Beyond the Ariel

Snowfall In The Afternoon

Just took this photo a few minutes ago, and fiddling around in Photoshop to get the color balance to match what it really looks like (cameras can get weird ideas of white balance in snow scenes). The white blobs are big fat Colorado snowflakes.

P.S. Another difference between underhung and overhung is that the field lines that flow through the coil are quite straight with an underhung voice coil (from pole piece to top plate), while by contrast an overhung voice coil has partly straight field lines (the part that is within the gap) and some of them curved (fringing flux outside the gap). Much of the development over the decades with overhung voice coils (as used in the vast majority of loudspeakers) has been around various systems to improve the geometry of the fringing flux.

When JBL came out the "LE" series of loudspeakers in the late Fifties, instead of the meaningless phrase "Linear Efficiency" promoted by the marketing department, I suspect "LE" simply meant "Long Excursion" - overhung voice coils, for those in the know. Altec, being the traditionalists they were, stayed with underhung voice coils for their high-efficiency loudspeakers, but the low-key Altec marketing department never thought to promote it as a distinguishing feature. Like all marketing departments, they never wanted to overestimate the intelligence of the buyers, and promoted the efficiency and weight of the magnet.

It kind of reminds me of Class A versus Class AB amplifiers. Class A is inherently inefficient, but is also inherently the most linear, particularly at low levels. Class AB is inherently more efficient, but there have been hundreds of patents issued over the decades to get around the low-level nonlinearities. When you look at all the many patents issued to linearize fringing flux, remember it's possible to have a voice coil that doesn't use fringing flux at all. You just have to give up some things to get that, just as you have to give up some power to get Class A operation of an amplifier.
 

Attachments

  • snowfall_in_the_afternoon.jpg
    snowfall_in_the_afternoon.jpg
    85.2 KB · Views: 1,560
Thanks a lot for your summary .

*If* you would use the same VC (winding length and wire length) and the same magnet - and given you would need to double the top plate thickness to preserve approximately the same x-linear range - what's the loss of sensitivity between the two – 3dB ? 6dB ?

simplified example:
overhung gap=10mm / VC-lengt=15mm / / X-lin=2.5mm
underhung gap=20mm / VC-length=15mm / X-lin=2.5mm

Not that I would like to start a business :) but to get a picture of that kind of trade off


Lynn, beautiful picture - mellow scene - is it the sun comming through slightly?

Michael
 
Yes, the sun was almost but not quite visible through the snow clouds, as a very diffused glow casting a soft light on the scene. Very hard to get that kind of subtle thing through digital photography - all the colors are off-whites and centered around the gray/white blue/yellow gamut.

To the eye, the sky was the faintest pale yellow around the sun (just out of the picture), and the snow looked white with the palest blues around the shadows. To the camera, looking at the initial RAW conversion, the color balance was violently blue and very dark, so there was a subjective white-balance adjustment in ACR, then a lot of Photoshop adjustments in Levels, Curves, and even Hue/Saturation on a per-color basis. Shooting through a double-glazed window with an IR-reflective film didn't help, but I didn't want to open the window when the current temps are 17F outside.

In photography, as in audio, the subtlest things are the hardest to do. Very small errors are surprisingly noticeable - anyone that's ever seen snow knows what it should look like, but the real thing has many small gradations around a narrow gamut, and in this photo, the same thing is happening in the sky. If I were a better Photoshop guy, the impression of luminosity in the sky would be more realistic - and I would have had the wit to take more pictures with wider range of exposures. The original pix was probably underexposed - despite a +1 stop compensation - and that meant the gamma curves in the camera were doing odd things to the tonal scale. I also need to get around to color-calibrating my 21" Samsung monitor - I suspect the sky looks reddish instead of yellowish to many people.

This was a picture that wanted to be taken, with a feeling of stillness and suspended time. Ansel Adams is right when he said that photography was about conveying emotional states, and all the elements of technique are in service of creating that state in the viewer.
 
nickmckinney said:


Underhung wastes gap flux since the coil is not using all of it at any given time. So for the same magnet size you usually end up with less sensitivity all else being equal even after the lighter mass of the VC is accounted for.
...



Thanks a lot for the insights.

But on the other hand, overhung voice coil also wastes some of the magnetic field generated by the coil itself because of part of it not being covered by the gap, doesn't it?

If so, for maximum sensitivity, it seems a short (concentrated) gap with an almost equal length coil would be the best combination. (For a short throw midrange of course) --- Oh no, this looks like a guitar driver (except for the cone).
 
diyAudio Member RIP
Joined 2008
CLS said:
But on the other hand, overhung voice coil also wastes some of the magnetic field generated by the coil itself because of part of it not being covered by the gap, doesn't it?

If so, for maximum sensitivity, it seems a short (concentrated) gap with an almost equal length coil would be the best combination. (For a short throw midrange of course) --- Oh no, this looks like a guitar driver (except for the cone).


An overhung wastes as well, just not near as much.

Having the same gap size as coil is a recipe for serious distortion the moment the coil tries to leave that gap. And yes guitar drivers are usually designed to distort, thats part of the sound.
 
CLS said:



Thanks a lot for the insights.

But on the other hand, overhung voice coil also wastes some of the magnetic field generated by the coil itself because of part of it not being covered by the gap, doesn't it?

If so, for maximum sensitivity, it seems a short (concentrated) gap with an almost equal length coil would be the best combination. (For a short throw midrange of course) --- Oh no, this looks like a guitar driver (except for the cone).

This is quite true. The underhung is very linear until it reaches the end of its travel then it has very high nonlinearity orders (read audible). It also suffers low efficiency and hence thermal problems (all stated already). The overhung does waste voice coil turns in a low B field but has a softer "clip" and hence better over-excursion handling, but how the magnetic field looks at the edges is a big factor here.

I did a study of how to maximize the third order non-linearity of the BL(x) because the lower the order the lower the perceived distortion. Third order is almost inaudible - certainly the lowest of the symmetric orders. A rather interesting result was to spread out the B field as wide as possible by tapering the gap - quite the opposite to what is usually done. This maximizes the B field usage (maximum BL) while minimizing nonlinear orders above the third. The total THD goes up (who cares - or at least I don't), but the thermal distortion and the perceived nonlinear distortion go down. To me the choice of design was a no brainer.

Had Ai survived we would be making speakers like that now.
 
gedlee said:


.... A rather interesting result was to spread out the B field as wide as possible by tapering the gap - quite the opposite to what is usually done. This maximizes the B field usage (maximum BL) while minimizing nonlinear orders above the third. ....

Wow! How interesting. That's really new to me. I guess none was made like that commercially.
 
diyAudio Member RIP
Joined 2008
gedlee said:
A rather interesting result was to spread out the B field as wide as possible by tapering the gap - quite the opposite to what is usually done. This maximizes the B field usage (maximum BL) while minimizing nonlinear orders above the third. The total THD goes up (who cares - or at least I don't), but the thermal distortion and the perceived nonlinear distortion go down. To me the choice of design was a no brainer.

Had Ai survived we would be making speakers like that now.


Sounds interesting indeed, are you talking about making the top plate king of pointed on its inner diameter?
 
gedlee said:


...
I did a study of how to maximize the third order non-linearity of the BL(x) because the lower the order the lower the perceived distortion. Third order is almost inaudible - certainly the lowest of the symmetric orders. A rather interesting result was to spread out the B field as wide as possible by tapering the gap - quite the opposite to what is usually done. This maximizes the B field usage (maximum BL) while minimizing nonlinear orders above the third. The total THD goes up (who cares - or at least I don't), but the thermal distortion and the perceived nonlinear distortion go down. To me the choice of design was a no brainer.

Had Ai survived we would be making speakers like that now.
Some people think that it's easier to compensate digitally.
 
A rather interesting result was to spread out the B field as wide as possible by tapering the gap - quite the opposite to what is usually done. This maximizes the B field usage (maximum BL) while minimizing nonlinear orders above the third.

By spreading the B field would that be at the fringe field? If so, I see two potential problems (I am thinking out loud here). a) Wouldn’t that geometry have the lines not be perpendicular to the coil? b) Wouldn’t the B field be less "dense" there? If that's the case the effective BL (x) would be reduced negating any net advantage from the larger B field.

What you state is interesting and I can see how it can be valid. Could you please elaborate a bit on the mechanics and/or theory of how this works.
 
nickmckinney said:



An overhung wastes as well, just not near as much.

Having the same gap size as coil is a recipe for serious distortion the moment the coil tries to leave that gap. And yes guitar drivers are usually designed to distort, thats part of the sound.

Some drivers have had great success with this, the Audax PR17OMO is a very excellent sounding mid, with a dinky .5mm xmax. The Fostex FF225k is another, great sound, .3mm xmax.

Case in point, I've thought of requesting one of your 6.5"s with a 1mm xmax underhung coil, keeping the efficiency pretty high but the distortion low within that (but also requiring careful control of the excursion).

We'll see how the budget looks after the silly season, but pairing one of those with the heil AMTs I have here, might work real trick.
 
nickmckinney said:



Sounds interesting indeed, are you talking about making the top plate king of pointed on its inner diameter?


I think that rounded is the better description, but yes, thats the idea.

If you are interested, I am going to do a seminar at ALMA in January on magnetic modeling and I will use this study as a case study in that class. I will show how to do the magnetics model, transfer those results into Mathcad and use Mathcad to calculate the BL(x) curve and how this design change minimizes all orders but the third, which, of course, has to increase to do this. So total THD goes up, but audible THD goes down.
 
diyAudio Member RIP
Joined 2008
gedlee said:



I think that rounded is the better description, but yes, thats the idea.

If you are interested, I am going to do a seminar at ALMA in January on magnetic modeling and I will use this study as a case study in that class. I will show how to do the magnetics model, transfer those results into Mathcad and use Mathcad to calculate the BL(x) curve and how this design change minimizes all orders but the third, which, of course, has to increase to do this. So total THD goes up, but audible THD goes down.


Better yet, let me talk John into making a couple.
 
Great comment from Dr. Geddes on a subjective-distortion optimized gap structure for overhung voice coils. Most interesting!

I'm curious what effect localized saturation of parts of the top plate and the pole piece would have on the field lines within and just outside the gap. In one sense, saturation might be desirable since it isolates the magnet structure from the reverse fields induced by the VC audio-frequency currents (similar to a current source in an amplifier's power supply). On the other hand, a partially saturated gap has the disadvantage that parts of the gap are in full saturation, while other parts are only approaching it, which will bend the field lines on a dynamic basis as the saturated area grows and shrinks under dynamic drive. (Saturated areas reject additional magnetic field lines and force them into adjacent areas that are less saturated.)

Mapping out the geometry of which parts of the core approach saturation first is a big preoccupation for audio transformer designers, since it has a strong effect on distortion.