach's math is by far the best way to judge the driver / horn relationship, it even holds for compression drivers once one has the right values for the “bits”.
Personally I think it is rather amazing that everyone doesn’t use it, especially since he did the work in 1979 or 1980 as I recall.
Reactance annulling is one place where this math does not work, or more correctly, it does not specify the horn “t” based on the driver. I use a different formula for that.
This was derived by an old horn friend named Richard Long (who has since passed away) who was a whiz.
One of my favorite memories of AES were the Lunches he, Dave Martin (Martin sound and also now deceased) and I had. I figured it would be the chain smoking that would get Dave in the end but it ended up being his business partner.If you are at a point where you are zeroing in on a design, I will run that calc if you wish.
At live-audio.com there is a public bass horn project which was designed with this math fwiw.
Cheers,Tom Danley
Tom :I am in the process of locating and installing a proper bass horn driver in my mod'ed Speakerlab K-Horns. According to the parameters I have gleaned from Asylum the 15" driver I'm currently using rates a paltry Fs/Qts = 40 (!!!). I have come to the opinion that only a pro quality driver would be satisfactory and am investigating 2 that are particularly attractive :
15" Driver # 1 15" Driver # 2
Fs = 53 Hz. FS = 40 Hz.
Qts = 0.33 Qts = 0.20
Vas = 88.3L Vas = 200L
Xmax = 5 mm Xmax = 4.6 mm
Fs/Qts = 160 Fs/Qts = 200
Magnet = 5.94 lbs. Magnet = 17 lbs.
Gauss = 11,000 Gauss = 13,500Both have die cast aluminum baskets. # 2 has carbon fiber diaphragm.
I have not yet developed pricing/availability on either but hope to have that soon.I would be very greatful to you for relaying Marshall Leach's horn throat parameters. I have never been able to access them on the AES website.
Please respond to pkell44@juno.com I DETEST MICROSOFTS E-MAIL PROGRAM.
Hi
Can you direct me to a web site with plans for this horn so that I can get some needed dimensions? (or if you can get them)
Throat area
Mouth area
Length of horn path
Rear chamber volume
Front chamber volume if presentCheers,
Tom
You can use the following free programs to do your calculation:
http://home.mn.rr.com/blitzweb/index.html/LeachHornAnalysis.xls
http://melhuish.org/audio/MLutil.zip
Thanks. I'd like to take you up on that offer. What numbers do you need for the calc?
HiWell you sound like your keen to build something, cool.
A few other thoughts about the math, M. Leach’s math does not deal with directivity or series inductance, both of these can effect the HF acoustic power or on axis SPL vs acoustic power. These effects are seen in a more detailed “physical model” of system.
I have Marshal’s math in mathcad set up to provide the following:
Best driver parameters for a given HF, LF and horn parameters.
Best horn parameters for a given driver, HF and LF
(+ a generally less useful “greatest BW version” of the above)One also has a choice of two HF corners, one extended upward (and having in theory a 3rd order roll off) by choosing the front volume to be an optimum size (given) and a second (lower corner with 12 dB /oct roll off) if the front volume is zero. Intermediate values can be examined using a physical model and as mentioned the RL corner and horn directivity can alter this.
Since efficiency goes down as bandwidth increases, picking a real HF corner is the first step. Also, since this IS real acoustic loading, one would generally want to stay away from compression ratio’s greater than about 3:1 for a normal cone driver.
This is because the cone is usually not strong enough to act like a piston when the pressure reaches some level. While probably not an issue in the home, it is important at high power levels.
Also one would want to operate the driver at least an octave below the cone breakup point AND at the hf cutoff, the radiator diameter must be smaller than about 1/3 wavelength at the hf cutoff (not a problem if you use a phase plug but more involved). For 500 Hz, this suggests about a 10 inch driver is the largest one would want to use here.Several questions then, do you have the 15-inch drivers already? Can I suggest one if you don’t have them?.
How loud do you need these to go? Finding a maximum acoustic power alignment is usually not the lowest cost or highest efficiency alignment although the “greatest power” alignment is often the lowest in distortion at a given SPL.So far as length, the shortest a horn will operate happily is at ¼ wl in length.
This is a little longer than the physical horn length, particularly as the mouth size gets larger. Acoustically, the “mouth” actually sticks forward of the physical one, something like .6 * mouth Radii and is the same “end effect” that ports exhibit as well.
The horn is not “fully efficient” until it is about ½ WL in length.
Hope this helps, kind of a data dump in no particular order on a rainy afternoon.
Cheers,
Tom Danley
Thom :Thanks for the info and quick response.
The MR & T horns/drivers I am running are E-V Pro HP940 horn & DH 1 driver, HPT 94 horn & DH 3 driver and they are excellent performers.
I spoke with Trent Keeling of Advanced Audio & Lighting Systems in Peoria, Ill this aft. He sends his regards. He has suggested what appears to be a gangbusters driver : the 15" McCauley #6342 :
Fs = 40 Hz.
Qts = 0.20
Vas = 200L
Xmax = 4.6 mm
Magnet = 17 lbs.
Gauss = 13,500
The voice coil is copper ribbon, Kapton coil former, die-cast aluminum basket, carbon fiber composite diaphragm, and double spider, etc. etc.
I do have the EVM 15L drivers already.I have also rough cut the material for the speakers to plans made a few years ago. I optimized the LaScala/Belle Klipsch style folding geometry using many geometric and trig equations solved
simultaneously in Mathcad. But I put the project on hold, and now I'm trying to go back through my notes to sort everything out again, to make sure before I proceed. (I'd rather lose the cost of the wood, than build it wrong and then lose the cost of the wood.) After getting back into the thinking process, I noticed the difference between the Leach and Keele/Edgar throats, and freaked.
The Keele/Edgar method gives a larger throat size for a shorter horn, and is apparently optimized for highest efficiency/least mouth reflections (all good, I think?). Not sure what Leach's model is optimized for.I did some modeling with Hornresponse software by David McBean.
The EVM 15L drivers have the following parameters:fs=43
Qes=.25
Qms=4.96
Qts=.238
Vas=.245m3 (8.65 ft3)
Sd=855.3 cm2 (132.6 in2)
Xmax=200 watts
Re= 5.2
Le=? no data for this. Guestimated about .6mh based on similar JBL D130 spec sheet.Using Hornresponse, the following were calculated:
Bl=18.10 T.m
CMS=2.35E-04 m/newton
MMD=43.93 gm
RMS=3.18 newton.sec/m
Modeling an 80Hz 1/4-horn with M=.665, length=81 cm, St=484 cm2 (Keele optimum), I could reactance annul at 80Hz with Vb around 27 liters. However, the response curve looked a little flatter overall with about 19-20 liters, with FL increased just slightly. The response was down about 3db at around 550Hz.I tried using MLutil to model with Leach parameters, but I could not find a way to play with many of the variables independently, especially St and Vb.
If Hornresponse is an accurate tool, the response looks pretty sweet. I don't know what the maximum SPL would be with this model, though. Is there a program that will allow me to find that with the "non-Leach" optimization? This system is for home hifi use.
HiI am not familiar with hornresp although I have heard of it.
I don’t know if it can show the electrical impedance or radiator motion but if so, the difference between the Keele and Leach alignments is clear.A quick physical model of your driver and horn (Keele alignment) shows that while the output goes up above 900 Hz, the electrical impedance shows that the efficient range of operation is below ~ 350Hz
The Leach alignment shows a somewhat lower efficiency (predicted to be about 30%) based on the impedance but the acoustic loading is present on the motor to above 500Hz.
Also, as a result of a smaller throat, one finds the cone motion is significantly less than the Keele alignment for a given SPL. Since motor non-linearity is the primary source of distortion in a LF / MF horn, one could expect that the Leach alignment with the same driver will have less harmonic distortion at that given SPL.
If driven by a passive xover or tube amplifier, having a broader range of acoustic loading means there is also less SPL variation due to a changing load impedance.In each case, the horn is a little too small (for 80Hz) and in both cases, an increase in front volume (over the computer calc for an ideal horn) lowers the hf corner acoustically down to the crossover region, such a move also lowers the distortion as it acts as an acoustic low pass filter.
To be accurate here, one needs the real Le .I am not clear on a few things though.
Do you have a horn built which is 81 cm long ? (in which case the best bet would be to try to find the best driver for your horn based on its physical size)
Or, are you wanting to build an 80Hz horn for your EV drivers?
Or want to know what the response will be with your drivers on that horn?So far as what Leach’s math predicts, it is FAR better at scaling the acoustic load to suit the driver and bandwidth than the “folk lore math” as I have called it a few times over the years.
This is why Leach’s paper IS in the AES loudspeaker anthology and Keele’s approach is NOT.
Understand, the math is a “starting place”, nearly always (at least for bass) a practical horn is usually considerably smaller than the theory assumes. Being able to model the real thing is the best way to further “adjust”, based on predicted response.
Cheers,Tom Danley
Hi Tom!You wrote:
> > So far as what Leach's math predicts, it is FAR better at scaling
> > the acoustic load to suit the driver and bandwidth than the "folk
> > lore math" as I have called it a few times over the years.Accurate horn modeling tools have been made accessible and very easy to use, so I don't think many people are operating in a vacuum of information. Both Leech's and Keele's models are avaliable in a variety of forms, from spreadsheets to computer programs as well as the older ways of calculators and slide rules.
As for implementations, you might like the " Hornresp " program that McBean has done. For the average do-it-yourself horn builder, it's an excellent tool and the price is right. I've found it is easy to use and I like to model a horn for suitability before building it. It's great for "what if" scenarios.
Edgar seems to prefer using low-mass diaphragms in horns having a flare constant of 0.6. He has also stated his preference for Keele's models to Leach's. That may or may not be as accurate but I'm not sure I'd call his approach or that of others as "folk lore math."
I would say that your preference of high-mass cones is a design choice, and not really a "new math." I know that you prefer to use heavy cones to tune the system lower, but this is really an application of the same physics and formulas that govern other designs and not the result of improved mathematical models.
The use of low-mass cones and low-inductance motors is also a design choice. It extends the bandwidth of the system by raising the point where pistonic motion minimum-phase output from the horn rolls off. So I don't think the use of a low-mass cone is the result of "neanderthal math" or anything like that. Both low-mass and high-mass approaches can be easily modeled using the same formulas, and each has its own sets of strengths and weaknesses.
Wayne
I wrote this response based on your fist post, which has been removed, sorry if it doesn’t specifically address your new post.Hi Wayne
What I try to do is use the best driver for the job, how one determines “that” is what the math is for, at least as a starting point.
What I refer to as “horn folk lore” are silly notions like “a horn driver should have a light cone” and thumb rules about Qt etc.
The right driver simply has the right parameters for the job your trying to do.
Depending on the frequency, the physical appearance and properties of that driver changes considerably to produce these needed properties.
The LAB sub for example has a very heavy driver, contrary to “horn folk lore” but it is about 100 times more efficient in the horn and appears to work “properly” AND more importantly, as designed.In the current thread example with the EVM15 driver, one has two horn configurations with dimensions specified by the Keele approach and the Leach approach.
The differences in this physical model are described in the previous post.I don’t see any real reason why would one choose to use the less accurate math when by using the Leach approach one can get a wider BW of efficient operation, less distortion at a given SPL AND a greater excursion limited acoustic power with the same driver?
I have known Don Keele personally for 15 years at least, he is a great guy but I don’t see any up side to using his horn math here.
Cheers,Tom
A few thoughts on your new post.
Again, this has nothing to do with massive cones or anything like that. I have no personal preference.
I prefer to use what is appropriate for the job. It is “how” one determines what is appropriate that is the issue.
What my comments do have to do with is which math gives the best track record of results when you build and measure the real thing, not who uses it.
I did use Don k’s approach up until about 18 years ago, but without doubt, Leach’s approach works better.
Agree with using the right driver for the job. In the subwoofer labhorn a heavy power tolerant driver cone/VC is desirable to allow the needed watts of bass output. That same type of driver for higher freqs may be fine for sound reinforcement but for home HIFI it may not be the best option. In the home HIFI where often flea powered amps are used the ultimate resolution not the ultimate SPL is the priority. A more resolving lighter cone/VC driver with likely lower power tolerance would be a better choice. It all depends on the manufacturer's market wether professional sound reinforcement or HIFI home use. So no need to disagree.Regards
Roland
Hi
I agree but from a different perspective.
In most things, non-linearity is an undesirable thing, it IS for speakers and electronics used to re-produce sound. This nonlinearity is usually directly related to motion or operating point so far as a devices maximum capacity.
In the case of the LAB sub, as long as you didn’t change the operating band, one simply see’s less and less distortion as the levels falls, nothing bad happens at 1/100 Watt or anything.
On the other hand, if one only had 4 Watts available, a much lower cost driver could be made which would also do well in that application.
Other things can then be altered, say use a lighter cone, smaller motor driver with less compression ratio and larger throat area to get the same response but less power output follow me? this kind of scales things including the appearance of the driver..
What the horn needs to have to have so far as a driver IS partly governed by how much power it needs to produce as linear excursion and thermal power handling set two boundaries.
A higher compression ratio produces more acoustic power with a given excursion but needs a more massive and stronger driver for the same BW. One way to look at it is where are you when one runs out of excursion relative to the thermal power or visa versa.
Mass alone is not a bad thing, only bad if you have too much or too little for what your trying to do.
CheersTom
I agree that a driver having great power handling capacity does not necessarily mean it won't perform well at low levels. Also agreed that non-linearities will be proportionally reduced, so the massive motor will work great at low levels. But one big issue for some of the tube-power guys is the impedance curve. The output transformer forms a voltage divider with the load and it has a significant source impedance. So most of the stuff with heavy cones makes a ton of back EMF and really needs an amp with good current sinking ability.There are some flea-power amps that are seemingly immune, but others are incredibly sensitive to the load. Not only that, but it may be that some of these kinds of amps are so sensitive to their loads that certain conditions can send them into oscillation. That would be a condition that most would say is an amplifier problem and not simply a trait, but it is something to consider, nonetheless.
Check out the links in the earlier post called " Speaker cables, feedback loops and Spice ," if you're interested in models that show this behaviour.
Hi WayneGood post, the electrical side of things is a good place to start looking at what a driver in a horn does.
Back EMF is one of those words, which strikes an ominous note in many peoples minds.
Most will say it has something to do with something “bad” and yes it can be a problem.An electrodynamic motor has a “force constant” that is it produces some force per Amp of current applied.
For a loudspeaker it is usually in Newtons of force per Amp (BL)
Related precisely to that is the lesser known (and common term in motor lingo but not in speakerland) Voltage constant.
The more force a motor produces per AMP, the more Voltage it produces as a generator for a given velocity. These are tied together, one cannot have a motor that has a high force sensitivity AND low bEMF.In an oscillatory system like a cone driver, the back emf IS what allows the driver impedance to be anything over Rdc. These two constants along with the radiator motion define what the impedance is in any box or horn. The radiator is governed by two things, it has mass, the mass is in parallel with a spring which when they are equal but opposite force, dictate the Fb.
Above Fb, mass is dominant, below it is the enclosure and suspension spring in parallel.This relationship is very clear in a sealed box. This mass and spring when reflected through the motor (or gyrator as some call it) show up as electrical reactance’s with moving mass being capacitive and the spring being the parallel inductance forming the tank circuit which produces the familiar box resonance impedance curve.
For a direct radiator, there is practically NO air load on the radiator, it is too small to “feel” the radiation load of free space. The motor is connected to the amplifier through a series R and L (Rdc and Le), these can be large or small and “hide” the full reactance of the motor from the amplifier.
For example the larger one makes the Rdc, the smaller the reactance’s are in total and the more like a resistor it looks. Unfortunately, as one increases the series R (while leaving everything else alone), the efficiency goes down as a larger amount of power is “lost” through I^2 R losses.
While very small the airload is also in parallel with the driver’s mechanical losses, the sum is what sets the driver’s Qm.
At resonance the magnitude of the radiation load and suspension losses set the high point of the impedance curve.
AT resonance, one notices several things, one, the impedance although high is resistive AND the acoustic output is “in phase” or not shifted RE: the input Voltage.
Either side of resonance and the load falls greatly and becomes reactive (this is what makes it fall).All of this does make the job of a passive crossover or non-zero impedance amplifiers job more difficult, neither wants to see a changing impedance or reactance. For the crossover, these changes are not accounted for in text book filter formula’s and need to modeled in Spice or other’s which simulate the “real thing”.
Enter the horn.
Take the sealed box, reduce its size until Fb is about 1/3 from the bottom of the frequency range one wants to cover and increase the motor strength until the impedance curve falls to around Rdc at those extremes.
Now one has a giant, broad impedance curve which spans the width of the BW you wish to cover (more or less).
This impedance curve shows the radiator has large mobility anywhere within that operating band.
Still though either extreme is set by mass at the high end and spring at the low end.
Now, add a horn.
A horn (as it is assumed in the math) provides a more or less constant impedance load to the small end, above the “high pass” filter point set by the flare rate, mouth size and length.
At the small end, one also has the option of using an acoustic transformer (compression ratio) to aid in the matching .Matching of what?
Since the air load is in parallel with the mechanical losses (and these can be made very small) and this shows up in series with the Rdc (remember the impedance at Fb). The air load can be sized so that it is equal to the Rdc.
Since it is resistive and so is the acoustic load, the system appears to be a resistance that is 2X Rdc.
Half the power is lost in Rdc, the other half as acoustic radiation or about 50% efficiency (in this example).
Now, one has an electrical load which goes from around Rdc below band, increasing to about 2XRdc in band and then falling back to Rdc above band (not including driver series Le here).
Within the band, the load is resistive and not reactive, the acoustic output has non changing group delay as it has no net phase shift (unlike the direct radiator).
It is only at the band edges that one see’s the reactance showing up.Also worth mentioning, one cannot tell from the impedance curve if the driver is light or massive, high compression or low, weak motor or strong “IF” it is sized properly.
Take one horn with a 1:1 compression ratio and feather light parts and another with a 5:1 compression ratio and massive driver/motor. If designed for the same BW, the would be indistinguishable from the impedance curve.
All this assumes that one CAN match the acoustic load to the drivers impedance curve (motor strength, cone area etc etc) so that one gets the condition described above.This is where Leach’s math is better than Keele’s. Keele’s math does a poor job scaling the physical parameters such that one has an acoustic load over the desired band.
Keeles math does work, you just don’t get what your asking for which is a problem if your trying to design something specific.Does leach’s math mean massive drivers? No nothing like that.
People do realize at one level that different frequency ranges need different parameters.
Yet at another level they say things like “horn drivers have to have light cones etc”.
While true enough in one frequency range, it is not true in another.
At one extreme, look at the motor strength to radiator area in a compression driver.
Then look at a woofer in a horn, then imagine what you want if you went down further in frequency.Why would one use a heavy driver?
If a large acoustic power were desirable, then a larger compression ratio is your friend as this is a transformer ratio. Working backwards from a desired acoustic power and low cutoff, one finds that with a practical excursion limit, a given compression is needed.
Given the compression ratio, one finds that as the compression ratio goes up, the driver is necessarily more massive and strong in order to have the proper acoustic properties.
Still though, that massive strong driver has the same load on the source as the dinky light driver, it can simply handle and put out an order of magnitude or more in power.
It is not any more difficult a load than any other alignment “IF” its acoustic properties are matched and it is this matching that the Keele math is less accurate with than the Leach math.
Cheers,Tom
Great! Now this thread is finally getting somewhere. This is the type of discussion I was originally looking for. I don't think there's a difference between Keele's math and Leach's math. I think that there's just a difference in priorities. Keele's priority was efficiency. The throat size he picked was to optimize efficiency (by picking the air load size equal to the Rdc). And Keele did state that condition may not result in widest bandwidth, which is clearly the case. But, it also gives a shorter horn; but, larger Vb, higher diaphragm motion, etc. Alternatively, Leach optimizes the widest, flatest, highest sensitivity bandwidth by picking parameters of the supposedly same math, differently, but requiring a much longer horn.That being said, maybe it's time to start another thread from this point, to discuss what parameters are required for best resolution in a home hi-fi situation, and the cost/benefit of parameter adjustment. While Tom makes clear the issues from sound reinformant standpoint, I think we're all agreed that the focus in that environment is not necessary identical to that in the home. As I stated before, a long horn is not as desirable in a home environment as a short horn, especially when folded, because of time alignment issues, etc. A long straight horn may be beneficial for time alignment issues, but get too long, and you lose the benefit of the rear wall. A home hi-fi would most desirably, IMO, be a quarter size or less. Then there's always the issue of system integration.... In sum, where is that happy medium in the continuum of parameter balancing for the home, as opposed to a stadium? The idea of a light cone and stronger motor does sound like a good thing for resolution, at least from an intuitive standpoint. There must be a limit where compression ratio gets too high to be a good thing, such as when air compression gets to be non-linear. Slopes at the bandwidth limits may be a factor for realizing crossovers with realworld components. Etc., etc., etc.
HiI am glad you enjoyed the horn discussion, thinking of it in terms of its mechanical / electrical operation is a good way to see it I think.
I guess what I see is that you ask Keele’s math for a horn that operated from 80 to 500 and you get a horn that covers 80 to about 350 Hz efficiently instead.
If one wanted to model each case and skew what you asked for to end up with the desired range of efficient operation, one could use the Keele math.
Rather, one could ask the Leach math for the same range but actually get it on the first try.While Keele does talk about “efficiency” in real life this is the difference between the radiated power and the heat produced in the drivers (I^2 R and mechanical losses), NOT the difference between Rdc and the impedance at a given point. A higher impedance does not necessarily reflect greater efficiency and more likely represents more reactance.
Using the Leach math going the other direction, let’s ask the following question.
What are the driver parameters for a 15 inch driver which has a 2:1 compression ratio
What are the driver parameters for a 15 inch driver which has a 4:1 compression ratio
Both intended to be efficient between 80 and 500 Hz.
2:1 4:1
Vb=400^3 in Vb=200^3 in
Vf=125^3 in Vf=63^3 in
Mms= 121 Gm Mms = 238 Gm
Rdc =5.6 Rdc = 5.6
BL=22.1 BL=31.6
Qm=10 Eff=66% Qm=10 Eff=66%Funny thing is, one could not tell which one was which from the electrical terminals or frequency response.
With a normal driver, the 4:1 would typically have less distortion, as the motor etc is the MAIN factors in Low frequency horns.If you want high efficiency, one must use the right driver for the horn and frequency range.
That is why I asked if you could use different drivers, the EV’s are not as well suited to the job as the “right” parameters would be.
What is resolution?
While that may have many answers, I would bet most would agree what it is not.
It would not be a result of anything, which altered the input signal relative to the acoustic output signal.
This would include errors in amplitude response, time (acoustic phase and reflections), “bonus sound” that is not part of the input signal but rather “noise” made by the driver including harmonic distortion.
Not usually thought of are level dependant effects, which govern dynamic linearity too.What one concludes when examining loudspeakers is that generally the lower in level you drive them relative to there maximum motion or power, the smaller ALL of the nonlinearities are.
This is why Commercial sound / Pro sound Woofers and compression drivers can be used in the home, even drivers like an M-4 (which can put out 100 acoustic Watts) has VERY low distortion at “home” sound levels.
Bass horns are no different than Hf, mid and mid bass horns in that regard.Looking at the two approaches to the horn design. One can see that the load less dominated by resistance HAS more reactance AND so a greater spreading of the input signal “in time” over the op BW.
The one with less loading also has more cone motion for a given SPL and so would likely have considerably greater distortion.
While this driver would have a lighter cone etc, the other properties would not suggest the sound would have a greater “resolution” unless that is something other than following the input signal as closely as possible. For a faithful reproducer, you want the speaker to take as few liberties as possible with what goes in.
Cheers,Tom
This thread is getting too long and scattered, so I've started a continuation thread.
A horn raises the impedance within the pasband, but below that the driver acts very much like a direct radiator. In the bottom octave, the horn has some impedance peaks from motor diaphragm resonances and reflected standing waves in the horn flare. These two combine to form pretty healthy impedance spikes. So down in the bottom octave, you can expect diaphragm mass to become quite significant.Try a LAB12 woofer on a Decware Zen amp or something like that. Try it both in a horn and in a box. Not only is the minimum impedance very low - 4.3 ohms for a LAB12 woofer, 2.2 ohms for a LABhorn - but also the resonant peaks are very high, so the the change of impedance through the bottom octave is high. At low frequencies, the system acts as a direct radiator, so at the 22Hz resonant frequency of the diaphragm, impedance will be very high, ten to twenty times that of Re .
Back EMF from the motor's mass swamps the output transformer at resonant peaks such as those found at 22Hz, and in the LABhorn, also at 33Hz and 60Hz. Even by shifting the Fo resonance up with the sealed chamber, such as is done for reactance annulling, you still don't get around these large impedance peaks. In fact, this tuning method requires the amplifier's current-sinking ability to work. That's the damping mechanism of a speaker at Fr, or when mounted in a box, at Fo.
There's no way tiny little tube amps can provide the necessary damping. This isn't true only of the LAB12 woofer, but also of most similarly tuned devices. Low-power amps aren't what the speaker was designed for. If the amp has good damping ability, a high-mass speaker will work great with very small amounts of power. But if the source impedance isn't very nearly zero, a device like this isn't appropriate.
Hi WayneSome good points Wayne.
Having a matched source and load (as was custom before the 60’s) has an advantage compared to a voltage driven horn. In this case, one sat at the top of an efficiency curve and should the horn load go up or down, one only moved to the side of center on a reasonably broad curve (like antenna mismatch)
This makes for a more constant amplitude response in non ideal conditions (if I could only convince the Pro sound community eh?).
With a voltage source, the impedance dropping or increasing by a factor of two causes a much larger change in delivered power.
It is no accident that (in the case of a non-ideal horn). there is usually an output dip associated with an impedance peak and output peak where there in an impedance dip.
With a matched source Z, the amplitude variation is greatly reduced.So far as a massive speaker the important thing to keep in mind is the reactance is not directly related to mass but rather mass coupled through a motor.
If the motor is strong, that mass is a small reactance. If the motor is weak, that mass is a large reactance. In this way, two very different drivers could present the same electrical load.
AS one moves toward the low frequency end, one finds the T&S parameters for a given box and cone size change towards heavier cones and stronger motors . AS with electrical filters (as a speaker is), as one lowers the frequency, the size of the inductors and capacitors increase.
This does not make them bad, just lower frequency parts.Until the Voltage source convention was adopted in the 50-60’s, it was difficult to “design” (as in predict what would happen in a given design) bass systems.
Sealed and Vented boxes both depend heavily on Electromagnetic damping to set the shape of the LF corner and with every amp being different it was iffy.
I had several Tube amps in the old days that had Damping adjustments, which allowed one to tune the shape by adjusting the source impedance.
Going to a (very low impedance) Voltage source made the goal at least a simple one and with the “cure all” (an entirely different rant) of –fb, making the impedance much lower was easy..
Once the amplifier output Z was about 1/20 the load, essentially very little further change was noticed but bigger numbers here sell.
Loudspeakers designed for this operation would be a tough load at any frequency and cause changes in the frequency response if driven by a significant source Z just as you point out.
No modern bass speaker would be happy on such an amplifier as they assume a low Z source and have significantly altered frequency response etc if driven through even a pureMagnepans are a very resistive load, but from the same thing which makes them low efficiency as well.
On the plus side, one could find box alignments which are over damped when voltage driver but “flat” when driver through the higher Z.
On the plus side too, a significant source Z also extends the HF of many speakers (when the roll off was the R / L corner, now Rs+Rdc / Le)
For a horn (at least with Leach’s math) one can also specify the source Z and arrive at the suitable parameters).I would agree load induced effects are a really important thing if one has a significant source Z, one the other hand, it is largely resolvable by accounting for it from the beginning.
I have enjoyed the discussion.
Cheers,Tom
The point is that a motor with high-mass may be good for VLF-band applications when used with amplifiers having high current sinking ability. But it is clearly not the best choice when the subsystem must be used at higher frequencies or with amps having source impedance that is relatively high, such as is the case with tube amplifiers. Those are situations where diaphragm mass and voice coil reactance become limiting factors. Motor strength is a function of current, and if the amp cannot source and sink a lot of current then cone can't be controlled. It is essentially made into a much less powerful motor with a very heavy cone.Honestly, the 4 ohm impedance and high mass of the LAB12 woofer should steer any tube guys away from the driver, but most of them use high-current amps for the bottom end anyway. Even drivers with much lower mass tend to peak down low in basshorns, but the 4 ohm high-mass thing is just an extreme case. Still, I don't see many people actually running tubes for the bass subsystem - Even the biggest proponents of tube/horn technologies sneak in a transistor amp with EQ for the bottom end because of these very issues.
There are many factors affecting power tolerance and the ability of a driver to thermally dissipate. Hi power tolerant drivers among other things have a heavier VC/cone assemblies. The VCs are often made out of metal sometimes with metal dust covers. The cones are built heavier to prevent flexing and star scars under hi power. Keeping all other horn design parameters constant other than power/SPL this extra mass should affect the resolution and clarity...especially under lower power conditions? Granted horns with their more limited cone excursion will be less affected than other cab designs but nevertheless should be affected to some degree. Keeping the same driver motor mechanism it is easier to accelerate and stop a lighter cone/vc than a heavier one. Yes you can make an 18 wheeler truck out accelerate and out brake a Corvette but not using the same size engine and same size brakes!! Now I am only referring to midbass, MR and higher freq here for bass and subs I can definitely see the benefit from the power tolerance and weight since literally you need to put out watts of power. I understand that in the world of sound reinforcement power and SPL output is the priority not necessarily ultimate resolution or is it that in horns cone motion is so limited that momentum plays a smaller role?? I do notice that Altec drivers designed for horns had massive motors light VC/cones had excellent resolution but did put out little bass power. The question then is are we sacrificing resolution at the expense of higher power/SPL output???Roland
Hi All, very interesting thread!Tom, based on your recomendations here, I obtained Leach's A.E.S. paper "On the Specification of Moving-Coil Drivers for Low-Frequency Horn-Loaded Loudspeakers". I have read through it a few times and I intended to do a comparison betweem it and the Keele and Small papers, so the comments by all here have saved me some considerable time. I originally wondered why I had'nt orderd Leach's paper earlier, but I guess it originally seemed a bit redundant with the earlier Small and Keele papers when you are perusing all of them just by the titles, it of course is not when you read it! Leach's approach departs from the elecroacoustical circuit models of Beranek and Olson in 2 ways: namely in that it uses a gyrator model for the voice coil of the driver to eliminate the combining of parallel elements; and that the output impedance of the driving source is assumed to be a negligable factor in Leach's approach as applicable to "modern amplifiers" (circa 1979). This last factor is of interest to us in 2003, as Leach states that "...the early analysis of horn loudspeakers driven from amplifiers of moderately high output impedances calculated the efficiency as the ratio of acoustic output power to maximum power available from the source.", and further from this that: "The maximum available source power is inversely proportional to the source output impedance.", and thusly that "...the maximum vailable output power is a meaningless specification." You don't have to read between the lines too much to realize that in 1979 "modern amplifiers" meant bi-polar transistor amps with considerable negative feedback (the power mosfets were just arriving on the scene), and amplifiers with "moderately high output impedances" meant vacuum tube amps which were definitely on the wane in the late 70's, except for a fringe group here, and the Japanese of course. It would seem that this is where the two design protocalls make a fork in the road. Now jumping in the time machine and going back to 1977 and telling a group of audiophiles that in 2003 audiophiles would be interested in single ended triode tube amps with 8 Watts or less of power would have got you a free fitting for a nice white jacket with dead end sleeves. The prevailing philosphy would have held that you should use the largest power amplifier available, and operate it at the lowest level possible where distortion would be lowest (making the various 200 W+ S.S. amps the best amps). The primary interest in horns in the late '70's was toward the exploding pro sound/rock concdert market. EV was still making Patricians but sending almost all of them to Japan. I started assembling my horns in '77 with the intention of using them for a sound system for rent, so I'm as guilty as anyone for flowing with the spirit of the times.
Anyway, the 2 (3?) approaches produce different results. I guess I would tend to agree with Wayne that neither one is correct in an absolute sense, but the Leach approach seems to be biased toward the pro sound persuasion, with Dr. Edgar's biased towards the hi-fi end. BTW as I recall Dr. Edgars's problems with Leach seemed to be based on the fact that you could'nt easily substitute different mouth sizes into the math, and that Edgar found that the formula for the back chamber was not accurate, but I don't have the references handy at the moment.
Your thoughts on all this would be most interesting as always!
Paul
Let me remind you of the 1994 work done by Dr. Putland, which references prior work of Dr. Leach. You and I discussed this in an earlier thread, just a few days ago.There's a link to Putland's paper in the post called " Webster Equation ." It's a long but very interesting read, and he goes to great lengths to define and discuss the one-parameter definition so important in discussions of the Webster equation. Putland's mathematical models are represented with finite-difference equivalent circuit methods, and implemented in C code and with Spice .
As an aside, if you're in the information sector, you'll find it fun to see the older C statements rather than the C++ and Java objects that everyone uses these days. Just like the 80's, there are strncpy , fget and fprintf statements all over the place.
Hi WayneYou're a stern task master! I have'nt forgotten, and I'm still working my way through Putland's paper as it's not exactly an easy read. I'm about half way now, and I'm trying to get the lay of the forest. The old programing references are as lost on me as they would be on a monkey though :) I promise I'll get through it!
Paul
Didn't mean to come across in a negative way. I just wanted to interject Dr. Putland's paper here because I feel it contains some really good new mathematical models.
Hi WayneWhen I first saw that paper and looked at the fact that it was over 300 pages long I thought “oh great, probably some guy’s thesis paper” but went ahead and downloaded it anyway.
Looking at it now I am struck by the same “oh great, some guys thesis paper”.I would ask, what did you learn from reading it?, I mean how did this help you become a better horn designer?
I didn’t see much of use to the DIY’r to be honest, also the Leach paper he refers to IS NOT the one we were talking about in this thread on horn and driver matching
He does have some things to say about constant directivity horns (chapter 5.6) which might be interesting reading for those laboring under the notion that conical horns have a “funny” pressure distribution.I also down loaded the hornresp program and played with it a bit, this I would strongly agree with you could be a very useful tool for the DIY’r (assuming it is accurate) as one can model what you really have.
The weakness of both Keeles and Leach’s math is that it assumes a full size horn, anything else and things change and “different “ conditions give better results.
Being able to make a physical model is the best thing in this case.
Cheers,Tom
I'm surprised you would shrug off this paper with such disregard. The models included are excellent. And you're right that Putland doesn't stop with the subject of horns. He also models other acoustic chambers, devices and properties. That's where he references Leach.This particular paper isn't one that would interest most do-it-yourself horn builders. But for those that are technically inclined, it is very interesting. The C code is valuable for making Spice models too. He doesn't merely discuss a topic and describe how it would be implemented. Compilable source to construct his models is included in the document.
So if you see this as "just another thesis paper," I suppose you're free to do that. But I suggest it's more significant than that. The participants here are implementing solutions using the mathematical models that others have written. But Putland - like Keele and Leach - has derived useful mathematical models, which engineers are then able to use.
Hi WayneLike most thesis papers it is long and involved, sort of a tour de-force in technicality that is designed to maximize the impression on professors and not necessarily be a practical or easily useable document.
Seldom are such papers of actual “landmark status” regardless of the number of pages or greek figures, this is someone who is just finishing the education process after all, producing a document that says “see how smart I am” to the ones who educated him.
To be sure it is an impressive work, beyond my doing but I wonder what he does for a living now.
Perhaps you just got more actual useful information from it than I did I don’t know.
To be fair I will slug through it another time.I do have a great respect for Marshal Leach even though I never met him, the fellow I had doing the scientific math stuff for me at Intersonics was a student of his when in college, Marshal is still a professor at Georgia tech I think.
I had written his math first in GWBasic and then when I got assigned a mathematician for another task.
I had Dan write Marshal’s math in a couple different forms in Mathcad for finding horn /driver stuff.
Later, I had him modify a different very involved acoustic transmission line program for modeling passageways in the acoustic levitators for loudspeaker use.
This effort lead to his paper on horn modeling being a published AES paper and then appeared in the Journal issue on horn modeling some years back. If you have back issues look for Dan-Mapes Riordan as author.
These tools are what I have used since then although I do use Spice like simulation for electrical stuff.
I have not compared his models to my program but I know Marshal also has Spice models for horns fwiw.For the DIY’r, I would say that the M.L. math in an easy to use form combined with hornresp would get most people on target, much faster, than they would be using this paper.
Certainly academia isn't immune to egos and "arm waving." But I often find even more "spin" in the private sector. It's all about public perception, isn't it Tom?
Hi WayneIf you mean “industry” or the sound business, I would say for sure.
It is a sad truism that a dollar of marketing produces more profit than a dollar spent in R&D.
We wouldn’t be able to joke about stupidly expensive boutique cables and so on if it weren’t true.At least in academia lying is frowned on, on the other hand Madison ave. types have unfortunately diluted many words to the point of meaninglessness and at times down right lie for profit.
This “profit at all costs” mentality is why we have Best Buy’s and Circuit City’s etc (not to focus on them specifically) stocked full of glitzy looking stuff that breaks in a few years, is not intended to be repaired and is environmentally costly to produce.In this discussion though, we were speaking of two design approaches which give different results given the same target response and driver.
It is easy enough for anyone to build and measure either design, I am simply saying that generally Leach’s approach gives measured results that are very close to the design inputs and Keeles less so.
Cheers,Tom
A couple of different horn design approaches have been discussed on this thread. That, in itself, makes for a very interesting discussion. Specifically, we've spoken about Leach's and Keele's mathematical models, and you've expressed an opinion - perhaps an observation - that Leech's methods were superior. But others have expressed the opposite view.I particularly like Leach's Spice models for horns and for electrodynamic and piezoelectric drivers. As you might guess, that's right up my alley and I've enjoyed making Spice models for analysis of my circuit designs, even going so far as to build analog computers to implement similar models in the analog realm, and reading their results with A/D converters.
I tell you this to let you know I'm with you on that part. I'm not discounting your appreciation of Leach's horn models. I don't take issue with your choice of high-mass diaphragms either. What I'm focused on is the issue of "spin" and your choice of the phrase "folk lore math." When people have recommended horns with highly damped motors and light cones, you often reply with the thing about "folk lore math." But there are very good reasons a person might want to design a horn with exactly that kind of motor.
So I don't find your approach of using a heavy diaphragm to be particularly unappealing, but I do think it means that the horn will be mass dominated. Having a lighter mass would obviously result in a design that is less so, and there's merit in making that choice. But the point is that this discussion isn't really just about Keele vs. Leach. For you, it's as much about defending your choice to use heavy cones vs light ones, and it's about using motors with higher Q - Stuff that really needs "engine breaking" from the current sinking ability of the amp.
Then again, you and I both know all too well how difficult it is to sit and watch the "other guys" doing their share of stretching the facts. I've sat and watched bogus demonstrations of the wavefront inside various horns, completely misrepresenting the shape of the pressure wave as it travelled down the flare. I've spoken with men who claimed not to find a resonant peak in their measurements of a series circuit containing a single coil and a single capacitor. And I've heard 'em claim that what amounts to an undersized basshorn will be flat down to 25Hz without the benefit of corner boundary conditions. Of course, the "fine print" shows the requirement for a powerful amplifier and a ton of EQ to bring up the bottom end. So that's spin too.
But putting all that aside, I'm not sure it is appropriate to talk of absolutes in this industry. After all, it is really an illusion that we are trying to achieve here - To reproduce sound as convincingly as possible. And there's room for plenty of different implementations - As you've said yourself, there are a continuum of solutions.
I did a little reading of that paper today, too; but I still haven't been able to read Keele's preprint on Optimum throat sizes. However, in in other papers Keele mentioned he optimized for efficiency and reduction of reflections. Upon reading Leach, it appears that he is optimizing bandwidth along with sensitivity. So it may be that none are wrong. They just have different priorities. Such is engineering. After playing with Leach's equations, one finds that St is proportional to 1/(a+1). Thus, as St goes up, Vb must go up, and a, Fc, and Fl*Fh go down. Therefore, Fh goes down with Keele's increased St and Vb, and diaphragm displacement is slightly higher than Leach. However, the horn is able to be shorter, and folding geometry has more flexibility. After modeling with both approaches in Hornresponse with the Evm15L driver using Leach's throat size and Keele's throat size, and keeping everything else identical except necessary changes in Vb and horn length, Keele's model had a much smoother response with less ripple in the FR, less phase anomolies in both the pass band and especially out of band, and less reflections showing up as impedance spikes. In both models, diaphragm travel stayed below about 10% of linear xmax through the entire frequency range. Thus, Keele's appeared to be a "cleaner" output, and Leach the wider bandwidth. The size of Vb could become a factor with Keele, though, since it is bigger. Don't know about peak power ability, but in the home, it would not be as big a deal as for PA. The other thing I'm not yet clear about it the slope at Fh area, the phase response there, and the implications for crossovers. Leach's seems to have a two pole critically damped response there. Don't know about Keele's. For home hifi, though, Keele's loading sure looked prettier with the Evm15L, if one is comfortable with the bandwidth that is down 3db at 500-550 area.
P.On diging through my AES papers to reference your comments, I realized I had grabed the wrong one: Optimum Horn Mouth Size (not Throat). Anyway, it's interesting from the point of view of the present discussion that the said "Mouth" paper arose from some computer modeling studies which surprisingly contradicted then current theories, namely that for maximum efficiency and minimum mouth reflections the finite (exponential) horn's mouth size must be equal to or larger than a certain fraction of the wavelength at the cutoff frequency, which was assumed to be the size where the circumference is equal to the horn cutoff frequency wave length (still a rule of thumb today), and from this that the implicit assumption that you can aproach arbitrarily close to the ideal infinite horn operation just by making the finite horn's mouth and length larger and larger, was found by Keele to be false. Keele's paper presented evidence that this was so, and that further there was a definite optimum mouth size which would minimize reflections from the horn's mouth. The optimum mouth size was found to be very dependent on the solid angle into which the horn radiates (this is all paraphrased from Keele's Intro). Thanks for the opportunity for the progressive optimization of (my) geriatric equiptment.
Twas ever thus!
Paul
I really goofed on that one. I mixed things up, too. I don't have the Optimum Mouth Size paper, so not having read it, misread the title when looking for the optimum throat size rationale. Keele's optimum throat size is mentioned in the paper below, which I did find. It is based on optimum EFFICIENCY. It has nothing to do with mouth reflections. Sorry for all the confusion. Keele does state that it may not be best for optimum bandwidth. I'll have to study the paper to see if it is actually optimum efficiency, as opposed to optimum sensitivity, as Leech distinguishes in his paper. However, as stated by Wayne, their is a continuum of optimizations. This does not mean that Keele was wrong. His was just optimizing for a particular condition. You have to trade off. If bandwidth is your bag, Keele's optimization is not the way to go. For me, I'd like to get high sensitivity so as not to have to pad the midrange and tweeter down so much to match the bass level, but also keep the horn as short as practical, to minimize time delay issues. A straight horn 3 1/2-4 feet into the room isn't too desirable, both for appearance reasons, and that you start to lose reinforcement from the back wall in a 1/4 horn. Thus, folding is in order, making time alignment not a simple matter of moving the midrange/tweeter to the back.
I think Keele's equation for optimum throat size appeared in "Low Frequency horn design Using Thiele/Small Driver Parameters," AES Preprint #1250, 1977.
Hi GuysA few further thoughts.
First, understand I /we do not sell anything at all to the DIY and I do not particularly want to upset someone else’s business in this area or upset people who are very comfortable with what they “know” to be true. This was for those curious enough to consider or try something else.
I responded to your inquiry about your horn project based on my experience designing horns and acoustical stuff for most of my life for a living and having been amused at the approach used to design horns for some time. I am a DIY'r at heart as well.
Part of the LAB sub project was in fact to show the Pro Sound folks that the same math was just as inappropriate in Pro sound and that one really could make a very good LF horn using a different (Leach’s) approach. Not only that but the drivers selected would not appear to be proper using Keeles math.
Believe me I would not have gone on that forum (with something like 50,000 Hits / month) and told everyone that they could make a horn subwoofer more powerful than about anything they could buy (from all the big companies) if I wasn’t sure I could design one, thankfully Leachs math held up again.I had enough track record using both Don’s and Marshal’s approach and my own modeling tools to have felt comfortable with what I was saying and I do have some history making horns (see link for example of some of my earliest horns, made by thumb rule)
You or anyone else is of course free to use Don’s math, Marshal’s math, your own or even none, some have been happy with the results using ears alone with no measurement or theory.My point was that in your design, you asked for a 500Hz upper cutoff, so far as proper horn operation the one using Keele’s math results in one which the horn load poops out at about 300-350Hz which is not a 500Hz horn. Keeles alignment allows too much mobility over too little BW to be a 500Hz horn.
Marshal’s math does give a high cutoff in the ball park of what was asked for.Yes the Keele alignment "works", any driver on any horn does something too but look more carefully.
In the pass band, ideally one wants a resistive impedance of about 2 X the Rdc (which is about a 50% efficiency and 30% would be more like Rdc + 30% of Rdc)
In the Keele design, the average impedance in the passband and is less resistively controlled (more average reactance, less resistive loading), for the Leach alignment, the passband Z is closer to a uniform level although the peaks suggest the horn (both cases) is not terminated properly and that is more visible (due to the greater loading on the cone)on the Leach alignment.
Want to try an alignment more like the Keele, use the Leach paper and design for a 300Hz cutoff instead of 500Hz.More importantly for the design stage, since the Leach formula produces results which are closely related to desired outcome it can also be used to zero in on what really would be the best driver for this job as well.
If your desire is the best performance then that needs to be part of the goal from the beginning including the selection of a driver and the size of the horn.So far as folding, the smaller dimensions at the throat make the Leach horn somewhat easier to fold (at that point) other wise they are the same.
So far as path length, this horn (81 cm) is acoustically too short, it needs to be about ¼ wl minimum at the low cutoff and the mouth size should be large enough to load sufficiently at the low end (which it would seem a bit small)
That is why I asked if you had an existing horn (which then can be modeled physically and then the best driver selected based on the real thing) or you were designing a horn for 80 to 500Hz and “had” to use the EV drivers.So far as “pretty” you are comparing two different horns, one having a 350Hz efficient cutoff and the other having one around 500Hz.
So far as “cleaner” I would disagree based on experience with measuring real horns, the greater compression ratio means less distortion (generally) using the same cone driver over the same f band. Motor linearity is the issue.
Having the efficient range of operation so much lower than the acoustical high cutoff (Keele) also means that more of the harmonic distortion will come through and out the horn mouth. If one is looking for a particular sound (coloration) then this may be a good thing but as a faithful reproducer it is less so.Yes Pro-sound is a different area, there is a premium on output, reliability and pattern control. At the levels and quanity the equipment is used at, it is often not as good as even a junky home hifi. Here self-interference can even turn a pile if individually good sounding speakers into a musical roar (although self interference is normally also present at Xover on most multi way speakers).
On the other hand, at home hifi levels, often Pro drivers and systems have far less distortion than hifi parts. The EVM15 is I think a Pro driver for example and most of the compression drivers people use are as well. Running at –20dB FS can make a big difference.
My interest IS and always has been hi-fi, although for larger groups of people most of the time. The company I work for does not spend market much marketing, our sales come from having products that work and win most side by side comparisons with our competition. No BS.
The stuff wouldn’t work as well if I had continued to use Don’s approach, it would still work though.To be honest, I do have a soft spot for someone who wants to build a speaker himself or herself and you are talking about making sawdust, and is why I replied.
Since proof is in the pudding, lets re-state your project goals.You want a horn, which is well behaved (in all respects) from 80 to 500Hz, yes?
You have wood rough cut but are not stuck on a flare rate, “t” or a particular size other than "not too big", yes?
You have some 15” EV drivers which you would like to use but are not 100% stuck on that either, yes?
You do not necessarily need “Pro-sound” sound levels (a few acoustic Watts is enough) but would like low distortion, yes?If these are true, how about, just for fun, I look at this with a clean sheet of paper and see what happens.
I will limit the design to “off the shelf” drivers which you will probably be able to buy new, selling your EV’s .Does this sound interesting?
Tom
You appear to be modeling free space horns. I am looking for 1/4 horn.
Sensitivity of 107-108db or more would be preferable, also.
HiYep, my quick model was a full space horn comparison.
This is for a corner horn then?
How big a box is the current wood cut for?
Tom
against floor and back wall.Wood is cut for a 22.5" x 25.25" mouth.
I agree that the EVM 15L is not ideal here. I think one of the problems is low Fs. I understand it's desirable to have Fs near the horn flare frequency. Wood is cut for a t=.665 and length of 32 or so inches.
One of the system issues is time alignment. Get the horn too long, and it gets more difficult to time align with the midrange, especially in the home setting.
You speak much of distortion. I fully agree that reducing it is ideal, and you are much more concerned about it in a pro setting. However, in a study of studio monitors by Newell and Holland, "Round the Horn", Studio Sound, March 1994, pp.59-70 (Reprinted in Speaker Builder, 8/94), they compared various styles of loudspeakers and tried to group them according to sound. Included in the study were the original Quads, horns, and cones. They found little correlation with regard to harmonic distortion or phase; mixed results on freqency response; and mixed, although generally good, correlation by horn length.
What they did find was a really good correlation when cepstral analysis was used as a comparison. The power cepstrum of a transfer function is the Fourier transform of the log of the transfer functions's amplitude. The plots are very effective in showing reflections existing as spikes aling the tine/distance axis. Thus, mouth or other reflections in a longer horn would be more delayed than that of a shorter horn, or one with fewer impedance discontinuities. Obviously, reduction of impedance discontinuities, especially at the mouth, appears critical to good sound. This is why I originally was interested in Keele's minimum reflection approach. I'm not knocking Leach. Leach is a better bandpass. What I'm trying to analyze, though, is what really matters, sound-wise, in the home environment. Is time alignment critical at 500Hz? I've heard it is. Long horns make this more difficult. It the natural rolloff of Keele at the high end good or bad for crossing over to the higher band? In other words, if Keele is down 3-db at 500Hz, is this a bad thing or a good thing for crossover? Keele would clearly have a different phase character at rolloff than Leach. Is it better to have bandwidth extended well above crossover and use a crossover, or allow the natural rolloff without one? This would need to be modeled also for a system integration standpoint.
Hi Tom,Very informative post, and timely. As I mention to Pooge below, I also have some EVM15L's (and EVM12L's) and have been working on an 80 - 500Hz flare to match my LABs. I'd be very appreciative of any assistance with a design too, hopefully one that works with the EVM, but I'll take it as a learning process either way.
There's no sawdust here yet.Looking forward to seeing what you come up with.
Thanks.
nt
That impedance reflection I saw in the Leach model is not due to the horn length, apparently. It was due to the length of rear chamber. It seems that shortening the length of the rear chamber is a benefit. Shortening it appears to raise the frequency aberations from its reflection way up out of band.Further observations: Keele Vb appears to be around double or more of Leach's; Keele's diaphragm displacement appears to be 2-3 times that of Leach around 100Hz, but very close to Leach at FL frequency (yet still only 10% of xmax); and Keele's frequency response is smoother at low cutoff (probably due to his minimum reflection optimization). The cost again is FH lowering, and Vb increase.
Edgar teaches to adjust Vb to make driver resonance frequency equal to Fo, or the flare frequency. While this optimizes FL to lowest frequency, it also lowers FH. Perhaps a tweak of this routine is to watch your bandwidth. In other words, look at Fh also while you tune Vb, so that the 3-db down points of both FL and Fh are where you want them. If Fh is higher than needed, then adjust Vb for lower FL, as long as you don't run out of physical volume to increase Vb.
Since Keele's St is double that of Leach, the compression ratio will also be reduced. This will have its own advantages and disadvantages.
Anyone know Le for EVM15L?"I am not familiar with hornresp although I have heard of it.
I don’t know if it can show the electrical impedance or radiator motion but if so, the difference between the Keele and Leach alignments is clear."Hornresp appears to do a Leach model, and it can show electrical impedance and radiator motion, as well as acoustic phase.
"A quick physical model of your driver and horn (Keele alignment) shows that while the output goes up above 900 Hz, the electrical impedance shows that the efficient range of operation is below ~ 350Hz. The Leach alignment shows a somewhat lower efficiency (predicted to be about 30%) based on the impedance but the acoustic loading is present on the motor to above 500Hz."
Could this Keele alignment therefore be beneficial as a high pass without a crossover?
"Also, as a result of a smaller throat, one finds the cone motion is significantly less than the Keele alignment for a given SPL. Since motor non-linearity is the primary source of distortion in a LF / MF horn, one could expect that the Leach alignment with the same driver will have less harmonic distortion at that given SPL. If driven by a passive xover or tube amplifier, having a broader range of acoustic loading means there is also less SPL variation due to a changing load impedance."
I realize less motion is better. Is there a good rule of judging when there is too much via a model? I might be able to lengthen the horn a bit and shrink St some.
"In each case, the horn is a little too small (for 80Hz) and in both cases, an increase in front volume (over the computer calc for an ideal horn) lowers the hf corner acoustically down to the crossover region, such a move also lowers the distortion as it acts as an acoustic low pass filter. To be accurate here, one needs the real Le."
What is front volume? Is this just another way of stating length? If so, isn't this what the Keele alignment reduces?
"I am not clear on a few things though.
Do you have a horn built which is 81 cm long ? (in which case the best bet would be to try to find the best driver for your horn based on its physical size)."No.
"Or, are you wanting to build an 80Hz horn for your EV drivers?"
Yes. I don't have a horn built. I can always get rid of the drivers and start over. But I do have the drivers, and panels rough cut. As I stated, I might be able to lengthen the horn and reduce St to somewhere between Keele and Leach. Before I scrap the project, I'd like to find out if I can still salvage the driver, wood, and folding plans. Leach models a smaller Vb, so I can move the driver back into this space and lengthen the horn somewhat. Also, I can move the lower cutoff up some, if it would help, although having Fs so low seems to make Va larger. The Keele model shows very flat FR, but I am concerned about maximum output and distortion. The diaphragm motion does peak in the Keele model. As I understand Keele's throat model, though, it optimizes mouth reflections. This would seem to be an admirable quality.
I can start over with the Leach model and/or with a new driver. Any driver recommendations?
However, I would like to see what best can be done with the present driver with an 80 Hz mouth 1/4-size. Understanding the max output and distortion of the Keele alignment would therefore aid a decision of whether or not it is "good enough". 107cm is the 1/4 wavelength of 80Hz. You have said to add up to .6*(mouth diameter) to actual length. At 81cm actual, and some slight lengthening, I might be able to acheive close to 107cm acoustical length. The question is, when does it become "good enough". I can see that Leach's Vb is much smaller than Keele's, and has less diaphragm motion. But Keele is supposed to have less mouth reflections. Then there's every alignment in between the two. Is there any rules for predicting the audible consequences between the alignments. In other words, if Keele has twice the distortion of Leach, will it even make an audible difference?
Anyone know Le for EVM15L?< I don’t know if it can show the electrical impedance or radiator motion but if so, the difference between the Keele and Leach alignments is clear.> >
Hornresp appears to do a Leach model, and it can show electrical impeadance and radiator motion, as well as acoustic phase.
According to the Drivers Ed database it's 1.24mH
Thanks! Where is that database found?
Thanks!Glad to help. This is an interesting thread, as I have EVM15, EVM12 and K140's to use to make an 80-500Hz midbass horn like the one you're talking about.
Where is that database found?
Link below; it's a download.
The Altec cab straight A4 horn measures well 80-500 Hz or higher with 108-110Db sensitivity with two altec 515-8G drivers. This on top of labhorn enclosures should provide the often requested connection to the MR/tweet horns. By the way those BMS coaxial neo magnet drivers sure look good!! The A4 cab can be cut down to 5 foot...only use the horn or if you want leave 6" below horn as reflex vent opening it will augment the bass below 80Hz..need to leave both horn flares open to provide 10 cu ft/driver bass reflex space.
HiGosh I love those Altec horns, I really have a soft spot for those monsters.
Way back when at the Aragon Ballroom in Chicago there was a sound company that had a wall of them with a row of A7’s on top for good measure.
I even made a smaller “copy” version a million years ago with 2 12’s back in the olden days. See the link but don't laugh too much, this was a long time ago.
Cheers,Tom
http://www.prosoundweb.com/lsi/hist/water.php
Those were fun days Tom!! I also played guitar in a band built their PA gear. Still have tons of good old gear...Fenders, Marshalls echoplexes, old synths...Was a hippie till I got hit in my lippie and eventually became an M.D. Now I build PAs for my tennis court/performing stage and have my own AV room/recording studio. What goes around comes around! Can not get rid of the Hi efficiency
speaker bug.Cheers
Roland
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
