question on baffle size

[augerpro]

How do you calculate what baffle diffraction effects will be excluded due the gate on an MLS signal?

In my file "Baffle Effects" I modeled my measurement baffle. Not so good. I'll be using a 6.1 ms gate for farfield, since I've got a bit a of room from the floor to ceiling in my garage and I want good resolution. Nearfield has a much longer gate but the point of nearfield is to "drown out" those kind of baffle effects correct? So that shouldn't be a problem.

Also I modeled another baffle, "Baffle Effects_woofer" This had the smoothest response I could come up with. So worst comes to worse this simulation can at least give me an idea what peaks and dips are not really in the drivers true FR.

So is this a problem or am I worrying about nothing? Now that I think about, maybe there is way to subtract this effect out from a driver's FR in SE....hmmm


EDIT: Files to big. Give me a bit to crop them.

[cjd]

Diffraction is almost exclusively something to worry about in the tweeter and upper midrange area. Woofers can almost always be centered and you'll never know the difference.

Somewhere out there is a simple formula for figuring how low your measurements are accurate based on gating, and I think you'll be surprised how low 6.1ms will be good. If you're doing a 3-way you'll run into trouble almost for sure, but for a 2-way you've got plenty of room. T/S modeling should be a more accurate representation of the low end anyhow, realistically. At least, good 'nuff.

C

[augerpro]

Here you go:

[augerpro]

Diffraction is almost exclusively something to worry about in the tweeter and upper midrange area. Woofers can almost always be centered and you'll never know the difference.

Somewhere out there is a simple formula for figuring how low your measurements are accurate based on gating, and I think you'll be surprised how low 6.1ms will be good. If you're doing a 3-way you'll run into trouble almost for sure, but for a 2-way you've got plenty of room. T/S modeling should be a more accurate representation of the low end anyhow, realistically. At least, good 'nuff.

C

Thanks Chris. And just to clarify this is not measurement for modelling the XO of an actual speaker, where you want the baffle effects included for accuracy. This just for measuring different drivers for comparison ala Zaph. In fact he uses a similar baffle I believe.

As you can see it's the woofer response that has the greatest swings though. If it was confined to less then 400 Hz, I wouldn't worry. Then I'd just splice the nearfield at 400 Hz, as I don't think it will be effected much by the baffle. But the effects are pronounced too far up the FR.

I believe 6.1 is good to about 168 Hz. Which is also the resolution of an MLS signal, so i wanted to get it as low as possible.

[cjd]

Ahh, I begin to understand. It's still frequency related.

And, as long as it's for *comparison* you'll find similar trends in ALL drivers measured on any given baffle, which are baffle induced. And you'll find trends NOT consistent, which are probably driver induced.

Best results are from a pretty crazy big baffle, which pushes diffraction anomalies from the baffle pretty far down in frequency.

C

[augerpro]

Ahh, I begin to understand. It's still frequency related.

And, as long as it's for *comparison* you'll find similar trends in ALL drivers measured on any given baffle, which are baffle induced. And you'll find trends NOT consistent, which are probably driver induced.

Best results are from a pretty crazy big baffle, which pushes diffraction anomalies from the baffle pretty far down in frequency.

C

I was hoping to either prevent the effects from being measured or have SE subract them out. If I put the results up a website I don't want to have a disclamer saying "look at my baffle effects image and mentally subract them out of the driver FR" ya know what I mean? I'll have to dig into the SE manual again, I've looked over it so much some things are just not sticking to my brain anymore 8O

[cjd]

You can model a baffle in the Edge or the FRD's BDS and then do a negative sum (FRD's FRC works really well for that).

C