Channel Folds

[sunnysal]

D-man, I hope mr. delgado chimes in but I assume that this is a bit of a non-issue since the JAES article clearly showed smoother and more extended upper limits for this design versus the K-horn, IIRC up to 1khz in the jubilee design so it would seem that they solved the "problem" mentioned or proved it was secondary. tony

[D-MAN]

Tony, yes, I was wondering why they would use a full channel splitter to effect a 90 degree turn and not on the next 90 degree turn(s). They know something... and I want to know what it is!

Dana

[rigma]

Bump. I am also curious about the pros & cons of the full channel reflector.

[D-MAN]

The general consensus from a conversation on AA with Fitzmaurice is that for frequencies above 300Hz, the full channel reflector should be used.

Below that, a radius (partial) reflector is fine.

Below 100, no reflectors are needed.

It depends on the upper bandpass that you want and the mass rolloff of the driver in question.

With that in mind, my personal opinion is that the full-channel reflector should be used in most cases, except subwoofers where it has no effect except to add weight to the enclosure.

DM

[jwc]

Yes I too would love to hear some input on this. When Dana suggested I place this in the Jubilee clone, I read up on it a little and couldn't find anything to suggest NOT putting it there. I was pleased with my results using the same drivers as the Jubilee.

Raw bass bin. I skipped a few frequencies to speed up my testing. The only dip really is the one around 200Hz that has been chronicled enough on another thread.

If I were to build another pair...I would again put the full reflectors in there.

jc

[DrWho]

Perhaps Roy and crew measured with and without reflectors and the difference was negligible? Or even better without? I would imagine Paul and Roy would have played quite a bit with the bass bin trying to optimize the new flagship. Theory only goes so far and assumes quite a bit to make the math work out nice, which is where measurement comes into play...

[rigma]

Yea, I wish Roy would chime in with at least a comment which he thinks is better[:$]

[D-MAN]

It only REALLY matters which approach YOU think is better!

Myself, I would also go with full-channel reflectors in the Jubilee front corners, since its design is particularily concerned with a high mid-and-upper bass frequency bandpass.

However, it certainly is not a show-stopper if its a radius-style reflector instead, as in the AES drawing.

DM

[bodcaw boy]

Been researching horn channel folds as of late, and I have a question for Roy who helped design the Jubilee.

The Jubilee AES paper discloses the use of a full-channel-width hard-surface 45 degree splitting wedge at the throat which accomplishes the 90 degree turn into the horizontal throat channels, which we all know is a good thing.

And the 180 degree fold at the tailboard is picture-perfect. The attention these particular folds received (and change from previous PWK designs) brings up this question:

Why was a partial-channel-width RADIUS technique (not so good) used at the front corner 90 degree folds? It seems like quite incongruent to me. It is clear that it was a purposeful choice, I was wondering what the reason for it was...

Thanks,

Dana

hi dana,

good question. first, think fluid not light rays. acoustic sound waves behave more like a fluid than a light ray and, although, at higher freqs, sound waves can bounce off of walls. they are acting that way because the wavelenths and their ability to "spread" (again, this one is for doc, think bubble) can get directional. but it has to work at all freqs.

so with that mindset, the idea is to turn and guide the bubble without disrupting it or causing it to break. the ideal and something paul and i really investigated whether it was feasbile (plastic tooling is expensive) is to have a curve in each bend. and so how do you create each curve? it is very important to stick to the area expansion as dictated by the equation that you are using.

thinking bubble again, as the freq goes up, the ability for the bubble to make the turn in a time period as related to wavelength gets shorter and shorter and so you get to the point that wave is going to low pressure faster than the bubble is completing the turn. thus the bubble collaspes and this is what determines the high freq limitation of a bent horn. it can be easily calculated by taking the path length difference in the turn. outside length of the turn minus inside lenght of the horn and then divide by the speed of sound, divided by 2 (i think if i remember correctly).

because of cost constraints, we added enough boards to approximate the curve as much as we could without really complicating the build.

sorry so long, but you asked...

roy delgado

[Islander]

Light waves are transverse waves, while sound waves are longitudinal waves, or compression waves. Using light as an analogy to sound is tempting but can be very misleading, since their behaviour is similar in a few ways, but very different most of the time, depending on wavelength.

Obviously, I'm not directing this to Roy, who works with these concepts every day.

[DrWho]

hi dana,

good question. first, think fluid not light rays. acoustic sound

waves behave more like a fluid than a light ray and, although, at

higher freqs, sound waves can bounce off of walls. they are acting

that way because the wavelenths and their ability to "spread" (again,

this one is for doc, think bubble) can get directional. but it has to

work at all freqs.

so with that mindset, the idea is to turn and guide the bubble

without disrupting it or causing it to break. the ideal and something

paul and i really investigated whether it was feasbile (plastic tooling

is expensive) is to have a curve in each bend. and so how do you

create each curve? it is very important to stick to the area expansion

as dictated by the equation that you are using.

thinking bubble again, as the freq goes up, the ability for the

bubble to make the turn in a time period as related to wavelength gets

shorter and shorter and so you get to the point that wave is going to

low pressure faster than the bubble is completing the turn. thus the

bubble collaspes and this is what determines the high freq limitation

of a bent horn. it can be easily calculated by taking the path length

difference in the turn. outside length of the turn minus inside lenght

of the horn and then divide by the speed of sound, divided by 2 (i

think if i remember correctly).

because of cost constraints, we added enough boards to approximate

the curve as much as we could without really complicating the build.

sorry so long, but you asked...

roy delgado

I understand the gradual turning

bit, but what about that first 90 degree sharp bend? Does this not

disrupt the "bubble" because the wavelengths (~1 foot at 1kHz) are

much larger than the width of the channel at the bend?

I've been

thinking of building myself a wave table to help visualize the wave

propagation . Will the effects seen in water be fairly accurate on a

smaller scale as long as everything is normalized to wavelength???

Technically, waves in water is still "sound" []

The

exponential expansion assumes that the wave front is a pure plane wave,

but everyone knows this isn't true (which is why the reflectors don't

work as they should). However, they seem to approximate pretty well for

the larger wavelengths. The tractrix expansion assumes a spherical

wave. In reality I would expect the actual shape of the wavefront to be

neither; perhaps more of a cardioid pattern getting more focused as

frequency increases. If that's true, then I wonder how the calculations

for expansion rates change when that's taken into account. [^o)] I

would also presume that the shape of the wavefront is going to change

every time the sound changes direction, thus requiring different

expansion calculations for each section of a bent horn...Or you could

just make life easier and keep the wavefront shape as close as possible

to the assumptions of the area expansion rate being used. Is that what

you mean by "preserving the bubble", Roy? Just trying to get a tangible

hold on what exactly you're referring to. I can see lots of ways to

destroy bubbles, but I think an engineer would be more interested in

finding the ways of preserving them []

I know this is a bit

off-topic, but the 200Hz dip was mentioned in passing (which is the

result of vertical standing waves from the height of the horn at the

exit). Couldn't this be mitigated by horizontal shelfs that divide the

height into multiple sections? Perhaps it's a bad approach because

harmonic distortion will stimulate the resonances? I suppose absorbent

material could then be added, but then you're talking even more cost...

[bodcaw boy]

hi dana,

good question. first, think fluid not light rays. acoustic sound waves behave more like a fluid than a light ray and, although, at higher freqs, sound waves can bounce off of walls. they are acting that way because the wavelenths and their ability to "spread" (again, this one is for doc, think bubble) can get directional. but it has to work at all freqs.

so with that mindset, the idea is to turn and guide the bubble without disrupting it or causing it to break. the ideal and something paul and i really investigated whether it was feasbile (plastic tooling is expensive) is to have a curve in each bend. and so how do you create each curve? it is very important to stick to the area expansion as dictated by the equation that you are using.

thinking bubble again, as the freq goes up, the ability for the bubble to make the turn in a time period as related to wavelength gets shorter and shorter and so you get to the point that wave is going to low pressure faster than the bubble is completing the turn. thus the bubble collaspes and this is what determines the high freq limitation of a bent horn. it can be easily calculated by taking the path length difference in the turn. outside length of the turn minus inside lenght of the horn and then divide by the speed of sound, divided by 2 (i think if i remember correctly).

because of cost constraints, we added enough boards to approximate the curve as much as we could without really complicating the build.

sorry so long, but you asked...

roy delgado

I understand the gradual turning bit, but what about that first 90 degree sharp bend? Does this not disrupt the "bubble" because the wavelengths (~1 foot at 1kHz) are much larger than the width of the channel at the bend?

it's not the width that matters in disrupting the bubble so much as the path length difference in the turn. but that is correct.

I've been thinking of building myself a wave table to help visualize the wave propagation . Will the effects seen in water be fairly accurate on a smaller scale as long as everything is normalized to wavelength??? Technically, waves in water is still "sound" []

actually i have thought of the same thing and water can approximate air but air can be compressed but water can't, but at least you can get a good approximation of what the horn will do in air.

The exponential expansion assumes that the wave front is a pure plane wave, but everyone knows this isn't true (which is why the reflectors don't work as they should). However, they seem to approximate pretty well for the larger wavelengths.

then why do high freq "bounce" off of walls?

The tractrix expansion assumes a spherical wave. In reality I would expect the actual shape of the wavefront to be neither; perhaps more of a cardioid pattern getting more focused as frequency increases. If that's true, then I wonder how the calculations for expansion rates change when that's taken into account. [^o)] I would also presume that the shape of the wavefront is going to change every time the sound changes direction, thus requiring different expansion calculations for each section of a bent horn...Or you could just make life easier and keep the wavefront shape as close as possible to the assumptions of the area expansion rate being used. Is that what you mean by "preserving the bubble", Roy? Just trying to get a tangible hold on what exactly you're referring to. I can see lots of ways to destroy bubbles, but I think an engineer would be more interested in finding the ways of preserving them []

the exponential assumes the wave starts a plane wave and remains a plane wave and the tractrix assumes it starts off as plane and begins to bugle into a spherical wavefront (if i remember correctly). the challenge (one i have been looking at and why we use a modified tractrix equation) is coming up with a continuous equation. the constant directivity concept uses three (most times) very distinct equations to take advantage of the positive qualities of each. however, in the process of doing that, you get some severe discontinuities between the equations. that causes more problems, in my opinion. yes it is best to keep to the area dictated by the equation, in my opinion. all those years of college just to play with clay and bubbles.....[]

I know this is a bit off-topic, but the 200Hz dip was mentioned in passing (which is the result of vertical standing waves from the height of the horn at the exit). Couldn't this be mitigated by horizontal shelfs that divide the height into multiple sections? Perhaps it's a bad approach because harmonic distortion will stimulate the resonances? I suppose absorbent material could then be added, but then you're talking even more cost...

i don't quite understand by what you means by horizontal shelves. ultimately, it comes down to the exit angle in that plane. in order to not increase the complexity of the build, we kept the top and bottom panels flat which means no taper in the vertical.

roy delgado

[rigma]

Hear is what I am thinking. The full channel reflector may extend the high
frequencies but at the expense of efficiency of the lows. So if you are
going to crossover in the 600 range wouldn't it be better to use the partial
reflector to preserve efficiency? Roy am I way off base?

rigma

[bodcaw boy]

Hear is what I am thinking. The full channel reflector may extend the high frequencies but at the expense of efficiency of the lows. So if you are going to crossover in the 600 range wouldn't it be better to use the partial reflector to preserve efficiency? Roy am I way off base?

rigma

well kinda.....you have can have fast flare rate to help the low freq stuff but i am not sure if it improves efficiency. but it is probably best to not radically change the flare rate in the bend. make sense?

roy delgado

[DrWho]

I understand the gradual

turning bit, but what about that first 90 degree sharp bend? Does this

not disrupt the "bubble" because the wavelengths (~1 foot at 1kHz) are

much larger than the width of the channel at the bend?

it's not the width that matters in disrupting the bubble so much

as the path length difference in the turn. but that is correct.

So the inside turn is going to be a length of 0,

and the outside turn is going to be a length around 6"? I guess (1130 /

0.5) / 2 = 1030Hz. [Y]

I've been thinking of building myself a wave

table to help visualize the wave propagation . Will the effects seen in

water be fairly accurate on a smaller scale as long as everything is

normalized to wavelength??? Technically, waves in water is still

"sound" []

actually i have thought of the same thing and water can

approximate air but air can be compressed but water can't, but at least

you can get a good approximation of what the horn will do in air.

The exponential expansion assumes that the wave front is a pure

plane wave, but everyone knows this isn't true (which is why the

reflectors don't work as they should). However, they seem to

approximate pretty well for the larger wavelengths.

then why do high freq "bounce" off of walls?

I thought the Huygen reflector assumed a parallel

wave front? I'm probably not describing it right, but the higher

frequencies don't diffract as strongly around the channel bends -

sending them straight into the boundary where they bounce instead of

wrapping around the bend like the lower wavelengths?

The tractrix expansion assumes a spherical wave.

In reality I would expect the actual shape of the wavefront to be

neither; perhaps more of a cardioid pattern getting more focused as

frequency increases. If that's true, then I wonder how the calculations

for expansion rates change when that's taken into account. [^o)] I

would also presume that the shape of the wavefront is going to change

every time the sound changes direction, thus requiring different

expansion calculations for each section of a bent horn...Or you could

just make life easier and keep the wavefront shape as close as possible

to the assumptions of the area expansion rate being used. Is that what

you mean by "preserving the bubble", Roy? Just trying to get a tangible

hold on what exactly you're referring to. I can see lots of ways to

destroy bubbles, but I think an engineer would be more interested in

finding the ways of preserving them []

the exponential assumes the wave starts a plane wave and remains

a plane wave and the tractrix assumes it starts off as plane and begins

to bugle into a spherical wavefront (if i remember correctly). the

challenge (one i have been looking at and why we use a modified

tractrix equation) is coming up with a continuous equation. the

constant directivity concept uses three (most times) very distinct

equations to take advantage of the positive qualities of each.

however, in the process of doing that, you get some severe

discontinuities between the equations. that causes more problems, in

my opinion. yes it is best to keep to the area dictated by the

equation, in my opinion. all those years of college just to play with

clay and bubbles.....[]

Sounds like someone should sit down and implement some hardcore computer simulations that don't ignore or assume any acoustical behavior...seems the profs are very confident that the fundamental theories are correct - just that the math quickly gets insane (which is what computers are good for). [^o)]

I know this is a bit off-topic, but the 200Hz dip was mentioned in

passing (which is the result of vertical standing waves from the height

of the horn at the exit). Couldn't this be mitigated by horizontal

shelfs that divide the height into multiple sections? Perhaps it's a

bad approach because harmonic distortion will stimulate the resonances?

I suppose absorbent material could then be added, but then you're

talking even more cost...

i don't quite understand by what you means by horizontal

shelves. ultimately, it comes down to the exit angle in that plane.

in order to not increase the complexity of the build, we kept the top

and bottom panels flat which means no taper in the vertical.

Here's a crude sketch as viewed from the front.

Basically, every cross piece in the horn mouth is a shelf extending

straight back as far as needed to keep the free air vertical height

short enough to prevent standing waves. (Yes, I know it's way more

complex to build since they all need to be perfectly parallel while

wrapping around the horn).

[bodcaw boy]

I understand the gradual turning bit, but what about that first 90 degree sharp bend? Does this not disrupt the "bubble" because the wavelengths (~1 foot at 1kHz) are much larger than the width of the channel at the bend?

it's not the width that matters in disrupting the bubble so much as the path length difference in the turn. but that is correct.

So the inside turn is going to be a length of 0, and the outside turn is going to be a length around 6"? I guess (1130 / 0.5) x 2 = 4520kHz. [Y

I've been thinking of building myself a wave table to help visualize the wave propagation . Will the effects seen in water be fairly accurate on a smaller scale as long as everything is normalized to wavelength??? Technically, waves in water is still "sound" []

actually i have thought of the same thing and water can approximate air but air can be compressed but water can't, but at least you can get a good approximation of what the horn will do in air.

The exponential expansion assumes that the wave front is a pure plane wave, but everyone knows this isn't true (which is why the reflectors don't work as they should). However, they seem to approximate pretty well for the larger wavelengths.

then why do high freq "bounce" off of walls?

I thought the Huygen reflector assumed a parallel wave front? I'm probably not describing it right, but the higher frequencies don't diffract as strongly around the channel bends - sending them straight into the boundary where they bounce instead of wrapping around the bend like the lower wavelengths?

The tractrix expansion assumes a spherical wave. In reality I would expect the actual shape of the wavefront to be neither; perhaps more of a cardioid pattern getting more focused as frequency increases. If that's true, then I wonder how the calculations for expansion rates change when that's taken into account. [^o)] I would also presume that the shape of the wavefront is going to change every time the sound changes direction, thus requiring different expansion calculations for each section of a bent horn...Or you could just make life easier and keep the wavefront shape as close as possible to the assumptions of the area expansion rate being used. Is that what you mean by "preserving the bubble", Roy? Just trying to get a tangible hold on what exactly you're referring to. I can see lots of ways to destroy bubbles, but I think an engineer would be more interested in finding the ways of preserving them []

the exponential assumes the wave starts a plane wave and remains a plane wave and the tractrix assumes it starts off as plane and begins to bugle into a spherical wavefront (if i remember correctly). the challenge (one i have been looking at and why we use a modified tractrix equation) is coming up with a continuous equation. the constant directivity concept uses three (most times) very distinct equations to take advantage of the positive qualities of each. however, in the process of doing that, you get some severe discontinuities between the equations. that causes more problems, in my opinion. yes it is best to keep to the area dictated by the equation, in my opinion. all those years of college just to play with clay and bubbles.....[]

Sounds like someone should sit down and implement some hardcore computer simulations that don't ignore or assume any acoustical behavior...seems the profs are very confident that the fundamental theories are correct - just that the math quickly gets insane (which is what computers are good for). [^o)]

I know this is a bit off-topic, but the 200Hz dip was mentioned in passing (which is the result of vertical standing waves from the height of the horn at the exit). Couldn't this be mitigated by horizontal shelfs that divide the height into multiple sections? Perhaps it's a bad approach because harmonic distortion will stimulate the resonances? I suppose absorbent material could then be added, but then you're talking even more cost...

i don't quite understand by what you means by horizontal shelves. ultimately, it comes down to the exit angle in that plane. in order to not increase the complexity of the build, we kept the top and bottom panels flat which means no taper in the vertical.

Here's a crude sketch as viewed from the front. Basically, every cross piece in the horn mouth is a shelf extending straight back as far as needed to keep the free air vertical height short enough to prevent standing waves. (Yes, I know it's way more complex to build since they all need to be perfectly parallel while wrapping around the horn).

the shelves essentially add the waves back at a zero degree angle so they would still see the full mouth and diffract....

roy delgado

[rigma]

Hear is what I am thinking. The full channel reflector may extend the high frequencies but at the expense of efficiency of the lows. So if you are going to crossover in the 600 range wouldn't it be better to use the partial reflector to preserve efficiency? Roy am I way off base?

rigma

well kinda.....you have can have fast flare rate to help the low freq stuff but i am not sure if it improves efficiency. but it is probably best to not radically change the flare rate in the bend. make sense?

roy delgado

Well not really.[:$] Are you saying the partial (45 deg. radius) reflector is the better choice? Does that maintain the flare rate through the bend?

For those who are interested scroll to results at the link below. There are animations of different reflectors and various frequencies.

http://ldsg.snippets.org/HORNS/waves.html

rigma

[bodcaw boy]

Hear is what I am thinking. The full channel reflector may extend the high frequencies but at the expense of efficiency of the lows. So if you are going to crossover in the 600 range wouldn't it be better to use the partial reflector to preserve efficiency? Roy am I way off base?

rigma

well kinda.....you have can have fast flare rate to help the low freq stuff but i am not sure if it improves efficiency. but it is probably best to not radically change the flare rate in the bend. make sense?

roy delgado

Well not really.[:$] Are you saying the partial (45 deg. radius) reflector is the better choice? Does that maintain the flare rate through the bend?

For those who are interested scroll to results at the link below. There are animations of different reflectors and various frequencies.

http://ldsg.snippets.org/HORNS/waves.html

rigma

i saw the website. impressive stuff. what i am saying is that a curved bend whose curvature is dependent on what the area expansion is doing does the best. have you ever analyzed a curved bend?

roy delgado

[Deang]

have you ever analyzed a curved bend?

Does my wife count?

[John Warren]

Take a look at US patent # 2866513. It's the patent for the Millenium 4D rear loaded horn. The 4D was driven by a pair of 15"JBL drivers. I heard a single unit years ago. The build quality was outstanding. A single unit weighed over 350 lbs.

http://www.stan-white.org/a_hist_adv_1.htm

I've discussed the design with White (by email, he's still around). All he said to me was that should I decide to build it, be sure to use the curved panels otherwise it will not perform very well. He had a single 4D unit on display at a the Chicage HiFi Expo and the engineers at Jensen were all over it, later to introduce the Imperial. The patent came out later (1958).

Isn't Janice cute? The costume is great, so typical of the time.

Stan White is a real character, a lesser known from Audio's "golden era".

Maybe I can get him to chime in here(?)

jw

Edit-

Note in the advertisments he specifically states the horns are curled and not folded to within 1% of the design formula.