'Very' OB Midrange Baffle Width Study

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'd like to give another installment in my development of an open baffle (OB) speaker. This part is a set of measurements of two driver's off axis response, used with varying widths of baffle, with the interest in seeing what happens on and off axis as the width varies. Particularly after reading Floyd Toole's recent book, I've been given more importance to off axis response.

In another thread, we have been exploring this, but I thought a thread of its own would be more appropriate. http://www.diyaudio.com/forums/multi-way/142015-my-open-baffle-dipole-beyma-tpl-150-a-13.html Posts 357 and 364 show two examples of drivers where we have been trying to optimize the width of the baffle to get the most regular off axis response within the driver's bandpass. In particular we are looking at midrange drivers.

First off, the thing I am looking for is what I call 'regular' off axis behavior for dipoles, as defined by their off axis attenuation at certain angles: -1dB at 30deg, -3dB at 45deg, -6dB at 60deg, and roughly -25dB at 90deg (should be more, often is less).

All the measurements here were done on the same day, all the same way, so they are fully comparable. Driver was 2.9m off the ground, microphone 1.9m away (my approximate listening distance). The graphs have a resolution of about 90Hz. Although they are "anechoic", there was noise polution, particularly below 200Hz. IGNORE everything below 200Hz. The graphs were all smoothed 1/3 octave, although that didn't make them look much different. There is no EQ, these are raw responses. Vertical divisions are 3dB.

The first driver was an Eminence Alpha 6a, a 6" driver, with an outer diameter of 17cm. The second driver was a Dayton RS225, an 8" driver, with an OD of 23cm. Each driver was measured with no baffle, a 'thong' baffle (a square the width of the driver frame, named by mige0 in the above thread), then a baffle 4cm wider, then another 4cm wider. For the 6" driver, the baffles were 17cm, 21cm and 25cm; for the 8" driver, they were 23cm, 27cm and 31cm.

So here are the 6" driver measurements: no baffle, then each following baffle wider. Measurements are at 0deg, 30deg, 45deg, 60deg and 90deg. Being able to flip between them easily is helpful.

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


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


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


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




Here are the 8" driver results, from least to greatest baffle width:

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


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


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


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




So this is what I am seeing: as the baffle gets wider, its baffle peak 'hump' moves down in frequency (starting around 1kHz, moving down to about 700-800Hz). Above the peak, its off axis radiation become more attenuated. So as far as I can see, if you want the widest bandwidth of off axis regularity, the less the baffle the better. The trade off is that you loose low frequency sensitivity/output (about 6dB between the baffle extremes).

I undertook this study after reading Michael's (mige0) post about the 'thong' baffle - look at post 357 in the above thread. In Michael's measurements, he shows that using a square baffle the width of the driver's frame is able to boost the off axis response above 1kHz, which would be helpful. This is an improvement over the unbaffled response, but in my measurements, I did not get any more useful bandwidth as defined above, the ideal dipole attenuation. In fact, by those stricter standards, there is less usable bandwidth. The reason I am concerned with this is a mismatch in directivity when crossed over to a tweeter (almost 10dB), which will happen somewhere between 1.5 and 3kHz, and the resulting suckout in the power response between 1-2kHz.

This brings up me to my final thoughts - it appears that above 1kHz, many/most drivers have a lot of trouble maintaining strict 'regularity' of their off axis response. At wider angles, the response is dropping too quickly above 1kHz (at 60deg, being -15 to -20dB down), but then at 90deg showing a massive spike in output. This must be due to the driver's backside frame/spider/magnet structure. Ultimately, if one wants to maintain strict control of the off axis response, I suspect one must look to a different driver arrangement. I will be exploring this soon.

;)
 
Thanks for sharing your measurement. It's true above dipole peak frequency the response falls apart. However my finding in previous measurements were a bit different

http://www.diyaudio.com/forums/1916323-post21.html

The response -6db clearly formed, not fall too quickly, and above the peak it's actually still uniform until the dip. Then it falls apart. That dip is about 2khz.

So crossing to the tweeters at 1.5kHz is ok. Which leaves the tweeter as the problematic frequency.
 
Hi gainphile, I'm glad you appreciate them!

It seems that off axis response is very driver specific - some looking better than others. Your driver is definitely more usable in the 1-2kHz area than mine.

But even in your measurements, there are still irregularities around 1kHz. Compare 500 and 1000Hz - at one point there is a spread of about 9dB - and at the other there is a spread of 20dB. I am looking at the worst part of it, but it is something I'd like to avoid. I haven't seen a driver yet that maintains strict regularity between 1-2kHz - the response is always dropping more quickly in this area. I hope to address this when PE gets some drivers back in stock.

Unfortunately, it seems impossible to get a driver that can do 100-2000Hz open baffle and maintain a strict dipole response...
 
While I'm waiting for more drivers to experiment with, I'll show what I've been thinking of.

I'm currently concerned with getting a tight off axis response with open baffle drivers up to 2kHz (tight being close to -3dB at 45deg, -6dB at 60deg). Between 1 and 2kHz, it appears that the magnet/frame structure causes premature roll-off off axis, combined with driver beaming, particularly for 8" and 6" drivers. I would like to find a remedy to this.

I'm waiting on a 4" driver that may do, but this requires making the speaker a 4-way. Kind of a pain. But something that occurred to me was to seal off the back of the frame, and then use two drivers as two close spaced monopole/omni source, then hopefully getting better response between 1-2kHz. So I tried it:

(again, the graphs are 0deg, 30deg, 45deg, 60deg and 90deg. The driver is an Eminence Alpha 6a)

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


I used tar paper (for roofing) to seal the backs up.

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


The results? Well, lets take a look. First the raw driver with no baffle, with its frame unchanged, for reference. The poor off axis response above 1kHz should be obvious. Using the driver like this will result in a big dip in the power response:

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


Here is the driver after sealing the back up. A couple of things to note: the response is almost omni up to 1kHz, where beaming kicks in, but off axis response above 1kHz looks better - especially at 90degrees. Also, the tiny back chambers have caused an fc of about 500Hz, below which things roll off at 12dB/octave.

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


Now, I add a second sealed driver behind the first (as pictured above), wired out of phase. The driver separation is about 15" on center, and the baffle peak appears to be about 700Hz. Below that things are a tight dipole response - but by about 1.3kHz, you can see the off axis has become a mess. Also, below 500Hz, the roll off is now 18dB/octave, due to the dipole pattern. Notice the peaky response on axis between 1 and 4kHz - this is classic dipole response above the dipole peak, and can be smoothed using a rectangular enclosure (measured, not shown), but doing so does nothing to improve the off axis performance.

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


There are a few generalities here - 6" drivers are still too big to get a tight dipole response up to 2kHz, due to driver beaming and the necessarily too large driver separation distance. So to pull off a 2kHz tight dipole, one would need to use a 4" driver (there are lots of issues with doing that though, which I will explore in a later post).

One might say that the dual sealed driver arrangement might have a more desirable off axis response compared to the raw driver response. While the raw driver really drops off above 1kHz off axis, the dual driver maintains its output off axis, just not its regularity - in terms of absolute deviation from an ideal off axis response, the sealed dual driver has less error. In fact, at 1.8kHz, it is almost back to an 'ideal' off axis pattern, which makes it a lot more useable for a crossover at that frequency. Still not ideal though.

So where I will go soon is take two 4" drivers, put them small enclosures that will hopefully give us output down to 200Hz or so, while being omni up to 2kHz, and use them back to back out of phase. The major issues here are low sensitivity, power handling, distortion, and complexity. The benefit would be a 3-way with tight off axis performance through its whole passband. Might be worth it, we'll see. The other option is a 4-way with a single 4" driver operating open/no baffle.
 
Wow! quite a price to pay for the perfect off-axis response.

And yes I agree as the narrow usable bandwidth of each driver, at least 4-way is needed to esteblish the whole system on dipole by this strict requirement of off axis response. I guess that's the reason for your another xover thread?

About the dual sealed drivers arrangements, could it be the shape of the drivers causing the off axis irregulations in higher mid? Maybe an enclosure with large radius round over might help (to an extend).

And, by pure intuition, I'd like to take omni configuration for the goal of uniform off-axis response instead of dipole. As a whole, a dipole system for this would be too complex, almost inevitably:( Oh, but I'm lazy.:eek: And it seems you are marching torwards your goal in a steady pace:D
 
Hi CLS - yes, things are getting complex. The crossover thread is definitely related - I want to make sure I don't do more damage than good by adding another driver. Roundover helps in some cases, esp box speakers, but what we are dealing with here is not part of that.

In a way, I've already explored omni - SL's Plutos were some of the first speakers I built. I haven't used them in a while. There is still a 'box' sound to them, and I find the sound field too reverberant (which leads to a loss of detail, I think) - I've never been fully satisfied with their imaging - I felt they lacked depth. They are really good speakers in other ways - except for the box sound, they sound exceptionally neutral. They definitely 'disappear'. The highs roll off some because the tweeter is relatively large. Really, they are excellent speakers - I'm just being very picky.
 
Good work!!! So much more enlightening than endless computer models. The real world is always so much more interesting than running another iteration of the software.

Still curious what an open mesh (chickenwire or similar) filled with loose filling (felt or cotton) would do for the open-baffle setup. I'm visualizing a pseudo-box made out of an open mesh with layers of felt on the inner surface - a semi-transparent box, in other words.

Even a moderate attenuation of the rear wave would make a significant dent on the depth of the cancellation nulls we're seeing here.
 
Good work!!! So much more enlightening than endless computer models. The real world is always so much more interesting than running another iteration of the software.

Still curious what an open mesh (chickenwire or similar) filled with loose filling (felt or cotton) would do for the open-baffle setup. I'm visualizing a pseudo-box made out of an open mesh with layers of felt on the inner surface - a semi-transparent box, in other words.

Even a moderate attenuation of the rear wave would make a significant dent on the depth of the cancellation nulls we're seeing here.

Yes, it i always good to see data, but the simulations tell us what to expect. The measuremets, of course, can account for the precise driver directional characteristics, but this isn't a problem, at low frequency.

As for the mesh box, it is really a matter of the details. Such a box would comprise a low pass filter applied to the rear wave. Thus there are two issues, the attenuation vs. frequency, and the delay introduced, again vs. frequency. The effect of attenuation will be as you say, reduction in the depth of the nulls. The effect of delay would be to shift the angular position of the nulls to the rear. Once the frequency is above the corner frequency of the LP filter the response will undergo a transition to that of the front radiation alone with little impact of the rear radiation on the frequency or polar response. Details will, as always, require measurement, but there is nothing new here that we haven't discussed before in the "Beyond ...." thread.

While I do appreciate the efforts made ot take these measurements, and agree that the details will be dependent of the specific drivers used, I fail to see why they are enlightening. I don't see anything here that would not be expected. The use of dual sealed drivers back to back provides the expected improved symmetery between front and rear.


In fact, last year at the New Engliand DIY event I presented such a speaker which I called the Bird House because the small, 4" midrange drivers were mounted back to back on what looked like the roof of a steep sloped roof Bird House. On the peak was a B&G NEO 3 planar tweeter in dipole mode. The woofer was a 5" unit firing out the bottom of the bird house. Radiation above the mid to woofer x-o was dipole, Through the mid/woofer crossover the radiation pattern morphed from dipole through cardioid to monopole. The design concept was to have full range dipole behavior above the modal region of a typical room, a transition to cardioid response in the frequency range where relfection form the wall behind the speaker would color the sound, and monopole response in the modal region. This was just a proof of principle design initiated form computer simulations. Nothing fance.
 
Let me just introduce some thoughts here. I visited Danmark Radio's new Concert Hall this last weekend and had an extraordinary concert experience.
Which might have some bearing on this discussion. However we measure and theorize we will all end up in our home rooms. And hear what there is to hear. This is what the Conert Hall looks like:

Interactive FullSreens 360° © John E. Kroll

Why am I bringing up this. Well The New DR Concert Hall is debated. First thing is : how is the reverberation time. It is short by any stadards and hence the sound is questioned. However what is heard from the podium is what you all expect to hear from a very good stereo record. All instruments placed on their place on the prodium both 2- and 3- dimensional. No blurring of images, no reveberation ratteling noices. There were no singing in this concert so I cannot judge that part. But otherwhise I would say it was perfect. But would we really hear this at home ?

But that doesn't mean we shouldn't go for perfection !

/Erling
 
Last edited:
Ah, nice to have some discussion! Thanks everyone for posting. Of course, JohnK is right, this is all to be expected from theory, I just hadn't seen it explicitly demonstrated with measurements here. I wanted to be sure that what was in my head was also in reality. I must have just missed the New England DIY last year, wish I had gone!

I actually suspect that using 4th small driver to preserve the off axis response between 1-2kHz won't make that much of a difference sonically - but there is another big reason for going 4-way: in the midrange, one typically has to apply 18dB of EQ to compensate for three octaves of roll off. This equates to 64W in around 200Hz to get the same SPL level at 1W in at 2000Hz - thermal compression anyone? Splitting the midrange into 2-ways reduces the needed compensation wattage a bit.

Who would want a speaker that is 70dBSPL/1W/1m? (which is what OB's are doing around 150-200Hz)
 
There is another observation I'd like to add: when I first got into open baffle speakers, one of the main 'selling points' was the uniformity of their off axis response - unlike box speakers, with their directivity narrowing with increasing frequency, OB speakers were supposed to have a figure of eight pattern through their passband. They do, but not the advertised -3dB at 45deg, -6dB at 60deg, etc.

The two (fairly different) H-frame dipole woofers I've measured were only down -3dB at 60deg. The 8" midrange drivers I've seen are typically -8 to -16dB at 60deg (between 700 and 1600hz), while dipole tweeters, if baffled, will be something like 0dB down at 45deg. So what we can see is that a full dipole speaker has a fairly variable polar and power response! Add the effect of drivers pushed to their max, both in terms of bandwith and power handling/displacement, and it is kind of a wonder they sound so good!

John, did your 'Bird House' have any 'box' sound to it? I'm curious how dipole versus OB speakers sound. What were your XO points? I wish I had heard them!
 
It would be interesting to know, whether Johns bird house had the midrange drivers in a common air volume or separated by a dividing wall.

What I am missing in Cuibonos measurements is exactly that configuration:
Both Eminence Alpha 6a in a common enclosure, which would "seal off'' the rear output of both drivers from the outer world.

Regarding the measurements already shown: It's always comforting to see how real life follows the theoretical predictions - and to see where real life gets too complicated to be simulated by our simple models any more. :p

Thank you very much, cuibono, for your effort. :)
 
The Bird House was just a proof of concept. Sound was ok, but it was never worked on as being a finished speaker. Still, those who heard it thought it showed promise. Bot mids were in a single stuffed chamber. Since there was no net change in volume the chamber didn't load the drivers and the dimensions were such that standing wave frequencies were above the x-o point.
 
So, here is where we are going next:

4" drivers in dual sealed enclosures:

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


Appearing today, a cheap poly from PE, and a nice Tang-Band. This is the money shot:

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


The first picture is with the 4" poly in back to back, small sealed enclosures, getting measured.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.