Two Driver Sealed Box Acoustical Measurement
Techniques
To Calibrate or Not to Calibrate
The alignment of a closed box system
drivers and their crossover usually consists of taking an acoustical measurement
then changing a crossover component or adding stuffing or bracing to the
box. This is then repeated until the systems response is flat. The
calibration of the sound card or the microphone may not be necessary for
this tedious work as long as all acoustic measurements are always done at
the same drive level. Once the relative response is flat then the system
can be calibrated for a single absolute response measurement.
Look at the response of the sound card
below. It is flat to within +1dB from 30 to 19000Hz when using an
FFT size of 8192. This is usually adequate for initial crossover
adjustment. To determine the response of the sound card connect the line
out to the line in and send an MLS signal through a SoundIO module in
Record/Play mode to a Spectrum Analyzer in FHT/FFT mode. Using the
SoundIO Repeat feature pre-stimulates the sound card greatly increasing
stability.
Figure 1: Typical Sound Card
Frequency Response
Look at the measurement microphone
response (green) and phase (yellow) curves below. The response only
deviates by 1dB at frequencies above 12kHz. Unless there are large deviations
in the tweeter response above 10KHz, calibration is not necessary as most
of the work on response adjustment for two way systems is usually done in
the crossover (1 to 4 kHz) region.
Figure 2: Typical Measurement
Microphone Frequency/Phase Response
Obtaining the Response Curve for Two-Way System
Adjustment
For most two way crossover adjustments
only the combined far-field woofer and tweeter response is required. It
is obtained by placing the microphone at tweeter level, 1 meter from the
tweeter dome. It is usually valid down to 250Hz which is usually well
below the crossover frequency. An MLS from a Signal Generator module is
sent to a SoundIO module in Record/Play mode. This is followed by a
Spectrum Analyzer in FHT mode to create a time domain impulse response.
An Oscilloscope module is used to gate the resulting impulse to mask room
reflections. It is gated at the point where the first reflection arrives
at the microphone, usually 4 to 5 mSec for a typical size room. The gated
impulse is sent to a Spectrum Analyzer module in FFT mode for generation
of the frequency response curve.
Figure 3: Sonic Beacon Process
Document to Evaluate Loudspeaker Far Field Frequency/Phase Response
Crossover Adjustment
Two drivers with the responses below
were installed in a 44 liter sealed enclosure. The crossover board was
suspended outside the enclosure so that components could be easily
changed.
 
Figure 4: Free Air Woofer and
Tweeter Impedance and Frequency Response Curves
Initially a 2nd order
Linkwitz-Riley crossover was designed for this system but the woofers response
(green) at the 2100Hz crossover frequency caused phase problems that left
a 4dB dip in the combined response (yellow) between 800 and 3000Hz. Here
woofer (green), tweeter (blue) as well as the combined far-field
measurements were separately taken and pasted into the Datalogger.
Figure 5: Four dB Dip in the
Sealed Box System Response Using a 2nd Order Linkwitz-Reily Crossover
This was replaced with second order
Butterworth network, with a 3dB peak at the crossover, in the hopes that
the dip would be eliminated. A 4dB artifact remained between 2 and 3 KHz.
The purple trace is the response with the tweeter polarity inverted.
Figure 6: Minus Four dB Artifact
in the Sealed Box System Response Using a Butterworth Crossover
with 3dB Peak
A 2nd order Chebyshev, with
a 6dB peak at the crossover frequency, replaced the Butterworth network.
Although this did not remove the majority of the dip it brought it more
in line with the rest of the tweeters response
Figure 7: Sealed Box System
Response Using a 2nd order Chebyshev Network with 6dB Peak
Once the tweeter attenuation was
adjusted the following far-field +2.85dB response resulted.
Figure 8: Combined Far Field
Sealed Box System Response Using a 2nd order Chebyshev Network with 6dB
Peak
Combining the Acoustical Response Curves
After the response in the crossover
region is flattened the system is then calibrated so that the system
sensitivity is measured correctly. In order to create a frequency
response graph for the dual driver sealed box system the two measurements
must be combined.
First the near-field woofer response is
obtained. This is done by placing the microphone at 0.25” from the woofers
dust cap with the Signal Generator, SoundIO, Spectrum Analyzer,
Oscilloscope, Spectrum Analyzer setup from above. Impulse gating is
usually done at about 50 mSec to get a response down to 20Hz.
The near-field response is pasted into
plot1 of a separate instance of a Datalogger module. Note the very high
SPL due to the close proximity of the microphone.
Figure 9: Near Field Woofer Sealed
Box Response Pasted into Datalogger (Plot1)
The combined far-field woofer and
tweeter response is obtained (gating at 4 to 5 mSec) and pasted into plot
2 of the datalogger.
Figure 10 Near Field Woofer Sealed
Box Response (Plot 1) and Combined Far-field Woofer and Tweeter Response
(Plot2) in Datalogger
The level of the near-field response is
brought into coincidence with that of the far-field by reducing the gain
of plot 1. Select Plot1 in the Sel: Combo box of the Plot Adj: group and
press the Gain: Dn button until the levels of the two plots are equal at
the proposed merge point. In this case the near-field data is valid up to
about 718Hz (fmax = 4311/ effective cone diameter) and the far-field data
is valid down to 250Hz. Both curves are reasonably flat between 200 and
300Hz so 250Hz might be a good merge point.
Figure 11: Determining the Merge
Points of the Near and Far Field Plots
Then the near-field response is
merged with the far-field response. Enter the merge frequency in the
Freq: edit control in the Mode group box and select Merge in the Sel:
combo box. Note the response dip below 500Hz due to spreading loss caused
by the enclosures front baffle. Also note that the system sensitivity is
that of the tweeters original free air graph.
Figure 12 Merged Near and Far
Field Plots In DataLogger
|
