"My First 3-way" crossover response re-worked *PIC*

[bpodruzny]

Ok, so I fiddled around with the crossover a bit and was able to get the response in the picture below. It seems a bit better and doesn’t change the OX topography much.

All I added was a series notch filter and shifted the mid low-pass back into the woofer a bit by increasing the cap value.

Let me know what you think ( reply box un-checked ;-) ).

Cheers
-Brian

[hulan]

Much better, but looks to be a little bright. I'm no crossover expert so wait for the others to chime in.

[bpodruzny]

> Much better, but looks to be a little
> bright. I'm no crossover expert so wait for
> the others to chime in.

I listen to a lot of rock and watch a lot of movies. So I tried to add a little more bass in the 100 hz range and I left the top a little hot. I may go back and add an L-Pad later.

Thanks for the input
Cheers
-Brian

[curt_c]

Brian,

You've done a lot of things right. This response would sound very good at low levels, but IMO, at higher levels it will have a little too much boom and sizzle for most. I would try to bring up the mid 3 dB or so, and perhaps try to cross to the tweeter around 2K to put the mid break up modes a little further down in the stop band. -This is assuming the response plots have been measured on the intended baffle, or modified to account for the diffraction effects. If these are anechoic responses, than as Hulan says, you merely need to attenuate the tweeter by about 3 or 4 dB.

C

[bpodruzny]

Thank you for the encouragement. I was hoping to keep as many of the original crossover parts as possible, but it looks like I’ll have to completely re-re-work most of it. That fine with me though, gives me something to do this summer.

Ok, so what I did. Dropped the crossover points to 200 and 2000 (as Curt suggested), changed the notch filter accordingly, and attenuated the tweeter (as Hulan suggested).

The baffle on the boxes are 20 1/4" Wide and 31 7/8” tall. From what I’ve been able to read and model in the various freeware programs the baffle step should be low, somewhere in the 225 hz region?

Thanks Again for all the help
Cheers
-Brian

> Brian,

> You've done a lot of things right. This
> response would sound very good at low
> levels, but IMO, at higher levels it will
> have a little too much boom and sizzle for
> most. I would try to bring up the mid 3 dB
> or so, and perhaps try to cross to the
> tweeter around 2K to put the mid break up
> modes a little further down in the stop
> band. -This is assuming the response plots
> have been measured on the intended baffle,
> or modified to account for the diffraction
> effects. If these are anechoic responses,
> than as Hulan says, you merely need to
> attenuate the tweeter by about 3 or 4 dB.

> C

[dlr]

Baffle step may be more of an issue as I see it, but I'm looking at this info of yours for the first time. First, I'd like to know what the 2-pi sensitivity of the woofer is from the manufacturer's specs and/or curves. That one should be operating almost exclusively in 4-pi space.

In any case, without knowing more detail (and this is predicated on that at the moment) such as the woofer 2-pi sensitivity rating, I'd be more inclined to say that the treble (tweeter) might need to be dropped just a bit. More importantly, that broad bump in the upper bass/lower mid is going to be very prominent.

Working the woofer/mid crossover is, IME, the hardest part of a 3-way, since even with measurements, most of us don't have the resolution desired due to measurement conditions. Expect it to take some iterations of building/listening to get that part right. All aspects require that, but by the most frustrating can be dialing in the W/M crossover.

dlr

[curt_c]

I'm still assuming you are using the PE plots in the PCD (which are anechoic). If this is the case, you will want your response to compensate for baffle step, and the diffraction effects of the baffle. IOW, your modeled response will have to look something like the sample plot below to actually resemble flat in the real world. (9" baffle width shown.) The better option of course is to sum the diffraction effects with the driver responses, then your modeled response will be flat and resemble the measured responses on the baffle.

-I sent you an email of a screen shot of the diffraction effects per the BDS software.

C

[curt_c]

[bpodruzny]