One size fits all? Where baffle size matters.

Rudolf · « **Reply #20 on:** 26 Jan 2009, 01:22 pm »

Pano, Graham,

as I concentrated on presenting the baffle width aspects of OBs, I obviously did not explain with the necessary precision my argument regarding large driver radiation. What did I want to say? If large drivers keep a wide radiation pattern beyond their geometric beaming "threshold", they only can do it, because those high frequencies are only radiated by inner parts of the cone (including wizzer cones or dustcaps).
As reactions have shown, I clearly missed to get that understood

Thanks Graham for explaining the resulting interrelationships some more. I actually never thought about the retracting acoustic center of radiation myself, but you certainly are right. But as I already have difficulties to explain simple things, I would not dare to bring this to the public too.

JeffB · « **Reply #21 on:** 26 Jan 2009, 10:02 pm »

I find this topic quite fascinating.
Today I was looking at a couple designs from the Zaph Audio site.
The first one is here:
http://www.zaphaudio.com/ZMV5.html
If you scroll down the page and look at the
Modeled on-axis frequency response,
you will notice a breakup in response above 10kHz that looks much like the diagrams Rudolf posted.
The tweeter is on a 8" wide baffle.

So I wondered if Zaph had any designs with a smaller baffle and I found this one:
http://www.zaphaudio.com/ZBM4.html
Here the breakup starts around 17kHz on a 6.25" baffle.

Perhaps what I see can be explained by things other than the baffle width, such as the design of the tweeter,
but it seems to match what Rudolf posted.

Of course, I have all seen flatter responses on wider baffles:
Ascend CBM-170 SE, 9" baffle.
http://www.ascendacoustics.com/pages/products/speakers/cbm170/cbm170meas.html.
There is still a hint of breakup above 10kHz though.

Rudolf · « **Reply #22 on:** 27 Jan 2009, 01:57 pm »

Quote from: JeffB on 26 Jan 2009, 10:02 pm

you will notice a breakup in response above 10kHz that looks much like the diagrams Rudolf posted.
The tweeter is on a 8" wide baffle...
Perhaps what I see can be explained by things other than the baffle width, such as the design of the tweeter,
but it seems to match what Rudolf posted.

Sorry Jeff,
but the effects I mentioned can´t be discovered at isolated on-axis-measurements. You can equalise everything to look good on axis. Too wide baffles get uncovered by comparing on-axis to off-axis response. And those irregularities have to start at frequencies, whose wavelength equals the baffle width - roughly.
On an 8" wide baffle for instance those effects had to start at ~1700 Hz.

And please keep in mind: We are talking about OBs. On boxes (without that inverted rear radiation) those baffle width effects do happen too, but MUCH less accentuated than on OBs.

Rudolf · « **Reply #23 on:** 27 Jan 2009, 06:17 pm »

I have some left-over cheap Visaton W250 woofers http://www.visaton.de/en/chassis_zubehoer/tiefton/w250_8.html, which are well suited for OB.
In combination with the small (and cheap) fullrange drivers FRS 8 http://www.visaton.de/en/chassis_zubehoer/breitband/frs8_8.html they result in a simple OB system, which will deliver 100 dB/1m from 50 to >10000 Hz before exceeding their Xlin or their rated power.

Following my previous considerations the OB will be built like this:

It´s a pyramid baffle with 50 cm width at the bottom and 1 m height. The FRS 8 is mounted at 88 cm.
But let´s look at the bottommost woofer first. Without any filter the response on a baffle 100x40 cm (Boxsim only does rectangular baffles) will be like this:

You can see the dipole peak at 700-800 Hz (40 cm = 857 Hz), from where on the controlled dipole radiation starts to get wrong. So we better cross over to the FRS at 500 Hz. Many people now would mount the FRS on that wide baffle. With the appropriate filter and the FRS at 88 cm height we get the following response:

The X over at 500 Hz is almost seamless. But between 800-5000Hz the radiation pattern is wrong. Above 5000 Hz the FRS radiation pattern does not care about dipole anyway, as Reply #7 already showed.

If we reduce the baffle width at the position of the FRS to 10 cm, the combined response will look like this:

Obviously it is much better controlled between 800 and 2000 Hz (or even 4000 Hz) than on the wide baffle - and that is as far as it get´s in this combination. Since the FRS8 itself is 3" wide, we can´t make the baffle smaller. We would need a dome tweeter without a front plate to optimise the response even above 2000 Hz.

If you are curious how the filter network for this combination would look:

HiFiNutNut · « **Reply #24 on:** 27 Jan 2009, 08:41 pm »

Great posts.

Here is my story. I "advertised" in the diyAudio forum "Free OB speaker design service Sydney" in this thread: http://www.diyaudio.com/forums/showthread.php?s=&threadid=135069

I had one response and was requested to help with the OB with B200 and Alpha 15. I really have great interests in that combination because it has such popularity and good reviews. I am just curious about how they could work well in a wide baffle. Unfortunately, the person who responded is from Melbourne and I don't think I am able to help given that I will only design the XO / EQ based on my measurements, not just simulations.

Anyway, I am thinking about applying damping materials to the wide OB panels to overcome OB ballooning. In theory it should work. I have described my idea here: http://www.diyaudio.com/forums/showthread.php?postid=1716918#post1716918

Regards,
Bill

Rudolf · « **Reply #25 on:** 27 Jan 2009, 09:58 pm »

Quote from: HiFiNutNut on 27 Jan 2009, 08:41 pm

Anyway, I am thinking about applying damping materials to the wide OB panels to overcome OB ballooning. In theory it should work. I have described my idea here: http://www.diyaudio.com/forums/showthread.php?postid=1716918#post1716918

Sorry Bill,
but the idea you described won´t work - at least not in theory. Sound does not just travel along the baffle surface and then around the baffle edge. By applying damping materials to the OB surfaces you will only reach a fraction of the acoustic energy radiated from the driver.
Think about a cape where the waves are turning into the lee side. Do you really believe you could stop those waves by braking the waves at the cape shore only?

If we really want controlled dipole radiation way above the bass region we must proportion baffle width to frequency. This means that we are operating along the dipole roll-off area alone -NOT beyond the dipole peak. And in doing that equalisation has to be your friend. In this place are many people who want to build dipoles and fear equalisation like the work of the devil. They may build OBs - whatever that means - , but certainly no dipoles.

HiFiNutNut · « **Reply #26 on:** 27 Jan 2009, 09:59 pm »

Rudolf,

Your pyramid baffle looks great. The idea is really good and I am very impressed. I am pretty sure the speakers will turn out good.

The dip at 3kHz would not be bad at all. If it was me I would probably apply a small notch just below 2kHz.

The curves look really nice.

Given the values of the capacitors, I guess it may work out cheaper to do it active. A pair of 47uF capacitor would cost $$$$ if you get something like Mundolf M-Cap Supreme?

Regards,
Bill

HiFiNutNut · « **Reply #27 on:** 27 Jan 2009, 10:11 pm »

Rudolf,

With regards to applying damping materials to the baffles, I don't mean to attach fabric or thin blankets to the panel. I am thinking about adding the damping materials so that they form a shape like a donut. The thickness of the high density wool batts on each side may be as much as 10cm or even more. This may have good attenuation down to 800Hz or lower. If the first dipole null is placed at 800Hz or higher (determined by baffle width), if the driver directivity is taken into consideration, and if the above described damping method is used, in combination, it would be possible to remove the dipole nulls and peaks above (and include) the first dipole null. I could be wrong. Since no simulation software does this modelling, the way to find out would be to try it and measure it. I speculate that as long as we use sufficient amount of damping materials, it should work.

Regards,
Bill

Rudolf · « **Reply #28 on:** 27 Jan 2009, 10:20 pm »

Bill,

I thought the same about that "BBC dip".

Any other EQ will have to be considered when prototypes are built - and whether the cost can be justified.

The W250 is sold in Germany for 25 Euros, the FRS 8 are 10 Euro each. This really is a low budget project. I hardly can see how a working passive dipole could be done at a lower price. So putting pricy Mundorf caps and coils on the X-over would really be overdoing it.

Doing it active would be much more elegant indeed. To get a frequency response that is even more linear than what I showed you only need a 6 dB LP at 80 Hz and a 6 dB HP at 500 Hz. No level adjustment necessary at all! That´s what the simulation says. Reality might vary ...

Rudolf · « **Reply #29 on:** 27 Jan 2009, 10:57 pm »

Quote from: HiFiNutNut on 27 Jan 2009, 10:11 pm

With regards to applying damping materials to the baffles, I don't mean to attach fabric or thin blankets to the panel. I am thinking about adding the damping materials so that they form a shape like a donut. ... I could be wrong.

Sorry again Bill, but you ARE wrong. You can do something about edge diffraction that way -yes. But with regard to the dipole dips and peaks you definitely have to think about sound as radiating in spheres - not in planes. The back wave will come around all the way between your baffle edge and the next room boundary. You can´t stop it. And if you can, it won´t be a dipole any more.

Sorry if I sound a bit harsh, but how about playing with the Ripple tank simulation http://www.falstad.com/ripple/ once more? It told me much about wave propagation issues.

Cheers

D OB G · « **Reply #30 on:** 28 Jan 2009, 12:29 am »

Hi Rudolf,

Quote

you only need a 6 dB LP at 80 Hz and a 6 dB HP at 500 Hz.

This all certainly works, as I found out with my project "an OB design". I wonder if a sloped baffle will work better than a specifically stepped baffle??? The Edge seems to suggest? that a square baffle of the size of the driver (ideally a square driver?!?!) works best.

I used a 32 mH inductor!, with appropriate parallel resistor, in series with two woofers on a baffle the size of the woofers. This meant the 6 dB LP compensated for the baffle roll-off, and gave a flat response down to 30 Hz.

This crossed over to a 4 inch Focal on a 4 inch square baffle. Inductor to match the woofer baffle crossover, giving a 6 dB HP at the woofer baffle frequency up to Fequal of the 4 inch baffle i.e. flat overall response.

The intention was to go on to do the same thing with the 2 inch Jordan, on a 2 inch square baffle, but it turns out that the rear radiation sounds terrible.

So I was forced to go to a dome tweeter, with an asymmetrical crossover.

The intention was that this project was to be specifically passive.

If I did it again , I would add a 12 or 18 dB symmetrical electrical slope at each baffle width, to the 6 dB acoustic slope, so as to get a smoother response above each baffle width. (I have some 3 inch Boston Acoustics, which together with a 7 or 8 inch would go well with the 15 inch drivers).

David

HiFiNutNut · « **Reply #31 on:** 28 Jan 2009, 12:33 am »

Rudolf,

Thought: how can a waveguide make the sound directional at higher frequencies?

Sound waves from the rear of the panel need to travel around the baffle to reach to the front half. If a large area in that travel path are covered by damping materials, then sound waves with a wave length less than 1/4th of the damping material thickness would be greatly attenuated. I am thinking about a donut shape which may extend 10cm each side from the panel surfaces front and back, and possibly over 20cm from the driver edge extending to over the panel sides. Depending on how we shape the damping materials, damping can be applied, especially at the rear, to cover most of the space / travel paths.

Of course, this would not work with lower frequencies. The damping materials would appear nearly non-existent to the lower frequencies, and they work as a low pass filter, for example, below 300-500Hz. So at those frequences, I expect the behavior to be the same as your modelling software. But at those frequencies we have no concern because they are below the first dipole peak. My target is to deal with the peaks and dips above the first dipole peak, and we are talking about possibly over 800Hz here.

Sorry I am not convinced.

I can fully understand your points of edge diffraction, radiating in spheres, etc. and they are right. But you have not considered the attenuation of sound by blocking a large space in the sound travelling paths.

I have my large U-frame speakers under development. In order to obtain a small amount of front wall sound reflections to enhance the "sound stage", I deliberately leave a big hole behind each driver. My MLS gated measurements show that the holes do not affect the frequency response above 500Hz, due to damping materials applied.

Regards,
Bill

panomaniac · « **Reply #32 on:** 28 Jan 2009, 05:14 pm »

This is a really good discussion - I have to thank Rudolf for bringing up the subjet.

Funny thing is, just as this topic was appearing - I was planning to go something similar with my next OB.
Several asymmetrical ideas, this being the lastest

This is with a 12" & 8". A 15" would fit, too. The baffle would rest on the floor leaning back slightly. Supported by one leg in the back.
The idea was for a fun, retro atomic look, and an asymmetric baffle.

gainphile2 · « **Reply #33 on:** 29 Jan 2009, 01:53 am »

That's really nice and original pano.... post them pics when done !!

Some people attempted waveguides for mid-high. I can only think this is beneficial for on and off axis response ?

Viridian · « **Reply #34 on:** 29 Jan 2009, 03:26 am »

Quote from: gainphile2 on 29 Jan 2009, 01:53 am

That's really nice and original pano.... post them pics when done !!

Some people attempted waveguides for mid-high. I can only think this is beneficial for on and off axis response ?

JohninCR did a waveguide for his B200's for that reason and runs them FR in that configuration. Prior to that he was experimenting with every possible permutation of OB. After that he kind of dropped off the boards. I asked him why and he replied that with the wave guides he is now satisfied and has moved on to other things he can experiment with. Of note is that he mirrored the waveguide front and rear.

Erik

D OB G · « **Reply #35 on:** 29 Jan 2009, 03:52 am »

JohninCR (I've seen the photos) would be optimizing the power response as well as getting a good polar response with his symmetrical waveguides.

I really think that both parameters are very important.

But even SL with his rear facing tweeter only mirrors the less than ideal power and polar responses at the front, at the rear, which waveguides improve upon.

And drivers that only radiate to the front at high frequencies are missing the full benefits of an even power response.

My "Audiosyncrasy" had symmetrical waveguides, but has morphed into something a bit more complicated now.

David

Viridian · « **Reply #36 on:** 29 Jan 2009, 04:02 am »

Dave, I've wanted to try the waveguide front and rear like John with my driver of choice but the task is a daunting one for a guy like me. I could pull it off but I don't understand the theory well enough to be confident of working up a suitable design. Those pics live on my desktop and I look at them now and then.

D OB G · « **Reply #37 on:** 29 Jan 2009, 04:42 am »

Hi Viridian,

I don't understand the theory well enough either, that's why I use commercial waveguides.

Three sources, at least, go into design parameters: Geddes, Zaph, and Rod Elliot, that I know of.

In all cases they are talking frequencies above about 1 kHz.

There are graphs for oblate spheroids around, and a thread on DiyAudio.

The interface with the throat seems to be the big problem, and Geddes goes into some detail about how it is almost impossible to get it right!!

David

scorpion · « **Reply #38 on:** 29 Jan 2009, 11:39 am »

This subject is interesting but I don't think it makes OB worse compromised speakers than any other type. And if I understand right: on axis listening
is unaffected by this argument. So it is the constant strive for perfection ? And for point sources only ? Rudolf's proposed design is however both imaginative and thoughtprovocing in that it would bring about clear improvement.

How does this argument relate to other possible OB-types like Line Sources and Electro- and Magnetostatic panels ?

I don't think it will change the 1st truth: OB rules !

/Erling

Viridian · « **Reply #39 on:** 29 Jan 2009, 02:27 pm »

Quote from: scorpion on 29 Jan 2009, 11:39 am

I don't think it will change the 1st truth: OB rules !

/Erling

And there is no question in my mind. You are right about that! I'll never go back to boxes.

Erik

Thanks Dave for pointing out those sources. I really need to study this thread and slog my way through those as well. If any of it can penetrate my thick skull there might be hope!