compression driver filter question

[Alien_Shore]

Calling all filter experts.
I'm working on a crossover design for a 2-way using an Altec 802-16G HF compression driver mounted on 511B horn paired with a 816 bass horn. I need a passive filter that will work for four of these 2-way enclosures. After lots of testing and measurements, I settled on a 900 Hz crossover point (seems best for the woofers I'm using). The trick here is that the compression drivers have some variance in their impedance curves (some are a little hotter than others).

One of the HF drivers has Fs at 893 Hz / Zmax 31 ohms, while another (with the same brand-new diaphragm) has Fs at 699 Hz / Zmax 36 ohms. With this variance, and both fairly near to my XO point, I felt like I really need a notch/contour filter that will accommodate these variances across the drivers. After reading a few different design books and web sites, I'm seeing different opinions on which type of resonance filter works best.

Some advocate a parallel filter (A), while others recommend a series filter (B). I've designed and prototyped the parallel filter, in the circuit below, and it sounds pretty good. But before I go to the trouble of putting together the (B) option, I thought I would check here to see if anyone has ideas, advice, suggestions. If I go with option (B), is that the correct placement (i.e., after the XO and L-Pad)? If it is, I'll have to go back and re-sim the XO values, since it will change my impedance curve...

Can anyone offer any guidance here?

[fpitas]

Seems like option B would give you the most constant impedance seen at the filter output, while also driving the filter directly from the amp. Frankly though, if you're willing to juggle values around in the sim, that may be kind of academic.

When you sim both, does either one present a smoother impedance to the amplifier? That's a desirable feature.

[ivica]

Calling all filter experts.
I'm working on a crossover design for a 2-way using an Altec 802-16G HF compression driver mounted on 511B horn paired with a 816 bass horn. I need a passive filter that will work for four of these 2-way enclosures. After lots of testing and measurements, I settled on a 900 Hz crossover point (seems best for the woofers I'm using). The trick here is that the compression drivers have some variance in their impedance curves (some are a little hotter than others).

One of the HF drivers has Fs at 893 Hz / Zmax 31 ohms, while another (with the same brand-new diaphragm) has Fs at 699 Hz / Zmax 36 ohms. With this variance, and both fairly near to my XO point, I felt like I really need a notch/contour filter that will accommodate these variances across the drivers. After reading a few different design books and web sites, I'm seeing different opinions on which type of resonance filter works best.

Some advocate a parallel filter (A), while others recommend a series filter (B). I've designed and prototyped the parallel filter, in the circuit below, and it sounds pretty good. But before I go to the trouble of putting together the (B) option, I thought I would check here to see if anyone has ideas, advice, suggestions. If I go with option (B), is that the correct placement (i.e., after the XO and L-Pad)? If it is, I'll have to go back and re-sim the XO values, since it will change my impedance curve...

Can anyone offer any guidance here?

Hi,

May be I have not understand the problem. I you have used 16-ohm L-pad, and suppose that in the mid L-pad position serial resistor has about 8 ohms, and the 'parallel' has about 16 Ohms (so at the driver res.frq. Z1=32 ohms) , so in the first L-C network would 'see' 8+10.67 = 18.67 ohms,
(as 1/(1/16 +1/32) =10.67) while on the second (z2=36 ohm) that would be 8+11=19 ohms ( as 1/(1/16 +1/36) =11 ohms).
It would be not expected that such small variations between the total impedance would make an influence in the network behaviors much.
I think that first you have to measure the drivers FR responses, to see are they differ too much , and then to try to compensate that with the appropriated network.