So far, we talked about frequency response and the importance of tuning drivers individually, and we talked about the importance and purpose for time alignment. In this post, I'll go over crossover points. My article on my site went into reasonable detail regarding what crossovers are and how active and passive crossovers work.
Since we aren't going to be buying a component set with a passive crossover bundled and will instead be buying raw drivers, we need to figure out what our crossover points should be. Since raw drivers rarely list an advised crossover point and slope, how do we figure this out?
Measurements. For this part, we use software called ARTA to measure harmonic distortion. Harmonic distortion affects a significantly large frequency range and will not come up in a frequency response measurement. The most offensive to our ears is 3rd order harmonic distortion. I can help anyone learn how to use this software since there's a small learning curve, but basically, you use a measurement microphone to get the reading you need. We can just skip ahead to the measurements you'll get. What we do is start with a low crossover point and take a measurement.
For the tweeter, we need a high-pass crossover filter, which means we need to let the high-frequencies pass through the filters and let all frequencies below the filter be blocked out. We naturally don't want our tweeters to be playing bass, but we want them to play as low as possible to lift our sound stage above and away from our feet. The purpose of harmonic distortion measurements is to determine just how low we can play them before they start to sound fatigued and strained, and the best way to do that is to measure them.
Here are some measurements I took while tuning a friend's Scion tC. In Red, I outlined the harmonic distortion (gray line). In blue, I outlined a particularly harsh frequency response peak that we either need to tune out with a parametric EQ, or filter out with a crossover. In this particular case, you'll see that peak reducing as we raise the crossover frequency.
We start our measurement at 800Hz. This way, we get a good idea of the big picture and have a starting point. I would advise no lower than 650Hz for dome tweeters. Note the large peaks in harmonic distortion and frequency response.
Going to 1500Hz reduced both peaks, but they're still too high for comfort, and I know better than to think this tweeter can do 1500Hz. Something also seemed to mask the distortion in the first peak at 2500Hz, so we'll just keep going.
Up to 2000Hz now. Our frequency response peak is more manageable, and we are seeing a more accurate picture of harmonic distortion. Our biggest concern at this point is the HD peak at 2800Hz.
Let's go to 2250Hz. Looking better!
At 2350Hz, we see a reduction in both Harmonic Distortion peaks!
At 2500Hz, we don't see much of a consequential change.
At 2750Hz, we see the first HD peak at around 2800Hz being reduced, but only because all output is being reduced now, not because our crossover point is causing harmonic distortion.
At 3000Hz, we see a notable reduction in total Harmonic Distortion, but we've now lost a lot of sound stage due to such a high crossover point. This is what you'd typically find in a component set's passive crossover.
Let's try to tie this together.
From 800Hz on up to 2250Hz, we see harmonic distortion at high levels, with peaks at ~2800Hz and ~4000Hz. As we increase the crossover point, we see those harmonic distortion peaks reducing, even though we are not filtering those specific frequencies! The intention of these charts and this post is to point out that we are getting distortion at 2800Hz and 4000Hz simply due to us forcing the tweeter to play lower than 2000Hz. To put it simply, we are playing the tweeter too low out of its designed range, and that is screwing up the rest of the tweeter's range.
It is typically assumed that if we have a tweeter that can play down to a given frequency such as 2000Hz, that if we cross it down to 1500Hz to have it play 1500-2000Hz, that only that range would be affected. It is very important to realize that in addition to hearing distortion at 1500-2000Hz in this hypothetical scenario, we will also be creating distortion above 2000Hz.
In this particular case, any crossover point above 2350Hz does not affect our harmonic distortion in this way, and every frequency below 2250Hz does create additional harmonic distortion. As a result, for this particular tweeter, I would recommend a crossover point of 2250-2350Hz, but no lower. This allows us to fully utilize the tweeter's capabilities without creating unwanted distortion.