My contribution to the community... a lobing error visualization tool

[RHosch]

I'm in the process of writing a general purpose visualization tool that will help in the analysis of lobing error/comb filtering for various design efforts. It's a relatively simple sheet in Mathcad format.

Thus far, it is functioning fairly well for three driver layouts... with the ability to model TM, MT, TMM, MMT, TMT, and MTM layouts, as well as widely spaced L/C/R (though manual rescaling of output plots seems unavoidable). The current modeling uses point sources. I'm trying to make it as flexible as possible, so to this point I've implemented driver spacing (whether vertical in a single enclosure, or horizontal in a grouping of enclosures), acoustic center offset, phase offset (to account for crossover phase shift, polarity, delay filters, etc.), room temperature, and frequency to be analyzed.

I'd like to implement crossover attenuated relative amplitudes of each driver (useful for vertical lobing investigation of a single enclosure), but can't seem to find a good explanation of the function (i.e., dB(freq)=xxxxxx) for the common crossover types. Most useful would be Linkwitz-Riley and Butterworth of various orders I would imagine. Much of the current worksheet's data entry area is for entering information regarding filter network description for each driver, but alas until I find the transfer functions I will not be able to get that part to work.

I think it could be easily expanded to look at multi-driver line arrays, or greater than 3-way enclosures. I'd like to increase the accuracy by replacing point sources with functions more appropriate for pistonic devices once the size becomes large enough, but have yet to figure a way to do this.

After some more debugging of the super-alpha-pre-beta version (), I'll post the worksheet file here for others to tinker with.

At this point, any suggestions on what you would like to see would be useful. I found in my research that there were various ways of calculating lobing error, or calculating the zero-axis angle, or rules of thumb explaining how close two drivers playing the same frequency should be, but I didn't find any good tool for visualizing what is happening. I wanted to know how much comb filtering increased if I changed a crossover slope from 24 db/oct to 6 db/oct with no other changes. Or how much improved the situation becomes if I move a tweeter 6mm closer to the midrange. Hopefully, this workeet will help me to see this in an intuitive way... and perhaps help some others as well.

[SY]

This sounds very cool (if only I had MathCAD!).

I've used a similar program (from FRD consortium) set up as an Excel spreadsheet, but it's... twitchy. If you can pull this off in a stable format, that would be immensely useful for a lot of us.

[sreten]

IMO two main points to consider :

1) crossovers add large phase shifts as well as amplitude changes.

2) drivers do not add acoustically the same way they add
electrically, see Linkwitz / Riley alignments that take this
into account, and should explain the phenomenon.

sreten.

[Timn8ter]

This sounds very cool (if only I had MathCAD!).

It's not always easy to pay out the money for full versions of software. Now, if this works using MathCAD Explorer as the ML-TQWT worksheets do, it'll really be something special!

[Paul W]

Well, while you're coding, I'd really like a visualization tool that allows us to look at angular lobing. In other words not just horizontal and vertical, but a 3D look at what is going on in the entire hemisphere in front of the speaker. Hope I'm making myself clear!
Paul
4-way

[RHosch]

I'll look into compatibility issues with MathCAD explorer. My hunch, never having used it before, is that it might not handle the 3D plots required to properly display results. But, I'll look into it.

sreten,

For now I have manual phase shifts that can be added to account for crossover contribution. Since the sheet only examines one frequency at a time (for now), at least it is possible to enter the correct data... although cumbersome. If someone can point me to a rigorous mathematical transfer function for the common crossover types (I'm checking my reference material tonight to see if I actually have that information already), then calculating the correct phase shift for a given XO point, slope, type, and specified frequency shouldn't be that difficult.

FRD tools has the equations built in... but, well, they are password protected in the visual basic scripts. I'm guessing that the original AES papers contain the information I'm seeking, but I don't have access to them yet. I'm aware that acoustic summing is quite different from electical summing. The model so far treats the drivers as spherical point sources... i.e, a perfect driver. Acoustic output is modeled as an undamped wave function spreading out from that point source. Multiple drivers can be located in space and their outputs summed.

I plan to incorporate a distance based decay that gives a first order approximation of point source, line source, and horn source driver types. It assumes omnidirectional monopole sources... hopefully dipolar sources can be fanagled in there later. I also hope to include an extraction function that plots the summation along a specified line, so that the data is easier to interpret at, say, 1 meter distance from the zero acoustic center plane.

Other suggestions are most welcomed.

[Bricolo]

Originally posted by RHosch
I'll look into compatibility issues with MathCAD explorer. My hunch, never having used it before, is that it might not handle the 3D plots required to properly display results. But, I'll look into it.

sreten,

For now I have manual phase shifts that can be added to account for crossover contribution. Since the sheet only examines one frequency at a time (for now), at least it is possible to enter the correct data... although cumbersome. If someone can point me to a rigorous mathematical transfer function for the common crossover types (I'm checking my reference material tonight to see if I actually have that information already), then calculating the correct phase shift for a given XO point, slope, type, and specified frequency shouldn't be that difficult.

FRD tools has the equations built in... but, well, they are password protected in the visual basic scripts. I'm guessing that the original AES papers contain the information I'm seeking, but I don't have access to them yet. I'm aware that acoustic summing is quite different from electical summing. The model so far treats the drivers as spherical point sources... i.e, a perfect driver. Acoustic output is modeled as an undamped wave function spreading out from that point source. Multiple drivers can be located in space and their outputs summed.

I plan to incorporate a distance based decay that gives a first order approximation of point source, line source, and horn source driver types. It assumes omnidirectional monopole sources... hopefully dipolar sources can be fanagled in there later. I also hope to include an extraction function that plots the summation along a specified line, so that the data is easier to interpret at, say, 1 meter distance from the zero acoustic center plane.

Other suggestions are most welcomed.

can mathcad deal with complex numbers?
If it can, you could simply use the equation of the filters (like 1/(j.c.w))

[RussianBlue]

Nice one !
This kind of visualization would be very helpful for teaching purposes. Once you got that down you could start thinking about "modeling", in other words, changing parameter and being able to "listen" to the results. This might be very esoteric and involved, but new powerful and cheap DSP processors are available (by TI for instance) that could work for this kind of (close to "realtime") modeling quite well. Off-line simulation should be much less problematic, but still a "bear" in complexity, I imagine. Sounds like this might become your day-time job...
What is your research based on at this point? I imagine you have looked at some of the work on clustering done by the big loudspeaker reinforcement companies such as Renkus-Heinz, JBL etc.?
Cheers

[Bricolo]

Any news about the worksheet, RHosch?

Such a file would be great! Especially since the FRD'sone don't work on my computer

[kelticwizard]

There is a whole thread on that French Excel problem.

Here is another one:

"1425 visitors downloaded the MS Excel Analysis Toolpak Translator 6.8 Utility - Cumulative total incl. All previous versions.

Note this update is introduced to solve an incompatibility with the French SECURITY ENCRYPTION system defined by French governemental regulations.

This edition should be compatible now with all Countries. For more info see the Download Page.

When applicable the incompatibility shows, when opening previous editions of the file, by following error message:
Impossible d'ouvrier ce classeur car il est protege par un mot de passe avec une methode de chiffrement qui n'est pas disponible dans ce pays. "

http://www.acoustics-noise.com/