History: A few years ago, I purchased a pair of S99 Lancers (w/o fretwork grills) on ebay and I use them on my computer (through vintage Marantz 7T preamp and 15 amp, other similar vintage gear, and an Echo Layla24 sound card). I recently turned the system on with the Volume cranked (thank you microsoft for screwing up the volume controls on Vista x64) and blew one of the LE20-1 tweeters. Back to ebay. After removing the grill, I notice the woofer surround is kind of hard - researched and decided to try the Brake Fluid (haven't actually done it yet), refoaming only if necessary. While I have the speakers partially disassembled, I might as well rebuild the crossovers.

Researching the various forums, I found the schematic for the LX4-1, the Thiele small signal parameters for the LE14A woofer, and tests on the tweeter (http://www.troelsgravesen.dk/JBL_LE20-25.htm) and started to do some calculations. I also read about the "dreaded" midrange hole and the mismatched drivers in the S99 (I actually like the sound!). The calculations and data proved interesting and informative. The JBL data sheet stated a 2kHz crossover frequency, but when I did the calculations using 8 ohms for the tweeter (that's what's on the LE20-1 labels), I came up with a crossover frequency of 2.5kHz. Wait a second - the tests at troelsgravesen.com show the actual impedance in the 2 - 2.5 kHz region to be 4 ohms, not 8 (although it does rise to 11 ohms at 20 kHz due to the voice coil inductance). When I redid the calculations using 4 ohms, I get a crossover frequency closer to 5 kHz!!!!!! No wonder there's a "Hole in the middle". The small signal parameters for the woofer shows that it starts to roll off at about 800 Hz due to it's relatively high voice coil inductance. The LE20 tests yielded an LE20-1 resonance spike at about 1380 Hz in free air, but the LX4-1 crossover 5 ohm, 8uF, 2.5mH impedance compensation circuit is a notch filter at about 1125 Hz. I'm going to assume that the JBL engineers designed this through testing the tweeter in the cabinet, and that accounts for the resonance frequency difference (if they were off that much, they completely missed the spot).

The tapped 3mh inductor really has no impact on crossover frequency calculations, as it hits 8 ohms down around 400 Hz (4 ohms at 200 Hz), well below the capability of the tweeter. It provides only L-pad attenuation.

What I'm going to experiment with, is to replace the series 8uF cap with 16uF, 24uF, 32 uF ... (doing this gradually, in steps) to get the crossover frequency as close to the 1125 Hz resonance freq as possible (or even going lower into the natural roll-off of the tweeter). Undoubtably, this will increase distortion, and I'll have to determine what sounds best to me and what I can live with.

Any thoughts anyone ????