dipole baffle idea ...

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
will probably look completly stupid
but i have some design i have to expose
( well ..cna't say if i can call it a "design " )

please don't laugh at me :D

http://img185.imageshack.us/img185/2160/dipoledesigna1uy5.jpg


just trying to remove all reflections within an as compact as possible open baffle design and trying to remove front baffle diffraction ??
the outside could be all rounded ..

let me know if i am waisting my time completly
( probably ) or would be worth the effort over a front flat baffle

thanks again all :)
 
it is a block or cylinder...keeps area constant without leaving any paralell walls
( i guess that a driver can't push out at back more air at a time then what it's castin/basket allows to pass through ??? )
and it has aesthetic purpose as open backed loudspeakers don't look too good from the back
( wife tested already ...:D )

mmm ..other than that i don't know what it does there!
:p


but seriously i thought about rear dispersion for higher fs ..but i can't say that it would work ( like waveguide?)
 
I'll have to brush up on it, but I don't think you'll get the classic dipole radiation pattern from it. In fact, due to there being some restriction on the flow out to the rear and a less restricted flow 'up' (unless it's designed to be from floor to ceiling), you might find it's behavious to have some aspects of horn and/or t/line enclosures.

Not sure. Have you built one? Not for real, but out of (say) an old refridgerator cardboard box. It's give you some idea of dispersion pattern...
 
i would think that this would be benificial at a frequency with smaller wavelength than the width of the cabinet.


below that i think you might just end up with a phase or dispersion issue at with the rear as the waves are pressured again at the mouth you have there.

not founded at all just a thought.
 
I agree with your comments.

I think that dipole speakers we usually see are either flat baffle or flat baffle with wings. If we cut the posted drawing and use the front half, it would work in a way similar to flat baffle with wings but with rounded edges at the front. Of course, the radiation pattern, phase would be different from a standard dipole with flat baffles.

The idea is to reduce the on axis, negative effect of diffraction as much as possible. I have been playing with Baffle simulation programs a lot lately. I hate to see the frequency response irregularities (+-2dB typical around the 1k - 4k region where human ears are most sensitive) and delay / phase errors generated by the wide, flat baffle, though according to John K, the diffracted source may be -15dB comparing to the direct radiation from the driver.

I have just got the A B C Dipole from John K and started playing with it. I already found it highly enlightening. I would recommend anyone to get a copy. It has a lot of dipole theory packaged in simple descriptions with illustrated graphs to make them easy to understand, plus John K's open baffle simulation program and more. It is really a bargain for USD$10.95.

I found the only way to reduce the diffraction effect is to make the front baffle as narrow as possible, which has an additional advantage of pushing the first, usually the largest peak and dip higher to the XO point so that it can easily be dealt with by the XO. Of course, most commercial speakers do just that. But to go one step further, a 2 inch radius rounded corners will reduce the peaks and dips to within +-0.2dB, and I would expect the sound be a lot more smoother and detailed. This is what I am aiming at. Of course, this generally applies to closed boxes.

For open baffle, narrow baffle is not practical. Adding angled wings will introduce a lot of unknown problems to be resolved and it does not reduce diffractions. So I am thinking about the narrowest baffle for the midwoofers, then using a 2" large radius rounded corners. The woofers will be centred horizontally. The tweeter will be offset slightly up to 1". With this baffle, the dipole source distance between the front and back are effectively increased over the standard flat baffle, pushing the dipole peak and dipole roll-off lower, making it possible to have less EQ and less demands on the midwoofers. The diffaction effects are hopefully reduced substantially due to the large transitional surfaces. It may no longer be a standard dipole, and the dipole null would be a bit behind the 90 degree axis. The radiation patterns will be changed a bit, and I can expect a stronger, narrower rear radiation, which may not be a bad thing. Below is a quick drawing I just did with MS Paint to illustrate the concept.

An externally hosted image should be here but it was not working when we last tested it.


I don't know if I could still use the A B C Dipole to model this baffle by calculating with an equivalent D. I think I will email John K and invite him to shed some lights on this thread if he has the time. Hey, I have to tell you honestly I am not a salesman for John K. I just admire his work and has built his NaO with much satisfaction - still tuning and playing with the XOs for my experiments.

Don't count on me. I am still learning. If this turns out to be good, where can I find a cabinet maker / wood worker to make the baffle for me!!!!! Any forum mates around Sydney area can help? I will do the electronic part for you for a swap.

Regards,
Bill
 
Your drawing could be done in alot of things other then wood, why do everybody stick to wood as the ONLY material ?

neway ..if you want the exterior to be almost sphere like, you could work probably 40% of you angle on a big lathe turning ... and do the rest with hand

you could also use the glued pannels technique and make all the pannels from 0.75" mdf with the perfect shape and then glue them together from bottom to top


so on to the design ..
the reason why i started this thread, is that i was wondering about the implication of using wings on a dipole

i undestand that it pushes the center of the radiation pattern backward and changes the front SPL pattern,
does it affect dispertion of the drivers also?

Then as said, consdering only the idea and not the actual drawing i made in a few seconds ..
If the area at the back between the "cyllinder" and the drivers output is more than what the driver needs i don't see why it could influence it much ( won't be restricted and shouldn't do some loading ) ????


Lets get a few things straight ..
( well for me, cause you all might know this already)

on a sphere like enclosure as we've all seen on few designs and Apple desktop speakers, is there any problem associated with diffraction ?

Then if we only use a slightly curved surface as per NutNut drawing, what is the implication on the refraction??


then what is the implication on the radiation pattern and accoustics if using wings on a dipole wich alters the dipole center offset ?

do we absolutly want a perfect centered pattern ?
 
JinMTVT,

What are you intending to use these for? Is it a single driver, 2, 3, more? MTM, WMTMW?

I was thinking (although in actual execution it would be hard) of something that looks a little like an hour-glass from the front, but is open backed. The shape, when looking from the top would always be hyperbolic, but would vary depending on where in the hourglass it is.

Sorta like this picture (but I'm lousy at pictures).

But, as I'm lousy at wood-working I wouldn't be able to do this anyway. Maybe if I make friends with someone who does formed plywood moldings...
 

Attachments

  • hg.jpg
    hg.jpg
    17.1 KB · Views: 830
JimMTVT,

You can download the Baffle Diffraction Simulator from the FRC site and play with it. Of course, that is only for closed boxes. For dipole baffle simulation, get the A B C Dipole.

For woofers with larger radiation area, diffraction is usually not a problem especially at lower frequencies. The problem is mainly with the tweeter that has a small radiation area.

Rounded corners reduce diffraction. The radius of the rounded corners is related to the frequencies. If you have a radius of 1", the response is a lot smoother for frequencies above 10k roughly, but it does not do much for frequencies below 10k. But if you have a radius of 2" / 50cm, frequencies from 2k upward can be made very smooth. Larger radius is not necessary.

Regards,
Bill
 
NutNut,

how great is the afflection by diffraction when using and not, radiu corners....and to what % or extent does the 2" radius remove or limit unwanted stuff ??

Cloth : sorry but i don't quite get your drawing ...

I'll try and think about it more
and see what i can come up with to illustrate better what i have in mind ...
 
I don't want to get into a long discussion about what baffle shapes and configurations are best for dipoles or to reduce diffraction. There are so many variations that will work fine, if carefully designed, that it's impossible to really say what is best. It obviously depends on what frequency range and what the intended application is.

For Bill, when I state that the diffraction signal is -15dB or so, remember that this is in reference to the response below the dipole peak. At higher frequency where driver directionality and asymmetric front and rear response enter the picture things can behave more like a typical box speaker baffle. With that in mind I will offer some comments. First, when wings or other baffle edge extensions are added to an open baffle consideration also has to be given to what happens to the acoustic loading of the rear response. Modest wings of short extensions can be modeled in my code but the results will only be meaningful at low frequency.

With regard to diffraction in general I have prepared the following figure. When thinking about diffraction it is probably easier to look at it in the time domain. The figure I posted shows three cases; a flat circular baffle with sharp edge, circular baffle with rounded edge, and a rectangular baffle with offset driver and sharp edges.




An externally hosted image should be here but it was not working when we last tested it.


The first thing to consider is that along the baffle surface a circular wave propagates outward from the source toward the baffle edges. The distance to the baffle edge represents a time delay between the direct radiation and the diffracted signal. If we consider an impulse response, in the first case, since the source is centered on a circular baffle, the surface wave will reach the baffle edge at the same time over the entire perimeter. As a result the diffracted signal is a close replica of the original signal, inverted and reduced in amplitude.

The center picture shows the same circular baffle but with a rounded edge. When the surface wave reaches R1 it will begin to be diffracted. But two things happen. Because the edge is rounded the diffraction doesn't happen all at once. Thus the diffraction is spread over time or smeared compared to the sharp edge. The strength of a diffracted signal is related to the angle it is turned by, and how quickly it is turned, so to speaker. In the share edge case, the signal is turns by 90 degrees all at once. With a rounded edge we can think if it as turning the signal through 10 increment of 9 degrees. If we assume the strength of the diffracted signal varies linearly with turning then the rounded edge is like tuning the signal 10 times, 9 degrees each, with a strength of 1/10 that of the sharp edge. Each of these 9 degree turnings happens sequentially in time as the surface wave propagates around the edge. So what we have done it to reduce a single sharp diffraction pulse to a series of lower amplitude pulses sequentially delayed in time. The result is, when we sum up all the diffraction signals over the edge and around the circumference, a longer period, lower amplitude diffraction signal.

Finally we have a rectangular baffle with sharp edges and offset driver. In this case when the surface wave reaches the baffle edge varies all around the baffle. In this case, at any point around the perimeter, the diffraction signal would be sharp like on the circular baffle, but since the surface wave reaches the baffle edge at different times as we move around the perimeter, when we sum all around the baffle the result is again a time smeared, lower amplitude net diffraction signal.

It really doesn't mater how it is done, the conventional diffraction treatment is to take a high amplitude, short duration result and transform it into a low amplitude, longer duration result in the hopes that the reduced amplitude will make the signal less obvious that the effect of the time smear. How we position the driver on the baffle and how we treat the edge affects the delays (or phase) of the diffraction signals around the baffle edge, and how all the diffraction sources will sum.

The things that have been omitted here is consideration of attenuation due to differences in propagation distance to the baffle edge, and dissipation. Without dissipation the energy contained in the diffracted signal is a fraction of that contained in the original surface wave (based on turning angle) but redistributed in time. To reduce diffraction we need to dissipate the energy of the surface wave before it reaches the edge.
 
If we assume the strength of the diffracted signal varies linearly with turning then the rounded edge is like tuning the signal 10 times, 9 degrees each, with a strength of 1/10 that of the sharp edge.

Hi John,
Within the context of the overall message here, I understand why this assumption is made. However, how accurate is it? I've been working with larger waveguides recently and find diffraction at the mouth to be a huge issue...and the roundover radius required to combat diffraction nearly impractical. In search of optimized size/performance, I wonder if the reflection may be significantly diminished as it travels 'round the bend...not only by distance, but also diffused by the curve.

So, for a large freestanding WG, can we "cheat" by progressively reducing the physically large roundover radius as it curves back around behind the plane of the mouth (like a french curve)? In other words, if the width budget for roundover is "x" inches what is the best way to spend "x"?
Thanks,
Paul
 
Paul W said:


Hi John,
Within the context of the overall message here, I understand why this assumption is made. However, how accurate is it? Thanks,
Paul

It will actually get progressively weaker thus you would start with a large radius and decay to a smaller one IF trying to make the surface pressure vary linearly (ignoring attenuation due to path length from the source). Not even sure that would be a good thing. And I'm not sure how this would relate to your wave guides. There may be other issues that generate what appear to be diffraction effects. I have never really looked into wave guides.
 
John,
Good, I needed a second opinion on the effect. For a given roundover width, the alternatives are: a uniform radius across the full width, or, a larger radius deeper in the mouth decreasing to a smaller radius as the mouth expands.

I've tried #1, so now on to #2. I don't know if it will work, but at least it seems worth giving the neighborhood another coat of MDF dust.

One other question. How "thick" is the surface wave? Molecule, 1/4"???


Jin,
Sorry for going a little OT with my issue...

Perhaps felt edge treatments will help dissipate the energy. For your baffle, I would tend to go wider and not so deep. Once the backside of the baffle begins to close in (anything sharper than 45 degree wings) you may have cavity resonance(s) to deal with...not impossible, but another issue to deal with. Something like HiFiNut suggests won't have that challenge.
Paul
 
Dear Paul,

I've been working with larger waveguides recently and find diffraction at the mouth to be a huge issue...and the roundover radius required to combat diffraction nearly impractical.[
/QUOTE]

Could you elaborate on this statement? Are you using the term "impractical" in the sense of "not capable of dealing with the issue in practice; lacking sense" or "capable of dealing with the issue, but not utilizable due to other prctical considerations"?

Could yo uelaborate on your testing protocol?

Thank you,

M
 
Thanks, John, for your comments.


"the conventional diffraction treatment is to take a high amplitude, short duration result and transform it into a low amplitude, longer duration result in the hopes that the reduced amplitude will make the signal less obvious that the effect of the time smear."


I think this is to way to go for me.

"To reduce diffraction we need to dissipate the energy of the surface wave before it reaches the edge."


What if we apply a thin layer of felt (1mm-3mm thick) on the entire surface of the baffle including the rounded edge and back surface? Would it reduce diffractions dramatically? Does it have any negative effects?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.