Tweeter Ball No.3181 for ESS Heil AMT Tweeter
Introduction
A fellow enthusiast reached out to me and asked if I could design a tweeter ball for the new reproduction version of the ESS Heil AMT tweeter. I was excited about the idea and so I cut him a 50% discount on the design since we could use this design for ongoing in-house design development.
The ball measures 32cm in diameter and proportions nicely with a 10" onken cabinet I designed and built last spring.
I printed the tweeter ball on my Elegoo Pro Max printer. I changed the nozzle from 0.4mm diameter to 0.6mm diameter to allow for a faster print time. Even with these changes the print time was 40hrs for a single half sphere. It took about ten attempts to print successfully because I was having bed adhesion problems. The solution was to raise the bed temperature from 60C to 64C and upgrade the filement from standard PLA to PLA+. I tried a variety of supports from tree to grid. I settled on a large square grid pattern.
The tweeter ball uses a wall thickness of x3 layer (0.6mm x 3 = 1..8mm thick) with 15% infill.
DFT RAW frequency response (unsmoothed) overlay (Tweeter ball RED) (Stock BLACK)
On-Axis Response overlay with 15,30,45 Degrees Off-Axis
Horizontal Off-Axis Colored Polar Map
Vertical Off-Axis Colored Polar Map
CSD (25dB vertical scale)
Burst Decay (25dB vertical scale)
85dB Harmonic 1m
95dB Harmonic 1m
SMPTE Distortion using 1kHz & 10kHz dual sine test tone (4:1 Power Ratio)
This test reveals that H2 is dominant with H3 extremely low in both. (See 2kHz and 3kHz respectively)
IMD products are shown at 9kHz and 11kHz.
Conclusion
Tweeter ball No.3181 improves the frequency response smoothness and resulting time domain performance (CSD and Burst Decay). Although directivity narrows with the tweeter ball, the overall behavior is much more linear indicating reduced diffraction artifacts. We see a sensitivity increase across the usable bandwidth with the resulting lower distortion (see SMPTE graph above).
For the tweeter ball, horizontal coverage is near-constant-directivity with a nominal coverage window of 120 degrees in the 2kHz region narrowing to 80 degrees at 10kHz.
It's hard to pin down an exact horizontal coverage window with the stock driver because the polar map is quite irregular. Generally we see a slightly wider coverage especially in the 2kHz, 4kHz, 8kHz, and 16kHz regions. However narrowing occurs in the 2.5kHz, 5kHz, and 10kHz. (all multiples of each other). This is likely a function of diffraction related to the physical dimensions of the stock tweeter.
