diyAudio (
-   Multi-Way (
-   -   Comb Filtering from "mesh" (

pjpoes 12th March 2009 08:02 PM

Comb Filtering from "mesh"
Ok while I have a very basic conceptual understanding of comb filtering in this sense, I would love a better one, so if someone would care to either explain or point me in a good direction, that would be great.

I'm most interested in the type of comb filtering that happens when something is placed in front of a speaker, such as with speaker grill cloth or more specifically an acoustically transparent screen.

I've read in places like AVS that moving the speaker farther away from the screen material reduces the effects of comb filtering. Is this true? If so, how come? If not, why do people believe it? I took measurements of some various materials and found that distance had no effect other than to move the peaks and valley's around. Does that make sense? Is there a relationship between comb filtering effects and frequency relative to the materials' weave or porosity?

Would comb filtering be considered a linear form of distortion? Based on the impulse response windowed to above 1khz and in the first 10ms, I found that delayed signals were coming through. Originally I thought it was reflections from the cabinet, but I suspect it's not. I say that because the speakers are Dr. Geddes Abbey's, with a large foam plug, so I would expect that absorbs some of the reflection (other tests suggested this too) and it's not seen with "more transparent" materials.

Can comb filtering cause non-linear distortion? I suppose that comment is contingent upon my earlier question being, "Yes it's linear distortion."

breez 12th March 2009 09:07 PM

A comb filtering effect is created by a delayed version of the original signal. I have no idea about generic 'acoustically transparent' screens' properties, but there must be a reflection. The sound bounces from the back of the screen towards the loudspeaker and wall and then back again through screen. Distance between the screen and loudspeaker will generate different delays and thus move the peaks and valleys around.

Longer the distance, lower in frequency the combing starts. The peaks/valleys are linearly spaced on the frequency scale and thus it helps to have longer distance because a larger amount of peaks/valleys then fall within the critical bandwidth of ear.

Comb filtering is entirely a linear phenomenon, although it should be noted that linear phenomena can create non-linear effects in hearing. I think this was the gist of Dr. Geddes' investigations on audibility of diffraction and very early reflections.

pjpoes 12th March 2009 09:20 PM

Ok well given what you are saying, it's happening from the signal bouncing off the material, back at the speaker, and then back through the material. The above image was with the material taped directly to the speaker, covering the foam plug of the horn, and measured roughly 1.5 feet from that. Can diffraction around the pores and solids of the material cause this, without actually bouncing off the speaker or wall, since that was largely eliminated from the equation, no?

As for the distance, I'm still a bit confused. Adding more distance causes the comb filtering to start lower? Ok, so doesn't that just move the comb filtering further down into the audible range, if it was concentrated at between 15k and 20k where it's probably benign, then moving it back would move it lower, into an area where it's not benign. Or am I mis-interpreting what you are saying.

breez 12th March 2009 10:10 PM

The peaks look like to be about ~800 Hz apart. This would correspond to a distance of 344 / 800 = ~43 cm. How deep is the horn? If it is around 20 cm deep it is then probably a reflection from the screen back towards the deep end of the horn and back again. Even if the screen is touching the foam a reflection will be produced.

About the distances and combing starting lower. When you have the frequency axis on a logarithmic scale the peaks/valleys are getting cramped together as the frequency gets higher. Now, for example 1/6 octave adjacent bands in the high frequencies contain similar amount energy while at lower frequencies there will be large variations. If the distance is greater the spacing between the peaks will be smaller and the peaks/valleys will more cramped together. More of the audible range will be in the 'cramped together' range.

pjpoes 12th March 2009 10:32 PM

Don't know how deep it is, but 8" could be right. If Dr. Geddes chimes in, it's his wave guide so he would know better.

Well I may need to go back to the drawing board to get better measurements then (reflecting real world performance). None the less, it's not seen with an actual acoustically transparent material like speaker cloth or open weave fabric.

How audible are the effects of comb filtering like this?

breez 12th March 2009 10:48 PM

Floyd Toole argues in his book "Sound reproduction" that comb filtering from a single reflection tends not to be audible in reflective spaces because all the other reflections with different delays tend to even out the cancellations and boosts. However, in a case like this with a very small delay there may be other factors than just the spectrum and coloration to be considered as Dr. Geddes' will surely point out if he comes around.

pjpoes 12th March 2009 11:33 PM

shouldn't Floyd's arguement play out with in room measurements at listening position distances. If I measure the speaker, with the material in place, at my listening positions, I should see less of an effect, right? If that is the case, I should give that a shot, get a better idea (assuming it's true).

While I didn't for this set of measurement's, I have taken phase measurements at the listening position. It often is surprisingly flat in this short window between 2-3khz and say 10khz. The comb filtering was, of course, very obvious in the phase, and I would think it would still be there at distance, given that I can measure a flat response in that area.

of course if a cricket farts in my room it will throw that phase measurement off, I tend not to trust them much given how unreliable the measurement is. I could take 500 measurements, changing nothing, and get 500 different results, mind you they are often very close (usually).

well I have set two goals now, though they will have to wait probably close to 3 months at this point (I'm presenting at a series of conferences, defending a proposal, and finishing up wave two of data collection over this time, and I'm so far behind I can't let things like this take up too much of my time. None the less, I need to get a full size sample of the material set up farther from the speaker, with the mic farther back yet to take measurements. Then I need to see if the visual advantage outweighs the acoustic disadvantage of the material (assuming this lessens the effect).

gedlee 13th March 2009 03:34 AM


I'm not sure where this is leading, but I would prefer to see impulse responses for this kintd of thing. That would tell you if its a coherent reflection from the screen or something flanking like diffraction arround the sample.

A reflection off of the screen is most likely going to audible to some extent. Its just like the reflection from the waveguide mouth and creates a standing wave and coloration. Floyds argument about comb filtering is more for a reflection off of a wall, that kind of thing and I think that its different for a direct reflection along the wav path of the direct sound. I look at it as very similar to mouth reflection and I've never found that to be a good thing.

pjpoes 13th March 2009 03:47 AM

I'm trying to figure out how bad the screen materials I have are on sound. I can show some impulse responses.

I was hoping to get a better understanding of the physics which cause this phenomena, how audible it is, etc. I mean, if my measurements are indeed accurate, my impulse responses indeed show problems of some sort, etc. then it would seem that the advertising claims of even the woven screen's are a bit misleading, and, from an acoustic standpoint, a more normal fabric material, such as a bed sheet, is better. I was really hoping the woven fiberglass and pvc materials would offer a better picture with equal sound to something like your bed sheet idea, but thus far, it's been a no go. That is of course, assuming my measurements are being taken in an acceptable way.

gedlee 13th March 2009 02:33 PM

Its basically a simple boundary refelction problem. The greater the reflection the bigger the problem. Just look at the impulse responses with and without and they will tell you all you need to know.

All times are GMT. The time now is 08:47 AM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright 1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2