JBL LE26 tweeters
Copyright 2007 Troels Gravesen

"Why do you buy all these old JBL tweeters?", one of my friends asked, having noticed quite a few LE26 tweeters on my speaker shelves. "Well, hmm....I just like to have a lot of them...", I replied. "OK, fair enough", he said. I like that attitude. Anyway, there's a good reason I've bought all the LE26 tweeters I can find on eBay at sometimes 50 US$/pair, sometimes 80 US$/pair. I now have 5 pairs and that'll be it - I think - and maybe I'll sell two pairs again, once tested and fitted with waveguides.

1. The LE26 is quite sensitive; average SPL/2.8V is around 94 dB. From 6-15 kHz sometimes up to 96-97 dB/2.8V.
2. It has a large membrane area, almost twice that of a standard 1" dome, so it doesn't have to move much.
3. Membrane material is paper and I've found it integrating well with e.g. Supravox drivers. Paper + paper....
4. A waveguide can relatively easy be made to raise the 2-6 kHz region further, thus reducing distortion towards point of crossover.
5. The LE26 is better than the LE25 - for some reason. Only difference appears to be the colour of the paper pulp used and coating of the surround fabric, extending the frequency range from 13 kHz to 17 kHz and getting rid of the 12-13 kHz peak of the LE25.
6. Last but not least, high-efficiency paper cone tweeters are becoming rare as hen's teeth these days.

Initially these LE26 tweeters were meant for the JA8008 high-efficiency 8" driver while seeking for a modern alternative, eventually being the Audax TW034 dome. The LE26 tweeters are obviously no longer available but quite often for sale at eBay. By coincident, the waveguide for the LE26 will also fit the TW034 dome. However, having ten LE26 tweeters at hand made a rare opportunity to report the spread in performance of these tweeters. The JA8008 driver has a midrange sensitivity of around 95-96 dB/2.8 V and tuned to a system sensitivity of around 94 dB I needed more than average sensitive tweeters to match this driver. I'd rather have a large dome with an upper limit of 15 kHz compared to most modern 1" domes doing well up to 20 kHz, but not particularly well below 3 kHz. Yes, I know some domes go below even 1 kHz, but they don't sound good - to my ears. The LE26 does well up to 15-17 kHz, more than enough. The current trend of claiming 30-40 kHz is ridiculous. We should rather worry about what our tweeters do between 3 and 12 kHz where "treble" is. Serious resonances between 15 and 20 kHz may disturb lower registers, as the terrible alu dome introduced in replacement of the LE 25 and 26. Actually this dome was removed again due to complaints about performance. Later on, a titanium dome was introduced doing better than the alu dome, but they could have stayed with the paper cone tweeter, which became very much out of fashion during the late Seventies and Eighties.

Instead of the LE25's foam surround, the LE26 comes with a plastic faceplate, slightly conveks, which is not such a great idea. Inverting the faceplate makes a better performance as seen here:

Red = conveks faceplate, blue = inverted faceplate.

Design before performance, but not a big deal. All measurements below are done with inverted faceplate.
Removing the faceplate takes a gentle twist by a screwdriver. Some sticky gue is used for mounting the faceplate. Soak it in xylene, leave for a minute or two and scrape it off. Take great care not to damage the fragile wires to the voice coil.


Left: Sample 1-4, SPL @ 0.25 meter to front panel. Inverted faceplate, flush-mounted on a 30 x 40 cm baffle.
Right: Sample 5-8, SPL @ 0.25 meter to front panel.
Input signal adjusted to resemble response @ 1 meter distance, 2.8 V.

Left: Sample 9-10, SPL @ 0.25 meter to front panel.
Right: Impedance of sample 1-4.
Input signal adjusted to resemble response @ 1 meter distance, 2.8 V.

Left: Impedance of sample 5-8.
Right: Impedance of sample 9-10.

As can be seen from the graphs above, these tweeters perform very well. Impedance is surprisingly uniform, making crossovers easy.
Also the frequency responses are excellent, basically flat from 1.5 - 15 kHz. This is as good as many modern tweeters.

Left: Sample 3 fitted with modified MCM waveguide.
Right: Sample 4 fitted with modified MCM waveguide.
The waveguide will raise on-axis response from around 1.2 - 6 kHz.

Left: Sample 9 at 0, 10, 20 and 30 deg.
Right: Sample 9 + waveguide at 0, 10, 20 and 30 deg.
As can be seen the waveguide makes a more even power response compared to the un-modified tweeter. "As-is", the roll-off starts already at 3 kHz, where the LE26/waveguide roll-off starts at 8 kHz and the loss at 30 deg. off-axis at e.g. 15 kHz is around 5 dB compared to 10 dB for the un-modified tweeter. Quite a difference.

Left: Distortion test on sample 9. Distortion curves raised 30 dB.
Right: Distortion test on sample 10. Distortion curves raised 30 dB.
Blue = 2nd, green = 3rd, yellow = 4th, purple = 5th.
These measurement were done at 0.25 m distance and with an average level of 97 dB, which makes some 91 dB @ 1 m distance.
From 2-7 kHz the 2nd harmonic is some 25 dB + 30 dB = 55 dB below average level, rising to 50 dB above 7 kHz. These are insignificant levels. Previous measurements on the LE26 displayed higher distortion levels, but were done at 6 dB higher input, so the good question is at what level these measurements are relevant. Playing music at 90 dB is pretty loud, so these measurements are well in accordance with my subjective evaluation of sonic performance from the LE26 tweeters.

Left: Cumulative spectral decay of LE26 sample 9.
Right: Cumulative spectral decay of LE26 sample 10.
I'm pleased to see these graphs, displaying a clean decay above Fs.

Fitting the MCM waveguide to the LE26 tweeter takes a simple flange, here made from 4 mm MDF. Dimensions seen below. As can be seen from the photo to the right, the tweeter corners were cut off and smoothed in a lathe.


Another flange for the MCM waveguide is needed. Dimensions seen above.

You need a lathe to do this. The hole for the tweeter diaphragm is 43 mm in diameter.