LIMP User Manual
LIMP User Manual
LIMP User Manual
User Manual
Version 1.9.4
Ivo Mateljan
Artalabs
J. Rodina 4,
21215 Kastel Luksic, Croatia
June, 2021.
Content
2
LIMP User Manual
1 What is LIMP?
LIMP is a program for the measurement of loudspeaker impedance and estimation of the
loudspeaker’s physical and dynamical parameters (also called Thiele-Small parameters). It is also a
general-purpose program for measuring impedance in the audio range.
The installation of LIMP is through a common setup program for the ARTA Software.
All Windows registry data for LIMP will be automatically saved at first program execution, and files
with extension ".LIM" will be registered to be opened with a program LIMP.
Results of measurement can also be saved as ASCII formatted files in .ZMA format.
LIMP does not dump graphs to the printer, instead, all graphs can be copied to the Windows Clipboard
and pasted to other Windows applications or saved to files in .BMP, .PNG or .JPG formats.
3
LIMP User Manual
4
LIMP User Manual
N −1
g (t ) = Ak cos(2 f k t + k ), k = random 0, 2
k =1
The pink PN has frequency components on all the FFT frequency bins (fk = k fs/N) and its spectral
magnitudes roll-off at -3 dB/octave (Ak =√2A2k ) after specified cut-off frequency.
In LIMP, a variable low frequency cut-off frequency can be changed in the 'Signal Generator Setup'
dialog box. The pink noise is usually used with a cut-off frequency set close to the frequency where
the loudspeaker has an impedance maximum (20-100 Hz). The pink PN signal has the statistical
distribution close to the normal distribution.
Multitone signal, as defined in the IEC 60268-21 standard, is the sum of random phase spectral
components that have equal magnitude (Ak =const.), but they are sparsely generated only on FFT
frequency bins that are approximately 1/n-octave apart (fk+1 = fk 21/n). In Limp the resolution of 1/24
octave is used if FFT size is less than 64K, otherwise the resolution is 1/48 octave.
Although multitone and pink PN are mathematically similar signal, they sound very different and are
used for different purposes. Pink PN is always generated as a wideband signal while the multitone
spectrum can be user defined in a restricted frequency range [Fstart, Fstop].
In LIMP multitone and pink PN signals are generated with a crest factor slightly larger than 12 dB.
The sine signal has crest factor 3 dB. Signals generated using D/A converters have limited peak signal
value, which means that maximal total RMS value of all multitone components is about three time
smaller (9 dB) than maximal RMS of sine signal (for the same peak value). If we generate multitone
with resolution of 1/48-octave it means that in audio range there will be 480 multitone components
each with RMS value that is sqrt(480) times lower than multitone total RMS, or 3*sqrt(480) lower
than sine rms value. This reasoning leads us to the conclusion that measurement in stepped sine
mode theoretically has 36 dB better S/N ratio.
5
LIMP User Manual
2. After one pre-averaging cycle, which is necessary to reach the steady state, voltages at both
ends of the reference resistor are measured as discrete time series u1 and u2 of length N.
3. The DFT is applied to time series u1 and u2 to get spectral components U1 (f) and U2(f). They
are used in the equation (1) to calculate the impedance Z(f).
Figure 2.2 A circuit for the loudspeaker impedance measurement, with the noise generator En
If we apply equations for impedance measurement (1) to this circuit, we get the estimated impedance
value:
U2 R + Z + Rg
Z estimated = R =Z+
U1 − U 2 1 + E g / En
What this equation shows is that estimated impedance differs from the true impedance Z by
the term that is dependent on the S/N ratio (Eg/En) and values of resistors R, Rg and the
impedance Z. By using additional knowledge from measurement process we can conclude:
6
LIMP User Manual
1. The simple theory shows that the signal generator must supply a high voltage, to assure high
S/N, but practical measurements have shown that it is better to use smaller output voltage.
Two facts prove that. First, a larger voltage drives the loudspeaker to highly nonlinear region.
Second, a larger voltage drive can enlarge driver self-vibration that can produce discontinuous
noise in interaction with mounting. AES recommends voltages in the range 0.1 – 0.2 V, but
even lower values are acceptable. This is especially important if we measure impedance to get
data for driver parameters estimation.
2. When we use the FFT method the measurement results are highly affected with noise. The
loudspeaker acts as a microphone with a highest sensitivity in the region of the membrane
resonance. It means the highest level of the noise is at low frequencies, so we must generate
the signal with high level at low frequencies, i.e. the pink noise, or multitone with spectral
components equally distributed on log-frequency scale.
3. Values of resistors R and Rg must be small, an optimum being a value close to the magnitude
of the measured impedance. Practically, we can use R = 10-27 ohm to get a very good
impedance estimation, but then we need a power amplifier to supply large current. If we use
the soundcard headphone output as a signal generator, then we can use R = 40-100 ohms. If
we use the soundcard line output as a signal generator, due to the limited current capability,
we must use R>600 ohms. In that case we can't get a good estimation with FFT method, but
we still can successfully use the stepped sine method.
R
Z esitimated =
U1U1*
−1
U 2U1*
Note: brackets <> denotes averaged values and star denotes the conjugate complex value.
better results than the synchronous averaging in systems with nonlinear distortion, but it needs longer
time for measurements.
Figure 2.3 Signal generation and acquisition during the synchronous time domain averaging process
Figure 2.4 Signal generation and acquisition during the asynchronous frequency domain averaging
8
LIMP User Manual
3. Hardware Setup
The simplest measurement configuration is shown in Fig. 3.1. The soundcard phone-out (headphone
output) is used as a signal generator output. The soundcard right line-input is used for recording the
reference voltage U1 and the soundcard left line-input is used for recording the voltage U2.
If the soundcard has no headphone output, then we have to connect an external power amplifier to the
soundcard line-out output, as shown in Fig. 3.2.
To protect the soundcard input from a large voltage that may be generated at a power amplifier output,
it is recommended to use the voltage probe circuit with Zener diodes, as shown on Fig. 3.3. Values of
resistors R1 and R2 have to be chosen for arbitrary attenuation (i.e. R1=8200 and R2=910 ohms give
probe with -20.7 dB (0.0923) attenuation if the soundcard has usual input impedance – 10 kΩ).
9
LIMP User Manual
10
LIMP User Manual
Figure 4.3 Status bar shows the peak level (ref. full scale) of left and right line inputs
Images of graphs and windows can be copied to Windows clipboard or saved to the file in a three
image formats: png, bmp and jpg. It is recommended to use png format.
Obtaining a copy of the full window picture is simple. User needs to simultaneously press keys
Ctrl+P. After that command the window picture will be saved in the System Clipboard, from there the
user can paste it in other open Windows applications (MS Word, MS Paint). Keys Ctrl+Alt+P
activate command to save that image in the file.
11
LIMP User Manual
To copy or save the graph picture, that is shown inside the window, user needs to simultaneously press
keys Ctrl+C or activate the menu command 'Edit->Copy', or press appropriate 'Copy' button. In the
main window toolbar, the 'Copy' button is shown as toolbar icon .
The Copy command opens the dialog box 'Copy/Save Image with Extended Information', shown in
Figure 4.5. Here user has to setup following options:
1) By using the combo box above ‘OK’ button, user chooses one of three modes of
saving the image: Copy to Clipboard, Save to File and Save to File + Copy to
Clipboard.
2) In the Edit box user optionally enters the text that will be appended at the bottom of
the graph.
3) Check box ' Add filename and date' enables adding text to the graph that shows file
name, date and time.
4) Check box 'Save text' enables saving entered text for the next copy operation.
5) Combo box ‘Aspect ratio’ – sets aspect ratio for copying graphs with fixed width.
6) Bitmap size is chosen by selecting one of following combo box items:
▪ Current screen size - variable width and height option
▪ Smallest - fixed graph width 500 points
▪ Small - fixed graph width 600 points
▪ Medium - fixed graph width 800 points
▪ Large - fixed graph width 1000 points
▪ X Large - fixed graph width 1200 points
▪ XX Large - fixed graph width 1500 points
▪ XXX Large - fixed graph width 2000 points
The button 'OK' copies the graph to the system clipboard or opens dialog to enter name of file in
which picture will be saved.
The button 'Cancel' cancels the copy operation.
12
LIMP User Manual
Overlay
Set as overlay- sets the current impedance curve as an overlay
Delete - deletes overlay
Load - loads an overlay from an impedance file
Set as target curve - sets current curve as target curve
Load target curve - loads an target curve from an impedance .lim or .zma file
Delete target curves - deletes all target curves
Make target curve from TSP – makes and shows target impedance curve calculated from
measured loudspeaker parameters
Edit
Copy - copies the graph and user defined text to the clipboard or saves that image to the file
Setup colors and grid style - sets graph colors and grid style
B/W background color - changes the background color to black or white
Thick pen - uses thick pen for plotting curves
Thick grid - uses thick pen for plotting grids
Subtract overlay - subtracts overlay impedance from current impedance
Add overlay - add overlay impedance to current impedance
View
Toolbar - shows or hides the Toolbar
Status Bar - shows or hides the Status bar
Fit graph range - changes graph top margin to show full impedance curve
Magnitude - shows only impedance magnitude curve
Magnitude+phase - shows impedance magnitude and phase curves
Record
13
LIMP User Manual
Setup
Audio devices - sets current input and output devices
Generator and Input Monitor - sets generator parameters and measures input voltages
Measurement - sets measurement parameters
Cable compensation - sets resistance and inductance for cable impedance compensation
Graph - sets graph margins
CSV format - opens dialog box for setup of decimal separator character in CSV files
Analyze
Loudspeaker parameters - Closed box method - shows the dialog box for the estimation of
loudspeaker parameters using Closed box method
Loudspeaker parameters - Added mass method - shows the dialog box for the estimation of
loudspeaker parameters using Added mass method
Loudspeaker parameters - Fixed mass method - shows the dialog box for the estimation of
loudspeaker parameters using Fixed mass method
RLC impedance values at cursor position - estimate values of impedance serial RLC elements
Help
About - gets information about LIMP
Registration - shows the user license registration
User Manual - shows the help file
14
LIMP User Manual
Activate the menu Setup->AudioDevices. You will get the 'Soundcard Setup' dialog box shown in
Fig. 4.6. In this dialog box you choose which soundcard will be used as an input or output device.
Generally, choose the same card as an input and output device.
Important notice: Please mute the line and microphone channels at the output mixer of the
soundcard; otherwise you might have positive feedback during measurements. If you use a
professional audio soundcard, switch off the direct or zero-latency monitoring of the line inputs.
15
LIMP User Manual
16
LIMP User Manual
17
LIMP User Manual
18
LIMP User Manual
Figure 4.11 Setting the native bit resolution and sampling rate in Vista
Note: There are a lot of drivers that are not stable under Windows 7. In that case please use ASIO
driver if it is available for your soundcard.
Figure 4.12 E-MU Tracker Pre ASIO Control panel for setting bit-resolution and buffer size
19
LIMP User Manual
In music applications user usually sets buffer size as small as possible with stable operation. That
gives the lowest input/output latency (system introduced delay).
In LIMP, the latency is not problem, as it is encountered in software, but it is not recommended to use
buffer with size larger than 2048 samples, or smaller than 256 samples. Some ASIO control panels
express the buffer size in samples, while other express the buffer size in time [ms]. In that case we can
calculate the size in samples using the following expression:
ARTA automatically sets the buffer size for signal duration of 10 ms (i.e. 512 samples for sample rate
48kHz, 1024 samples for sample rate 96kHz and 2048 samples for sample rate 192kHz)..
LIMP always works with two input channels, and two output channels, treating them as a stereo left
and right channels. As ASIO support multichannel devices, user has to choose in a dialog box
‘Soundcard Setup’’ which pair of channels will be used (1/2, 3/4, …).
User chooses signal type in the top dialog bar or in the dialog box 'Generator and Input Monitor
Setup' shown in Fig. 4.13. It opens by clicking the menu Setup -> Generator and Input Monitor or
by clicking the toolbar icon . Besides signal type, the user can choose other signal characteristics
(output level relative to maximum level and pink PN cut-off frequency).
The dialog box also contains a two-channel virtual voltmeter for monitoring voltages on the generator
output (usually left input channel) and on measured impedance (usually right input channel). Input
monitoring has no influence on the measurement of impedance; it just serves to set measurement
signals to levels that will be treated as small signal or large signal excitation.
The virtual voltmeter should be calibrated. The calibration steps described below sets the input
channel sensitivities. We define sensitivity as follows:
The digital measurement system records sequence of discrete measured values D[i] in a restricted
dynamic range. In ARTA software, discrete values D[i] are floating point values in the range from -1
to 1. Virtual measurement system maps internal presentation D[i] to external discrete voltage values
V[i], that has been recorded, using sensitivity factor:
Sensitivity = V[i] / D[i], where |D[i]| <= 1, i = 0,1,2,…
The unit of Sensitivity is the Volt as D[i] is dimensionless. The same hold for RMS values of periodic
signals;
Sensitivity = Vrms / Drms.
Maximum possible value of discrete sequence D[i] is 1. That gives us an alternative definition of
sensitivity as the maximum peak value (or full scale) voltage that can be recorded by digital
instrumentation;
Sensitivity = Vpeak-max (Volts).
20
LIMP User Manual
Generator section
Type - chooses the excitation signal: Sine, Multitone or Pink PN
Sine freq. (Hz) - enters test frequency of the sine generator
Pink cut-off (Hz) - enters the low-frequency cut-off of the pink noise generator
Stepped sine level (dB) - chooses the output level (0 to -30 dB)
Multisine level (dB) - chooses the output level (0 to -20 dB) for multitone or Pink PN
Start/ Stop button - starts/stops the generator with current settings, also capture signals on both channels
Ok button closes dialog and saves dialog state.
Cancel button closes dialog and saves dialog state.
Peak meter - shows peak value ref. full scale. Meter bars are shown in following colors:
Green - for levels below -3 dB,
21
LIMP User Manual
With LIMP the calibration is simple. It requires that user chooses sine signal and sine frequency in the
range 300-500 Hz. The circuit for impedance measurement should be connected to the sound card, as
shown in Fig. 3.1 or 3.2. The process starts by clicking button Start/stop. If recording starts
successfully we can do calibration by clicking buttons Calibrate Left and Calibrate Right. That
opens dialog box shown in Fig. 4.14.
Then, user should measure voltage using voltmeter connected instead of measured impedance and
enter that value in the edit box. The calibration finishes by clicking button ‘OK’. If user click button
‘Cancel’ the calibration is rejected.
For usual measurement setup shown in Fig. 3.1 or Fig. 3.2 this procedure will give voltage
measurement accuracy better than 1%, which is more than enough for monitoring purpose.
Note: If system uses voltage probe, as in Fig. 3.2, calibration encounters sensitivity scaling
automatically. Input impedance of voltage probe should be 100 times larger than reference resistor to
retain accuracy below 1%.
Recommendation:
• For most reliable results use sine generator, but don't push loudspeaker into large
displacement (the largest displacement is at frequencies below the loudspeaker resonance
frequency). Small signal regime requires sine voltages on impedance smaller than 100mV
rms.
• For measurements of bass or mid-bass loudspeakers set the Pink cut-off frequency close to
the loudspeaker resonance frequency (20-100 Hz).
• If measuring impedance for TSP estimation use low level signals. If you want that estimation
of TSP using stepped sine mode is comparable to FFT mode lower the sine level 13 -15 dB.
• Standard AES2-2012 recommends measurement of impedance with 0.1 V rms of sine
voltage. If we apply multitone or pink with 0.2 V rms we will get comparable results.
22
LIMP User Manual
23
LIMP User Manual
If you set averaging to Linear, measurements will be repeated and averaged until the number of
averaging reach the predefined value for Max averages in the 'Measurement setup' dialog box. You
can stop the averaging at any time by clicking the menu Record->Stop or by clicking the toolbar icon
. You will get the graph like one shown in Fig. 4.17.
You can also choose the exponential averaging. It differs from linear averaging in a way that it gives
more weight to results from last five measurements.
24
LIMP User Manual
View All – enables the view of all spectrum components that are used in impedance estimation.
View Phase – enables a phase plot.
Update – updates the graph with a new setup.
26
LIMP User Manual
Graph colors can be changed in two ways. The first one is to change the background color from
"Black" to "White" by clicking the menu command View->B/W color or by clicking the toolbar icon
.
The second way to change graph colors is a "user mode". User sets an arbitrary color for every graph
element using the 'Color Setup' dialog box, shown in Fig. 4.18. This dialog can be activated by
clicking the menu Edit->Colors.
Clicking the left mouse button on colored rectangle opens the standard Windows dialog box 'Color'
(Fig. 4.19). Button 'Default' restores default colors.
A check box 'Use dotted graph grid' enables drawing of grid with dotted line style.
We can browse the graph by moving the cursor. At the bottom of the graph the label 'Cursor:' denotes
values of the magnitude and the phase, at the cursor position.
27
LIMP User Manual
In section Generate Pink PN we set the output volume for Pink PN signal which sampling rate and
sequence length was set in Measurement setup dialog. Clicking the button ‘Generate’ activates the
signal generation and monitoring of input levels on the Input Level Meter. If necessary, we can adjust
the output volume by using the combo box 'Output volume'.
Finally, in section Calibrate we enter the Number of averages (recommended value is 1, as we
measure high level signals) and click the button ‘Calibrate’. After a second, we have the calibrated
system.
Attention: If we change the periodic sequence length (FFT size) or a sampling rate, we have to repeat
the calibration procedure.
The last section shows a report of the calibration procedure. If the Channel difference is larger than 3
dB LIMP gives a warning, and calibration is rejected. This precaution is required because some
soundcards change input impedance if gain changes.
When we click the OK button, the dialog will be closed but the calibration process to be fully
effective needs one more thing:
It is recommended that instead of the current value of reference resistor R in measurement
setup we enter instead the value R || Rin, where Rin is the soundcard passband input impedance,
which is a real value. This is necessary if we need accuracy better than 1%.
Note: calibration procedure is changed from LIMP version 1.9.2. In older versions only channel gains
differences were compensated and measurements were restricted to impedances values up to 200
ohms.
To explain how calibration is implemented let’s look at Fig. 4.22. which shows real measurement
condition.
28
LIMP User Manual
Figure 4.22 Electrical circuit for impedance measurement and system calibration
We assume that voltages are measured using soundcard input channels which have input impedances
equal to Zin. Reference channel (usually left channel) measures voltage U1. The other channel is
supposed to measure voltage U2 on measured impedance Z, but value of voltage on Z is a U2, where a
is the ratio of possible different input channels sensitivities. The generator has output voltage e and
output impedance Rg, but that values are not important for current discussion.
The circuit has switch CAL that should be closed during measurement of impedance Z and open
during the calibration.
The current through reference resistor R is I = (U1-U2/a)/R. During impedance measurement
following equations hold:
U2/ a U2
Z || Zin = =R
I aU 1 − U 2
here Z || Zin denotes parallel circuit of Z and Zin. It is obvious that even if we have equal gains in both
channels (a = 1) we measure Z || Zin not the impedance Z alone.
After some algebraic operation we get expression for impedance Z:
RZin 1
Z=
R + Zin U 1 a Zin − 1
U 2 R + Zin
This expression is different from basic impedance measurement expression from chapter 1. It is biased
with values of gain a and input impedance Zin.
Now we will define calibration as process which give us the value of factor: a Zin / (R+Zin).
In calibration procedure we measure voltages U1 and U2, but with measured impedance disconnected.
From circuit on Fig. 4.21 we get:
U 2cal a Zin
=
U 1cal R + Zin
This value can be memorized and applied in expression for impedance, which now gets form:
29
LIMP User Manual
RZin 1
Z=
R + Zin U 1 U 2cal − 1
U 2 U 1cal
Most soundcards input impedance is a real value in the whole audio range (Zin = Rin). If we know the
value of Rin we can enter in LIMP measurement setup value R || Rin instead of reference resistor R,
and we will get very accurate measurements.
Let’s se what will be relative error without accounting for Rin.
R R Rin Rin
= (R − ) / R = 1−
R R + Rin R + Rin
Let Rin = 10k and R = 47 ohms, then relative error is 0.46%. If input impedance is 100k error is
0.05%. Practically, even without correction for Rin we will measure with error lower than 1%.
Note: The measurement impedance range is also limited by input channel noise when we measure
small and large impedances. If we measure a small impedance, the voltage U2 becomes small. If we
measure a large impedance, voltage difference U1-U2, that determines current, becomes small. As a
rule of thumb, we can define measurement range from R/100 to R*100 for most soundcards and signal
types. If we use high quality soundcards with low noise inputs and stepped sine measurements that
range can be much larger.
Two edit boxes are used for entering cable resistance (in ohms) and inductance (in nH). The check box
"Automatically subtract cable impedance from measured impedance" must be checked if user
wants that cable compensation be applied during the measurements.
30
LIMP User Manual
Textual files have extension .txt or .zma. They contain lines of text with three numeric literals:
frequency (in Hz), magnitude (in ohms) and phase (in degrees). The line ends with new line character.
The only difference between .txt and .zma files is that .txt files can contain comment lines. Comment
lines start, after arbitrary number of space or tab characters, with nonnumeric characters, while data
lines start with digit or dot character.
The file can be loaded as regular impedance data or as data for the overlay plot . Only one overlay is
followed.
Binary file .LIM format is as follows:
char id[4]; // four characters id {'L','I','M','\0'};
unsigned version; // version number - started from 0x0101
unsigned reserved; // 0
int numdata; // number of data points
int cursorpos; // last position of cursor
int fftlen; // length of FFT in FFT mode
float fs; // sampling frequency
float data[3*numdata] // contains frequency, magnitude and phase
int infolength; // info string length
char string[infolength]; // info string data
31
LIMP User Manual
5 Loudspeaker Parameters
This chapter gives some definitions and measurement procedures for the estimation of loudspeaker
parameters.
Electromagnetic parameters:
RE - voice coil DC resistance ()
ZLE - impedance of voice coil inductance and coupling to the pole piece ()
Bl - force factor (Tm)
Equivalent circuits used to characterize the impedance ZLE are shown in Fig. 5.1c, and denoted as: L2R
[10], [11], L3R [7], and L2RK [12], [13].
Mechanical parameters:
S - effective area of membrane (m2)
CMS - membrane mechanical compliance (m/N)
MMS - mechanical mass of membrane MMD plus mass of air load on membrane MMA (kg)
RMS - mechanical resistance plus membrane radiation resistance (kg/s)
For the unbaffled driver the membrane air load mass is MMA = 0.5658 S3/2, which is half of the air load
mass in an infinite baffle.
The membrane effective area S, also called piston area, is normally obtained from a cone diameter
measurement that includes 1/2 of the surround width [6], [11]. This value is good estimation for S if
motion of surround is linear, but in practice it can vary. Program MLSSA [10] suggests diameter value
that include only 1/3 of the surround width.
Fig. 5.1 shows wideband and low-frequency equivalent circuits of an electro-dynamic loudspeaker that
is mounted in a free air unbaffled or in an infinite baffle. The circuit for definition of low-frequency
input impedance uses the following elements:
32
LIMP User Manual
L2
Bl LE
L2
RE LE
L CES R ES C MES R2
L2RK K
Z LF
Using these analogous circuit elements, Thiele and Small ([6], [7]) introduced dynamical loudspeaker
parameters. They are defined in the Table 5.1.
1 1
Resonant frequency in free air (Hz) fs = , s =
2 M MS CMS M MS CMS
S M MS 1
QMS = =
Mechanical Q-factor RMS S CMS RMS
S M MS RE
QES =
Electrical Q-factor (Bl )2
RE RES Q Q
Total Q-factor QTS = S CMES = MS ES
RE + RES QMS + QES
Table 5.1 Thiele-Small dynamical loudspeaker parameters ( 0 = 1.18kg/m3, c = 345m/s, pref = 20Pa)
33
LIMP User Manual
Thiele and Small have shown that by using these parameters it is easy to express the low-frequency
response of a closed loudspeaker box as 2nd order high-pass filter and response of bass-reflex box as
4th order high-pass filter. Today, almost every loudspeaker manufacturer provides physical and Thiele-
Small parameters in their loudspeaker data sheets.
An old industry standard assumed that the voice coil electrical impedance contains resistive (RE) and
pure inductive (ZLE = jLE) components. Voice coil inductance (LE) was usually specified as inductive
component of driver impedance at 1 kHz. In modern instrumentation that model is abandoned, and the
voice coil is treated as “lossy” inductor whose impedance ZLE depends on frequency. Physically, ZLE
represents the input impedance of the "transformer" whose secondary winding is the pole piece. Eddy
currents in the pole piece give rise to primary circuit resistance.
The simplest model [8] for a primary input impedance is Kj = K (1+j) /2, where K is constant. It
is expressed in unit called semiHenry (sH). The impedance of such semi-inductance increases with
rather than . This model is not practical as engineers in many numerical simulations use some
form of an equivalent electrical circuit that closely matches measurement data.
The most commonly used equivalent circuit for the electrical voice coil impedance in modern
instrumentation ([10], [11]) is serial connection of resistor RE, inductor LE and parallel connection of
resistor R2 and inductor L2. It is denoted in Fig. 5.1 as L2R model. For many years it has been proven
as useful model in many simulations.
LIMP also enables estimation of two additional and more accurate equivalent circuits for lossy
inductor: L3R and L2RK models. They are also shown in Fig. 5.1. The L3R model extends L2R
model with one more parallel circuit L3||R3, while in L2RK model, as defined by Thorborg [12], a semi
inductance K is added parallel to circuit L2||R2.
Originally, Small [6] defined parameters for loudspeaker mounted in an infinite baffle, but he allowed
the same expression to be used for unbaffled loudspeaker. The same is allowed by the IEC 60268-5
standard [15], with requirement to report which type of mounting is used in measurements. Instead od
using an infinite baffle, the IEC standard requires the use of a baffle of dimension 135x165 cm with
driver mounting hole offsets from center 15x22.5cm. Standard AES2-2012 [16] exclusively advocates
measurements with unbaffled driver. LIMP procedures support both type of driver mounting.
Thiele-Small parameters can be used to measure and analyze the response of a loudspeaker mounted in
a closed box. The analogous circuit have the same elements as circuit on fig. 5.1., but values of
elements change: CMS is replaced with parallel connection of CMS and equivalent box volume
compliance, MMS changes in air load mass term, and RM changes due to added dissipation inside the
box. The air load mass of loudspeaker mounted in closed box is between values for unbaffled driver
and driver mounted in infinity baffle. This is probably the main reason why both types of mounting are
accepted in practice - both are equally accurate and valuable to predict box response sensitivity and
efficiency.
For the estimation of Thiele-Small parameters (TSP), LIMP offers two methods:
1) Thiele-Small method and
2) Nonlinear Least Square Error (LSE) minimization method.
LIMP also uses LSE minimization for estimation of lossy inductor equivalent circuit elements. It is a
first step in TSP estimation because it helps to remove influence of inductance element from analog
circuit before using it in TSP estimation.
34
LIMP User Manual
= Z M ( f ) − Z LF ( f ) − Z LE ( f )
2
where ZLF(f) is a low frequency loudspeaker impedance estimated by Thiele-Small method (which is
described in the next Section).
By default, LIMP estimates elements of L2R model by stable, linearized LSE error minimization
method – sometimes also called a linear regression method. In addition, user can choose more accurate
nonlinear LSE minimization methods for L2R, L3R and L2RK model circuit elements. For full
discussion of these and TSP estimation methods it is recommended to read the paper [13].
(1 − f 2 / f S2 ) + j f /( f S QT )
Z ( f ) Z LF = RE
(1 − f 2 / f S2 ) + j f /( f S QMS )
The impedance has maximum value at the resonance frequency:
Z max = Z ( f = f S ) = RE QES / QT = RE + RES as QES >QT.
Figure 5.2 Typical impedance curve of a loudspeaker that is mounted in free air
At frequencies f1 and f2, (where f1<fS<f2, and f1 f2 = fS2) impedance values are of equal magnitude;
Z ( f1 ) = Z ( f 2 ) = r1RE , if f1 f 2 = f S
2
35
LIMP User Manual
If we substitute this expression in the expression for the input impedance, we get:
+ (QMS ( f1 − f 2 ) / f S )
2 2
Z ( f1, 2 ) = r1 RE = RE
2 2 2 2 r0
.
1 + (QMS ( f1 − f 2 ) / f S )
2
r0 − r1
2 2
fs
QMS =
f 2 − f1 r1 − 1
2
= Z M ( f ) − Z LE ( f ) − Z LF ( f )
2
where optimization variables are RE, fS, QMS and QT. ZLE(f) is impedance of lossy inductor that is also
estimated by LSE minimization.
If the value of RE is known, optimization variables are: fS, QMS and QT.
In some rare cases a convergent LSE minimization cannot be realized. In that case the modified
standard Thiele-Small procedure would be used.
36
LIMP User Manual
Three methods will be used for the estimation of physical parameters MMS, CMS, RMS and Bl;
1. Added mass method,
2. Closed box method,
3. Fixed mass method.
All these methods can be used for measurement of a loudspeaker mounted in an infinite baffle or in
free air. The results for two mountings will differ slightly as air load mass is not the same. The air load
in a large baffle is two times larger than in free air. Although results will differ slightly we argue that
they can be successfully used in box design.
There are two standards that define measurement of physical loudspeaker parameters. IEC 60268-5
[15] defines closed box method. The other standard AES2-2012 [16] extends IEC standard with added
mass method. AES2-2012 also specifies that mounting should be done with firm clamp in vertical
position so that gravitational force does not offset membrane.
For estimation of physical parameters LIMP uses and improves methods promoted by Small [6] and
measurement system MLSSA. They are described in next sections.
M added
M MS = “constant force factor Bl”
f S QEM
−1
f M QES
When we know MMS and fS, it is easy to get the mechanical compliance CMS, resistance RMS and force
factor Bl. By using equations that are defined in Table 5.1 we get
1 S M MS RE S M MS
CMS = , VAS = 0c 2 S 2CMS , Bl = , RMS =
M MSS
2
QES QMS
This method is used in LIMP by default. It is called “constant Bl” method. Standard AES2-2012 uses
“constant compliance” method, where expression for CMS with and without added mass are equated to
get:
M added
M MS = “constant compliance”
f S2
−1
f M2
37
LIMP User Manual
In most cases assumption of constant compliance gives worse results than “constant Bl” solution. Why?
In low frequency region membrane has large displacement which generates nonlinear behavior of
compliance and force factor Bl, but compliance also has rise toward low frequencies due to
viscoelastic creep effect. In both cases it is important to measure impedance with small signal
excitation.
V AS 1 S M MS RE M
C MS = , M MS = , Bl = , RMS = S MS
0c S
2 2
CMSS
2
QES QMS
M MSVAS S2
S (cm 2 ) = 10 = 0.1677 f S M MSVAS
o c 2
where MMS in grams is measured using the added-mass method while VAS in liters is measured using
the closed-box method. Frequency fS is driver resonant frequency that is measured without box.
38
LIMP User Manual
1 S M MS RE S M MS
CMS = , VAS = 0c 2 S 2CMS , Bl = , RMS =
M MSS
2
QES QMS
This method is most accurate, and simplest as we need only one impedance measurement.
LIMP procedures for the estimation of basic loudspeaker parameters are as follows.
To estimate Q-factors of a loudspeaker we need:
By clicking the menu ‘Analyze->Loudspeaker parameters – Added mass method’, we get the
'Loudspeaker Parameters - Added Mass Method ' dialog box, shown in Fig. 5.3.
39
LIMP User Manual
In this dialog box, user enters values for Voice coil resistance and Membrane diameter in
appropriate edit boxes. Click on button ‘Calculate’ gets the report shown in the left edit box. This
report can be pasted to the clipboard by pressing the button ‘Copy’. The click on button Export saves
data in file using the Excel .csv format.
In the section "Nonlinear LSE Optimization" user sets three check boxes to allow more accurate
nonlinear LSE optimizations:
Estimate TSP by LSE minimization
Estimate voice coil resistance (if voice coil resistance is unavailable)
Estimate lossy inductor model
To apply LSE method, impedance measurement should be done in full frequency range (20 kHz), or at
least to frequency that is a hundred times larger than loudspeaker resonance frequency.
Figure 5.4 Impedance curves (the one with lower resonance frequency is obtained by adding mass to
the membrane)
40
LIMP User Manual
Then we click the menu item Analyze->Loudspeaker parameters – Added mass method to get the
'Loudspeaker Parameters – Added Mass Method' dialog box, shown in Fig. 5.5.
Figure 5.5 Dialog box for the estimation of loudspeaker parameters using Added Mass Method
User enters values for Voice coil resistance, Membrane diameter and Added mass in appropriate
edit boxes. Click on the button 'Calculate parameters' gets the report shown in the left edit box. This
report can be pasted to the clipboard by pressing the button 'Copy'. The click on button Export saves
data in file using the Excel .csv format.
Check box 'Driver mounted in infinite baffle' sets condition of measurement. Here it is empty which
means that driver is unbaffled.
Check box 'Use constant compliance method' enables alternate equation to estimate physical
parameters. The default is "constant Bl" method.
In the section "Nonlinear LSE Optimization" user sets three check boxes to allow more accurate
nonlinear LSE minimization.
Estimate TSP by LSE minimization
Estimate voice coil resistance (if voice coil resistance is unavailable)
Estimate lossy inductor model
Three lossy inductor circuits:
Le+L2||R2
Le+L2||R2+ L3||R2
Le+L2||R2||K
are selectable by using the list box at the bottom of this section (+ denotes serial connection, || denotes
parallel connection).
41
LIMP User Manual
Then we click the menu item Analyze->Loudspeaker parameters – Closed box method to get the
'Loudspeaker parameters – Closed Box Method' dialog box, shown in Fig. 5.6.
Figure 5.6 Dialog for the estimation of loudspeaker parameters using Closed Box Method
Finally, we enter values for Voice coil resistance, Membrane diameter and Closed box volume in
appropriate edit boxes. Press on the button 'Calculate' gets the report shown in the left edit box. This
report can be pasted to the clipboard by pressing the button 'Copy'. The click on button Export saves
data in file using the Excel .csv format.
In the section "Nonlinear LSE Optimization" the user sets three check boxes to allow more accurate
nonlinear LSE minimization:
42
LIMP User Manual
are selectable by using the list box at the bottom of this section (+ denotes serial connection, || denotes
parallel connection).
Then we click the menu Analyze->Loudspeaker parameters - Fixed mass method to get the
'Loudspeaker parameters - Fixed Mass Method' dialog box, shown in Fig. 5.7.
Figure 5.7 Dialog for the estimation of loudspeaker parameters using Fixed Mass Method
43
LIMP User Manual
User enters values for Voice coil resistance, Membrane diameter and Static membrane mass in
appropriate edit boxes. Click on the button 'Calculate parameters' gets the report shown in the left
edit box. This report can be pasted to the clipboard by pressing the button 'Copy'. The click on button
Export saves data in file using the Excel .csv format.
In this dialog the check box 'Driver mounted in infinite baffle' is not for informative purpose only,
like in previous cases. Here it determines the proper value of air load mass shall be used in
calculations of MMS.
44
LIMP User Manual
6 RLC Measurement
LIMP can be used to measure value of resistors, capacitors and inductors, simply by calculating
resistive, inductive or capacitive parts of the measured impedance. For example, Fig. 6.1 shows
impedance curves of an inductor with nominal value of 110uH.
By clicking menu command Analysis->RLC Impedance value at cursor position we get the dialog
box with report as shown on Figure 6.2. LIMP reports that measured impedance has resistive part of
0.15 ohms and imaginary part is inductive with value of 112 uH.
45
LIMP User Manual
Figure 6.3 Wrongly estimated impedance of plastic capacitor 4.7 uF/250V (measured without
the calibration)
Figure 6.4 Correctly estimated impedance of plastic capacitor 4.7 uF/250V (measured after
the calibration)
46
LIMP User Manual
Literature
[1] I. Mateljan, "Signal Selection for the Room Acoustics Measurement", Proc. 1999 IEEE Workshop on
Applications of Signal Processing to Audio and Acoustics, New Paltz, New York, 1999.
[2] I. Mateljan, K. Ugrinović, "The Comparison of Room Impulse Response Measuring Systems", Proc. AAA
2003, Portoroz, Slovenia, 2003.
[3] D. D. Rife, J. Vanderkooy, "Transfer Function Measurement with Maximum-Length Sequences”, Journal
of the Audio Engineering Society, vol. 37, June 1989.
[4] J. S.Bendat, A.G. Piersol, "Engineering applications of Correlation and Spectral Analysis", Wiley, New
York, 1980.
[5] L.L. Beranek, Acoustics, 1986 Edition, Published for the Acoustical Society of America by the American
Institute of Physics.
[6] R.H. Small, "Direct-Radiator Loudspeaker System Analysis", Journal of the Audio Engineering Society,
Vol. 20, June 1972.
[7] A.N. Thiele, "The Thiele-Small Parameters for Measuring, Specifying and Designing Loudspeakers",
PALA International Conference, Singapore, 2004.
[8] J. Vanderkooy, "A Model of Loudspeaker Driver Impedance Incorporating Eddy Currents in the Pole
Structure", Journal of the Audio Engineering Society, Vol. 37, March 1989.
[9] W. M. Leach, "Loudspeakers Voice Coil Inductance Losses: Circuit Models, Parameter Estimation and
Effect on Frequency Response", Journal of the Audio Engineering Society, vol. 50, June 2002.
[10] MLSSA SPO – Speaker Parameters Option – Reference Manual ver. 4WI, DRA Laboratories, Sarasota,
USA, 2001.
[11] Klippel Analyzer System – Linear Parameters Measurement, Klippel Gmbh, Dresden, Germany, 2003.
[12] K. Thorborg, C. Futtrup, "Electrodynamic Transducer Model Incorporating Semi-Inductance and Means
for Shorting AC Magnetization", Journal of the Audio Engineering Society, vol. 59, September 2010.
[13] Mateljan, I., Sikora, M. "Estimation of Loudspeaker Driver Parameters", Proc. AAA 2012, Zadar, Croatia,
2012.
[14] J. J. Moré, B. S. Garbow, K. E. Hillstrom, User Guide for MINPACK-1, Argonne National Laboratory
Report ANL-80-74, Argonne, Ill., 1980.
[15] IEC 60268-5 standard, “Sound system equipment – part 5: Loudspeakers”, IEC, 2003.
[16] AES2 – 2012 standard, “Methods of measuring and specifying the performance of loudspeakers for
professional applications - Drive units”, AES, 2013.
47