Asymmetrical crossover

[LightBit]

Hi, I have a question about asymmetrical order and frequency crossover.

I have Visaton BG 20 "full-range" driver, but since it is very poor above 10kHz I have decided to add tweeter Visaton TW 6 NG.

BG 20 has bump from 1kHz to 9kHz so I came up with idea of using 6db low pass filter at 1kHz on BG 20 and 12db (LR2) high pass filter at 4.5kHz on TW 6. It works quite well in simulation, but I'm worried it might not work well due to some side effects (phase, ...). BG 20 would definitely need notch filter, if used above 1kHz.

Here is simulation:

An externally hosted image should be here but it was not working when we last tested it.

Would this work?
Do you have any considerations or suggestions about this?

 

[eriksquires]

Hi !

Sure, you can use your low pass filter to also minimize or remove resonances in the woofer.

Your crossover design needs to include the acoustic offsets before you can deal with phase matching properly.

Best,

E

 

[eriksquires]

I should have answered your question a little more.

Assymetrical crossovers are very common, but the details are driven by phase AND amplitude matching. You will need to examine both closely as you design.

There is a thread in the software section for XSim, which includes some designs. You can use them to learn about what is involved, or you can start here:

https://speakermakersjourney.blogspot.com/2016/02/the-lm-1-bookshelf-version.html

Look at the last two links under "Design Documents"

 

[LightBit]

I get similar results in XSim.
So, I have to measure acoustic offset first and include it in simulation (as delay).

What about vertical distance between drivers?
For 1kHz it should be 34.6 cm, for 4.5kHz it should be 7.7 cm.

 

[eriksquires]

Right!

Also make sure the FRD data includes phase, XSim won't create it, and it must be there. Same for your impedance data. So, Distance, FRD/acoustic phase and impedance with electrical phase, all must be in your sim to get accurate results.

There is no absolute requirement of distance. What there is are rules of maximum distance at given frequencies, but these are not set in stone, and often challenged, since they will integrate better with longer listening distances.

Best,

E

 

[eriksquires]

When you have all of this in, reverse the tweeter in your sim and examine the FR. Should create a dip that is deep, wide and symmetrical.

 

[eriksquires]

BTW, I did not mean to say you should only use XSim. Rather, that any simulation you use to design a crossover should include all this data to reach accurate conclusions.

My examples area all in XSim however.

 

[system7]

I don't see why you should use anything but Visaton Boxsim for this.

I had a look at the issues here, and they are fierce.
BG 20 - 8 Ohm
The TW70 seemed a much better idea, BTW.
TW 70 - 8 Ohm

Best idea I saw was a Troels Gravesen project:
Monacor SP200X +

But you're not going to like it. 4th order on First Draft! Lots of impedance correction too. Nothing wrong with that, but hard to build.

 

[LightBit]

I didn't use Boxsim and XSim, because they are not for Linux.

TW 70 only gets to ~15kHz, TW 6 NG gets to 20kHz easily.

Hm, looks really good, but such crossover would be too expensive (for cheap drivers). Why two TW 70? For better power handling, because they are crossed low?

 

[LightBit]

There is no absolute requirement of distance. What there is are rules of maximum distance at given frequencies, but these are not set in stone, and often challenged, since they will integrate better with longer listening distances.

As close as possible would be best than?

 

[LightBit]

Do I understand correcty, this value in Bassbox is acoustic center?
It is 50 mm for BG 20 and 14 mm for TW 6 NG. Values seem a bit deep.

 

[eriksquires]

As close as possible would be best than?

Yes. Exactly.

Some get pretty serious about keeping the center to center distance less than some value based on the crossover point, but at listening distances I feel that matters much less than the angle, which is usually very small.

Your brain/ear/head/body combination is able to discriminate among different angles, which cause comb filtering effects. To me, that's much more important to worry about than "I have a crossover point of 3kHz, therefore my tweeter needs to be less than X cm away from the midrange"

What will matter, without argument, a great deal is the acoustic offsets, because that will create major fr issues and lobing if not handled well.

Here is one link on how to find this:

https://speakermakersjourney.blogspot.com/2016/02/lm-1-testing-driver-distances.html

Best,

E

Best,

E

 

[Juhazi]

When you have all of this in, reverse the tweeter in your sim and examine the FR. Should create a dip that is deep, wide and symmetrical.

This does not apply to asymetrical xos! Just look at step response.
An example attached, Amphion Helium, bass polarity switched. Sorry I don't have separate responses for bass and tweeter

 

[eriksquires]

The shape of the dip has to do with the electro-acoustic phase matching, not just electrical. If this were all op-amps you'd be correct.

I have 2 assymetrical crossover design on my screen, and it pretty much applies. It's not "as perfect" as my symmetrical design, but it is pretty close.

Best,

E

 

[eriksquires]

Compare the phase of the individual drivers on the normal position. I think you'll find they are pretty far away from matching through the crossover.

Again, before you can use the phase charts you _must_ set the acoustic offsets correctly.

Best,


E

 

[eriksquires]

Almost forgot, inverting the tweeter to check phase matching applies only to "normal" crossover designs. There is one kind who I can never recall the name of (Horsh?) that will not meet these requirements.

 

[LightBit]

I measured acoustic offset to be 1.15 in (29 mm). According to Boxsim data it is 36 mm (1.4 in). Difference is probably due to vertical distance (mic was more on woofer's axis).

I simulated some crossover in Boxsim and wanted to double check it in XSim. I have extracted frequency response from Visaton's images to *.frd and added phase with phase extractor. Copied crossover from Boxsim to XSim and imported *.frd and *.zma.
Result was very similar with no offset, but different with offset (no matter 1.15 or 1.4 in). When I selected FRD phase source derived in XSim, it matched with offset (1.15 or 1.4 in), but not without offset.

Somehow frequency response must already contain that information. What SPL at 1m/1w actually means? 1m from speaker or it's acoustical center?

 

[eriksquires]

You measured with the cap in place??

 

[LightBit]

Yes, 5uF as you can see on screenshot.

 

[system7]

50mm back for the BG 20 and 14mm for the TW6NG is right.

Filter I came up with in Boxsim is below. 30L closed box. Whether it sounds any good? No idea. Seems to push the tweeter hard.