The little I know about cross overs

[Mongrel714]

I have been reading and studying cross overs for the past few days, I have kearned what they are made up of, what thoses parts do, the different circuits, the orders and how they are made up, and some info that I don't understand, my problem is with the math and formulas used to determine the value of the componenets and theyre relationship with the speaker measurements, my question is, will programs like "Passive cross over designer", or "Speaker workshop" do the math and formulas or do I need to get a math tutor for this??

[Anthony]

Passive Crossover Designer is a fantastic tool as is Speaker Workshop.

I have not used SW to design a crossover, only to model one that I prototyped in PCD.

PCD is powerful and complicated. You need accurate frequency response (frd file) and impedance plot (zma) for each of the drivers you are using.

Be sure to follow the instructions in order. Load the drivers first, then set the type (2 way, 3 way), then set what's in series or parallel, enter offsets, and go.

Once the preliminary setup is done, you need to look at where you are considering crossing over for each section. Then you set a "target". There's a reference volume level, frequency, and slope. Slope is based on order and Q (bessel, butterworth, linkwitz-reilly). PCD will then overlay a curve showing what your target is.

Next, set the crossover frequency in the appropriate box and select which order you want (1st, 2nd, etc) and click the Textbook Values to get a pretty good guess. In some cases, I've seen the "guess" hit the target spot on. most of the time, though, it needs tweaking.

That's where PCD shines. There are little arrows on every component field and you can increment inductance and capacitance values up and down and immediately see the effect on the crossover. You can't do this for resistance, but those are used less. You can add resistance at different legs to change the slope away or steeper, quiet the driver universally, or tame a resonance between driver and crossover. Those are usually trial and error.

With PCD you can also put in things like Zobel circuits to tame rising impedances, notch filter to hide a breakup mode or localized peak, or contouring circuit for baffle step compensation. If it sound intimidating, it is. But the best way is to just find some drivers with the frd file already done for you at Parts Express and play around with it.

Note: you have to modify the PE text files a bit to work with SW or PCD. They have header lines to say what the columns are and the impedance file has the T/S parameters at the end. Edit the files so that all that's left is the frequency/level/phase numbers and rename it appropriately (.frd or .zma).

PCD needs Excel to work properly, SW is standalone. PCD only does more traditional designs, although it is very powerful. SW is more like SPICE and can design just about any circuit and test it -- even ones that don't have a prayer of working as a crossover Actually, it's pretty handy if you want to try a weird crossover design like an elliptical filter.

SW has a broad array of actual test features that I'm just now exploring. PCD is more of a crossover only tool, but it is designed to work with a whole other array of tools from the FRD Consortiium. Be sure to get the Passive Crosover Calculator as well. It is good for finding values for notch filters, padding, and zobel circuits. It's handy to get some values for it, plug into Passive Cross Design and see the results.

Bottom line: you don't need to know the math. PCD and SW will help you out there. However you should play around with the values and you will develop an intuition as to where more capacitance is needed or an inductor, or the best place for a resistor to tame that nasty spike.

The best part is: they're free!

Good luck.

[Mongrel714]

Wow , Anthony, thank you for all the info, you have made this so much easier, I thought PCD was going to be a handy tool, I can't wait to get a grip on it, I've read for hours on the subject and it is amazing what can be accomplished with a few caps and coils, I even had our web master at work give me a crash course on some of the math and formulas today so I'm starting to understand that also, thanks again for the help and I'll be posting more questions soon!

[Anthony]

No problem, glad to help.

PCD also has the ability to model box effects, active equalization, and compensation circuits, but I have not used any of those yet.

Right now I'm both learning the programs and working on some open-baffle home theater speakers.

I settled on a design: BG Neo3PDR tweeter, HiVi B3N midrange, and Dayton 8" reference series woofers. All told I should get good extension below 80 Hz. If all goes well, I may even get down to 60, where I can run them as large, but I won't celebrate yet.

My only complaint about PCD is that it does not have an easy way to set an overall frequency target. The natural tendency is to model for flat at the SPL you want, but that's never the case. For open baffle, you have severe rolloff at some point (depending on baffle size), for box speakers you have baffle step loss at the midrange and room interaction for the bass. So if you designed flat, your real SPLs would actually be all over the place!

FRD Consortium has a box designer, there's also unibox and some others to help. The Edge program allows for modeling some baffle step/loss effects to help you get a handle on things. There are some good writeups of projects from concept to product (for both FRD tools and Speaker Workshop).

Another good resource is diyaudio.com -- some die hards (read: VERY opinionated), but a lot of good information there. I only ask questions there if I'm truly stumped. Otherwise, it's more searching for things people have done before.

Good luck.