What would you consider the limits of speaker response in terms OF +/- specs? I've noticed many manufacturers list speaker response between x and y, but on their own graph it looks like it's definitely with a +/- of 10 dBs sometimes. What do you consider the audible limits? I always thought that +/- 3dBs was the norm.

 

It ultimately depends on what your source material is. If it is a crap MP3 its tolerance is less than a CD. It is also a formula so what you do on one side of a formula directly impacts on the other side. These can be manipulated to give the spec that will sell the most units.

 

Denonite - as you have noticed, a GOOD manufacturer lists the +/-, a decent one keeps it to +/-3db even if they don't specify it, and the rest go to +/-10db just to try to extend their claimed range.

I've seen woofer manufacturers that use enclosures, and sometimes small field measurements, to try to improve their advertised low end extension.

How about sensitivity measurements (that are always suspect anyway) that are at 2.87volts even though the driver is 4ohm. Um, that's two watts, not one.. OR measured at half a meter rather than a full meter. If they do both, 2.87v @ 4ohm @ 1/2m, their sensitivity spec is likely 6db overstated..

All number games

 

I've very rarely seen this published, but I've found 2nd and 3rd order harmonics, and driver distortion in general to be a determining factor in how great a driver sounds. Sometimes a "flat" driver sounds like crap because the spectral response is all over the place.

For higher frequencies, above 1khz, the baffle that drivers are tested in also has a profound effect. For example, you can flush mount a tweeter in a 1m wide baffle and get a certain response, and then flush mount the same driver in a 0.5m wide baffle and get a slightly different response. Then you ask how relivant is THATwhen we are installing a tweeter in an A-pillar. Then all of a sudden the waterfall plot falls back into the forefront....

I have seen many drivers from well known manufacturers test within the limits the specify. +/- 3db is often met...but beware the smoothing applied to the plot, and the number of sample points.

Dukk brings a GREAT point up about sensitivity. One of my pet-peevs on this topic is drivers that are spec'd by their peak output at 1w/1m. So what about all the rest of the spectrum you play with these drivers at say 6 db below the peak. Not very accurate...but part of that can be blamed on a complete lack of proceedures or guidelines when specifying efficiency. Personally, I'd like to see efficiency rated as a mean average across the +/- 3 db range.

Darn...I was hoping if I ranted on long enough...we'd end up off topic. :P

 

How 'bout them Leafs??.....

 

My cat's breath smells like cat food...

 

Hmmmmmm