Speaker Designing

Designing speakers takes skill and a lot of experience, which is why very few people, even in DIY do it.

The idea is not to measure the drivers in open air, you want to measure the drivers in the baffle where they will be operating. Then design the crossover based on the measured data.

SW isn't user friendly at all. There are no easy to use freeware programs that allow gating of the signal.

If you want to do a screen capture of the measurements, ETF 5.0 is more user friendly than SW. The demo doesn't allow saving of data but a screen print gets you a copy of the plot. Nice thing about the ETF demo is that you can get free tech support for it via the tollfree number on the website.
http://www.acoustisoft.com/

He isn't kidding. I'd hate to add up the hours I've spent learning, measuring and building. Its a deep subject that takes a multi-disciplinary knowledge base. You also need tools that have a learning curve and will spend a lot of time making bad speakers before you make good ones.

Some hints to get you on the way:

#1. Measure your drivers in the enclosure you are going to use them in. You should have a basic idea of how they are to be used from Manf. specs so build the enclosure. Do your measurements outdoors, with the enclosure about 8ft off the ground. I built a support for doing this so that I get good clean measurements. You have to start with good clean data to get good repeatable results.

#2. Measure FR for the drivers in the enclosure they are to be used in @ about 2M (more for something like a line array). Make sure your well into the far field of the loudspeaker. Set your levels and take one measurement, then swap to the other driver without moving the mic or changing levels. Take the second FR, if your doing a 3-way take the third driver FR measurement, etc....

#3. Measure the Z in the same set-up but obviously you can take your proto -box off the stand.

I also like to get off-axis measurements for the drivers while its in the same position. That way you can model the off-axis data with the real off-axis measurement. I can swap between measurement data with a given crossover and see how it looks with both the on/off axis measured data (the Z doesn't change ).

Then you build something and re-measure to see how close you are to the model. Then you start the listening process to "voice" it to taste. I re-measure multiple times in the voicing process.

Lots of fun but it takes a ton of time and the better you understand the reasons for the FR & Z the better you are at making good choices in terms of design tradeoffs. The better you understand the underlying reasons for non-linear distortion and what is & is not audible, the better your chance at getting a good design.

^X2

Also, if you want to really get your feet wet, I recommend getting a digital crossover to mess around with, I have the DCX2496 and it is a very powerful piece of hardware.

I really like Praxis & LSPCad. Liberty Instruments is run by great guy.

Speaker Workshop is not all that hard, comparatively. Getting it all set up, gated properly, etc. is only slightly more miserable in Speaker Workshop than some other alternatives. And it doesn't have all the tools you need built in, so it's a little more round-about. Slow, but not impossible.

There are a couple Linux packages that might work (I have yet to get them to compile though I've only put marginal effort into it), and then we get into pay software. Fuzzmeasure for the Mac is great (no crossover work with it though). LSPCad, Soundeasy are probably the next steps up (though LSPCad doesn't measure, you use some other software for that - I forget which at the moment).

C

I have been using Praxis for some time to do all measurements and then use SoundEasy for all my crossover modeling. You can also do all your measurements SoundEasy though. John Kreskovsky has an excellent guide that walks you step by step through the design process using soundeasy for everything. I haven't seen another step by step guide as good for any software.

http://www.musicanddesign.com/Guide.html

john