One area that gets overlooked quite often, by DIY speaker builders, is the effect that component matching has on imaging. Many folks know what imaging is and how to identify a speaker that images well from one that does not, but not too many folks can readily identify the factors that determine how well a speaker images. I'm not going to go into a long discussion of these factors (in this post anyway), but one area I wanted to mention is component matching.

Most of the components you purchase for your crossover construction have a stated "tolerance" -- how much they may vary from the stated value. For a 150 uF electrolytic capacitor with a 10% tolerance, this means that you might use a capacitor in one crossover that is as low as 135 uF and another in your other channel that is 165 uF. When you're all done, you may have two crossovers that use the same parts (as labeled) but perform differently. Enough to drastically change the frequency response? Probably not (what's a few bumps and wiggles, right?). But is it enough to have an audible effect on the speakers? Probably not to the point where there are going to be any objectionable peaks or dips, but you will end up with very slight differences between your speakers that degrade the imaging of the speakers.

Think of this. Imagine a note being played by a trumpet that the sound engineer balances equally between the channels to place the instrument dead-center on the soundstage. Also, imaging that your left speaker has a little 1/2 db rise at the frequency range where the fundamental of that note occurs. In your right speaker, maybe there's a 1/2 db dip. So what's the effect? The note from that trumpet gets moved from dead-center of the soundstage to just to the left. This has nothing to do with how flat the overall response of the speakers are -- just how well matched one channel is to the other channel. Imagine how undefined the image becomes when you have numerous regions where the speakers don't match very well? Can you hear a difference in response from each channel? Probably not. Will you notice the difference in imaging? In most cases, yes. Maybe you'll just attribute the poor imaging to the room placement (which is always a likely suspect), the crossover design (another likely suspect) or poor consistency between the enclosures and drivers. Again, these are all likely suspects when poor imaging is involved, as are some of the more elusive ability of drivers (some drivers supposedly image well, while others don't -- I don't ascribe much to this notion, since I've heard systems with the same drivers that image exceptionally well and other systems with the same drivers that don't).

Other than carefully measuring and matching your components (drivers, crossover components, enclosures, etc.) I don't have much more advice on how to improve your imaging, just that by doing these things, you'll eliminate a few variables that will help when you tackle the most important contributing factor to imaging -- room placement. Below is a picture of a crossover I built over the weekend for my "project X" speakers. Each of the inductors was matched to +/- .005 mH, the caps were matched to within 1% and the resistors, for all practical purposes, were found to be identical to within the tolerance of my Fluke multimeter.

The impedance graph below is actually an overlay of the impedance of each crossover. Notice how it appears to be a single curve? I'll post a typical one with unmatched components soon for comparison. What the graph below, however, shows is what you want -- two crossovers that will produce identical response

Finally, you'll find that the previous updates are now in the archives (if you haven't already, they've been there for a few weeks) and I've started to create the user project page. I've got many submissions showing the great work you people do and plan to put the links and pics up soon (so keep them coming!) Also, the final write-up of the DB MTM's is coming.