High Frequency
Driver Far Field Response Measurement
Here we measure the acoustical frequency response of a common
double chamber 1” dome tweeter unit. This unit has a voice coil
resistance of 6 ohms and has a nominal power rating of 100 Watts. Its
resonant frequency is 750 Hz making it a good candidate for a two-way
system. The driver is flush mounted in a box with front panel dimensions
of 11" x 25". The room is semi-reverberant and measures 6.5m(l)
x 5.5m(w) x 2.4m(h).
Modify the "32768_MLS_Impedance_Measurement.process"
to measure frequency response.
Wire the measurement circuit for level and
frequency calibration.
Set the level of the sound card and
optional amplifier to 2.83V.
Calibrate the sound card and optional
amplifier input level and frequency response.
Run the process and window out floor and
ceiling reflections so that we can observe the relative SPL vs.
frequency response of the driver in the spectrum analyzer.
Observe the absolute SPL vs. frequency
response as we gradually move the microphone closer to the driver.
Present the SPL vs. frequency response vs.
microphone distance from the DUT in the data-logger.
Loudspeakers are
typically specified in terms of dB SPL for 2.83 volts peak input. The
sound pressure level is measured on-axis in anechoic conditions at a
distance of 1 metre from the loudspeaker. 2.83 Volts corresponds to the
voltage across a standard 8 ohm speaker driven at 1Watt. The current
required to supply 1 watt to 8 ohms is 0.353 Amps.
Although +12V
DC is available on all computer PCI backplane connectors, the analog output
stage of a typical AC97 audio interface is powered from +5VDC. At best
these -10du (40-300 ohm) single ended line outputs can produce a full
scale output voltage of about 1.25Vrms which falls short of the 2.0Vrms
(0.7071 x 2.83Vpeak) required to produce 1 Watt across 8 ohms. Better
quality sound cards may have +12V powered, swappable, in common
8-Pin Dip packages (as below) op-amps in a balanced configuration on the
line outputs but these at best can deliver 8V into a 600 ohm load (13mA
~100mW).
The table below
shows the maximum output voltage of two motherboard based audio codecs
into 4 load impedances.
Load
Impedance (ohms)
Output
Level ALC888S Codec (Vpeak)
Output
Level AD1885 Codec with LF353 Amp (Vpeak)
10
0.115
0.800
100
0.550
2.322
1000
1.610
2.786
Infinity – Open
Circuit
1.863
2.900
In addition to the
line level outputs, some older sound cards such as the sound blaster AWE
32 had speaker outputs that could easily drive 2 Watts into an 8 ohm
load. If you have such an output this would be the one to use for
frequency response measurements.
The four pin USB
2.0 port found on most computers can output 500mA at +5V or 2.5 Watts.
Most USB powered audio interfaces only have balanced international studio
line level outputs (+4dBu or 2.19Vpeak nominal). These 300 ohm outputs
can produce about +10dBu or 3.49Vpeak maximum but are not suitable for
driving 8 ohm loudspeakers. Their headphone outputs are only good for
about 100mW into 32 ohms. Although eight pin +12V and +24V powered USB
and six or nine pin Fire-wire audio interfaces could easily meet the
above standard these types of audio interfaces are usually equipped with
the same +4dBu outputs as the USB types.
If you already have a consumer HiFi
receiver or amplifier you can purchase a 3.5mm TRS to ¼” RCA
male adapter cable (as below) and connect the sound cards line-out
to the equipments Aux-In input and drive the DUT with the amplifier.
You only need a single
amplified channel to measure loudspeaker on axis frequency response.
The +6V split supply powered, +15dB gain class AB amplifier
on the left below can be inexpensively constructed and can deliver 4
Watts to a 4 ohm load at less than 0.1% THD at 1 KHz. The +12V
single supply amplifier on the right can do the same but is not
recommended due to the large output decoupling capacitor required to
isolate the DC from the loudspeaker. At 20Hz this 2200uF capacitor
would have an impedance of 3.6 ohms. This has the effect of reducing
the voltage at the loudspeaker terminals as frequencies are lowered
so that 1 Watt is no longer dissipated across the voice coil.
Although the IC’s in the circuits below have thermal shut down
a 6C/Watt heat sink is required.
Just
connect the amplifier in series with the audio interface outputs and
adjust the mixer levels to produce an output of exactly 2.83Vpeak
(2.00VRMS) into the loudspeaker.
If you are measuring impedance or relative responses (such as you
would when you are adjusting cross-over components) you do not need to
calibrate levels. If you are measuring absolute responses like we are
here you should calibrate levels as you probably would like to refer your
measurement to some manufacturers data sheet.In order to
calibrate a sound card line input you need to compare it to a reference
standard. An inexpensive way to do this is to purchase a pocket
multimeter such as the 4 1/2 digit EXTECH DM110. It costs about $34.99
USD or 27.41 EUR. Its AC RMS voltage accuracy from 40-400Hz on its 4.000V
range is +1.0% + 10 digits. This translates to about 0.05V. More accurate
6 1/2 digit meters such as the Agilent 34401A with VAC accuracies of
better than +0.06% are available. You must ensure that your meter
measures AC voltage at the frequency (~200Hz) that sonic beacon outputs
during level calibration. For instance, during voltage calibration for a
sample rate of 96000S/S at an FFT Size of 1024 a 187.5Hz waveform is
output.
We will modify "32768_MLS_Impedance_Measurement.process"
to perform the frequency response measurement. This process ships with
the release version of this product. When we are done it will consist of
six modules. The first is the signal generator, which generates an 8192
length MLS stimulus to excite the DUT. Second is the SoundIO module,
which plays the stimulus and records the response of the driver. Third is
the Arbitrary Filter, which will invoke our microphone compensation
curve. Fourth is a Spectrum Analyzer, which will perform an FHT on the
MLS time domain data in order to convert it to an impulse response. Fifth
is the Oscilloscope module, which will allow us to window the impulse
response in order to remove room reflections. Finally another Spectrum
Analyzer, which performs an FFT on the impulse and allows us to view
amplitude vs. frequency and phase vs. frequency graphs.
1.Install the driver in the desired enclosure or baffle. Bypass any
driver crossover unit. Place the enclosure on a stand about 1 meter from
the floor.
2.Open “32768_MLS_Response_Measurement.process”
from the applications File…Open… menu. Press OK
if the “No Compatible Calibration File Present”
message box appears.
3.Select Options…Process from the applications menu.
The Process Select dialog will open. Highlight Oscilloscope
in the Module List box. Highlight Arbitrary Filter in the Available
Modules list box. Press the Insert button. Highlight Oscilloscope
in the Module List box. Highlight Spectrum Analyzer in the Available
Modules list box. Press the Insert button again. The Process
Select dialog should appear as in Figure 2.
Figure 2: Driver
Measurement Process Select Dialog
4.Press the Ok button in the Process Select dialog. The
modified process will open as in Figure 3. Press OK when the
“No Compatible Calibration File Present” message box
appears.
5.Open the FFT Options dialog from the applications Options…FFT…
menu and change the FFT Size to 8192. Press OK in
the FFT Options dialog box. Press OK when the “No
Compatible Calibration File Present” message box appears.
8.Select the Step: 6 Spectrum Analyzer module. Select Log100
from the Xaxis Sel: combo box. Select dBRel from the Yaxis
Sel: combo box. Select 5dB/div scale from the Yaxis Sel:
combo box. Enter 1000 into the Ref1: Edit control and press
the enter key. Select FFT from the Type combo box and check the Apply
Freq. Cal. checkbox in the Options group box.
9.Wire the circuit for calibration as shown in Figure 1. Use short,
low resistance or shielded wiring. Note that external amplifiers should
be in the calibration loop. These devices will produce more output swing
than is tolerated by the sound cards line inputs so keep levels low.
Figure 1: Driver
Measurement Process Calibration Wiring
10.Press the Open
Mixer button in the SoundIO modules Options group.
Select the Volume Controls Options… Properties… menu.
Choose the sound card from the Mixer Device and press the Recording
radio button in the Adjust Volumefor group. Press the OK
button.
12.Select the Record
Controls Options… Properties… menu. Choose the sound
card from the Mixer Device and press the Recording radio
button in the Adjust Volumefor group. Press the Playback
radio button in the Adjust Volume for group. Press the OK
button.
13.Mute all Playback
mixer gain settings except the MasterVolume and the Wave.
Set both volume sliders to one quarter and equalize the balance settings.
Press the Close Mixer button the SoundIO modules Options
group.
14.Press the Calibrate
button in the SoundIO Options group. The Calibration dialog
box will open.
15.Select Input
from the Calibration Type Select: combo box. Select Vpeak
from the Measure Select: combo box. Connect your AC Voltmeter or
oscilloscope across the driver inputs.
16.Press the
Calibration Run button. A sine wave will be output for 30 seconds.
Its frequency is the maximum of the closest harmonic to 200-Hz as the
sample rate and FFT size of the SoundIO module will allow; or the first
harmonic of the selected FFT Size (Sample Rate / FFT Size). This low
frequency is to allow slow AC voltmeters to correctly measure the signal
amplitude. For instance, for a sample rate of 96000S/S at an FFT Size of
1024 a 187.5Hz waveform is output. Measure the peak value of the driver
signal with the external instrument. Gradually increase the Master
Volume and Wave sliders in tandem until a peak value of 2.83
volts is measured with the external instrument. Press the Calibration Run
button again to restart if Input Signal Levels Calibration has stopped.
Enter the peak level measured by the instrument into the Ch1(L): and
Ch2(R): Level edit controls in the Input Signal Levels
Calibration group box. Check both Apply checkboxes. This
action establishes a known reference level for the remainder of the
calibration process. If the sound card will not output 2.83 volts the
microphone distance from the loudspeaker will have to be reduced in the
later SPL measurements or the actual drive level will have to be
specified with the SPL measurements.
17.Select Auto
from the Calibration Type Select: combo box. Select MLS
from the Freq. Cal. Type Sel combo box.
18.Press the
Calibration Run button and wait for the hour glass cursor to
disappear. If a “No data in record buffer” message
appears increase the mixers Playback Master Volume and Wave settings.
You may also increase the Recording Line slider or reduce the
SoundIO Trig. Level(%F.S.) level.
19.Check both Apply
check boxes in Frequency Response Calibration group box. Press the
Save button to save the calibration file to disk. Enter a file
name when the Save Calibration File dialog appears. Long filenames
are permitted. Press the OK button in the calibration dialog and
select Yes when the “Calibration Parameter Has Changed.
Save To Process File?” message box appears. Do not change
the Recording ControlLine-In gain settings after this
point or re-calibration will be required. Changes to the Playback
mixer gain settings may be made and will not affect signal level
measurements.
20.Now rewire the
circuit as shown in Figure 9. Place the driver enclosure on a stand about
1 meter from the floor. Place the mike about 1 meter away so that it is
in line with the dust cap of the driver.
21.Press the Run
button. Four bursts of MLS sequence will be sent to the driver. Now
observe the resulting trace in the oscilloscope. You should see a single
large aberration at the beginning of the trace followed by a group of
aberrations about 3 mSec later. This trace represents the impulse
response of the driver and the room. The first aberration is the
driver’s impulse response and the later aberrations are room
reflections. They will distort the response curve and must be
excluded from the measurement. Place the mouse at the beginning of
the oscilloscope trace and press the left button. Now sweep the
mouse to the first room reflection and release the left mouse
button. Only the trace highlighted in inverse video will be included
in the measurement. See Figure 10. The distance (d2) for any height (h)
and microphone placement (d1) may be calculated as follows.
The time to first reflection can be
calculated as follows.
where: c = speed of sound (344.5 m/s at 20°C @ sea level)
Figure 10: Windowing the Drivers Impulse
Response
21. Now press Run again and
observe the spectrum analyzers channel 1 trace as in Figure 11. This is
relative response of the driver with respect to a 1KHz reference
frequency.
15. Now we will enter
microphone sensitivity in order to make true sound pressure level
measurements. In the second spectrum analyzer module select dBVSPL from the
YAxis: Sel: combo box. Find the microphone sensitivity in its
manufacturers data sheet. It is usually expressed in terms of output
voltage (mV) per Pascal (Pa). One pascal is equal to 94dB SPL. The
spectrum analyzer dBVSPL scale is referenced to 94dB SPL. If the
microphone sensitivity is expressed in some other terms, conversion is
necessary (see the conversion factors below). Our microphone sensitivity
is 0.031V/Pa. Now enter 0.031 into the Spectrum Analyzers YAxis: VRef:
edit box and press the enter key. Now press the Run button and observe
channel one of the second spectrum analyzer. Figure 12 is the absolute
sound pressure level of the driver at 1 meter.
16. Now we introduce the
microphone frequency and phase correction data to the process. If your
microphone does not come with an ASCII format correction file you may
enter one manually in the Arbitrary Filter edit control using the
manufacturers supplied frequency and phase response curves. See “To enter an arbitrary response envelope manually via
the keyboard” in the arbitrary
filter section of the user manual. Our microphone comes with an ASCII
file so we simply open the file in the arbitrary filter module. The
arbitrary filters edit control can translate almost any correction file
that has the form shown in Figure 13.
A note of caution; most correction files come with the actual
response curve of the microphone element. In this case the response curve
must be inverted. You must press the Invert button in the
arbitrary filter dialog bar for these files to give the correct response
output. The arbitrary filter multiplies the input signal amplitude
response by its amplitude response and adds its phase response to the
input signals phase response. Highlight the arbitrary filter module and
press the Edit button on the dialog bar. The edit control should
open on the right side of the module. Press the Open button on the
dialog bar. Select the All Files(*.*) option from the Files of
type combo box. Select the correction file from the Open
dialog and press the OK button. Press the Update button on
the dialog bar. If a Frequency Element out of Response Range
message appears, go to the given line number and place a semicolon in
front of it. This will cause the line not to be parsed. You may also
delete it altogether. This application has a maximum response range of 0
Hz to 22050 Hz. Now look up the microphone manufacturers frequency
response curve. If the plot in the graph window is equivalent to the
curve press the Invert button. Now press the Save button
and enter a file name in the Save As dialog File Name edit
control. Press the Save button.
17. Now we can measure the
response verses the microphone distance from the driver. Each time the
distance is halved the response should increase by 6dB. We will create a
new process consisting of only the DataLogger module to illustrate this
point. Press the New button from the applications File menu. Open the FFT
Options dialog from the applications Options…FFT… menu and change
the FFT Size to 8192 and Sample Rate to 44100. Press OK in the FFT
Options dialog. From the Options menu select Process... The Process
Select dialog box will open. From the Available Modules list box select
DataLogger and press the Insert button. The DataLogger will appear in the
Modules List box. Press the OK button. When the DataLogger opens change
the Xaxis scale selection to F1Log100. Change the YAxis1 scale selection
to dbVSPL.Enter the microphone sensitivity in the YAxis1 Ref: edit
control. Change the YAxis2 scale selection to None. Press the Title
button and enter a plot title in the Title edit control.
Sonic beacon is a Canadian
organization founded in 1997. It is located in Pakenham Ontario which is near Ottawa, Canada. Ottawa is also home to
the National Research Council of Canada's anechoic chamber, a key
facility for acoustics testing. The chamber has been instrumental in the
development of Canadian loudspeakers, hearing aids and microphone arrays.
July
8, 2009:Sonic beacon Version 1.1.0.6 released. Data-Logger
can now save in .FRD and .ZMA file format. Data-Logger and Spectrum
Analyzer can perform .FRD and .ZMA clipboard copy transactions. Signal
generator can output signal complements on each sound card channel to
allow bridging in many consumer audio interfaces. Module status bar shows
current data type, FFT size and sample rate.
