I wanted to know if it is possible a single driver, used open baffle (OB), could cover the 20Hz to 500Hz range, and maintain off axis polar regularity. So here are some raw measurements. Each graph is the frequency response of a single driver with no baffle, measured 8 feet off the ground. Each graph shows the 0deg, 30deg, 45deg, 60deg and 90deg response. The low end response (<40Hz) is accurate for most graphs, but environmental noise caused some to 'tail up' at 20Hz - just correct for this mentally. If it isn't parallel with the rest of the graphs, its due to noise.
There are a few things I'm looking for: a regular dipolar response (-1dB at 30deg, -3dB at 45deg, -6dB at 60deg, and about -20dB at 90deg) as high as possible (500-700Hz, assuming a 300-400Hz crossover to a small midrange driver, as in my current system design). I'm looking for a driver that can be used very low, as close to 20Hz as possible - this requires a lot of volume displacement, and low nonlinear distortion. Most of the drivers I tested have too high Fs to even reach close to 20Hz. Similarly, most drivers are not designed to have low distortion at the high displacements required by OB woofers. I've listed the nominal xmax of each driver, just to give an idea of what it is capable of - really, xmax is of limited value, and the real proof of a driver's suitability comes from nonlinear distortion measurements made with the driver in its final baffle.
Mostly, I just wanted to look at the polar response of different sized drivers, just to get an idea of whether or not it is possible for a single driver to cover the whole low end of an OB design. So I looked at a 15" driver, a 12" driver, two 10" drivers, and an 8" driver. Although these are just meaurements of individual drivers, the trends in polar response change with diameter are fairly consistent, and I feel they can be generalized.
Eminence Alpha15a, a 15" driver (xmax = 3.8mm):
MCM 55-2332, a 12" subwoofer no longer available (xmax = 12.5mm):
Goldwood GW-210/8, a 10" driver (xmax = 3.5mm):
Exodus Audio DPL-10, a 10" driver designed for OB (xmax = 19mm):
and here is the response normalized to the on axis response:
Visaton B200, which is an 8" driver (xmax = 3.5mm):
Here is what I get from this: 15" and 12" drivers have regular polar responses up to 500Hz, and 10" drivers up to 600Hz. Good news. The two subwoofers, the MCM and Exodus, probably have enough excursion to handle decent SPL levels. A 15" driver with a lot of displacement could easily do it too, but I'm not sure I want a 15" driver in my bedroom.
I did distortion measurements on the MCM and Exodus, but they were faulty done with errors, so nothing to report there.
I was hoping the Exodus would be a winner, but its voltage sensitivity is extremely low - with 50W in, it was still very quiet.
Anyway, interesting stuff. Questions anyone?
[A couple of side notes: astute readers might notice I've switched to ARTA - OMG it is a million times easier to do polar measurements than in SoundEasy. I will never use SE for this again! Unfortunately, ARTA doesn't do much for nonlinear distortion measurements...
Also, I want to point something out with the Visaton B200 8" driver. It looses polar regularity above 700Hz. It is highly likely that this shows up in Linkwitz's Orion loudspeakers as an off axis loss of output between 800Hz and 1.7kHz (the problem would be worse for other designs that cross the tweeter even higher). For a speaker that is all about polar response, I would consider this a problem. This is the reason I use a 4" midrange driver rather than something larger, and subsequently cross it higher.]
There are a few things I'm looking for: a regular dipolar response (-1dB at 30deg, -3dB at 45deg, -6dB at 60deg, and about -20dB at 90deg) as high as possible (500-700Hz, assuming a 300-400Hz crossover to a small midrange driver, as in my current system design). I'm looking for a driver that can be used very low, as close to 20Hz as possible - this requires a lot of volume displacement, and low nonlinear distortion. Most of the drivers I tested have too high Fs to even reach close to 20Hz. Similarly, most drivers are not designed to have low distortion at the high displacements required by OB woofers. I've listed the nominal xmax of each driver, just to give an idea of what it is capable of - really, xmax is of limited value, and the real proof of a driver's suitability comes from nonlinear distortion measurements made with the driver in its final baffle.
Mostly, I just wanted to look at the polar response of different sized drivers, just to get an idea of whether or not it is possible for a single driver to cover the whole low end of an OB design. So I looked at a 15" driver, a 12" driver, two 10" drivers, and an 8" driver. Although these are just meaurements of individual drivers, the trends in polar response change with diameter are fairly consistent, and I feel they can be generalized.
Eminence Alpha15a, a 15" driver (xmax = 3.8mm):
An externally hosted image should be here but it was not working when we last tested it.
MCM 55-2332, a 12" subwoofer no longer available (xmax = 12.5mm):
An externally hosted image should be here but it was not working when we last tested it.
Goldwood GW-210/8, a 10" driver (xmax = 3.5mm):
An externally hosted image should be here but it was not working when we last tested it.
Exodus Audio DPL-10, a 10" driver designed for OB (xmax = 19mm):
An externally hosted image should be here but it was not working when we last tested it.
and here is the response normalized to the on axis response:
An externally hosted image should be here but it was not working when we last tested it.
Normalized graphs are easier to read, as they directly show the relative response. I would do them more, but they require a little more work, so I don't do them much. Visaton B200, which is an 8" driver (xmax = 3.5mm):
An externally hosted image should be here but it was not working when we last tested it.
Here is what I get from this: 15" and 12" drivers have regular polar responses up to 500Hz, and 10" drivers up to 600Hz. Good news. The two subwoofers, the MCM and Exodus, probably have enough excursion to handle decent SPL levels. A 15" driver with a lot of displacement could easily do it too, but I'm not sure I want a 15" driver in my bedroom.
I did distortion measurements on the MCM and Exodus, but they were faulty done with errors, so nothing to report there.
I was hoping the Exodus would be a winner, but its voltage sensitivity is extremely low - with 50W in, it was still very quiet.
Anyway, interesting stuff. Questions anyone?
[A couple of side notes: astute readers might notice I've switched to ARTA - OMG it is a million times easier to do polar measurements than in SoundEasy. I will never use SE for this again! Unfortunately, ARTA doesn't do much for nonlinear distortion measurements...
Also, I want to point something out with the Visaton B200 8" driver. It looses polar regularity above 700Hz. It is highly likely that this shows up in Linkwitz's Orion loudspeakers as an off axis loss of output between 800Hz and 1.7kHz (the problem would be worse for other designs that cross the tweeter even higher). For a speaker that is all about polar response, I would consider this a problem. This is the reason I use a 4" midrange driver rather than something larger, and subsequently cross it higher.]
I'm sure you know about this, but have you tried the Polar plot generator in ARTA under Tools->Directivity Patterns ? I haven't tried it myself because as far as I can see it's not possible to use it in the unregistered version of the app (since you can't save multiple impulse responses to files, a prerequisite of using this feature) but if yours is registered give it a try if you haven't already.
There is an appnote covering it's use:
http://www.fesb.hr/~mateljan/arta/AppNotes/AP6_Directivity_Measurements-EngRev1.01.pdf
Not sure what you mean here, how about Analysis->Frequency Response and Distortions in ARTA ? No, it doesn't give distortion in simple percentages, (although you can work that out by converting dB to percentage at a given frequency) but rather a far more useful distortion vs frequency graph that will show each of the main distortion products vs frequency, which is far more revealing.
There is also Record->Distortion vs Amplitude in STEPS which will measure distortion vs amplitude for a set number of frequencies. See the following app notes for a really good example of using this:
http://www.fesb.hr/~mateljan/arta/AppNotes/AP7-EstimationOfLineardisplacement-IEC62458-EngRev01.pdf
I agree about ARTA in general though - I've tried a lot of different software measurement packages over the years and ARTA just blows me away. It does have a few small niggles though, mainly minor UI stuff, so when I have the spare money to register it I'll be sending the author a little note with some feedback
Last edited:
Hmmm.... they are not the same. I guess, maybe the different characters of breakup affect the polar responses around their higher end dispersions.
If the outer area (or the whole cone) could be de-coupled with the inner area (or dust cap), then the effective radiating area is no more the whole cone, thus smaller, and is getting wider dispersion....
If the outer area (or the whole cone) could be de-coupled with the inner area (or dust cap), then the effective radiating area is no more the whole cone, thus smaller, and is getting wider dispersion....
Only if they're acting as ideal piston radiators at the frequencies in question. If not, all bets are off as you no longer have a uniform radiating surface.
Progressive carefully controlled cone breakup is a technique that can be used to improve off axis response at higher frequencies (and extend on axis frequency response) well beyond what an ideal piston radiator would have by essentially reducing the cone diameter progressively at higher frequencies.
Full range drivers are the most obvious example of this, and various approaches are used including decoupling rings (pressed creases in the paper) near the perimeter of the cone which help isolate the edge of the cone at higher frequencies, thus reduce effective piston area at those frequencies.
Another approach is radially alternating standing wave patterns where at high frequencies only the centre part of the cone is in phase all the way around the radius, while the outer part of the cone alternates between in and out of phase radially around the cone, causing a net cancellation of radiation from the outer parts of the cone, thus an effective smaller cone area, resulting in wider dispersion at higher frequencies than the diameter would suggest.
A good example of this is the Kevlar FST midrange driver used in the B&W Nautilus 800 series, and there is a great white paper on the design of the driver which goes into a lot of detail on how this works.
Even more conventional paper drivers have some improvement in dispersion over the theoretical rigid piston case due to transmission losses in the cone at higher frequencies, however if the driver isn't carefully designed this improvement can be accompanied by unacceptable cone breakup distortions - eg very ragged frequency response and audible resonances...
At the end of the day there are so many things in the design of a driver other than just it's diameter that can affect the off axis response, (many of them "invisible", like cone material properties) that you just have to measure the actual off axis response of the driver rather than try to predict it.
Last edited:
For the measurements you took I presume you used "Impulse Response / Signal Time record" mode to record an impulse response (with swept tone or noise) then chose Smoothed FR or Dual-Gate Smoothed FR, and saved each response as an overlay ?
The only problem with that is the overlays only save what is visually shown on the graph, not all the information that was saved in the impulse response that was used to generate it. (For example you couldn't later on take one of those overlays and generate a cumulative spectral decay, step response, polar response etc)
For the polar response what you do is this - record the impulse response for the speaker on axis as before, but instead of going to Smoothed FR or Dual-Gate Smoothed FR to display it, just save the impulse response directly to a file (*.pir extension) with File->Save as, with the name of the driver with a suffix, eg:
DriverA_deg+0
Now take another measurement at 5 degrees off axis and save the impulse response with the same driver name but with a different suffix, eg:
DriverA_deg+5
And so on. The appnote recommends 5 or 10 degree steps for best accuracy. If the driver is symmetrical on a symmetrical baffle you can just measure the response to one side and get it to mirror the response for negative angles by ticking a box.
When all the measurements are done go to Tools->Directivity Patterns, File->Create Directivity pattern file, and then Load Files, to load all your correctly named saved impulse response files, select a few options and it should generate your responses.
Another good thing about saving the response at different angles directly as an impulse response is that ALL measurement data is captured in that file, so at a later date you could open one of those impulse response files directly, then view anything about it you want to - frequency response, phase, cumulative spectral decay, step response, and so on.
They are close, but the frame/spider structure changes things. At least thats what it looks like to me.
DBMandrake, good point, I should definitely be saving the .pir files. Being able to do post-processing would be nice.
Of course you are using both sides.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.