2. speakers
  3. 136577-20253

Comb Filtering - Quick way to Tell

Hi Everyone,
Lots going on here, with this odd Golden Ear vs. Tekton debates. I personally don’t care, except there’s a lot of theory being bandied about as proof of the superiority or inferiority of a design. The arm-chair experts are really giving me a headache because my eyes can only roll back so much before my optic nerve gets kinked.

I have no interest in either brand, really, but let’s talk about it in practical terms.

Comb filtering happens when you have simultaneous arrivals of the same signal which are slightly out of time alignment. Really, any multi-way speaker has to deal with this, but arrays in particular. Much has been written and debated, but let’s boil it down to this:
Comb filtering affects what your ears hear in a particular location in space. The frequency response at location A can change rapidly a few inches away at B. In fact, we can use this in a process called interferometry to get very precise measurements of acoustic offsets.
The point?

If you are concerned about an array having comb filtering, move your head. That’s it. If it is a problem, small changes in your listening location will cause significant changes in the frequency response. Stand up, sit down. Is the frequency response changing much? is it pleasant or uncomfortable/weird sounding? Do this for any speaker.  Listen at the center and both ends of the couch.

But please, listen realistically! Listening 1’ away to an array may yield vastly different results than at 9’ so it is important to understand the manufacturer’s design and intention, which is really true for all speakers.
Which speaker should you buy? The one you like the most. For heavens sake, listen to them and look at them.

We should also point out, as I alluded to elsewhere, there are lots of interesting ways to handle potential issues with comb filtering, so the idea that you can measure center to center differences, and diaphragm diameter alone will tell you what the speaker will sound like at the listening location is simply not enough information. Your head and ears however can do an excellent job of discerning these issues.
erik_squires
10 responses Add your response
erik_squires OP
16,056 posts
 
Another way to tell about the quality of the speaker in terms of dispersion and combing is the off-axis plots.

I haven’t seen one for the Tekton Double IMpact, but for the Enzo, which uses a much simpler array, the vertical and horizontal plots look really good:
https://www.stereophile.com/content/tekton-design-enzo-xl-loudspeaker-measurements

Based on that, I’d say the designer achieved the goal of wide horizontal and tight vertical dispersion, without evidence of comb filtering. I would expect a nice wide sweet spot and very usable horizontal listening location with enhanced detail thanks to limited reflections from the ceiling.
None of which means you should buy it, but that the issues being discussed regarding Tektons and comb filtering were not shown in these measurements.

Again though, your ears are going to be just as good at hearing those issues as measuring devices, and I have no particular interest in whether or not you like them or not. Just want to remove a lot of techno clutter from the debates.

For heaven's sake, listen! Charts, theory and calibrated microphones don't enjoy music, nor do they save, budget and work hard to have the money to afford our dream speakers.

All I want to do is remove the FUD, and let you know with confidence that you can tell if any of this debate is a problem for you or not.
shadorne's avatar
Ag insider logo xs@2x
shadorne
7,477 posts
 
Comb filtering really sounds bad. It is best avoided.

i think the Enzo only uses 1 of those tweeters in the top octave.

example
https://youtu.be/4qgmY8jIGi0



erik_squires OP
16,056 posts
 
@shadorne

Yep, agreed to all of that. The crossover has a great deal to do with the results. 

Best,

Erik
Post removed 
erik_squires OP
16,056 posts
 
. If you look at the horizontal response plot in the link, that is exactly where they begin to decouple and soon after their off axis response plummets

Except with a vertical array, horizontal "decopuling" as you call it (not a term I’ve ever heard in speaker design) is exactly what would never happen even in the most poorly designed arrays. What you are seeing is the normal off-axis drop off of the tweeter, combined with the unusually large mid-woofers (see JA’s comments in the measurement texts). We usually refer to this as lobing, and is a type of comb filtering inherent in almost all multi-way speakers. A combination of the woofer’s output, crossover and rapidly changing acoustic offsets with angle. We can argue about who does it better, and whether speaker A vs. B’s coverage sounds better, but to call this a fault of the line array itself is simply not accurate.

A better place to see this line-array comb filtering is the vertical plots, and comparing it to how a similar 2 way would function. In this sense the Enzo’s array offers superior coverage below axis, and, like many other speakers, above gets wonky.

As I try to mention, comb filtering of the type you are complaining about happens very very quickly, in a matter of a few inches, and goes back and forth. That is, as you move the microphone the nulls and valleys shift rapidly. This smooth drop off across the horizon is anything except that. It does, overall, suggest a narrow dispersion which in many rooms will be a good thing.

But based on the charts, I don’t see any of the types of array / comb filtering mistakes that have been discussed for it.


Best,

E
timlub
2,039 posts
 
Nice thread Erik and good, simple explanation for Comb Filtering.  
Way back in my Marcof/SpeakCraft days, we built a tower with a 15" in a 2way and 3way version and 4 Audax HD's on top.  Comb filtering was minimal and this was a fairly smooth and refined speaker. 
On the tweeters, we tried different variations of series parallel wiring as well as taking specific drivers out of phase and changing the slopes on these tweeters in sets. The end affect was that Comb filtering/lobing effects were minimal. So my experience is closer to Eriks.  
Tim 
Post removed 
erik_squires OP
16,056 posts
 
Kosst,

Comb filtering due to array specific issues will be shown in the vertical, not the horizontal plots. The reason is simple: The acoustics offsets to all tweeters is identical across the horizontal plane.

Across the vertical plane, the acoustic distances to each tweeter is changing. That is where comb filtering would appear.

Everything else I see is due to "normal" 2-way design choices.

However, I go back to listening. Listen to see if the off-axis performance is what you like or not.

@timlub

Thanks for the vote of confidence! 

Tim, sounds like you are discussing something similar to a Bessel Array. One of several ways to deal with real comb issues. Another simple way of course is to use smaller number of tweeters as the frequency goes up, which is what Tekton seems to do.


Best,

Erik
Post removed 
kalali's avatar
kalali
1,977 posts
 
"I don't think so."

Everything I've read about speaker design/acoustics seems to agree with Erik. Can you point to anything more scientific instead of just your "opinion"? Everyone has an "opinion" on everything and words are free...
Log in to respond
More to discover
80x80
KEF LS50 Meta
80x80
Genesis Advanced Technologies 350SE
80x80
Von Gaylord Audio VG-8 MK II Speakers