Wierd x-over modeling

[AndrewM]

I've been forever threatning to tweak this old Aria 7 TLR kit that I purchased years ago, I've never been really happy with the sound of them, they sound good, but I think they should sound quite a bit better than they are. The first realization came awhile ago when I discovered that the x-over schematic was wrong and that the tweeter should be out of phase, the fix provided some relief.

So I've been playing around with the FRD tools and speakerworkshop just to learn what's going on, playing with the FRD and ZMA files I come across (from PE mostly) to see what works and what doesn't and why. I still haven't invested in any measuring equipment yet for a whole host of reasons, so instead I decided I would use the SPL Trace program from FRD just to give me a rough idea of what's going on with my setup, now I realize it won't be 100% accurate using manufacture supplied data, and I'm certainly not going to purchase any x-over parts based on those published curves.

But now I'm confused/suprised by some of the results I've seen. I'd heard that Focal published fairly accurate FR and Imp plots for their drivers (one of the reasons why I decided to do the trace thing), but here's what I found;

This is the "stock" x-over that came with the Aria 7 TLR kit, I already fixed the x-over in this schematic (in terms of wiring one driver out of phase), simple and quaint, isn't it?


And using the traced FR and Imp data this is what the x-over and predicted response looks like;


The above looks downright scary (and I've checked, double checked everything, even re-traced the published graphs a couple of times). Could this famous speaker designer design an x-over really be that far off/bad? Could the published graphs be that far off?

Using the traced data I am able to get some nice looking results.

Any thoughts, ideas, tips, tricks?

Andrew

[Saurav]

You have a strange vertical line going through your source circle, which makes me wonder if you have some extra links between components that you can't see because the lines are overlapping on each other. Try moving each component around in the schematic to see if any extra links show up, or start with a clean schematic and build it up carefully. Not saying that there's an error, but it's a possibility, because I haven't seen that line through the circle before.

> Could the published graphs be that far off?

That might be a factor. Not that the published graphs are off, but rather, the drivers behave differently on the speaker's baffle than they do on whatever standard baffle they use for published plots.

What software are you using for the lower two plots?

[AndrewM]

Sorry about that, I fixed the schematic to make it a little clearer. I should also mention that I'm not a newbie when it comes to electrical type "stuff", it's not my day job, but I've been around it all my life (father was an EE for many years), and we managed to build at least 70% of the projects heathkit sold

What software are you using for the lower two plots?

That's FRD's passive x-over designer worksheet (looks a little cleaner than what speaker workshop produces...at least that I could find). The baffle idea is interesting, but would that really account for such a difference (that graph has almost a 15db swing in it)? I also though about BSC in there, but that would be just a little overkill and the rising tweeter response wouldn't make sense either.

[Doug Lockwood]

What are the woffer and tweeter models?
I can try to check it in SounsEasy.

Also, double check the crossover values.

Doug

[AndrewM]

There are two Focal 7k4411's per side along with a Focal TLR tweeter per side. The cabinet is around 9w x 24h x 15d (the 15 is a maybe, the enclosure is about 1.1-1.2cu/ft total).

I'm sure the x-over is modeled properly as I get the same result in both speakerworkshop and the FRD software. In the absense of measured data it's all theory talk anyway, but I'd just never seen such a wierd response before and I can't easily explain the results.

I say that because I don't think I'm hearing what that is showing, a 10db I would think would be pretty audible, that simulation should sound pretty bad, bass heavy and extremely bright. So while I think the speakers sound "off" I don't think they sound that far off (of course I could just be used to the way they sound now as well), so perhaps a combination of things?

Maybe tonight I'll load up some of the measured data from say the Dayton RS 7" and the RS28 tweeter, duplicate one of the many x-overs published out there and see how close the IEC measured information comes to "real" measured data from a project (say the project Darren did for PE), and see if I get strange results with it as well.

Andrew

[cjd]

You should run a Hilbert Transform to get minimum phase. It looks to be modeling with zero phase at the moment...

C

[AndrewM]

Thanks CJD, I think that may have been it, I grabbed a copy of FRD's Minimum Phase Response Modeler, only breezed through it real quick and re-simulated everything and I would say I'm within an error tolerance level of the traced data and the quick-n-dirty phase model. For instance the woofer's response went from a 15db swing to 5-6db swing, the tweeter still looked a bit ragged, but I breezed through that even quicker.

Andrew

[Dennis H]

Don't forget the baffle step. That woofer curve looks like it was done on an infinite baffle.