Correcting gradual roll off

[nzlowie]

Hi guys
I have a Yamaha 6681b CD driver on my LeClech 340hz horn which is meant to be pretty flay across the range but it rolls off gradually across the range.
Is there a way to even out this response in the xover? Currently its a simple 2nd order but I'd like to correct the roll off if possible. Impedance is flat across the operating range.

Thanks for your thoughts.

 

[Sonce]

It is possible to flat the response even with the simple 2nd order filter, but you must sim that. For the first step, change the original capacitor with a smaller one.

 

[nzlowie]

Thanks Sonce, this is a second order now with 22uf and 5mh coil, this probably looks sharper than second order due to the extra roll off induced by the horn.

I just decreased the 22uf cap in XSim and it just raised the xover frequency. (16ohm)
Will try it for real tomorrow.

Cheers

 

[eriksquires]

+1 Sonce.

Could probably fix this with the right values of C and L, but definitely won't be standard values. Simple to tinker something together with XSim, but ...

Don't overcompensate, a gently descending FR at the listening location is ideal.

 

[Borus]

highs will be to soft. No ?

a gently descending FR at the listening location is ideal.

I'm not sure I agree with this, people hear better in the middle. Loader in the middle will overemphasize it and the highs will be to soft. No ? I would love to know you thinking here.

 

[eriksquires]

Hi Borus,

You can always tune a speaker for low volume listening, which is as you say, but for full volume listening this speaker will be too bright and bassy. Here's more for you:

An Acoustic Basis for the Harman Listener Target Curve | InnerFidelity

Best,

Erik

 

[nzlowie]

Thanks guy's, this measurement was taken at close range so should be flat if possible. The LC values were worked out in XSim, maybe horns respond a little different? Will just start trying different values and see where that takes us.

Cheers

 

[nzlowie]

Ok, lots of trials today. Had quite a few xover parts around so I just started trying different combinations and here's what we have.
A lot flatter curve but is it right?
Reduced coil and cap in the second order filter which brought down the lower part of the response then added a 10uf and 8.2R in series across the driver to even out the higher part of the response. Nice but the phase response isn't as flat.
Attached are the FR and phase plots from REW and the phase and impedance from DATS. Green was the original xover and blue / red is what I've come up with.

What do you think?

 

[Juhazi]

Are you sure, that green was before modifications? It looks pretty good!

 

[nzlowie]

Are you sure, that green was before modifications? It looks pretty good!

Yea, the blue trace has an even fall as frequency rises. Dropping to around 7R at around 6khz. On the original xover the green trace dips around the xover frequency then rises to 16R which is the impedance of the driver and is pretty constant. So it looks like my correction has completely change the impedance .
The phase change in the REW sweep looks excessive, is that an issue? The DATS phase doesn't look as bad. Once the frequency goes past the xover range the phase flattens out. Green trace of the original xover. The red trace is the new version and does have a lot larger swing.

Which is the correct phase response to study, REW or DATS?

Cheers

 

[Juhazi]

I meant spl response curves. I don't understand much about phase curves but they look OK to me. Wiggles above 2kHz is either interferences or xo. How is acoustic xo, individual slopes for bass and CD?

REW measurement has 500ms or 1000ms IR gating as default it is way too long to analyze separate ways and phase.

 

[nzlowie]

No, green spl trace was the final incarnation of my trial xover. Yea it now looks good but a flat trace isn't everything. Impedance and phase also have a big impact on sound quality.

 

[nzlowie]

I've had a couple of comments on the phase of this revised xover saying its too large a swing.
Any final thoughts before I order the parts to build it for real?