Open Baffle musings

[charlielaub]

Re: Open Baffle musings

I went back and changed the crossover to 2k Hz, and tweaked a few other things. More measuring. More tweaking. This is good for now. Except where the pattern "narrows" around 1k-2k Hz, this is well behaved off axis. Not bad for a 2-way that doesn't use any exotic drivers...



-Charlie

[charlielaub]

Re: Open Baffle musings

After listening to the version with the 2k Hz crossover, I didn't like it as much and decided that I should try something lower so I've implemented yet another version of the crossover, this time with a 1.2k Hz 5th order elliptic crossover.

The trickiest response to flatten is the tweeter between 1k and 3k Hz - it requires some complicated EQ because of the odd combination of a large peak superimposed on the OB response below. Since I know that the distortion performance of the tweeter is best above 800 Hz, so I would like to cross it over no lower than that. Dropping the crossover point from 1.2k Hz to 800 Hz would require a bunch of power boost to the tweeter, and I am trying not to go in that direction with this system.

I ran a set of horizontal off-axis response measurements, as well as a vertical set. In the horizontal response you can really see the difference between the woofer and tweeter off axis responses. I'd like to address the area between about 700Hz and 1.2kHz where the woofer is very tightly grouped, but don't see a way to do so without adding another driver. Another thing that I tweaked after pondering the horizontal off axis response family is the tweeter response. I ended up changing the response on axis so it is now rising. This has the effect of making all the other off axis curves "flatter" and I anticipate that the listening position will actually be at 22 degrees off axis when you set up two speakers. The response at this angle is quite flat.

Here are the off axis plots for this latest revision. First the horizontal off-axis responses:



Next, the vertical off-axis responses. Sorry the legend is wrong - it should read (top to bottom) up 20 degrees, up 10 degrees, on axis, DOWN 10 degrees, DOWN 20 degrees.


I have addressed the rising response below 800 Hz but that is not shown here. I really need to haul the speakers outside and do some low frequency measurements there, but this is not a high priority for now. I am satisfied after listening for awhile that this is good enough.

The next thing that I will play with is the baffle size. As I mentioned before, I just grabbed an off-the-shelf 2'x4' panel and used that to create the prototype baffle. With some more modeling, I think that I can reduce the height of the baffle to about 36-40" (keeping the 24" width) and get essentially the same response. The vertical off axis response measurements shown above shows that the response stays pretty much unchanged above the tweeter plane, and not until 22 deg below does the frequency response start to degrade. This tells me that I can shorten the baffle without a problem, since this is shifting the potential listening positions "upwards".


-Charlie

[charlielaub]

Re: Open Baffle musings

After being shelved for some time, I've revisited this project. I decided that I just couldn't get it done with two drivers, so I added a third: a Seas MP12 midrange operating from 750 Hz to 3k Hz. This has helped a lot by removing some issues I had with the Tang-Band W2-800SL around 2k Hz, and I can now cross over from the Exodus Anarchy before it gets to the first dipole peak.

Today I did a set of measurements, and then spent awhile coming up with a crossover. There were also a couple of iterations back and forth where I uploaded the biquads into my MiniDSP, took a measurement, wasn't happy with what I saw, and then went back to see just where some odd peak in the response was coming from. I now have a frequency response that looks like this:



Since I have been working on the crossover, I have NOT been working on the speaker itself. I took the particleboard monstrosity that I had before, and glommed on the midrange using an Irwin quick clamp. The clamp holds the magnet, and the clamp bar (which is a ferrous metal) is fastly stuck to the side of the Anarchy magnet and just resting on the top of the Tang Band. A little Rube-Goldberg-ish, but it works well enough to get this prototype up and running!




-Charlie

[JasonP]

Re: Open Baffle musings

Hmmm, how did just the Anarchy measure on the baffle like that? I only ask because in Jeff's diffraction and boundry simulator, placing the driver on the edge of the baffle in dipole mode removes most of the peaking. Its one of the reasons the wavefront turned out the way it did, with the SB Acoustics 5" pretty close to the top edge.

[charlielaub]

Re: Open Baffle musings

Hmmm, how did just the Anarchy measure on the baffle like that? I only ask because in Jeff's diffraction and boundry simulator, placing the driver on the edge of the baffle in dipole mode removes most of the peaking. Its one of the reasons the wavefront turned out the way it did, with the SB Acoustics 5" pretty close to the top edge.

Here's a set of off-axis measurements for the Anarchy on the baffle, near the top, as shown:



Although the pattern is not as well developed as when the driver is mounted in the middle of the baffle, you can see the first dipole peak around 800-1800 Hz. To avoid the peak in power response that would result from using the Anarchy there, I am now crossing over to the midrange just below this frequency (at around 750 Hz). I use the other drivers nude to push their first dipole peak as high as possible in frequency. On the other hand, I want the Anarchy to be able to play a little lower in frequency, and it's only a 6.5" driver so excursion concerns come in to play. Thus, the large flat baffle. Starting at 800 Hz, the response is flat down to about 300 Hz as shown, then it transitions to the 6dB/oct open baffle slope until about 80 Hz, where it rolls off to the ultimate 12-18dB/oct. Using the driver for bass below about 80-100Hz requires a LOT of excursion, so I will probably cross over around 80 Hz to some conventional subs.

With my original 2-way design I wanted to reduce the peaks and dips to a minimum, and I came up with the location for the Anarchy (at the top edge) after doing sims with Jeff's diffraction modeler. This could have worked, if I had been able to find a small driver like the Tang-Band that really worked well in a 2-way, but I haven't been able to find something having an open frame that also has high enough sensitivity. Now that I am moving to the 3-way, I can move the woofer down on the baffle. The high location results in a floor bounce that creates a big dip around 150Hz on axis at the listening height. The lower location (e.g. 12" above the floor) doesn't have this problem, and I can probably still get good integration with the midrange with this driver-to-driver spacing, so that is the next thing I will try.

-Charlie

[charlielaub]

Re: Open Baffle musings

Today I took a set of horizontal off axis measurements on the new 3-way configuration. Here's the data plotted up:



The response curves looks very smooth out to 45 degrees (orange curve). Going further off axis, the dip that appears at 2.5k is due to the drooping off axis response of the midrange. This will just drop the power response in this region which is not a problem. Up top, the pattern of the tweeter (which is a 1.5 diameter cone) starts to narrow, but not until about 9k or 10k Hz, and that is just fine with me as well. All-in-all it's looking like a keeper.

This definitely an improvement of the pattern that I got with the 2-driver setup, and it's definitely worth having the extra complexity of another driver.

-Charlie

[fjhuerta]

Re: Open Baffle musings

Charlie, don't you feel the speakers are too bright / forward? I started with curves similar to yours, and had to really dial back the high mids / treble in order to get the tonal balance right.

[r-carpenter]

Re: Open Baffle musings

very interesting and informative thread.
Thanks Charlie.

[charlielaub]

Re: Open Baffle musings

I did a couple of vertical off-axis measurements to check out what is going on in that plane. Here's the result:



I measured at +10, +20, and +30 degrees and then at -10 and -20 degrees. That's how the legend is set up, so please note the colors listed.

Things look pretty good. +30 degrees is pointed up quite steeply so I am not too concerned that there is a null around the 3k Hz mid to tweeter crossover point. Again, this will just drop the power response in this area, which is not a problem since this is often a desired effect. Other than that, it's pretty flat in this angular range.

There really is very little to no sign of the crossover at 750 Hz in both the horizontal and vertical responses. The only thing I don't quite understand is the difference in the -20 and +20 degree responses between about 700 Hz and 2k Hz. This frequency range comprises the passband of the midrange more or less, but why this difference exists is not clear to me. Perhaps the tweeter is "in the way" in the "down" direction?

-Charlie

[charlielaub]

Re: Open Baffle musings

I don't perceive excessive brightness in the sound, but I need to do more careful listening. Tomorrow I might be able to put another one together so that I have a stereo pair and that will probably be more instructive. I know that SL advocates a down-tilted frequency response, and I prefer that kind of balance with other speakers that I have developed recently (conventional boxed speakers) but this one seems fine at the moment.

Were you able to take a set of off-axis measurements in both the horizontal and vertical directions with your speaker? What you "hear" at the speaker is both the direct sound from the speaker and the "room response" (the sum of all the off axis radiation and the adsorption/reflection properties of the room itself). The room response can be as high as -3dB or so compared to the direct response IIRC. So, even if you speaker looks flat on axis, if there is a peak in the off-axis response, you will hear it via the room. A related factor is the "listening axis" that you designed the speaker on, versus the one that you sit at when listening to the speaker. For instance, if you design the speaker to be "flat" when you are directly on axis with the baffle but set up the speakers as a stereo pair not toed in, you will be listening at something like 15 degrees off axis. I've tried to keep this in mind while working on this project, and you can see from my earlier post that the 20 degrees off axis response is relatively flat or downtilted by 1-2 dB above 1k Hz, but it's very subtle. This should give me some flexibility in placement. Finally, the interaction of the speaker with the room can cause peaks and dips that can make it sound off one way or another. Dipole speakers are not immune from room interactions, they just excite the room in a different way compared to a monopole.

-Charlie

[charlielaub]

Re: Open Baffle musings

I kept thinking about how to check into your thoughts that the speaker might be too bright sounding. Tonight I did a couple of in-room measurements: I did the usual impulse measurement but then used a LOT of the impulse, processed into a frequency response, and then used 1-octave averaging to get a general picture. Here's the result:



This looks a little different than the gated impulse measurements below 1k Hz. The in-room measurement shows that the lower frequencies are elevated by a couple of dB. This provides a kind of treble down-shelving like SL advocates, if you believe the measurement. It is also more like the model response of the loudspeaker in my Active Crossover Designer tools, where I did the crossover development, so it gets a nod there. This might be why the speaker doesn't sound bright to me, but rather balanced (missing the lowest bass notes, however).

If anything, I might add an EQ band of +1-2 dB around 175 Hz to fill in the broad dip and then do a little more listening to see if I notice anything (your standard old "not blind" experiment).

This all might be due to the room, and so this might be just EQ-ing the system to the room and in another room there will be other slight changes of a couple of dB here and there to the response.

-Charlie

[fjhuerta]

http://www.libinst.com/SynRTA.htm). That's what I use for my in-room, in my listening seat tests.

I have this feeling the balance is still a bit on the hot side (I don't see a downward tilt, but more of a bit of extra bass and a very, very flat midrange and treble), and I see you also have a bit of extra response around 400 Hz or so, like my own design. I haven't still found where it comes from, but I'm assuming it's a combination of baffle shape / size and dipole response.



My room is very bright...

Anyway, you can compare both our in room responses to the big OB Jamo R907 on Stereophile:



And it's interesting to see they also have a little hump at around 400 Hz. I wonder if that's part of the dipole nature.

[charlielaub]

Re: Open Baffle musings

Anyway, you can compare both our in room responses to the big OB Jamo R907 on Stereophile:



And it's interesting to see they also have a little hump at around 400 Hz. I wonder if that's part of the dipole nature.

Here's the last paragraph from the measurement section of the Stereophile article on the Jamos:

John Atkinson

I'm sticking to my guns... my MWOB-3 still doesn't sound too bright to me in my "speaker development room". I need to make the matching speaker, and get them set up in a larger room to check the sound there. I will post the results when I have them in a couple of days when I can get around to it.

-Charlie

[fjhuerta]

Re: Open Baffle musings

Well, brightness is in the eye (or ear!) of the beholder. If they sound OK to you, then that's what matters!

[Face]

Re: Open Baffle musings

Well, brightness is in the eye (or ear!) of the beholder. If they sound OK to you, then that's what matters!

It depends on the room too. If his room is on the dead side, IMO, those wouldn't sound bright at all.