diyAudio (
-   Multi-Way (
-   -   'Very' OB Midrange Baffle Width Study (

cuibono 8th December 2009 01:16 AM

Again, I've chosen 4" drivers because, as could be seen above, 6" drivers are beaming by about 1kHz. It takes a 4" to be good past 2kHz, which is what I'm shooting for.

The purpose of dual sealed enclosures it to remove the rear radiation from the acoustic output, while keeping the dipole action. Apparently, the magnet/frame structure causes a 90deg reflection of frequencies in the 1-2kHz range (wrt its rear radiation), disturbing the driver's 'ideal' dipole pattern. This causes excessive attenuation of off axis radiation between 1-2kHz, but then a spike in output at 90deg off axis, wrt the forward axis. This can be seen from the above data (dual 6" experiment), and will be shown again in this set of data.

So we'll start with looking at the cheap 4" poly driver (not a bad driver actually!):

Here is its nearfield response of the bare driver. Not bad, some problem at 3.3kHz:

Here is what happens on and off axis when used as a dipole with no baffle. Measured at 1m, rotated 0deg, 30deg, 45deg, 60deg, and 90deg.

What I'm interested in the above is seeing that the off axis polar irregularities have moved up to 2kHz (the 8" and 6" had them around 1kHz): excessive attenuation at 45deg and 60deg, then a spike at 90deg. I'm fairly sure one of the main causes of the frequency of the off-axis irregularities is the width of the magnet structure, but the frame ventilation makes a difference too as we will see. This suggests neo drivers may have some advantages when it comes to dipoles and regularity of the polar response.

Next is what happens when the driver is put into a small sealed enclosure (I'm guessing its about 50 in sq., plus there is stuffing inside). This was measured at 1m, but was measured out to 180deg off axis:

Looks pretty different than the no baffle dipole huh? What we can see is that it is basically a monopole up to 2kHz. Also, you can see the baffle step taking place between 1-2kHz. Not bad overall. I gave it a listen. More on that in a second.

Here is what happens when you put two of these back to back:

Yikes, that won't do. That notch/peak thing at 1.1kHz is classic dipole action. Really, the response looks like garbage from 800Hz up. The on center separation between the front cone and back cone was about 14", so to get the peak up well past 2kHz, the separation would have to be reduced to 4" or less. SOOO, long story short, there is pretty much no way one can get two drivers to cover 200-2000Hz, as far as I can see.

I listened to this configuration too, and it was kind of interesting (I did some digital EQing to try and roughly smooth things out). When used with a single driver and chamber, the presentation was classic 'box' sound: boxy. When switched into dipole mode, it became much more tonally neutral (despite a lot of other flaws). For me, the 'take home message' was that polar pattern matters. I did another experiment I'll write up that really emphasized this too.

cuibono 8th December 2009 01:38 AM

Here is where the TB driver comes to the rescue:

This is the bare driver, no baffle, measured at 1m, at 0,30,45,60,90degs. This actually looks really promising. Looking at 2kHz, at 45 and 60deg, the level is much less attenuated wrt other drivers (even the 4" poly), and the 90deg measurement shows no great spike in output. The diameter of the magnets of 4" poly and TB are about the same, but the rear ventilation is much greater on the TB. My guess is that this leads to a much more regular polar response.

Just to compare, look at what the Eminence Alpha 6a was doing under the same conditions:

Much more irregular between 1-2kHz. Really, the TB maintains very good polar response out to 4kHz, which is great! Actually, the only polar response error below 4kHz is the bump at 2kHz, which I'm guessing is the baffle peak. There is no way to avoid this, but it doesn't look to be much of a problem. At least, it will be less of an error than with other drivers.

So here are the caveats - a small driver can't play as loud or as low, and it has a bit lower sensitivity (about 5dB), compared to the tweeter (Neo3PDRW). I did a bit of math cad sims though, and things may be okay - in terms of volume displacement, the driver isn't particularly limited with LR2 or LR4 crossovers between 250-300Hz. My woofers can handle the high XO point. But the big limitation is power handling - it might be maxing out with 50W in, which might be about 100dB/1m. In fact, virtually all 4" drivers are power limited for our needs.

In reality though, things aren't so clear. The amount of power the driver is dealing with depends on the bandwith. Also, Xmax based calculations don't really tell you when the driver will have excessive nonlinear distortion. A quick look at the driver's NLD showed acceptable but possibly problematic distortion levels. SOOO, only time will tell if the driver is power or NLD limited. Luckily, they aren't too expensive, so I don't mind burning one or two. There are a couple other affordable 4" drivers with possibly better power and NLD characteristics if I blow these.

I haven't gotten a chance to listen to them yet because I killed both my Neo3PDRW's due to switching transients :( :( They developed a rasping buzz after an accident, but they actually might be easily fixable. We'll see.

cuibono 8th December 2009 02:00 AM

And here is dessert:

I had a few cheap, small neo tweeter laying around, and when my Neo3s went out of commission, I thought I would give them a go.

These are two Dayton ND20FB, back to back:

And here is their polar response:

If you look closely, it is actually kind of problematic: it is only -3dB down at 60deg. Also, it goes to pot between 3.5 and 8kHz. It's on axis response is fairly uneven, although 30deg looks good. It also had asymmetrical side to side polar response, I don't know why. The Neo3PDRW looked a lot better.

So I redid my crossovers and got the dual dayton tweeter up and running, and gave it a listen - it sounded bad! Considerably worse than I had expected. Let me show you something else: the mid and tweeter combined polar response...

This is explains a lot of the bad sound, IMO. The mid is still the Alpha 6a - and between 1-2kHz, the off axis response is dropping too rapidly. This can probably be expected from any driver with a large magnet. Then, above 2kHz, the polar response rises way too much. Notice the on axis response is EQ'ed almost flat (the XO was LR4 at 1700Hz). It really sounded like garbage, very tinny. I think it is safe to say back to back tweeters won't work (on a large enough baffle, things might look fairly different. It would be interesting to see).

So what I did was put a mic out in the room, and measure the 'reverberant' field. It clearly showed the dip at 1.5kHz, and peak at 4kHz, (and a couple others in the lower midrange). I then roughly EQ'ed it to have a flat power response - and low and behold, it sounded much better, even on axis 1m in front of the driver. So the 'take home message': polar response matters a LOT!


gainphile 8th December 2009 08:25 AM

Hi Cuibono, your measurement shows similarity with my back-to-back tweeter measurement. Especially that massive drop.

back-to-back Hivi K1

I am convinced that the performance ceiling of typical dipole setup is the tweeter and we need to find an alternative to make it better (back-to-back waveguide i.e. Mige0...)

I wonder why SL argues against directivity of the tweeter.

I have 10" dayton waveguide and will see how it performs. I dislike to wander this way to unknown territory though.

cuibono 9th December 2009 02:53 AM

Hmmm, back to back waveguides? Actually, a bare Neo3PDR is an excellent tweeter. Couldn't ask for more, really.

I agree though, the tweeter XO point plays a big part in how the rest of the system unfolds. I wish I could find one that would do 1kHz.

sendler 9th December 2009 03:05 AM

Neo 3 measurements

Originally Posted by cuibono (
Hmmm, back to back waveguides? Actually, a bare Neo3PDR is an excellent tweeter. Couldn't ask for more, really.

I agree though, the tweeter XO point plays a big part in how the rest of the system unfolds. I wish I could find one that would do 1kHz.

Do you have any directivity measurements of your open Neo3 before they died?

cuibono 9th December 2009 04:01 AM

Yeah, they should be here. The second graph is without baffle.

cuibono 26th January 2010 06:05 PM

So, almost 2 months later, I finally gave up on PE getting the NEO3PDR back in stock, and went with Meniscus (got here quicker too!), and I'm back at it.

Here is what I've been getting at:

The Mids are the Eminence Alpha6a and the Tangband W4-656sc that were discussed above. I'm comparing them because of their differences in acoustic output off axis, particularly above 1kHz. I've added tweeters to each, and put together an crossover (with EQ) so that they are as close as can be, on axis at 30inches. They are XO'd to LR4 at 1700 acoustically (the overlay feature in Soundeasy makes it a cinch matching acoustic responses!) Here is an overlay of the final responses:

The magenta curve is the TB with Neo3, and the green is the Alpha6a with Neo3. On axis, they are almost identical. The differences come with off axis response, with the smaller, 4" TB driver maintaining regularity off axis, and the 6" Eminence not (see above posts).

I then did several days of listening to answer the big question: IS THERE A DIFFERENCE???

The answer - yes, a small one. On the first day, it was hard to detect. I couldn't do quick A/B comparisons, because my setup wasn't sophisticated enough, but I thought I could perceive a difference. On day 2, I listened to only the mids, without the tweeters, and I set up quick A/B switching - then the difference became very obvious - it was definitely interesting to hear. Then, day 3 I hooked the tweeters back in (and lost the quick A/B capabilities), but I could much more easily hear the differences now.

How were they different? The smaller TB driver sounded more natural and especially more coherent. The larger Eminence sounded a little 'hollow' in comparison - more like a speaker system. It was especially noticeable on acoustic guitar. The TB driver had slightly more ambience, and had clearer, fuller mid-highs. It sounded more balanced overall, and was easier to listen to. Also, walking around the room, the TB driver had less change in sound. All good stuff.

So was it worth it? Would I recommend it? I'm glad I did it - it is an improvement. Not as massive as the difference compared to moving to the bare Neo3PDR tweeters, but still a small but definite improvement in the 'naturalness' of the sound. I would recommend it to those trying to really craft their speaker - all in all, it is a modest but appreciated change.

Next is to see how it does with time - will it handle high volume situations without burning or sounding strained? We'll see.

As a final tidbit, here are two graphs of the in room, ungated, 1/3 octave spectrograph - the red bar charts in the images below (thanks to JohnK for teaching me how to do it!). It shows the differences in total output (on and off axis combined) of the two drivers - first the Alpha6a, then the W4. Although they are EQ'd to be the same on axis, notice how much more output the W4 has above 1kHz. I find these graphs correlate well to what I hear.

Thats all for now! Enjoy!

Saurav 26th January 2010 07:24 PM

4 Attachment(s)
I've recently embarked on a similar adventure. Attachments show a sketch of my initial concept, a test jig to see how my midrange (Audax PR170M0) performed with no baffle, and the response curves (front and back, test jig rotated by hand so the angles aren't accurate or even). If I believe UPS my Neo3PDRs should be here Friday, so hopefully I'll get to play with them over the weekend.

I was concerned about diffraction problems around the 'nude' tweeter, and I see you used foam to address that. How much of a difference does that make, both audible and measurable?

Also, how did you kill your earlier Neo3s? Would be good to know what I shouldn't do :) And were those PDR or not (the measurements shown in the other thread that you linked to)?

cuibono 26th January 2010 09:17 PM

Hi Saurav,

I've always used Neo3PDR - I just don't always write out the name, I guess. I killed them either with too loud/long/low frequency test signals, or more likely, transients from switching interconnects/power plugs while the amps were on.

What I do now is put a cap in series with the driver, something like a 50uF electrolytic (film caps are expensive at that size!) to protect the driver. The foam around the edges smooths the raw response - I could do without them, they would just require more EQ (I use a computer with a lot of control for fine tuning). With the foam, they require very little EQ. In terms of audibility, I'm not sure the foam does that much. OB speakers are inherently devices with maximum diffraction - the back and front radiation intermixes almost immediately - and that isn't inherently bad. I guess the trick for OB is getting the two sides to interact as equally as possible.

Those measurements of yours look pretty good!

All times are GMT. The time now is 11:24 PM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 17.65%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio