shaunp said:
How does the frequency response of a speaker change as you crank up the volume?
Loudspeakers played at high power levels exhibit a phenomenon known as
thermal power compression. Loudspeakers are very inefficient at converting electrical input to acoustical output. On a good day, this conversion factor might approach 5% (100 watts in = 5 acoustic watts out). So, what happened to the other 95 watts? It is dissipated from the voice coil as heat and heat is an enemy of good sound reproduction.
Now, the wire (typically copper) used to wind the voice coil has what we call a thermal coefficient of resistance...meaning that as the temperature of the wire increases, the resistance of the wire increases proportionally. If you're familiar with Ohm's Law, you know that one method of calculating power is the square of the voltage divided by the resistance, E^2/R. So, if our voltage is constant but our voice coil resistance is rising with temperature, the linearity of the power output of the speaker decreases as the voice coil temperature increases. Hence the term thermal power compression.
Well, you're thinking "that's all well and good...but what does it mean in the real world?" OK, take a look at the attached file, PC1.jpg. It shows the frequency response of a 4" cone midbass driver "cold" and after being driven with a 35 watt MLS stimulus (wideband test tone similar to pink noise) for 45 minutes.
What we see after 45 minutes is an average loss in sensitivity of 3-5 dB from the speaker's resonance frequency up to 2 kHz. Above 2 kHz the loss in sensitivity is as great as 15 dB (more on this later). This is a significant loss in sensitivity as the rule of thumb says you need to double your input power to gain 3dB of output from a loudspeaker.
Back to the loss of output above 2 kHz. This particular design used an aluminum voice coil bobbin and the transmission of heat from the bobbin up to the glue joint at the cone/bobbin junction changed the compliance of the adhesive (made the glue softer) which damped the high frequency transmission from the coil to the cone. Actually, this is probably the only positive effect of the voice coil heating.
What would happen if we could lower the operating temperature of the voice coil? Well, in a past life, I was a product manager for Ferrofluidics/Ferrotec, a company that manufactures
ferrofluid for loudspeakers so I spent a fair amount of time dealing with loudspeaker thermal issues. I took the same loudspeaker under discussion and installed some ferrofluid in the magnetic air gap around the voice coil and repeated the measurements. The results are shown in the attached file PC2.jpg.
As you can see, the presence of the ferrofluid not only reduced the amount of thermal power compression, but it preserved the linearity of the speaker's output...i.e. if you scaled the hot curve up to meet the cold curve, there would be much more commonality between the two than what we saw in the non-ferrofluid results. In particular, the response above 2 kHz has been preserved since the heat from the voice coil was now being transfered through the ferrofluid to the magnet structure rather than up the bobbin and into the adhesive joint.
So, to summarize, heat is an enemy of good sound reproduction at high power levels. Power compression effects can be minimized not only through the use of ferrofluid but also but also through other design elements such as larger diameter voice coils and large magnet structures which also help to dissipate heat away from the voice coil and minimize the effects of thermal power compression.
. I'd like to eventually tie this knowledge into my micing setup (which mic to use for quiet vs. loud? where should it go?, etc.). For now though I'm mainly interested in the speakers themselves. Thanks for the help!