Quote:

Originally Posted by

**Dange**
That is what I thought, more than one driver and introduce a time delay/phase shift in one

Yes, or you could process it a bit more complex than that, but still it is very much the same thing.

Quote:

Originally Posted by

**Dange**
Because I'm not quite sure of the difference to be honest! I've be used to using the term directivity.....Q and dispersion are somewhat unknowns to me, comes down to terminology possibly

In Norwegian Q is referred to as "the goodness factor" (it is a very bad translation of the not much better expression "quality factor). Noone seems to understand what that means heh The easiest way to understand Q is to use the expression "Q is the ability to resonate".

If you look at an LCR-chain of passive components, it is pretty easy to determine a quite clean value for Q. In other situations, Q is a bit harder to determine precisely.

If you look at a parametric EQ, Q is defined as the divider for frequency when determining bandwidth. An EQ point at 1kHz with Q=0,5 normally means a bandwidth of 2kHz (there are some different definitions here as well). This is more or less indipendent of the level at which you EQ. In the passive components example, the level would connect directly to the Q value.

This difference is simply because Q tells us different things depending on where we use it within the same circuit.

Example:

If you use a woofer/BR-enclosure combination that gives a very smooth and soft roll off, Q is beyond doubt very low (this is the Qb, the Q value for the driver/enclosure-combo. The driver Q is mathematically a part of this equation, but Qb does not directly tell us anything about the driver Q). If you tune the driver so that it rolls off with a sharp edge at the resonance frequency (Fb) the Q is high.

However, this normally means that the level at Fb is much higher too. But if you increase the cabinet volume and reduce Fb, you will still get a very sharp roll off edge, but the level at Fb is now reduced. If you determine Q based on the sharpness of the edge, it will turn out pretty high. If you determine Q based on the level at Fb compared with the reference level at 200Hz or higher, the Q will turn out pretty low.

So throwing Q around without further definition is not very useful.

If you look at a PEQ band, decreasing the Q value should be the same as increasing the dispersion of this band in the frequency domain.

Dispersion pattern, at least what I have been talking about, is the polar plot of a loudspeaker, both vertically and horizontally.

If you put a loudspeaker in an anechoic chamber on a controlled turntable, you can measure the on axis response, then turn the speaker a few degrees, measure the first off axis response, turn a bit more, measure again and continue like this until you have mapped the entire 180 or 360 degrees around the speaker. That is what I am refering to as the dispersion pattern, polar response or energy response.

Dispersion pattern = all the info about the speakers on and off axis behaviour.

Polar response = frequency specific single plane response. However, it is quite usual to put the vertical polar response in one half of the diagram and the horizontal response in the other half.

Spectrum polar response = a version of polar response covering all frequencies by showing level as colour and angle/frequency on the X and Y axis.

Energy response = the sum of all on and off axis measurements plotted in the frequency domain showing the total amount of energy the speaker emits at each frequency.

EDIT: I see that SAC has brought directivity Q into this as well. I guess that answers my question of why you mix it with dispersion. When working with Q in room acoustics, I normally think of room nodes.

It is an old way to express directivity that very much looks like an EQ if you use the X axis as angle and the Y axis as level. The peak you get is the "eq point" and the X axis always have the same extent as the frequency which is being analysed.

The problem with using this is loobing. It very seldom adds up precisely to real life situations and is therefore pretty unuseful. The last 30-40 years it has been completely replaced by polar plots, and later, spectrum polar plots.