Comb Filtering

[werewolf]

So we are wanting the difference between the hypotensuse and the opposite side to find out where the comb will be?

three points form a triangle. May, or may not, be a "right triangle" (there may, or may not, be a "right angle" involved. doesn't matter if there is, or if there isn't).

two points are where the point sources are. third point is the listener.

draw three straight lines to connect the dots. the three lengths of these lines are defined above : D, D1, D2

The distance D does NOT matter ... so, for two point sources, throw away all the "ctc" rules everybody quotes but nobody understands. The distance D1 does not matter. The distance D2 does not matter.

What DOES matter is this : D1 - D2 = DD

When you know DD ... and ONLY when you know DD ... you will know what the comb filter frequency response is.

The SMALLER we make DD, the HIGHER in frequency the first comb null will be. This is a GOOD thing.

[winslow]

So for my car using an array spaced 6" apart and the adjacent side being 36" (that is pretty close from the head to the pillar), we do: a^2+b^2= c^2.

Assuming symmetry of the array and triangle to make this easy...

We get 9+1296= 1305

So hyp-adj= opp

So that gives 1305-1296= 9

So the sqrt is 3...which is what we started out with.

So you get roughly 2250 hertz.

[winslow]

Which getting 2 speakers 6" apart in a straight line is pretty bad if that is the best you can do with 2 2" speakers.

Place those 2 2" speakers next to each other where the difference in distance between the voice coils is 2", then you puch the combing up to 6750 hertz.

[werewolf]

So for my car using an array spaced 6" apart and the adjacent side being 36" (that is pretty close from the head to the pillar), we do: a^2+b^2= c^2.

Assuming symmetry of the array and triangle to make this easy...

We get 9+1296= 1305

So hyp-adj= opp

So that gives 1305-1296= 9

So the sqrt is 3...which is what we started out with.

So you get roughly 4500 hertz.

is there a "right angle" involved? if not, can't really use pythagoras ... well, maybe as an approx. But why complicate things?

This is VERY simple. Measure the distance from your nose to one speaker. Call that D1. Measure the distance from your nose to the other speaker. Call that D2. The only math you need to do is :

DD = D1 - D2

Typically, we would want DD to be as close to zero as possible. This will push the first "bad artifact" of the comb filter (aka the first comb null) HIGHER in frequency.

Note : the speakers could be ten feet apart, and it's still possible for DD to be zero. On the other hand, the speakers could 4 inches apart, and DD could be 3.8 inches.

[werewolf]

Those 2" speakers could be right next to each other, and the combing might be 10 MEGAHERTZ. Could even be 10 GIGAHERTZ, with the speakers two feet apart. OR, with the speakers 2 feet apart, it could be 1.7KILOHERTZ.

The only thing you know, with the speakers "x inches" apart, is the LOWER LIMIT on how bad the comb can be ... because that sets a maximum on DD (simple geometry). But without even knowing how far apart the speakers are, i can tell you that the first comb null can be as high as 247 MEGAHERTZ ... depending on where the listener is positioned (DD).

[winslow]

I just picked it to help make it easier to picture in my head using a common listening distance that could be real in cars...and I was thinking vertical D'Appolito with the tweeter in the center.

Might have been thinking too far a head, but the MTM with say a Whisper and a tweeter is (again) a popular idea on the forums.

[winslow]

If anyone wants to see an analysis of a 2 speaker array (and bessel arrays), here is a link to a paper by Don Keele:

http://www.aes.org/tmpFiles/elib/20110220/6012.pdf

[werewolf]

I just picked it to help make it easier to picture in my head using a common listening distance that could be real in cars...and I was thinking vertical D'Appolito with the tweeter in the center.

Might have been thinking too far a head, but the MTM with say a Whisper and a tweeter is (again) a popular idea on the forums.

no problemo

The MTM is popular, and a good learning tool.

First step : ignore the tweeter (for now). Let's see how far up (in frequency) we can run the midrange drivers.

Let's say the midrange drivers in the MTM array are 4 inches apart from each other. Where's the first "bad artifact" of comb filtering (aka the first comb null)?

a. 6 kHz
b. 13 kHz
c. 190 kHz
d. any or all of the above, depending on DD ... not enough information given

[winslow]

Not enough info.

We don't know how far we are away from the speakers and we don't know the orientation of the speakers in the array. But the later may not apply.

[winslow]

Cable Asylum

My head was hurting after banging it on the desk from reading that.

The orignal poster is serious...Steve Eddy was not in his explanation.