Ingemar Johansson, IJData
Luleå, Sweden
Email: ingemar.johansson@lulea.mail.telia.com
Homepage: http://hem1.passagen.se/ijdata
Unauthorized copying of this manual or the associated software is a violation against applicable copyright laws.
This demo manual is intended as a help when testing the demo, every now and then you will encounter references to figures that are missing, when you buy LspCAD you will get a printed manual along with the software however.
1 Introduction *
2 Installation *
3 General comments *
3.2 LspCAD directory structure *
3.3 Backward compatibility *
3.4 Technical support *
4.2 Common menu picks and dialogs *
4.3 Closed box *
4.4 Bassreflex box *
4.5 Double tuned bassreflex box type 1 *
4.6 Double tuned bassreflex box type 2 *
4.7 Passive radiator box *
4.8 Bandpass box type 1 *
4.9 Bandpass box type 2 *
4.10 Bandpass box type 3 *
4.11 Bandpass box type 4 *
4.12 What do the diagrams show? *
4.13 Nonlinearity simulations *
4.14 A few construction hints... *
4.15 Performance of the various boxes *
5.2 Menu picks *
5.3 The network dialogs *
5.4 The driver dialogs *
5.5 The general settings dialog *
5.6 The music filtering function *
5.7 What do the diagrams show *
5.8 A few hints *
6.2 The driver dialogs *
8.2 Export/import file formats *
Congratulations to your purchase of LspCAD.
LspCAD is a software that helps you to construct and model loudspeaker boxes and passive crossover networks. The name LspCAD is short for Loudspeaker Computer Aided Design.
LspCAD includes four utilities. The box utility that works as a box modeling program, the passive crossover utility and the advanced passive crossover utility that works as a filter modeling program with passive components and the active crossover utility. LspCAD does not only present ready to use solutions, a number of diagrams shows virtually all information needed in order to construct passive and active loudspeaker systems.
The box utility
The box utility manages to model 9 different kinds of loudspeaker boxes with dynamic loudspeaker units, these are:
LspCAD has a powerful tool for modeling room response and cabinet diffraction effects which will further help you as a user to create good sounding loudspeaker systems.
The simple passive crossover utility
The simple passive crossover utility can model 2-way and 3-way loudspeaker systems. Besides the rudimentary filter calculator existing in most loudspeaker modeling software the user can here import measured SPL and impedance data and thus model a real system. With the help from a unique feature one can filter a music sample through the loudspeaker systems frequency response and listen to the result through a pair of high quality headphones.
The advanced passive crossover utility
The advanced passive crossover utility can model 1-way through 4-way loudspeaker systems with up to four loudspeaker units in each network. Modeling capabilities are much more extensive than for the simple passive crossover utility. A major benefit is the capability to optimize the filter components in order to achieve a predetermined transfer function, this makes construction of complex filter structures as easy as a walk in the park.
The active crossover utility
The active crossover utility allows for the construction of active crossover networks built up with operational amplifier circuit elements. The optimization possibilities described for the advanced passive crossover utility exists for the active crossover utility as well.
System requirements
The demands on your IBM compatible are:
The software is easily installed with the enclosed installation program.
[x]:install followed by OK.
[x] is replaced by e.g. a or b i.e the name of your 3.5" disk drive.
The software for LspCAD for Windows is developed under Borland C++ 5.02. The installation program is used with permission from INSTALLSHIELD.
In order to not be worse than the big companies. It is hereby stated that no responsibility is taken for damages (personal and/or other) caused by using this software.
LspCAD is very straightforward and simple to use, however it is advisable to at least glance through this manual quickly before using the software.
Most interaction with the programs are made through dialogs, the dialogs are covered in detail in this manual, moreover the same and sometimes even additional info is revealed by clicking on the question-mark button that is visible in most dialogs, then a notepad document with help info will pop up. The extensive use of dialogs is quite neat as one does not need to cram every piece of information onto the screen, the drawback may be that the user only sees a blank screen on start up.
There exist a number of buzz words in the MS-windows world, some of which are covered below. By clicking is meant clicking with the left button on the mouse.
When LspCAD is started what comes up is only a blank screen. The user must here select what he/she wish to do. The menu pick File | New | Box… and
File | New | Filter… lets you select which kind of project you want to start.
Fig. 1 shows the looks of the menu picks for the box projects.
For instance if we wish to model a closed box the menu pick File | New | Box | Closed will create a closed box project.
3.2 LspCAD directory structure
When storing driver unit files and measured data it is strongly recommended that you put these data in a subdirectory of the LspCAD directory, the simple reason is that LspCAD then strips away the first part of the path in the path. For instance the path:
C:\PROGRAM\LSPCAD\DRIVER\DYNAUDIO\17W75.UNT
will be truncated to become:
DRIVER\DYNAUDIO\17W75.UNT
This makes it simple to share project files with other LspCAD users.
The driver unit files created under LspCAD 2.10 and later can be used in LspCAD 4.0.
LspCAD project files created from LspCAD 3.0 and later can be used by LspCAD 4.0.
Additional upgrade information can be found at :
http://hem1.passagen.se/ijdata
Technical support is given only to registered users of LspCAD. To become a registered user you must fill in an send the registration form (remember the registration number). Questions from unregistered users are not answered.
When you have problems or questions please send an email to:
ingemar.johansson@lulea.mail.telia.com
Remember to give you name and the LspCAD registration number in the mail.
The box utility is the part of LspCAD which lets the user model various type of loudspeaker boxes. The main object here is to model the low-frequency part i.e we aim mainly at boxes for the bass region, however boxes for midrange units can be modeled here too.
The project files created by this utility has .box as standard extension but one may use other extensions as well.
A multitude of diagrams can be viewed here, they are:
There also exist an information window that gives an overview of the project.
The menus and dialogs that are common are described first after that the different box projects are described. Finally the different box types are described in detail.
4.2 Common menu picks and dialogs
The user interface is arranged such that only the Box menu pick differs between the different box types, this makes it easy for the user to get a grip on the user interface very quickly.
See below a description of the menu picks that are common for all box types:
4.2.1 Driver unit configuration
In this dialog you can set number of driver units, isobaric configuration and electrical connection of multiple driver units and double voice coil electrical connection. Here you also import measured data.
Electrical conn...
In the listbox that pops up you have the opportunity to select how two or more driver units are electrically connected, see fig. 2. There is always the option to select series, parallel and separate feeding of the driver units.
Edit driver unit parameters. You can create you own loudspeaker parameter files which are used by LspCAD. The parameter files use .unt as default extension, but nothing stops you from using another extension.
Below follows a description of the various menu picks.
In this dialog you set general stuff such as number of analysis points, input level etc.
Frequency settings
Here you setup the appearance of the transient response diagram. See also chap 4.12.7 and 4.15.
Resolution
Select time resolution from 25ms to 2 ms.
Input signal
4.2.5 Room and cabinet impact (Room & Cabinet)
The room and cabinet simulation is controlled by means of a dialog that can be left open on the screen.
The room and cabinet feature gives you the option to watch the loudspeaker system behaviour under "real" circumstances, i.e in a room and with a normal limited-sized cabinet. This gives you the opportunity to simulate cabinet edge diffraction and also the nearest reflections from floor, ceiling and walls.
The model assumes that the loudspeaker is placed in a room given by the drawing in fig. 9.
Fig. 9 Loudspeaker in listening room
Room dimensions
Set the dimensions of the listening room
Speaker position
Set the position of the lower left corner of the front baffle of the loudspeaker.
Listener position
Set the position of the listener
All positions are in a three dimensional coordinate system where origin is denoted (0,0,0) in fig. 9.
Front baffle dimensions.
Set the Width and Height of the front baffle.
Source position
Set the position of the different visible acoustic entities (driver, ports e.t.c) of a box, for a bandpass box, only the ports are visible. The positions (dX, dY, dZ) are relative to the Speaker position.
The room reflections that are possible to simulate are:
The menu pick Absorbtion coeffs. enables you to set the absorbtion coefficients for the different reflecting areas of the room.
Measured SPL data
Here you import measured SPL (frequency response) data. The imported data is merged together with the simulated data so that the imported data occupy the frequency range below the Transition frequency in the diagrams.
The room reflections are weighted so that reflections up to 20 ms after the direct wave are weighted equally while reflections between 20 ms and 40 ms are gradually de-weighted. Reflections later than 40 ms after the direct wave are not considered at all. The reason to this is twofold.
Fig. 10 shows a comparison between a gated measurement and a room simulation. It is obvious that the correlation between the graphs is quite good.
The snapshot feature allows you to save up to five frequency responses on one diagram. It can be used to compare the performance of different constructions and also the frequency response for different power levels. Fig. 11 shows an example of how the snapshot diagram may be used.
The snapshot diagrams are saved separately for the simple reason that it should be possible to load them into other projects.
The snapshot dialog controls the appearance of the diagram.
The menu picks of the dialog are:
The Yankee tool is a neat feature, which is NOT intended as a yab at Americans, the yankee tool is a simple conversion tool between e.g mm and inch.
The iterative optimization dialog varies slightly in appearance depending on which boxtype you work with.
This feature is, when properly used, a good means to achieve an optimal construction for a given driver unit. For natural reasons one can not achieve everything. The algorithm does its best to get you where you want.
During the optimization you will probably have to modify the desired upper and lower cutoff frequencies in order to achieve a good response curve.
As always, when it comes to optimization, the algorithm can deteriorate, one therefore should check volumes and tuning especially for the bandpass boxes during optimization. Another good thing is also to check the port lengths so that they are not unnaturally short.
The start frequency should lie 1.5 to 2 octaves below the lower cutoff frequency and the stop frequency should lie 1.5 to 2 octaves above the upper cutoff frequency.
What might happen during optimization is that the construction does not get better (the mean error does not decrease). The reason may then be:
The dialog controls are:
This is the most simple box type available. One of the shortcomings of this construction is that the output power at low frequencies is very limited unless you use very large or long stroke loudspeaker units.
This boxtype has no advanced demands on driver unit Qts. However a very low Qts may give problems with a high cutoff frequency. Damping material can be used extensively in this box but should not be placed too close to the driver unit as this may increase distortion.
Here you may enter data about the box, the controls are:
QL that is the Q-value due to unwanted leackage from the box.
QB that is the Q-value due to damping materials in the box. An undamped box with rigid walls has a QB of 15, while one may reach a QB of 3 if one stuffs the box completely with damping material.
This type of box is probably the most common. The possibility to get higher output levels is greater than for a closed box.
The driver units should not have too high Qts. Damping material should be used with restriction and with the objective to reduce standing wave modes within the box. Much damping material may of course be of some help if driver Qts is high.
The box volume and damping is described in chap 4.3.
The port is described by (see also fig 12 and chap 8.1):
In this dialog you can also select from the menu picks:
4.5 Double tuned bassreflex box type 1
This type of box has been described in a book (title unknown). In the description the rear box should have twice the volume of the front box.
It is hard to say whether this box is useful or not. It’s extremely difficult to achieve a good frequency response. The idea is probably to widen the port resonance and thus limit the cone excursion over a wider range. In reality this seems to be very difficult, one may however use the rear port as a muffler that will block the standing wave modes caused by the front port. In this case one can tune the rear box as an ordinary bassreflex box and give the front box a relatively small volume and tune it to about 100-200Hz.
The dialog is divided into four major groups.
The menu picks in this dialog are:
4.6 Double tuned bassreflex box type 2
This box type is a variation of the double tuned bassreflex box type 1. It has the advantage that one can achieve a frequency response that helps to reduce the rise in sound pressure toward lower frequencies caused by ordinary listening rooms (see example)
The rear box should be filled completely with damping material while the front box should be moderately filled.
When making a construction using this box type one can start from an ordinary bassreflex box with a volume Vb and tuned to a resonance frequency Fb. One third of Vb is assigned to the rear box while two thirds of Vb is assigned to the front box. The front box is tuned to Fb and the rear box is tuned to about 100-120Hz. After these steps one only has to do a little trial and error. One can also use the Quick box pick to get a starting point fast.
The menu picks for this box is the same as for the double tuned bassreflex box type 1.
The principle behind this box is almost the same as for a bassreflex box. The difference lies in the fact that the vent is replaced by a loudspeaker without a motor and magnet, also known as an auxiliary bass radiator (ABR). Alternative names are drone cone or passive radiator.
The effective area of the ABR should not be less that that of the driver unit as the former has to move a lot of air at the resonance frequency.
An interesting point is that the frequency response shows a notch below the cutoff frequency. Damping material should be used with care and only to limit standing wave modes in the box.
The dialog is divided into two major groups.
This type of bandpass box is the simplest. The efficiency depends on the bandwidth in the aspect that the higher the bandwidth the lower the efficiency. The normal bandwidth is 1 to 2 octaves. A benefit of this boxtype is that it is quite simple to achieve high output at low frequencies, even though the total volume is small. A drawback is that the port is prone to generating standing wave mode resonances, the first at one half wavelength, so the bandpass box is best used in combination with a lowpass filter. The port should have a rather big area as all energy emits through it. The Qts of the driver units should not be too high as the required box volume increases with Qts. Suitable Qts values range from 0.3 to 0.8. Damping material can be used in the rear box but should be used with care in the front box.
When it comes to iterative frequency fit optimization, it can be said that it is quite hard to fail. Just set the desired upper and lower cutoff frequencies and start optimizing.
The dialog is divided into three major groups.
The menu picks are:
If the driver unit has a high Qts it may be impossible to select all alignments.
This boxtype looks like the bandpass box type 1, however an extra port is inserted between the boxes. This port serves to reduce the cone excursion at low frequencies and also to extend the frequency response towards lower frequencies, bandwidths of 2-3 octaves are possible.
The resonance frequency for the rear port should always be lower than the resonance frequency for the front port. Damping material should be used with care and with the objective of preventing standing wave modes in the boxes. A suitable Qts for the diver units lies between 0.2 and 0.45.
The dialog is divided into four major groups.
Note also that you cannot directly set the resonance frequencies for this type of box. The reason is that the ports are mutually dependent, however you can indirectly set the frequencies by changing e.g. port lengths.
The menu picks are
This boxtype looks like the bandpass box type 1, however an extra port emanates from the rear box. This port helps to reduce the cone excursion at low frequencies and to extend the frequency response toward lower frequencies, bandwidths of 2-3 octaves are possible. It can be noted that BOSE corporation, who also holds a patent on this boxtype also uses it in their acoustimass systems.
An interesting thing with this boxtype is that if both the volume and tuning of the two boxes are identical. The total output will in theory be zero !.
The front port should preferably be tuned to the higher resonance frequency, the acoustic phase will then pass zero in the passband and the signal is reproduced in correct phase.
Damping material should be used with care and only with the objective to reduce standing wave modes in the boxes.
The driver unit should preferably have a Qts in the range 0.2-0.45.
The dialog is divided into four major groups.
QB, simulation of QL is not supported for this box
type.
The menu picks are
This boxtype is actually a variation of bandpass box type 3 with the exception that it is built up with three chambers and has a minimum of 2 driver units, the performance is however the same. Note that you may only use 2, 4, 6 or 8 driver units with this box.
This boxtype is often used in car stereo system, mostly because it generally uses 2 driver units.
The front port should preferably be tuned to the higher resonance frequency, the acoustic phase will then pass zero in the passband and the signal is reproduced in correct phase.
The dialog is divided into four major groups.
The menu picks are
4.12 What do the diagrams show?
Speaker box has the diagrams in separate child windows or MDI windows as they are also called. A diagram consists of one or more plots. Diagrams that are iconized may be opened by means of a double-click on the icon. The diagrams can be resized and maximized.
Note that you can select Window | User note to insert a line of text at the bottom of the diagrams.
4.12.2 Free air SPL at 1m distance
This diagram shows the sound pressure level at 1m distance as a function of frequency and input power. The speaker is assumed to radiate into 4p or free field space i.e it is mounted at the end of a long tube. The sound pressure level if the speaker is placed in an infinite baffle is up to 6 dB higher. What this may look like you can look at in the SPL in room & cabinet diagram.
The dynamic range i.e. the difference between the lowest and the highest reading on this diagram can be set in the Measurement setup dialog (see chap 4.1.3).
This diagram also shows the phase response.This phase may show abrupt +/- 180 degree changes, this is however only caused by the atan2 function used when computing the phase.
This diagram shows the cone excursion (mmp-p) for each driver unit as a function of frequency and input power.
Remember that the cone excursion in this program is given as a peak to peak value. One important thing is that if the cone excursion exceeds the Xmax value then distortion will increase abruptly and sound pressure will quit rising linearly with input level. This is visible if the simulation of the BL non-linearity is turned on in the Driver unit config dialog.
This diagram shows the port air speed in the port(s) (m/s RMS). Important here is that if the port air speed exceeds some 15m/s then problems with chuffing is very likely to occur. If many ports exist in the box, e.g. a bandpass box type 2 then 2 plots are shown which are also tagged to avoid confusion.
If simulation of nonlinearity in the port is turned on the plots will show a knee at 15m/s. Non linearities in the port occur when the flow in the port goes from laminar to turbulent flow, this is meant to occur at about 4.5% of the speed of air. At these high airspeeds chuffing will also occur. High air speeds in the ports should be avoided if possible, not only because of the compression, but also because this chuffing noise excites the organ pipe resonances.
The impedance curve shows the loudspeaker systems total load upon the driving amplifier. Note that the program will automatically compute the impedance depending on the electrical connections of the driver units, even if there are many. This diagram also shows the phase
The group delay is defined as -(dF/df) i.e. the acoustic phase differentiated with respect to frequency. It may simply be expressed by how tones of specific frequencies are delayed with respect to nearby frequencies. As group delay magnitude is normally much smaller for higher frequencies than for lower frequencies one may have problems in seeing what happens at high frequencies due to the autoscaling function in LspCAD. With the Page Up and Page Down keys on the keyboard one can set the magnitude scaling of this diagram.
For the inexperienced this diagram may be difficult to understand. In this case the transient response may be easier to understand.
4.12.7 Impulse- Step- Tone burst response
This diagram shows the loudspeaker performance in the time domain.
The start frequency of the analysis should lie at least 2 octaves below the lower cutoff frequency of the loudspeaker system. This means that if we have a cutoff frequency of 60 Hz then the start frequency should lie at 10-20 Hz.
In the dialog that controls the appearance of this diagram one can choose between impulse, step or tone burst response.
The step response shows the response if you apply a fixed voltage to the loudspeaker terminals and measure the response with a microphone and an oscilloscope.
The impulse response shows the response if you apply a very short pulse to the loudspeaker terminals
The tone burst response shows what happens if we apply a tone burst with a specified frequency and specified number of periods at the loudspeaker terminals. This shows what happens when the loudspeaker tries to reproduce a hit on e.g. a kettle drum.
See chapter 4.2.5 for further description of this. This diagram also shows the phase.
4.12.9 SPL with port standing wave modes
This diagram only exists for the bandpass boxes and is intended to make the constructor aware of the problem that may arise with bandpass boxes. See also chap 8.1, the section about the organ pipe resonances.
See chap 4.2.6 for further description of this.
The information window is not a diagram. It is a document that shows all setups in the program and also some extra information such as -3 dB frequencies. This document may be printed out and copied as the can diagrams. If the diagram is copied the contents of the clipboard will be in ASCII format, the reason is that it should be simple to import the data to your own documents.
The so called jw method is used in this software, this assumes that we are dealing with linear relationships, this is the fact at low to moderate power levels. There are some possibilities to model non-linear relationships in LspCAD, they are described below. The models are not 100% proof but anyway they give a quite good idea of what happens at high power levels.
The model assumes that the model has a temperature of +250º C at Pmax and that the voice coil is made of copper. The function cannot be used if Pmax is not defined.
This feature is at best only commented in the product sheet, and it is virtually never mentioned what the stiffness curve looks like.
The model in this program assumes that the stiffness starts to increase at Xmax/2 and is doubled at Xmax.
This function cannot be used if Xmax is not defined.
· Port air speed non linearity:
Simulates the non linearity that occur in a port when the air speed exceeds 15 m/s, which in turn leads to turbulent airflow which will obstruct the flow in the port.
For the cabinet or enclosure it is very good to use Medium Density Fiberboard (MDF). MDF is easy to work with and has very good acoustic properties.
For the front baffle it is good to use double 19 mm board for medium sized boxes. Remember though to cut the loudspeaker hole of the inner board bigger than the mounting hole on the outer board. This will limit the risk that the mounting hole forms a long tube that may cause resonances.
It’s generally better to use a high and narrow box rather than a square shaped box as the medium and high frequency range is better reproduced if the driver units of the latter is higher above the floor. Of course one does not need to follow this advice if the loudspeaker is mounted on speaker stands.
The most common damping material is fiberglass. The main purpose of damping material is to limit standing wave mode resonances in the box which occur when one inner size of the box is equal to a multiple of a half wavelength. For example if the inner height of a box is 1m then one gets the first mode as low as 170 Hz !. This mode is actually best dampened out with heavy stuffing of damping material halfway up the box as the particle velocity is highest at this point for this mode. Damping material along the inner walls seldom does much for the damping...
4.15 Performance of the various boxes
The boxes presented in the previous chapters all have their pros and cons, performance can measured in a number of ways. For maximum sound pressure reflex loaded cabinets are to prefer over the closed box.
One can for instance compare the cone excursion between a closed box and a bassreflex box. While the cone excursion for a closed box increases up to a certain limit as frequency gets lower the bassreflex box shows a notch in cone excursion at the port resonance frequency. This has implications on both power handling and distortion and makes the bassreflex box able of handling higher power levels than a closed box.
The above statement applies only to steady state sinewave signals however. If a toneburst signal of the same frequency as the port resonance is applied to a bass reflex box, the cone excursion will initially be quite high and will reduce to a small value after one half to a couple of periods. This is due to the fact that the port-box system can be viewed as a spring with a weight connected to it. At first we need quite an effort to get things going, but after a while the system will oscillate by itself. With this in respect we realize that for transient rich bass sounds the cone excursion is initially equal to or sometimes greater than for a closed box!. Also if the bassreflexbox is subject to inputs with frequencies lower than the port resonance the driver is very easily overloaded with large cone excursions even at low power levels.
Another drawback with ported boxes are that the vents are prone to producing standing wave resonances and also chuffing noise at high power levels.
Speaking in general one can say that the higher the system order, the poorer the transient response with this at hand we can make a list of transient response quality with closed box being the best:
Transient response can be measured in a number of ways, in LspCAD one can model transient response in three ways: impulse, step or tone burst response.
The step response should die off as fast as possible, closed boxes normally shows one overshoot while a bass reflex box may show an oscillating behavior depending on how well it was designed.
One interesting thing is that a bass reflex box with a bump in the frequency response may in some cases show a better step response that a bass reflex box with a flat response.
A tone burst should be built up fast and die off fast. A closed box builds up and stops this tone burst very quickly while e.g. a bandpass box type 3 may show a very slow behavior.
How does a good transient response sound ?
The answer may be that a good transient response should not be heard at all. The major problem with boxes with poor transient properties is that the punch in the transients is smoothed up. The ringing after the transients are not an equally big problem as the ear is not as sensitive to this due to temporal masking effects in the auditory system
5 The simple passive crossover utility
The filter utility is a simple filter calculator with the ability to import measured SPL and impedance data, thus one can simulate and model "real" 2-way and 3-way passive crossover networks. The crossover networks consists of 2 or 3 networks for the bass, mid and treble region. Each network can model up to fourth order highpass and lowpass filters, furthermore the bass and mid networks support baffle diffraction compensation (a resistor in parallel with an inductor). All networks have support for attenuators and zobel networks. The number of analysis points is 500 in the frequency range 5-20000 Hz.
The project files created by this utility has .flt as standard extension but one may use other extensions as well.
There exists a schematic window that shows the looks of the passive crossover. The complete schematic for the mid network with all components in use is shown in fig. 14 below.
The component values in the schematic are hidden if they are set to zero (undef.) in the network dialogs. Note that the first digit in the component names are (1) for the bass network, (2) for the mid network and (3) for the treble network.
The user can view 8 diagrams, they are:
Below are described the menu picks for the filter utility is described.
- SPL (total in different
angles or on axis for each individual network)
- Filter gain for each individual
network
- Input Impedance (total
or for each individual network)
- Group delay (total or
for each individual network)
- Impulse response
The network dialogs serve as placeholders for filter parameters, the network dialog for the mid network is described in this chapter, the dialogs for the bass and treble networks are similar
The dialog is divided into a number of groups:
Attenuator
Attenuator circuit
Lowpass filter section
Highpass filter section
Impedance equalization
The driver dialogs serve as placeholders for driver parameters, the driver dialog for the mid network is described in this chapter, the dialogs for the bass and treble drivers are identical
The dialog is divided into a number of groups:
In this group one set the parameters for each driver unit, if only 1 driver unit is in use the one can only set the parameters for Driver unit 1.
5.5 The general settings dialog
In the general settings dialog various parameters are set. This dialog consists of a few groups.
Display frequency response
The internal frequency span in the crossover utility is always 5-20000Hz. In order to be able to view a smaller part of the frequency span one may here set the Lower and Upper limits.
IFFT (impulse response)
In this group one set the number of points and the sample rate of the impulse response calculation. One can also disable the calculation of the impulse response just to save time.
Measurement distance
One can here choose how the frequency response should be calculated, either at an Infinite distance or at a predefined distance from the loudspeakers reference point. In all cases the measurement is along the Z axis in the coordinate system.
Off axis plots
Here one can select which vertical off axis angles to measure from. For the surface plot one can select up to 16 angles besides the 0 degree angle which is always default. For the overlay plot one can select up to 7 angles besides the 0 degree angle. The angles for the surface plot is selected from the leftmost list, while the angles for the overlay plot are selected from the surface plot list. With the Flip up/down checkbox the user can select if the positive angles should be upper or the lower half of the surface plot. The inhibit check box will cause the plot of the off axis diagrams to be inhibited when it is checked, this is valuable if one wish to e.g optimize the networks without having to deselect all rays. The Hide on axis plot checkbox makes it possible to hide the on axis plot in the off axis diagrams.
Diagram scaling
The scaling of the diagrams that present impedance, groupdelay filter gain and sound pressure level can be altered here.
5.6 The music filtering function
This function has been the most difficult one to name. The purpose is, however to filter a music sample through the frequency response of the loudspeaker. After this operation one can listen to the result with a pair of high quality headphones. The file format of the input and output files is raw 16 bit stereo in Intel format i.e (LSB,MSB) format.
The Browse buttons in the Infile and Outfile groups are used to select the in/out files.
The Start/Stop in finally used to start and stop the filtering function. If the output level is too high one can reduce the level by means of the Scale edit field. Note that the filtering must be stopped before the scaling is changed.
The Simulate HRTF makes is possible to simulate the head realated transfer function when checked. The effect is that the perceived sound image is more alike a loudspeaker sound image although one listen with headphones.
For optimum speed a block filtering function using overlap-add FFT/IFFT operations is used. Therefore the sample rate in the IFFT group in the general setup dialog must be set correctly. Very short IFFT lengths may give an audible block effect, therefore it is recommended that the #points field in the general setup dialog is not less that 1024. The block length in the FFT operations is twice the #points, i.e if #points is 1024 then the block length is 2048. Note that this function is quite slow, filtering 1 minute of CD music takes about 5 minutes on a Pentium 200MHz PC.
For optimum performance one should use high quality sound boards in combination with this function. Examples of high quality soundboards are:
Also a couple of high quality headphones with a flat frequency response is a must, good headphones are e.g :
As the input and output files must be in raw 16 bit format (PC byte order). A shareware software called CoolEdit is recommended for recording and listening purpose. CoolEdit is a shareware which means that if you whish to use it you should pay the registration fee although the unregistered version can be used for recording and listening purposes.
A last word about recording music. To my knowlegde copyright rules allow one to copy own CD records for own purposes. Therefore it should be legal to record music into the harddrive and use it for simulation purposes. If you feel uncertain about the copyright rules check out with the record companies.
The filter gain lets you see the transfer function of each network.
The input impedance diagram shows the load on the driving amplifier, the individual impedance curves for each network is shown as is also the total impedance and the phase thereof.
5.7.3 Summed frequency response
The summed frequency response on axis is shown. Also the individual contributions and the total phase is shown.
5.7.4 Individual phase response
This diagram shows the phase response for each network, including driver phase response. This diagram is good to look at when optimizing for the best group delay characteristics.
This diagram shows the group delay for each network including driver and the total group delay. Interesting to see here is that if e.g the bass and mid regions does not overlap well at the crossover frequency the group delay will peak at this frequency. As group delay magnitude is normally much smaller for higher frequencies than for lower frequencies one may have problem in seeing what happens at high frequencies due to the autoscaling function in LspCAD. With the Page Up and Page Down keys on the keyboard one can set the magnitude scaling of this diagram.
Impulse response of the total system. The length and sample rate is set in the general setup dialog.
5.7.7 Vertical off axis response, surface plot
This diagram looks much like a carpet when many viewing angles are selected. Here one can view how uniform the loudspeaker is in performance for other angles than on axis. Redrawing of this diagram is quite time consuming, therefore this diagram should be minimized when not in use.
5.7.8 Vertical off axis response, overlay plot
This diagram allows the user to see the frequency response for a limited set of viewing angles.
5.8.1 Optimizing frequency/phase response
For the beginner it is advisable to start with a first or second order filters as higher order filters are quite tricky to optimize by hand. When optimizing two regions e.g bass and mid, it is recommended that the phase response of the two regions are overlapping i.e that the phase values are nearly the same at the crossover frequency. If the above proves difficult then one might consider moving the crossover frequency up or down or change the filter order of one or both networks, note that badly overlapping phase response will instantly show up as bad group delay characteristics.
5.8.2 Optimizing the zobel network
The most simple way to optimize the zobel network is type in Re and Le and then click on Calc in the zobel network group. This gives fairly good results but of course this can be done better. The steps below shows how:
5.8.3 How to use the notch section
The notch section can be used as a notch section if all three components R15, L15, C15 (for the bass network) are defined.
If only R15 and L15 are defined one can use it as a diffraction rise compensation link. A diffraction rise compensation link can be used when one whish to compensate for the step-like increase in sound pressure between 300-1500Hz that generally occurs due to baffle diffraction. The steps below shows how to utilize this feature in the best ways:
6 The advanced passive crossover utility
This utility allows for the construction of complex crossover networks. For those that are familiar with the IMP3wVR software by G.R Konce it can be said here that many ideas from that program are stolen with pride (and also with kind permission). Up to four way systems (four networks) can be modeleled, each network can consist of up to four driver units. The simple crossover utility is a good start for the one who feels that he or she is less experienced in this area. The advanced crossover however makes no assumptions about what is bass, midrange or treble. Instead for each network one can select from the same set of components.
The major advantage of the advanced passive crossover is that one can optimize the components of each network in order to achieve a predetermined frequency response.
In order to get all things working both impedance and SPL data must be imported for all networks.
The menu picks that differ from the simple passive crossover utility:
The topology of the network is such that each network can consist of up to 12 branches connected after one another.
Branch 1,3,5 and 7 are series branches i.e they are in series with the loudspeaker load.
Branch 2,4,6, 8 and 11 through 14 are parallel (shunt) branches.
Fig 16 below shows a simple schematic with one driver unit where all branches are used.
The naming of the components follow the convention
XIJJK where:
X is the component type, C for capacitor, L for inductor and R for resistor.
I is the network number (1..4)
JJ is the branch number (1..8,11..14)
K is either 1 or 2, in cases where 2 resistors are used in a branch one resistor will have suffix 2.
One example:
R3082 is the second resistor of the eighth branch of the third network.
It is possible to select from 10 different branch types (see figure 17 below), they should cover most of the needs.
The network dialogs differ from
the simple passive crossover utility in that one can select how the network
should look like by altering the branch types. For instance if one double
click on
Branch 08 in the left listbox another
dialog pops up (not shown) were one may alter the branch type.
If one double click on a component in the list box a smaller dialog pops up that allows the user to enter a new value. One may also alter the component values by selecting a component (single click) and then press arrow left/right to alter the values 5% up or down.
The network dialog box dialog box consist of two listboxes, the left is the main listbox where one can alter the branchtype and the component values. The right listbox is the optimization listbox. In this listbox one can only alter component values.
All components that one want the network optimizer to optimize are selected from the left listbox and added to the right (optimize) listbox by clicking on the Add button, to remove components from the optimize list one select the component in the optimize listbox and click on the Remove button.
Not exactly a new Houdini but anyway the Wizard might help the user to get a first starting point. The Wizard is accessed by a click on the Wizard button in the network dialog.
Here the user makes a first wild guess of the components in the network. In the example given in the figure a filter will be computed that would achieve a 3rd Butterworth bandpass alignment with cutoff frequencies 500 and 3000 Hz. The nominal load (here 5.5 W) is set so that the match with the loudspeaker load is the best possible.
In the insertion point list box one may select where the filter should be put in the network.
Note that if only a low pass filter is wanted the order of the highpass filter (HP) is simply set to zero. The opposite applies if only a high pass filter is wanted.
Very quickly one will notice that the filter components that the Wizard propose does not match a real loudspeaker unit very well, especially as the driver units frequency response need to be compensated for, a fact that the wizard does not care the least about. That is why we need the optimization tool that is described in the next chapter.
The optimization tool makes it possible to compute filters that match a predetermined target curve.
With this tool it is possible to model and construct complex structures such as 4th order bandpass filter alignments, a task which would take weeks or even months to do by hand.
The optimization tool is accessed by a click on the Optimize button in the network dialog.
Before the optimization session can start we must first add the components that need to be optimized to the optimization list (see chap 6.1).
The components that are subject to optimization are available in the lower left corner of the optimization dialog. In the diagram up to three plots are available:
Solid red (thick): The frequency response of the network+driver unit.
Blue: The target frequency response.
Red dotted: The unfiltered frequency response.
The target curve is "floating" along with the solid red plot, i.e the target does not show any target SPL, it is only the alignment with the target curve that is important here.
A few controls exist in this dialog.
Here one set the properties for the part of the target curve that describes the low pass alignment.
Here one set the properties for the part of the target curve that describes the low pass alignment, see above for explanation.
Optimization range
Here one select over what frequency range the optimizer should compute the difference metric, this is useful as valuable components are otherwise wasted on unimportant parts of the frequency range. A recommendation here is thus to set the optimization range so that the working range is included and also points in the frequency range that are 20-25dB below the peak sound pressure level or higher.
A few words….
With the optimization feature the construction of complex filter structures is really simple.
One problem that one may face is the case where the drivers for the different networks are not acoustically aligned.
By acoustically aligned is meant that the sound from each driver should arrive at the listener at the same time.
In the case where the drivers are not acoustically aligned the minimum phase property that makes it possible to optimize each network to a certain target response and then "glue" together the result will fail to various extents depending on how badly aligned the driver units are. This problem is larger in the treble region. There are many reasons to why the driver units should be acoustically aligned, the main reason is that the sound image fits together much better if the drivers are aligned.
The driver dialogs are almost the same as for the simple passive crossover utility.
When driver units 2 and up are selected two extra checkboxes appear with the caption "Same as driver 1". The idea behind this is that one checks this checkbox if one wish to use the data that is imported for driver unit 1. Thus one avoids the need to import the same data over and over again. These are by default checked and will be unchecked automatically as soon as one import a file. Moreover the Scaling input fields are dimmed out to signify that the scaling properties of driver 1 apply if the checkbox is checked.
7 The active crossover utility
The active crossover utility makes it possible to construct and optimize active crossover networks. As in the case of the advanced passive crossover utility it is possible to model up to four way active systems with up to four different driver units in each network. In order to get all things working SPL data must be imported for all networks. Impedance data can also be imported but there is no need for them if branches 11 through 14 are not used.
The topology of the network is such that each network can consist of up to 12 branches, each connected after the other.
Branch 1 through 8 are series branches i.e they are in series with the signal to the power amplifier. Each of these branches are equipped with an operational amplifier. Please note that each branch assumes that it is fed with a low impedance source, this especially applies to the first branch.
Branch 11 through 14 are parallel (shunt) branches and they are placed after the power amplifier.
The naming of the components follow the same rules as given for the advanced passive crossover utility. There is no impedance diagram available as the need for it is quite small. Fig 19 shows the available branch types.
When using the Q boost and Q cut one may quickly end up with enormously large inductors. The trick is here to use a gyrator circuit as shown in fig 20. With the use of a gyrator circuit one gets a simulated inductance that eliminates the need for a real inductance.
The network dialogs are essentially the same as for the advanced passive crossover. An extra edit control allows the user to set the relative Gain of the network. The network optimizer is described in chap 6.1.2.
The Wizard for the active crossover
utility is similar to the advanced passive crossover utility. The procedure
is simple, select lowpass (LP) or highpass (HP), filter order, cutoff frequency
and where in the network the filter section should be placed.