Was the set up info CARA provided valid?
Having modeled the room with reasonable care, including speaker position/toe-in, furniture, wall hangings, etc. CARA was set to determine the optimum speaker position. What was quickly learned was that in this room it was better to put some constraints on the software. Given too much lattitude the outcome was a "speakers-in-the-face" result. But this is a very complex room with angles, peaked ceiling, etc. In a simple, rectangular room the results were predictable as they were very close to those provided by the Cardas method. Being somewhat limited by the usuable space, CARA was run with those limitations considered. The end results are surprisingly good. After considerable tweaking the speakers are now within 1-2 inches from the positions defined by CARA for distance from front, side wall and each other. The model used, BTW, was to optimize for three listening positions side-by-side on a sofa. In this scenario CARA takes an average of the three. The variances may well come from my having made adjustments from the center position only.
Next, the calculation module, CARA Calc, was run to computing the sound fields of the room. The results were quite interesting. When CARA calculates sound fields it is as though virtual test tones ranging from below up to above the audible range (5Hz to 20.48kHz) were being generated within the modeled space. The effects are display in a 3-D animation using color variances to indicate changes in sound pressure in dbs. The animation can be started, stopped, and stepped backwards and forwards via a VCR-style control. This makes it quite easy to visually locate probable problem areas in a room.
Two observations derived from actually listening to my system and room were supported by CARA's sound field calculations. The first, and worst, is a bad bass problem in the left front corner. In CARA this shows up as a large red region. In reality simply putting one's head in the corner quantifies that the problem exists. The other was more telling of CARA's usefulness. A friend noted there was a shifting image in the mid-bass on the right side. The sound field visualization in CARA shows a small peak at exactly that position.
One downside to CARA Calc is the time necessary to run a complex model. My room has 150 visible polygons. This is high and it may be possible with further experience to bring this number down. The maximum reflection for this calculation was set to 5. According to the online documention "The maximum reflection order describes the highest number of reflections from any surface of a sound ray passing from a loudspeaker to the listening place. In a rectangular room, a reflection order of 1 yields 6 possibilities, one reflection for each wall surface. The order 2 yields 18 possibilities. For a realistic result of the sound field calculations a reflection order of at least 10 should be used. Increasing the maximum reflection order increases the calculation time, order 20 may take hours to complete."
Even though my model was run at a low maximum reflection order, the calculations took three days to complete. Thus, this is probably a step best taken when the user is satisfied with the model's completeness unless the number of visible polygons is low. Unfortunately, this limitation (actually a function of the machine as much as the product; mine is an AMD 850 with 384 megs of RAM) also renders CARA less useable for "what if" scenarios such as determining optimum placement of acoustic treatments.
All in all I have found CARA to be an interesting, useful tool. A side benefit is that now, as my system and room set up evolves, I will have a visual record of the changes for comparison.
Two things I would change in CARA are: 1) Add support for inches and feet 2) Make all fields, such as speaker placement, editable so the values can be edited instead of static displays. Dragging icons around works, but for accuracy nothing beats numeric input.
Next up: Long range impressions.
Having modeled the room with reasonable care, including speaker position/toe-in, furniture, wall hangings, etc. CARA was set to determine the optimum speaker position. What was quickly learned was that in this room it was better to put some constraints on the software. Given too much lattitude the outcome was a "speakers-in-the-face" result. But this is a very complex room with angles, peaked ceiling, etc. In a simple, rectangular room the results were predictable as they were very close to those provided by the Cardas method. Being somewhat limited by the usuable space, CARA was run with those limitations considered. The end results are surprisingly good. After considerable tweaking the speakers are now within 1-2 inches from the positions defined by CARA for distance from front, side wall and each other. The model used, BTW, was to optimize for three listening positions side-by-side on a sofa. In this scenario CARA takes an average of the three. The variances may well come from my having made adjustments from the center position only.
Next, the calculation module, CARA Calc, was run to computing the sound fields of the room. The results were quite interesting. When CARA calculates sound fields it is as though virtual test tones ranging from below up to above the audible range (5Hz to 20.48kHz) were being generated within the modeled space. The effects are display in a 3-D animation using color variances to indicate changes in sound pressure in dbs. The animation can be started, stopped, and stepped backwards and forwards via a VCR-style control. This makes it quite easy to visually locate probable problem areas in a room.
Two observations derived from actually listening to my system and room were supported by CARA's sound field calculations. The first, and worst, is a bad bass problem in the left front corner. In CARA this shows up as a large red region. In reality simply putting one's head in the corner quantifies that the problem exists. The other was more telling of CARA's usefulness. A friend noted there was a shifting image in the mid-bass on the right side. The sound field visualization in CARA shows a small peak at exactly that position.
One downside to CARA Calc is the time necessary to run a complex model. My room has 150 visible polygons. This is high and it may be possible with further experience to bring this number down. The maximum reflection for this calculation was set to 5. According to the online documention "The maximum reflection order describes the highest number of reflections from any surface of a sound ray passing from a loudspeaker to the listening place. In a rectangular room, a reflection order of 1 yields 6 possibilities, one reflection for each wall surface. The order 2 yields 18 possibilities. For a realistic result of the sound field calculations a reflection order of at least 10 should be used. Increasing the maximum reflection order increases the calculation time, order 20 may take hours to complete."
Even though my model was run at a low maximum reflection order, the calculations took three days to complete. Thus, this is probably a step best taken when the user is satisfied with the model's completeness unless the number of visible polygons is low. Unfortunately, this limitation (actually a function of the machine as much as the product; mine is an AMD 850 with 384 megs of RAM) also renders CARA less useable for "what if" scenarios such as determining optimum placement of acoustic treatments.
All in all I have found CARA to be an interesting, useful tool. A side benefit is that now, as my system and room set up evolves, I will have a visual record of the changes for comparison.
Two things I would change in CARA are: 1) Add support for inches and feet 2) Make all fields, such as speaker placement, editable so the values can be edited instead of static displays. Dragging icons around works, but for accuracy nothing beats numeric input.
Next up: Long range impressions.