Midrange drivers dips above 1Khz...

[Graham Maynard]

Hi Inductor.

Ooops - I wrote foam surround - I meant roll surround.
Was comparing cone/edge shapes only, but slipped up there, especially as I have those Peerless drivers here !

My observation is that the crossover from dip to bump arises where a half wavelength matches the cone diameter.


Hi Mr Twister.

Quite a dip on the Usher.


Cheers ......... Graham.

[dlr]

Originally posted by critofur
The dip I'm talking about is the one you see on manufacturer's charts, sometimes followed by a bump. This would typically be measured on-axis on a large baffle in an anechoic chamber.

Jay_WJ is correct, it's almost certainly due to a cone/surround impedance mismatch. This is found most with PP and doped paper cones, though not all. It's seldom found in hard cones such as metal and ceramic cones.

For true midranges it can be exacerbated by a restriction as mounted on the baffle that often impacts the same frequency range. It's very important to open up the air gap by chamfering the rear of the baffle around the driver opening.

When I was working with someone who designs and builds his own custom drivers for his company, I referred to it as the dipbump, with our drivers it was around 2Khz and not as severe as in some of the nice Vifa, Peerless, and Audax drivers I'd been looking at.

I think that foam surrounds are actually better in this area. Those by Scan-Speak are sealed foam that your supposed to last 20-30 years. I've got an old pair of 21W/8554 midwoofers with them that are still fine. You'll find that the better drivers also have a damping compound applied at the base of the roll on the surround that significantly reduces or eliminates the dip, if it was an issue to begin with. Seas, Vifa, S-S and many of the Tang-Band drivers use it, it's very effective in many cases.

I've applied what I think is the same compound with very good success, purchased from Parts Express. One can is about $18 and will treat possibly hundreds of drivers if you're careful to work quickly and not let it dry out, meaning cover the can as soon as you're through with a treatment. A caveat, as many tweaks are, this one is not reversible. The first coat, kept light, will show if it's helpful or not. I've tried it on some drivers with almost no change, primarily those with little in the way of a surround issue, though.

Some examples, the first is a 6.5" coxial (very much like the old Seas units). The second is a smaller "midwoofer" (hate to call it that) about 5.5".





It's not guaranteed, no driver mod is quaranteed not to be detrimental.

I've got complete pages on these mods at my site: Dave's Speaker Pages

Dave

[Inductor]

Originally posted by Graham Maynard
...My observation is that the crossover from dip to bump arises where a half wavelength matches the cone diameter.

But with it's reason being from the materials (differently bending) and impedances mismatchs as Dlr says and not an anti-phase being born there (?). So that the driver is between give or break(?)...

[dlr]

Originally posted by Inductor


But with it's reason being from the materials (differently bending) and impedances mismatchs as Dlr says and not an anti-phase being born there (?). So that the driver is between give or break(?)...

Graham is right in one respect, that is, it's related to the diameter in a way, but more accurately it's related to the distance between the former and surround attachment points. It's also related to the speed of sound in the material. This means that when analyzed in the frequency domain, two drivers with precisely the same dimensions, but different materials, hence different speeds of sound in the material, will exhibit the anomaly at different frequencies if it shows in both. You usually won't see it in hard cones, though.

It's a resonance due to insufficient damping of the wave by the surround. All resonances of this sort are dimensionally related. As a general rule, the smaller the driver, the higher in frequency the fundamental resonance.

Dave

[ShinOBIWAN]

Originally posted by dlr


Jay_WJ is correct, it's almost certainly due to a cone/surround impedance mismatch. This is found most with PP and doped paper cones, though not all. It's seldom found in hard cones such as metal and ceramic cones.

For true midranges it can be exacerbated by a restriction as mounted on the baffle that often impacts the same frequency range. It's very important to open up the air gap by chamfering the rear of the baffle around the driver opening.

Dave

Very interesting Dave. Thanks for that.

I've seen this numerous times with designs I've built. One thing I've noticed is that when taking a near field measurement ie. 40cm or closer to the driver axis, its quite pronounced. However, at a typical listening distance of 3 meters it disappears almost in my room. I take it this is because of a combination of reflected energy from the reverberant field reaching the mic.

Here's an unsmoothed measurement illustrating this:



Black shows 3m listening position measurement and green is at 40cm on axis with the mid driver.

[Graham Maynard]

Hi Dave,

Where I mention the cone diameter, you mention the former to surround dimension with regard to the dip-bump effect. Generally that dimension would then correspond to a quarter wavelength wrt the mid frequency between the dip and the bump.

This though is a dimension relating to the wavelength in air, not to wave propagation through the cone itself, and this being at frequencies well below break-up ?

The Usher and your plots showed more of a dip without the following bump. Maybe the driver enclosures were modifying the initial response and making the measured characteristics different to those normally presented by a manufacturer, though the effect of your cone edge/surround treatment is obvious.


Hi ShinOBIWAN,

Fortuitous floor reflection etc. ?


Cheers ......... Graham.

[dlr]

Originally posted by Graham Maynard
Hi Dave,

Where I mention the cone diameter, you mention the former to surround dimension with regard to the dip-bump effect. Generally that dimension would then correspond to a quarter wavelength wrt the mid frequency between the dip and the bump.

Considering low to high frequencies (decreasing wavelength relative to the fixed former-to-surround distance), any reflected wave starts off almost fully in phase. But most resonances such as these are relatively high Q, so the problem really only is an issue close to the center of the primary resonance, Fc, so we need to focus on that area.

When the distance is 1/4 wavelength, the reflected signal will be 180 deg. out-of-phase (round trip of 1/4 + 1/4), so a null will be strongest at that point. When the frequency is a bit lower, it's moving towards being in-phase, so there may be a small peak somewhere below Fc. When the frequency is a bit higher, it is again moving towards being in-phase again. At some point, it may be fully in-phase and create a peak, but that would be at 2*Fc. I haven't seen this much, if at all. I suspect that the damping is usually very effective outside of the primary Fc region, until breakup.

The caveat is that there are other variables that enter into it, such as cone internal damping and surround damping being frequency related and not linear, the geometry is significant, since there is a phase issue with the wave launch from the former to the surround having different distances to travel in air with the former area having more distance, thus more phase shift, cones have varying shapes and materials for any dust cap, dust caps will have their own resonances, cones and dust caps may inter-modulate, there may not be a dust cap, the air behind a dust cap may resonate significantly, the air gap behind the cone through the frame opening may be significant having its own resonance that back-modulates the cone, etc. Then there is some level of damping due to former modulation that creates a counter-emf generated in the coil, reducing the Fc peak. All of these are highly variable.

All of the above for distances and wavelengths are only related to the resonance (standing wave) in the cone. The cone radiates energy due to the resonance over the entire surface of the diaphragm, so at any given distance out, there will be a concentric ring of equal energy transmission into the air. At nulls that ring energy will be zero. At other rings, the total energy will differ from any other rings, since a ring is a length of driver surface that is length = pi * d. A ring of energy closer to the former has less total contribution to the total output of the driver at that frequency than those farther away. But there will have been additional damping of the wave as it moves from the closer ring to the farther ring, so it's highly variable.

This though is a dimension relating to the wavelength in air, not to wave propagation through the cone itself, and this being at frequencies well below break-up ?

The air has nothing to do with the resonances other than air mass loading of the diaphragm. The laser resonance images posted by John K in the enabl thread show how much variability there is in the resonances created in a diaphragm with mass loading changes, but that is for breakup. The main resonance is primarily a function of the diaphragm and surround. In a vacuum these resonances would still occur, but they would simply be altered due the change in air mass loading.

The Usher and your plots showed more of a dip without the following bump. Maybe the driver enclosures were modifying the initial response and making the measured characteristics different to those normally presented by a manufacturer, though the effect of your cone edge/surround treatment is obvious.

Yes, that bump might easily be from the mounting conditions. That's why one shouldn't try to read too much from any single measurement without having a full description of the test conditions. Determining the root cause (source) of any irregularity requires very careful testing by isolating the changes as much as is possible to single conditions, such as baffle mount cutout or changes resulting from some mod or tweak. Before/after measurements are indispensable.

Dave

[dlr]

Originally posted by ShinOBIWAN


Very interesting Dave. Thanks for that.

I've seen this numerous times with designs I've built. One thing I've noticed is that when taking a near field measurement ie. 40cm or closer to the driver axis, its quite pronounced. However, at a typical listening distance of 3 meters it disappears almost in my room. I take it this is because of a combination of reflected energy from the reverberant field reaching the mic.

Here's an unsmoothed measurement illustrating this:



Black shows 3m listening position measurement and green is at 40cm on axis with the mid driver.

It's hard to make any reliable predictions as to audibility. The direct response is very important, but so is the power response (in-room). The farther away, possibly the less audible I would suspect, I don't know for sure. Some folks have reported being able to hear this type of resonance on some specific drivers. It's certainly going to depend on many factors, including where in the spectrum it occurs with each person having differing ability to detect this stuff. It will also be music-dependent, of course.

I've found some things undetectable, then on other music a problem jumped right out. I recently had a 13m/8636 re-coned. There was a wandering soundstage, nothing I could pin down. Then I put on an a capella song that had just plain awful distortion from one driver not readily apparent in more complex music. It had to be replaced, the coil was not centered properly. The SPL measured just fine, the problem was in the distortion profile.

Dave

[Svante]

Originally posted by critofur
Seems like almost every midrange driver has a dip somewhere between 1-2Khz, even very good ones. Anybody know what's the consistent cause for that from driver to driver?

Examples?

[Graham Maynard]

Hi Dave,

Thanks for your reply - very clear.

I still can't help thinking that cone drivers used for anything above bass frequencies should be elliptical, and not circular, in order to reduce errors arising due to radial symmetry.

You wrote >>
A ring of energy closer to the former has less total contribution to the total output of the driver at that frequency than those farther away. <<

Yes, and I see this as a situation where the centre radiation of a cone driver can become cyclically lagging/leading wrt to the cone, thus another reason for peaks and troughs developing with changing higher frequencies, especially with full range drivers.

Cheers ....... Graham.