Horizontal Menus in CSS by Css3Menu.com

Project-XF18

Project-XF18 is a prototype subwoofer used for evaluating the effectiveness of the DSP Servo. It consists of an IST Mach-5 Audio 18-inch high-excursion subwoofer driver mounted in a 100 litre sealed enclosure that is constructed from 33mm veneered MDF with extensive bracing.The driver has been modified and an accelerometer has been installed beneath its dustcap on a specially designed bridge arrangement to provide feedback to the DSP servo controller. Unlike other vendors of servo subs we don't just test the small signal frequency response curves under controlled conditions that could be obtained from any speaker system. We push the system to the absolute limit and test it's time domain response in order to verfiy its performance capabilities and design objectives. In the following you will see tests that are rarely done on speakers simply because most will fail without servo control.

 

 

 

DSP Servo Setup

The DSP Servo was setup to drive the speaker with or without feedback applied so a direct comparison could be made of its effectiveness. In both cases the amplifier was driven just before the onset of clipping. At the bottom this page we include comprehensive specifications of the driver, accelerometer, and amplifier as well as providing a simulation of the driver in the box with no servo and/or equalization applied.

DSP Servo Setup

Feedback: Yes
Feedback Loops: 2
Nominal Loop Gain: 30 dB
Maximum Loop Gain: 47 dB @40Hz
Current Servo: Yes
Nominal Power during testing: 800 watts
Low frequency cutoff: 20Hz
Low Frequency Alignment Type: 2nd Order Butterworth (Q=0.7071)

 

Toneburst Results

In these tests we applied a 20Hz 5 cycle-on and 10 cycle-off toneburst to the speaker with and without feedback. The results were then compared with the results of the ideal case of a simulated high pass filter with a 20Hz and 36Hz cutoff which corresponds to the speaker with and without feedback respectively. We measured the output from the accelerometer rather than a microphone in close proximity to the speaker diaphragm which would have added it's own distortions. We shall repeat the tests at a later date using a microphone in a ground-plane testing environment.

Toneburst Test with NO Servo

Top Yellow trace is the input signal. Bottom Blue trace is the output from the accelerometer. As you can see the output of the non-servo controlled speaker is heavily distorted and contains severe ringing compared to the ideal case shown below in the red waveform !

The simulation to the right shows the case from the output of an ideal 2nd order Butterworth high pass filter with fo=36Hz, Qo=0.718 which matches the ideal simulated case from WinISD. The green waveform is the input stimulus and the red waveform is the simulated output of the speaker. As you can see from the output of the speaker above that there is barely any similarity between the actual speaker and the ideal case !
Toneburst Test with Servo Enabled

Top Yellow trace is the input signal. Bottom Blue trace is the output from the accelerometer. As you can see the output of the servo controlled speaker is almost indistinguishable from the text book example shown below in red and shows the effectiveness of the servo in controlling the diaphragm motion !

The simulation to the right shows the ideal case from the output of an ideal 2nd order Butterworth high pass filter with fo=20Hz, Qo=0.7071. The green waveform is the input stimulus and the red waveform is the simulated output of the speaker and is indistinguishible from the actual speaker output !

 

Sinewave Distortion Tests

In these tests we applied a test tone to the speaker with and without feedback applied and compared the spectrum from the output of the accelerometer. We measured the output from the accelerometer rather than a microphone in close proximity to the speaker diaphragm which would have added it's own distortions. We shall repeat the tests later on using a microphone in a ground-plane testing environment.

Distortion Summary Table:-

Frequency/THD
THD (Servo-OFF)
THD (Servo-ON)
Net reduction in THD (dB)
15 Hz
30.2 %
0.67 %
33 dB
20Hz
40 %
0.76 %
34 dB
100Hz
3.84 %
1.55 %
8 dB

20 Hz Sinewave Distortion Tests

20 Hz Sinewave Distortion Test with NO Servo

The oscilloscope snapshot to the right shows severe distortion in the measured waveform. The yellow waveform shows the input signal whilst the blue waveform shows the output measured from the accelerometer. The distortion is clearly evident !

The spectrum of the blue waveform in the oscilloscope measurements above confirm the gross distortion (40%) which is audiby very apparent contrary to popular beliefs.
20 Hz Sinewave Distortion Test with Servo Enabled

The oscilloscope snapshot to the right shows almost perfect reproduction in the measured waveform. The yellow waveform shows the input signal whilst the blue waveform shows the output measured from the accelerometer. Apart from the expected phase shift of the output compared to the input the output is virtually a perfect replica of the input and distortion is clearly non-existant !

The spectrum of the blue waveform in the oscilloscope measurements above confirm the clean reproduction and shows the effectiveness of the servo in reducing distortion from 40% down to 0.76%. This corresponds to a net reduction of distortion by 40 dB to negligible levels compared to the servo switched off !

 

15Hz Sinewave Distortion Tests

15 Hz Sinewave Distortion Test with NO Servo
The 15Hz sinewave test is considered a torture test for any subwoofer because it places extreme demands on both the amplification and the speaker driver. As can be seen in the spectrum plot to the right, even at 10 dB less output compared to the 20Hz test the speaker produces nearly as much distortion (30%) !
15 Hz Sinewave Distortion Test with Servo Enabled
With the high gain servo switched on the distortion artifacts are reduced to negligible levels of 0.67% !

 

100 Hz Sinewave Distortion Tests

100 Hz Sinewave Distortion Test with NO Servo

 

At higher frequencies most of the motor distortion (3.84%) comes from the non linearity in the voice coil inductance.

100 Hz Sinewave Distortion Test with NO Servo + Current Drive
With current drive only the high frequency distortion is reduced by 6dB. As well as reducing distortion by about 6 dB it also flattens the frequency response so you don't get that typical high frequency rolloff characteristic that usually plagues high inductance drivers such as this one.
100 Hz Sinewave Distortion Test with Servo Enabled
Because loop gain rolls off after 100Hz most of the distortion reduction comes from the constant current drive which essentially bypasses the parasitic voice coil inductance. Even so, a total reduction of 8dB in distortion has been achieved from a combination of current drive and the servo.

 

Specifications

The specifications of the driver, accelerometer, amplifier and power supply used in the tests is shown below. The specs on the DSP servo controller can be found elsewhere.

Driver Specifications

Manufacturer:
IST
Model:
Mach5-IXL18.2.2
Size:
18 inch
Cone Material:
Thick pulp cone with heavy foam surround
fs:
17.7 Hz
Re:
3.4 ohms
Qes:
0.39
Qms:
5.59
Qts:
0.37
Mms:
434.3 grams
Rms:
8.69 kg/s
Cms:
0.184 mm/N
Vas:
274.1 litres
Sd:
1029.2 cm2
Xmax:
22 mm (one way)
Xlim:
30 mm (one way)
Power Handling:
800 watts

Accelerometer

Manufacturer:
Measurement Specialities
Model:
ACH-01-03

Amplifier

Manufacturer:
Power Physics
Model:
Combo Chassis C-1014
Amplification:
Class-D
Output Stage:
Full Bridge
Power Output:
1000 watt @ 4 ohms
Maximum Power Output:
1300 watt @ 1% THD
Power Supply:
Universal Voltage - Switch mode with Power Factor Correction

 

Simulations

From WinISD the system response function of the driver in the box without the servo or any EQ is shown below. As you can see in the ideal small-signal case it should behave like a 2nd order electro-acoustical high-pass filter with a cutoff frequency of 36Hz and a Q of 0.718. In a perfect world the large signal should behave in identical fashion ! The droop in the high frequency response is caused by the voice coil inductance because it was driven using voltage drive rather than current drive as used by our servo system.

 

 


Last Updated November 2, 2014