A Phase Linear Transmission Line Project

Section 7: Crossover simulations


Figure 7.1 shows an early version of the tweeter response mounted in the same 35 cm wide box as the woofer. The severe peaking is diffraction oriented. (Solving this required a small box on top of the transmission line.)

More interesting in Figure 7.1 is the comparison of the 1st order 500 Hz slope and the driver combined with an 500 Hz 1st order active filter. The slope is really a 3rd order slope below crossover, which results in the dip around 750 Hz (Figure 7.2).

cross1_tn.jpg (25279 bytes)

Figure 7.1. Comparison of JX53 to 1st order function


meas1_tn.jpg (20837 bytes)

Figure 7.2. Crossover induced dip at 750 Hz


cross2_tn.jpg (65157 bytes)

Figure 7.3. Calsod predicted dip


cross3_tn.jpg (65157 bytes)

Figure 7.4. One JX53, no crossover


cross4_tn.jpg (65157 bytes)

Figure 7.5. Two JX53s, no crossover


cross5_tn.jpg (65157 bytes)

Figure 7.6. Four JX53s, no crossover


Simulating the setup in Calsod (Figure 7.3), the dip is also predicted here.

The constructor of the drivers, Ted Jordan, recommends using several drivers with no crossover. A single driver with no crossover is simulated in Figure 7.4. The dip is completely eliminated. Sadly, the driver needs at least the attenuation of the 1st order 500 Hz filter to be able to play music at acceptable levels.
     

cross6_tn.jpg (65157 bytes)

Figure 7.7. Calsod simulation of 250Hz 2nd order LP Bessel filter for JX53


Adding 2 and 4 drivers are shown in Figure 7.5 and 7.6. Off-axis, the response gets quite distorted from what the measurements shows. Still, near field effects will come into play, and the power response might even it out. So it might sound quite good. Now, it's getting the money to try it out...

An alternative solution was thought out. What if a higher order filter was placed a little lower in frequency? This will allow a better alignment to the 500Hz 1st order response around crossover. In Figure 7.7 we see the filter effects the off-axis response rather evenly. This might sound good, but the filter remains to be tried out.


[ ContentsIntro | Measurements | Optimization | Integration | Step Compensation | Time Align | Crossover | Construction ]