Great Curves

[Patrick Bateman]

What drivers offer the BEST BL curves these days?

In a recent thread about Danley's tapped horn, we got a bit off-topic and began discussing BL curves. It's my sincere belief that a flat BL curve is all but essential for a subwoofer, since excursion will be huuuuge at high power levels. As Dan Wiggins has evangelized, flattening the BL curve reduces distortion. As I understand it, flattening the BL curve also keeps the frequency response consistent at low & high power levels. On top of that, it reduces power compression.

Npdang has done klippel measurements on two VERY similar woofers, demonstrating the difference between a flat and an elliptical BL curve.

First, a elliptical BL curve, which is typical of 90% of the woofers out ther:
elliptical bl curve


Second, a flat BL curve, which is unique to modern woofers that use a number of motor configurations to flatten the curve. In this case, Adire's XBL^2:
Flat BL Curve

[CeramicMan]

If you're talking about extreme loudness, could it be a question of how "hard" you want the mechanical clipping to be?

Should the cone bottom-out suddenly with high order distortion products, or should it have softer distortion that starts earlier?

[Sabbelbacke]

This sure is interesting. Sadly, a Klippel or Dumax-Machine is very expensive. Would be great to find ways how to inexpensively measure BL curves.

[Patrick Bateman]

Originally posted by CeramicMan
If you're talking about extreme loudness, could it be a question of how "hard" you want the mechanical clipping to be?

Should the cone bottom-out suddenly with high order distortion products, or should it have softer distortion that starts earlier?

This has little to do with dynamic clipping. It's a fact that the thiele small parameters of a speaker change as the BL curve changes. In other words, the actual frequency response changes depending on how far the voice coil is out of the gap.

Have you ever noticed that some speakers sound radically different at high volume then at low? Part of this is due to the Fletcher Munson curves, but a big part of it is due to BL distortion and dynamic compression.

Make sense?

[Sabbelbacke]

Yes, TS Parameters change quite a bit at high powers. Apart from BL getting weaker, the heating of the voicecoil changes, hence Re changes and therefore Qes and Qms. CMS also changes and I´m sure more effects concerning the "motor" are there as well. The Cone also breaks up in more distortion and looses stiffness...
As far as I´m informed, most simulation-programs don´t take those effects much in account. LEAP has a extended TSP-Set where some extra measurements are necessary at high levels and LPSCad simulates things as BL-Change, Temp, CMS-Nonlinearity but does this mainly from a theretical model (e.g. temp is derived from the watts you enter). An additional measuremt of Le and Re at higher voltage helps LSPCad to "guess" better. But AFAIK there is no complete simulationmodel taking all known faktors into account. Maybe Mr. Klippel has something?

A flat BL Curve sure helps

Measuring distortion over excursion helps a little to clear things up how good a speaker behaves at high levels, but so far I found no way of deriving a BL-value directly from this. It´s more like finding xmax(lin) like most companies do (Eminence has an easy to understand short description in their Page).

So if someone has a DIY Dumax/Klippel-Method only using a Measurement Mic, a PC with a little easy to build elektronik and some Software and mountings for the speaker as sold by klippel.... Would be great to "develop" such a thing to investigate ourselves in drivers' BL.

[DevilDriver]

I stumbled upon this during a search, but there is a method in which you can calculate BL yourself utilizing a DC power source, a force gauge, a current meter, and a ruler. Attach the scale to the voice coil, apply current, measure the force generated, and divide by the current to get BL. Pull the scale to move the cone 1mm. Measure force again. Continue to do so until you have enough resolution to identify a point where BL is down by 30%. This is relatively similar to the process used by DUMAX.

You can account for suspension limits as well using similar methods.