Thanks motokok, this is some very good information. I downloaded edge and was able to do some quick modeling of very simplistic baffles, which brought up several questions. 1) it appears all you can model are 2D baffle designs. There are a few 3D designs I'd like to look at. For example, the F117 inspired design would have 3D Facets positioned around a circular driver that stands off the baffle a bit. Each facet would have height, length, and width. I don't see how that could be modeled. Another example would be a driver mounted in a bezel causing it to stand off the baffle an inch or so. How does that compare to a driver flange mounted in a recess (Most common design) or one that stands off the baffle by the thickness of the driver flange (That's the way my Thiel drivers are mounted)? 2) Is the compensation circuit real? Even the default baffle on a new file shows an inductor value in the 1000s of mH. Seems a bit unrealistic. 3) Once modeled, what do you do with it? I know you can save it as and edge file, but I couldn't find any other ways of using it. What am I missing? Anyway, thanks again for all the info.
Thanks JReave, this is exactly what I was looking for. Now I'll just have to learn to manipulate the crossover to achieve results like these and I'll be golden.
More great info. I'll look into the VituixCAD. I like the looks of Avalon speakers, but am not planning to build anything like them. But even if that's not the design, understanding how to model the facets of that design would be helpful. Although, I'm more inclined to ignore the upfront modeling and skip to build/testing. While risky from a waste of time/money standpoint, seems that would be the best way to ensure the real baffle effects are included in my ultimate crossover design (I'm assuming the box impedance manipulation can be done after testing with new FRD/ZMA data - right?)
- how to model complex 3D baffles? I don't know; not sure any hobbyist software can do this; hence, just build and measure - compare responses vs a rectangular test baffle; probably the easiest and only way to determine viability - maybe you can just buy dirt cheap drivers of similar diameters for this purpose (e.g. Parts Express buyouts or ebay fare - $10 or so)
- something to be aware of, e.g. on a tweeter with high frequencies (shorter wavelengths than the woofer) - every single nearby edge will diffract the sound to some degree - this is why complex, many angled baffle shapes are thought to be undesirable, or rather, avoided cause they are difficult to predict - empirical results will close debate
- I don't know what was done in the example. BSC circuits follow a distinct formula, but they can also be omitted, and similar results obtained with a bigger series inductor in the bass low-pass filter to still match a target system SPL and roll-off
http://www.quarter-wave.com/General/BSC_Sizing.pdf
- yeah I'm not sure how to export The Edge's things into XSim. Try and find Jeff Babgy's Diffraction Modeler excel file - this lets you export the response (as db loss) into say Vituixcad. Then you can use the calculator tool in VCad to multiply the diffraction curve with the datasheet curve to get a proxy.
--
All said, it's so much faffing around!
I'd just consider getting measurement gear, build some test baffles with dirt cheap proxy drivers 8", 5", 1" tweeter.
Your critical path is not becoming a master of XSim and passive crossovers, aligning phase and modeling baffle step loss; it's getting measurements of your complex cabinet design to determine viability.
Last edited:
Yes you are right. Either use a higher Q sealed box or higher tuned vented to create a peak at the cost of possible F3/6/10 extension, which some might prefer depending on listening environment and low end support (subwoofers).
I tend to like my bass (my equal loudness curve) and find it lacking and music less satisfying when missing.
I'm trying to do away with my current in corner subwoofer and just rely on my 3 ways (15" RSS390 being used)
Faceted baffles work. They reduce diffraction and improve off-axis consistency. For a simple design you can copy DXT-monitor:
https://heissmann-acoustics.de/en/dxt-mon-182/
Complex shapes benefit from BEM modeling.
You need AKABAK/ABEC or Comsol for 3D acoustic baffle simulations. Quite a high bar for novice, leave it alone for now.
Also don't bother Frequency Response Modeler 3.0, it's an awfully outdated tool, and exporting-importing results to Xsim is bothersome.
This thread is kinda heading towards clusterf-🤐 complexity bomb with an option of being stuck in analysis paralysis forever. I see two options for you:
A. Just build your speakers using linear response modeled in Xsim and using robust best practices: rounded or faceted baffle edges, 4" mid, 400 and 3000 Hz XO frequencies, L-pads for tonality adjustments. You won't be intimidated by the complex software and have the fruit of your work now, not in the indeterminate future. Most likely they will sound good, and you can always improve or replace crossover later.
B. Dive into theory and software. Learn VituixCAD, read Toole's Loudspeakers and Rooms, take a lot of mesurements, model your drivers, enclosure, room reflections. VituixCAD allows to model a speaker to such accuracy that real-world build will mimic digital model to a 1-2 dB, even closer for most of the range. If go this way, you'll likely end with a great speakers on par with, say, Revel F206, but the road is long, life might get in the way or you'll get burn out, without any speakers. Well, with your knowledge you'll be able to choose a good non-DIY system as a consolation prize.
Another option is to invest in an DSP active crossover setup. that way you can tweak to your hearts content, do A/B comparisons much quicker than with a passive crossover setup, get going quicker and not have to sweat "perfection" before ordering (passive crossover parts I mean).
DSP does not shortcut good measurement and design methodology, but it provides some learning flexibility.
DSP does not shortcut good measurement and design methodology, but it provides some learning flexibility.
Also, Mofi's faceted design -- note that it's still basically a box. I don't know how closely this matches the mention of drivers "standing off the baffle"
Heissmann also has a short article on baffle designs that may be of interest
Heissmann also has a short article on baffle designs that may be of interest
Totally agree. That's the path I'm going to do. That said, my next step is to update a few things on my enclosure/baffle design based on a few nuggets I got from the folks on this site and then build one (Maybe both). Test, test and do some more testing. That should be fun. Thanks for all your help.
It's a good thing to use large roundovers but not so easy to find a simulation for them, measuring is preferred anyway. I would be wary of models that round the sides without doing the same to the top or any place that requires it because the result can be misleading.
Last edited:
The stealth bomber tries to limit angular coverage from spectral reflections. Not at all the same as a baffle.
The main benefit of facets as used here is that they're easy to cut. Roundovers are the ideal in this situation.. However there is a special case use of facets called a 'secondary flare'.
I'm a fan of building one to begin with. I rarely get everything right the first time. And to me the first prototype is basically sacrificial, fast, and ugly. It's something to measure, adjust, and listen to to decide if I like it enough to really make it. The more time/money in a prototype, the less likely you are to chuck it or seriously hack on it when it has problems.