Thinking of building a pair of JBL CBT 50LA clones

[Aijan]

I'm thinking of building a pair of JBL CBT 50LA clones for surround duty in my home theater. These are passive speakers that use a line array of eight 2" full-range drivers.

The drivers are arranged into four groups in the following configuration: 1-8, 2-7, 3-6, and 4-5. The drivers in the middle (4-5) operate at full level, while the surrounding drivers' frequency responses are shaded (tapered) to minimize interference caused by the center-to-center spacing between the drivers. Also, JBL uses a clever trick to make these straight columns of speakers behave as if they are curved by applying delays to the drivers near the edges, further minimizing interference. These are supposed to be great as surround speakers because their level doesn't vary much even if you sit very close to them.

Fortunately, JBL published the crossover schematic in a technical manual, and I've whipped up a quick model in VituixCAD using the exact crossover filter values. To my surprise, the CBT (constant beamwidth technology) effect seems to be working fine, even though I plan to use a different 2" driver: the Lavoce FSN020.71F. I believe driver selection isn't overly critical, as long as all drivers are identical and correctly attenuated and delayed.

Here's the vertical directivity (about +/- 15 degrees nominal):

The crossover and its effect on frequency response and phase can be seen below. The notch filter at the top left wasn't part of the original design, I added it to tame a peak.

Please let me know if I'm missing something before I go ahead and build these. Thanks!

 

[sneakmasterG]

If you're lazy like me, just wait until a batch of b-stock becomes available and order the real thing. You won't learn anything, but you won't wonder how DIY compares. Handy mounting hardware is available, including flush-mount brackets.

The price differential for b-stock vs. DIY parts may not be much.

https://www.jbl.com/pro-b-stock-sale/CBT+50LA-1-WH-Z.html

I'm using four b-stock CBT-50la for surround speakers in my 7.4.4 setup and like them. Either way, have fun!

 

[kemmler3D]

The vertical directivity looks like things are working as intended, but the overall FR looks concerning. That dip at 1Khz and peak at 2Khz will definitely be audible and harmful to SQ. Probably the dip around 7khz and peak at 10khz too. Luckily (?) they are reflected in the off-axis simulation too so maybe you can just EQ them out with the AVR? Personally I wouldn't proceed with this without trying to correct the FR more, but I recognize the crossover is already somewhat complex.

 

[Aijan]

Thanks for the feedback. Buying the originals crossed my mind, but where I live they're not very affordable. So, building these myself out of baltic birch plywood instead of ABS for maybe a quarter of the price is tempting.

I've tweaked the crossover and included the 45-degree driver measurement from the official spec sheet to better model directivity. Here are the updated graphs (note the 500 Hz–20 kHz range and the 0–45 degree intervals). These speakers will be ported, so they should end up flat down to maybe around 120 Hz. I think the vertical directivity is a bit narrow after 3 kHz at only +/- 10 degrees, but the original has this as well. What do you think?

 

[Aijan]

According to a diyaudio thread, the drivers JBL used were likely 4-ohm units, combined with a 2:1 transformer to bring the impedance back up to 8 ohms. Since the drivers I plan to use are 8 ohms, I'll need to modify the crossover by doubling the inductors' values and halving the capacitors' values, if I'm not wrong. These changes improve vertical directivity, and the filter slopes now make more sense. Please let me know what you think.

New vertical directivity:

New crossover:

Old crossover:

 

[Aijan]

On further thought, it appears that adjusting only the first inductor-capacitor pair to accommodate 8-ohm drivers is sufficient, since the following crossover sections depend on this initial stage. After the adjustment, the frequency responses for the D3–D6 and D2–D7 pairs don't change significantly, and they are attenuated by about 3 dB and 6 dB respectively. The primary role of this rather unusual crossover is to gradually increase delay for the drivers flanking the middle pair, causing the column to behave as if it's curved. Finally, vertical directivity is better behaved this way.