Open Baffle simulation in Boxsim

I don't know how you came to that conclusion. It can be done, but that's not the default for a non-zero enclosure depth. Here is a plot from an example project. I took the default two way project that is in Boxsim every time you open it and made a duplicate of each driver on the back cabinet wall to produce a bipolar speaker. In the example below I set the reference project ( in the Edit menu ) to have a 10 cm deep enclosure. I then changed the enclosure to have a 50 cm deep enclosure. The reference is 10 cm deep box plotted with dashed line, the current 50 cm deep is plotted with the sold line. Check out the 7 dB difference at 200 Hz from the delay of the rear driver due to the 50 cm box depth.

The acoustic center of the rear driver can be moved in the " Driver N, enclosure and baffle menu". Driver 3 is the rear facing woofer in this project. See the last parameter in the lower screen shot is defaulted to -50 cm for the rear driver of the 50 cm deep cabinet. So a delay would not be required if it was set to 0, effectively moving it to the same location as the front facing driver on a non-zero thickness baffle. Including the baffle thickness allows for proper modeling of the diffraction and makes possible using a chamfer.

I attached the project file so people can try it for themselves. To model a dipole speaker flip the phase on the rear drivers in the crossover and set the "to listener" distance to the same value for front and rear drivers.

1747237056447.png


1747237807159.png
 

Attachments

Last edited:
  • Thank You
Reactions: jaapVL
Just designing and simulating the above system took me quite some time. Therefore, I cannot imagine that it is possible to build a passive open baffle system without simulation support.
In a subsequent post, I will describe step by step how I designed and simulated the above 4-way system.
It is possible, it is just not easy. I have done two pairs without simulation. That said, even though I think they are both good sounding and enjoyable, it is certainly possible that they could be better. From my experience, one of the biggest factors, once the frequency response is basically balanced, is the room the speakers are in, and their positioning within the room. I don't know if there is a program that can both design a di-pole or bi-pole, and also predict how it will work in a given space.
My systems are multiway, and taper gracefully toward the top, so they look good and take advantage of the varied baffle width already discussed here.
 
  • Like
Reactions: jaapVL
It is possible, it is just not easy. I have done two pairs without simulation. That said, even though I think they are both good sounding and enjoyable, it is certainly possible that they could be better.
If you want to make a passive system, I can also imagine the following procedure:
Build your speaker active initially. Measuring and correcting then is still relatively easy. If you are satisfied with the result, you can convert the active filters to passive ones (as I did in my simulation). Just be careful not to use delays or FIRs.
 
  • Like
Reactions: howardg
Check out the 7 dB difference at 200 Hz from the delay of the rear driver due to the 50 cm box depth.
Where does Boxsim put the microphone?
It shows the necessity of describing the context of a measurement (or a simulation?) that the delayed simulation looks a lot like the measurements of the bass range in a room---huge variations due to where the mic is in relation to standing wave modes.

Our hearing/neural processing must be very forgiving to allow us to enjoy reproduced music in domestic rooms. Given this concept, it is even more interesting that a loudspeaker (or pair) that have a relatively even octave/octave response anechoically, sound better for long term listening than similar speakers with uneven response/directivity. Our hearing process/perception must be able to make forgiving adjustments for rooms in a very sophisticated way(very valuable skill/no value judgement) .
 
Last edited:
If you want to make a passive system, I can also imagine the following procedure:
Build your speaker active initially. Measuring and correcting then is still relatively easy. If you are satisfied with the result, you can convert the active filters to passive ones (as I did in my simulation). Just be careful not to use delays or FIRs.
Yes I think that would be the shortest/smartest route to a decent sounding speaker.

Everybody learns differently/uniquely. I learn/understand/enjoy best when I have my hands on the parts and quickly interface the results of circuit changes with hearing/measuring. I think I end up with simpler crossovers than I would if I was using a simulation. That said, it probably takes longer than with the efficient use of simulation.

I sometimes joke that I am not in a hurry to finish a design, because then I will just need to find another project to get into. Obviously, I am mostly retired.
 
Last edited:
Where does Boxsim put the microphone?
It shows the necessity of describing the context of a measurement (or a simulation?) that the delayed simulation looks a lot like the measurements of the bass range in a room---huge variations due to where the mic is in relation to standing wave modes.

Our hearing/neural processing must be very forgiving to allow us to enjoy reproduced music in domestic rooms. Given this concept, it is even more interesting that a loudspeaker (or pair) that have a relatively even octave/octave response anechoically, sound better for long term listening than similar speakers with uneven response/directivity. Our hearing process/perception must be able to make forgiving adjustments for rooms in a very sophisticated way(very valuable skill/no value judgement) .
Hi Howard,

It seems to me that the response is simulated to be that of a microphone located in the very far field in an anechoic chamber with the measured level adjusted to what it would be with the microphone at 1 meter.

I found a description that matched my expectation in a user generated manual here:

http://www.dipolplus.de/downloads/boxsim_manual.pdf

"
2.3 Concept
Boxsim does all simulations for a listener in infinite distance. That means: there is no crossing of driveraxis’ at any point – they are always parallel. Consequently you can’t choose any simulation distance or height. SPL levels are calculated for 1 m distance though. The simulation is based on measured sound pressure, phase and impedance values. Measurements maybe taken on infinite baffle, on a DIN/ISO standard baffle or on the OB under construction. If you don’t have any measurement data, the TSP (Thiel Small Parameters) alone will suffice for a basic simulation.
"

To model the effects of a near field microphone aligned at different height or horizontal positions relative to the baffle ( like straight out from the tweeter) I believe VituixCAD models this detail. The software also allows users to simulate the loudspeaker's performance in a specific room or listening environment, taking into account the effects of room boundaries. At some point you are then sort of transitioning to designing room treatments or correction rather than speaker design.

Using VituixCAD is more complex and requires performing a specific sequence of user initiated calculation steps, with intermediate files generated using specific naming conventions for the results to flow from one calculation to the next. After making a change many of these steps must be repeated in the correct order.

In contrast Boxsim simply performs all calculations producing all of the results anytime any parameter is changed. There are no manual steps that must be performed by the used to produce results.
 
  • Like
Reactions: jaapVL and howardg
I read the paper written by Finke 14 years ago. I found three things particularly interesting:

First: what you have already enplaned on the position of the microphone.

Second: Finke proposed a simulation method for open baffle speakers. His method uses a vented enclosure with a large vent radiating backwards. I did the simulation of an open baffle speaker following his method. The results were practically the same as the results of my simulation with added backward radiating virtual drivers. His method is obviously simpler than mine because you don't have to add additional drivers.

Third, Finke stated that radiation from the back of the cone differs greatly from radiation at the front. This is obvious a downside of the simulation.
In the method I proposed there is perhaps an (partial) escape for this defect: If you measure the spl response of the driver separately at the front and back of the cone, you can insert this data separately for front- and back-driver in the Boxsim simulator. However, I really don’t know if this (partly) solves the problem.