In this blog post I test my newly developed compression driver No.2118-10. The driver is in the prototype stage requiring sound quality validation.
The driver is required for an upcoming system build for a customer who has commissioned me to design and build the complete system. The driver attaches to mid-bass horn No.2118 which is a large Biradial covering 100Hz to 650Hz. I wanted flexibility however on the crossover topology and so I needed to ensure the compression driver was trouble free up to 2kHz, allowing shallow crossover slopes with the ES-450 Biradial above.
2118-10 Driver Build Details:
- Custom designed compression driver
- Utilizes 10" Midrange from 18Sound, model 10NMBA520
- Sealed rear chamber
- 3.5 compression ratio
- 3D printed
- Heavy print settings
- 30mm thick walls
Application
- Large scale horn speaker systems, home listening, 2 channel stereo
Specifications
- 100Hz-2kHz bandwidth
- 117dB sensitivity (1w/1m, 2.83V)
- 8ohm
- 125mm diameter throat
Test Setup
Since the 2118 100Hz Biradial horn is not ready, I decided to test the driver on a 10' long plane wave tube. This length ensures that a full 100Hz wavelength is absorbed by the damped line.
I designed and printed an adapter to connect to the ABS pipe, ensuring a smooth transition. The measurement mic is placed through a hole and sealed with tape.
Measurements
A plane wave tube will allow me to see the linear region of the frequency response. My goal is to provide a smooth frequency response in the 100Hz-2kHz region. If the phase plug introduces any issues it will show up in the frequency response as irregularities.
Impedance Sweep
I began my testing my measuring the impedance sweep. This will tell me where the fundamental mechanical resonance (FS) lies with the sealed rear chamber.
The FS measures at 40Hz with a broad Q flattening nicely by 100Hz. The surround resonance at 900Hz is also seen in the published data.
Frequency Response
I measured the frequency response (red) with the mic placed inside the tube immediately in front of the compression driver exit. I've overlaid the impedance in green for reference. The response is smooth until 2kHz where we see a cancelation dip.
Calculating Sensitivity
In order to get a sense of the compression driver's overall sensitivity I decided to remove the plane wave tube and measure the driver's output using a 400Hz test tone. With the compression driver operating free-air I placed my decibel meter at 1m and set the voltage to 2v. This represents 0.5w input power. The decibel meter read 105dB SPL 1m. I then calculated the predicted sensitivity as follows:
- 105dB at 1/2w is 108dB at 1w (double the power = +3dB)
- Adding the horn theoretically adds +9dB which becomes 117dB
The predicted compression driver sensitivity is 117dB@1w 2.83. This would be a record as far as I know for high sensitivity.
Testing Distortion
When I normally measure distortion I set the DUI test SPL using a calibrated decibel meter. However the plane wave tube method does not allow me to set an output SPL. Instead I have to rely on the electrical input voltage to the driver as a reference. I can then predict the SPL by working backwards from my sensitivity calculation.
Test Voltages versus predicted output with horn:
0.06V = 83dB
0.25V = 95.6dB
1.00V = 110dB
2.00V = 113dB
2.83V = 117dB
Normally I would test at my standard 85dB and 95dB test SPLs. This corresponds to the 0.06V and 0.25V test voltages from the above tabulation.
Harmonic Distortion
0.25V (95.6dB SPL @1m predicted)
Distortion for the 95.6dB test signal level appears very low across the operating bandwidth of the driver. For the 300Hz region H2 is 0.056, H3 is at 0.016%, and H4 is only 0.002%.
Intermodulation Distortion
Testing IMD at the 0.25V (95.6dB predicted SPL) is shown below. IMD is at -65dB for the 300Hz region and -70dB for the 1kHz region.
Gedlee Metric (Gm) Distortion
I then measured the Gm distortion at the 0.06V and 0.25V test signals. Click on the image below to enlarge.
GM distortion for the 0.25v (95.6dB) test signal is only 0.005 for the 1kHz region. This compares favorably against the RCF ND950 which achieved 0.0141 which is an order of magnitude three times higher. 
Conclusion
Compression driver No.2118-10 achieves acceptable numbers so far. It will be interesting to see how the driver does when mounted on its dedicated horn.