Calling Vuki!! 8)

[Seadweller]

Hi Vuki,

I installed your XO design, which you were kind enough to share, in my GPA604 based system, and while it has tamed the shoutiness somewhat, I?m still not out of the woods. I don't know how to describe the sound character overall, other than to say the sound is very "thick" in the midrange frequencies. Basically, instead of hearing individual instruments; vocals, bass, etc., there's just a "clump" of sound, with everything blended together in a single ?mass.? There seems to be excellent detail at the frequency extremes, but the midrange frequencies sit out in space like a ?blob? of sound, which heavily impacts the frequency balance of the system. Midrange detail is very poor, and some instruments, such as a xylophone, for example, will jump out of the mass and still pin your ears back. Also, bass lines completely lack detail.

I believe your bass circuit is a second order design, meaning it will have a 12db slope. As I try to learn this stuff, I did my own research, and found that with an 8ohm driver, the 3.6mh inductor provides a cut-off frequency of right around 500HZ. I?ve seen it mentioned that the capacitor determines the ?knee? of the slope, but don?t know what that means unfortunately. I was under the impression that higher order XO?s determine the slope; for example a 3rd order XO providing a 18db slope, and 4th order a 24db slope.

Here?s where I get totally lost? The charts show that it would take around a 1.6mh inductor to cross at 1,500hz, on an 8ohm driver, so I?ll ask an ignorant question; given that you would double the inductor value to get the same effect at 16ohms, was this XO designed for a 16ohm driver?

If not, then I think I understand where this is going; if the 3.6mh inductor cuts at around 500hz, then that is the beginning of the slope. Therefore, in a second order design, is the response down 12db at 1,000hz, and another 6db at 1,500hz?

Now, I?m assuming that the capacitor somehow impacts the roll-off rate, but can it be higher than 12db in a second order design? What effect would increasing/decreasing the value of the capacitor have? Why not just go to a 3rd or 4th order network if a sharper slope is desired?

Sorry for all the questions?.

[donaldpatten]

Seadweler
I'ts not so simple. The woofer impedance curve dips to 7 ohms at 150 Hz and at 1500 is at 12 ohms and rising. So we call it an 8 ohm speaker but in reality it is all over the chart.
This rising Z curve changes the frequency of the knee . The point the roll off starts. The shape of the curve is a property of the filter design. Butterworth filters are popular, but others have their place too. My latest low pass is 10th order Bessel, to get the 467 usec delay for a 605 . The expected rolloff is 6* 10= 60 dB / octave. In the octave around 1600 Hz it s about 12 dB. As a word to you that may measure a crossover frequency curve, do it with the proper speaker attached . If its done with resistors as the load, the resultant curve is meaningless.
Don

[vuki]

Hi seadweller,

I'm sorry to hear that it doesn't sound good. The only thing that I can think of is that perhaps you didn't correctly connect drivers. Have you noticed that hf driver has to be connected with reversed polarity? Wrong polarity can greatly influence the midrange. Anyway - try it with reversed polarity on hf driver.

Regarding to your question about slopes - as Don said - it's not so simple. Except impedance thing there is also frequency response of the driver. 604 (as most bass/midrange drivers) has rising response which crossover has to "flatten". So, the crossover works in combination with impedance curve and frequency response of the driver to get desired response.

Hope this will help you. Please report results.

[Seadweller]

Thanks gentlemen...I had a suspicion it wasn't so straightforward...

I think it's obvious by now that I should stick to my day job, but I'm still quite curious!

Don, with your 10th order Bessel design, I assume it's possible to reduce the response at 1600HZ even further, say 24db rather than 12db?

Vuki, what effect would changing the capacitor value on your second order design have?

I guess the real question is, how do you idenfity the troublesome frequencies, and is it really possible to flatten the response of a 15" woofer that's trying to reproduce midrange frequencies? I have to assume the response of the driver is not a smooth, predictable slope...

[vuki]

Hi seadweller,
first thing to try would be to reverse polarity on the hf drivers.
If it doesn't work for you then something else is wrong and it won't be fixable with minor element value variations. Because flaws that you have described seem pretty serious.
Easiest thing to identify freq. response anomalies would be to buy some microphone and use it with any of the freeware acoustic measurement programs available on the net.
But it is quite strange to me why it sounds as you are describing, because your loudspeakers are quite similar to the ones of my friend (same drivers, similar enclosure volume) and his sound beautiful -for example; much better then Focal Electra 1037 or 2way ScanSpeak Revelator loudspeakers that he had.
In the near future I will go to his place and do some measurements and photos and post them here.

[Seadweller]

Thanks Vuki...

I'll try reversing the connections on the horn to see what effect it has on the sound. I've got things situated so connection changes are simple. Currently, I have the horn circuit connected using the typical positive-to-positive, negative-to-negative terminals, so I didn't know if the polarity was reversed internally by the XO, or if I physically needed to reverse the connection to the driver.

Also, when I built this MLTL cabinet, I was told that 3ohms of series resistance should be applied to the woofer, which I currently have in place. I'm assuming your XO design did not take this into account, so I will also try the system without the series resistance. Is it possible that this could impact the midrange as well?

[vuki]

Try it first without 3 ohm resistor. Then try reversing the tweeter connection.