An Epiphany on Driver CTC Spacing & Comb Filtering

[Xaborus]

Hey guys, I think I just had an Epiphany on a subject that I've never seen talked about from this standpoint.

So we all know that the CTC spacing for two speaker drivers (say a tweeter and woofer) should be within one wavelength at the highest frequency of interest, in order to reduce comb filtering.

But... what about the whole concept of stereo? Yes, the music signal differs from left/right signal, but what about center panned signals, and how different must the left and right signal be to not comb filter?

Using two speakers that are 8 feet apart over an entire bandwidth of frequencies. Eight feet is equivalent to 141Hz, so nearly all frequencies will be comb filtered from a set of stereo speakers...

But wait a second, is this what "crosstalk" is? That magical blending of left and right signal that causes speakers to have a soundstage which no headphone can achieve.

So the very thing that we try to eliminate by keeping our tweeters, mids, and woofers as close as possible to each other... may it be a secret to increasing the perception of soundstage?

Let's talk the effects on frequency response first. Obviously we all know a maximally flat and smooth frequency response is ideal. We also know that comb filtering causes high Q and high amplitude peaks and nulls in the frequency response. Wait one second though, doesn't the psychoacoustic research show that the higher the Q of a resonance, the less audible it is? Maybe comb filtering isn't as bad as we thought after all...

So I'll be completely honest here guys, this is a bit of a quacks theory. The only way to know is with vast experimentation. I have no money- I can't even afford to spend a small $40 sum on two cheap tweeters to test my theory. I wanted to post this idea in the public domain so that if it is revolutionary- nobody can patent it. I don't have enough money to develop and profit off of the idea even if it is a successful way to perceptually increase soundstage.


So if anyone out there wants to try out a wacky idea and report back your findings, please do! Possibilities include two or more tweeters spaced at relatively long distances compared to X-Over frequency equilivalent wavelength, and obviously large distances between various speaker drivers.

Anyone crazy enough to try this out?

 

[planet10]

So we all know that the CTC spacing for two speaker drivers (say a tweeter and woofer) should be within one wavelength at the highest frequency of interest, in order to reduce comb filtering.

I always think of it as keeping the C-C to ¼ wl at the XO frequency.

But... what about the whole concept of stereo? Yes, the music signal differs from left/right signal, but what about center panned signals, and how different must the left and right signal be to not comb filter?

Placement of speakers is very important if you want a pin-point image.

dave

 

[madisonears]

The primary difference between driver comb filtering and inter-channel filtering is that the two channels were recorded to be separate, whereas we are trying to make two drivers into a single point source. Stereo or multi has already introduced comb filtering as part of the recording process, which our brain easily interprets as "space" IF the reproduction is accurate.

Peace,
Tom E

 

[midrange]

The primary difference between driver comb filtering and inter-channel filtering is that the two channels were recorded to be separate,

But isn't anything that is pan potted usually recorded in one channel mono? It is an interesting thought that the reproduction of that mono sound from 2 separated speakers would create comb filtering.

 

[Patrick Bateman]

At low frequencies, our perception of location is dictated by interaural time differences. IE, if you have two loudspeakers that are eight feet apart, there's going to be comb filtering when you're off axis and below 141Hz. Understand that comb filtering is caused by pathlength differences. If you have two sources that are playing 141hz (8ft long) and you get to a point where the path length difference approaches about 2-3 feet, there will be comb filtering. The comb filtering is caused by the fact that one source is more than one quarter wavelength further away than the other source.

Having said that, if you get out your microphone and you measure the response of a stereo loudspeaker at your chair, you're going to see comb filtering at the high frequencies. The reasons for this is becasue it's hard to get EXACTLY equidistant at high frequencies. For instance, 5000Hz is 6.8cm long. That means that if you're 1.7cm to left or to the right, there's going to be comb filtering at 5000hz. And the problem only gets worse as you go higher in frequency.

A student of Ken Pohlmann did his PHD dissertation on this, and I was inspired by that to do this:

Anyone Tried Using One Tweeter? - Car Audio | DiyMobileAudio.com | Car Stereo Forum

I was pleasantly surprised by how good a single tweeter can sound! It's quite a trip; you would thing that using a single tweeter at high frequency would collapse the stage, but the opposite happened; I found that the stage was wider AND the center was 'filled in.' It's really quite bizarre, but understandable if you think about how we perceive the location of high frequencies.

Keep in mind though, this is only appropriate at very high frequencies. I'd high pass it above 10khz, possibly as low as 5khz. No lower.

 

[GDO]

Stereo with 1 tweeter only.

Stereolith - Expérimentez le son tridimensionnel

 

[Patrick Bateman]

The Hartman paper is different than this; here's how it works:

When you have two loudspeakers and those loudspeakers aren't EXACTLY equidistant, you're going to get comb filtering. The comb filtering is caused by the fact that the two signals do not arrive at EXACTLY the same time.

This isn't a huge problem at 1000Hz. 1000Hz is 34cm long, so a pathlength difference at 1000Hz won't be apparent until the two loudspeakers are 8.5cm apart, or more.

But at 10,000hz, it's a huge issue; at 10,000Hz a pathlength difference of just 8.5 *millimeters* will cause comb filtering!

So the Hartmann paper proposes a unique solution: put that tweeter in the center.

Basically we're taking the lesser of two evils; we're accepting the fact that we can't put our head in a vise. The illusion of a solid center image in two channel listening is caused by the fact that we're equidistant from both loudspeakers. And since it's challenging to be EXACTLY equidistant at high frequencies, we just stick the tweeter in the center. (Note that the high frequencies are omitted from the left and the right speaker of course.)

The thing that's particularly eye-opening about the setup is that it still sounds convincingly "stereo" in fact I think it might be better.

This is different than a typical three channel setup. The reason that it's different is because the lows and the midrange sounds of a three-channel setup are 'pulled' toward the center. This is a known problem with multichannel; it's one of the reasons that the processors for center channels allow you to high pass the center.

 

[scottjoplin]

Very interesting and makes sense. What's the lowest frequency do you think it would be worth experimenting with this? (due to age related hearing loss)

 

[Patrick Bateman]

It really depends on how close you can get to "equidistant"

5000Hz is 6.8cm long, so a pathlength difference of 1.7cm long creates comb filtering. So if you can sit at point where the pathlength difference between the two speakers is 1.7cm or less, than I would try having a single tweeter in the center, and that tweeter would be highpassed above 5000Hz.

If you CAN'T sit at a point where the pathlength difference is 1.7cm or less, than you might consider increasing the bandwidth of that center tweeter.

Note that the left and the right speakers are low-passed at the same frequency that the center speaker is high passed. This is different than "traditional" three channel, where all three speakers are playing full-range.

 

[rayma]

But isn't anything that is pan potted usually recorded in one channel mono?
It is an interesting thought that the reproduction of that mono sound from
2 separated speakers would create comb filtering.

Yes, pan potting a sound is mono. Hold a 1' - 2' square board (or thin pillow) vertically,
with the edge touching your nose, when sitting in the center. You'll then hear much less
crosstalk and comb filtering.

 

[chrisb]

and certainly join a very exclusive club

 

[GDO]

I am personaly ready to "forget about"/sacrifice stereo at usual 2 way xovers freqs, lets say 2khz. At least would try this...🙄

ITD and IID[edit]

Interaural Time Difference (ITD) between left ear (top) and right ear (bottom).
[sound source: 100 ms white noise from right]

Interaural Level Difference (ILD) between left ear (left) and right ear (right).
[sound source: a sweep from right]
From fig.2 we can see that no matter for source B1 or source B2, there will be a propagation delay between two ears, which will generate the ITD. Simultaneously, human head and ears may have shadowing effect on high frequency signals, which will generate IID.

Interaural Time Difference (ITD) Sound from the right side reaches the right ear earlier than the left ear. The auditory system evaluates interaural time differences from: (a) Phase delays at low frequencies and (b) group delays at high frequencies.
Massive experiments demonstrate that ITD relates to the signal frequency f. Suppose the angular position of the acoustic source is θ, the head radius is r and the acoustic velocity is c, the function of ITD is given by:[9] {\displaystyle ITD={\begin{cases}300\times {\text{r}}\times \sin \theta /{\text{c}},&{\text{if }}f\leq {\text{4000Hz }}\\200\times {\text{r}}\times \sin \theta /{\text{c}},&{\text{if }}f>{\text{ 4000Hz}}\end{cases}}} {\displaystyle ITD={\begin{cases}300\times {\text{r}}\times \sin \theta /{\text{c}},&{\text{if }}f\leq {\text{4000Hz }}\\200\times {\text{r}}\times \sin \theta /{\text{c}},&{\text{if }}f>{\text{ 4000Hz}}\end{cases}}}. In above closed form, we assumed that the 0 degree is in the right ahead of the head and counter-clockwise is positive.
Interaural Intensity Difference (IID) or Interaural Level Difference (ILD) Sound from the right side has a higher level at the right ear than at the left ear, because the head shadows the left ear. These level differences are highly frequency dependent and they increase with increasing frequency. Massive theoretical researches demonstrate that IID relates to the signal frequency f and the angular position of the acoustic source θ. The function of IID is given by:[9] {\displaystyle IID=1.0+(f/1000)^{0.8}\times \sin \theta } {\displaystyle IID=1.0+(f/1000)^{0.8}\times \sin \theta }
For frequencies below 1000 Hz, mainly ITDs are evaluated (phase delays), for frequencies above 1500 Hz mainly IIDs are evaluated. Between 1000 Hz and 1500 Hz there is a transition zone, where both mechanisms play a role.

Sound localization - Wikipedia

What i find hard to understand is how soundbars manufacturers insist in selling products with 2 tweeters, and worst again in the case of sound towers...😱

 

[scottjoplin]

You are looking worse than usual....

 

[GDO]

and certainly join a very exclusive club

Ha! +1😛

The one tweeter theory only makes sense (imho) for soundbars, stereoliths, or ambio stuff like that, in fact small convenient stereo ersatz...

Comb filtering and crosstalk are no brainers, as life itself...🙂

 

[GDO]

You are looking worse than usual....

🙄

 

[bentoronto]

There clearly are unalterable obstacles to the complete enjoyment of listening to sine waves.

Music reproduction (and when people are chatting in a room), on the other hand, is sound in constant flux and none of these sine-wave interference textbook diagrams mean a thing.

Hearing, like all the senses, is a mental construction using the available cues (generally, but not always counted according to their usual validity).

B.

 

[phase_accurate]

In case of listening to stereo the comb-filterig problem at higher frequencies is slighty ameliorated by the HRTF.

Regards

Charles

 

[midrange]

Stereo with 1 tweeter only.

I am rather intrigued by this idea. How in practical terms can one try this out? (With bass/midrange one can use a single twin coil driver (eg Monacor sph 135 tc), but how do you do it with a tweeter?)

 

[Patrick Bateman]

Stereo with 1 tweeter only.

I am rather intrigued by this idea. How in practical terms can one try this out? (With bass/midrange one can use a single twin coil driver (eg Monacor sph 135 tc), but how do you do it with a tweeter?)

It's trickier than it looks!

I use the "minidsp 2.1 advanced" plugin

That plugin is designed for a mono subwoofer, but there's nothing stopping you from using it for a mono tweeter

I think I posted some pics of how to do that on the diyma thread that I linked yesterday

 

[GDO]

I am rather intrigued by this idea. How in practical terms can one try this out? (With bass/midrange one can use a single twin coil driver (eg Monacor sph 135 tc), but how do you do it with a tweeter?)

Providing the amp allows it ( not a BTL for example), what's wrong with this or similar? 🙄

My only concern is about W to W distance. Imho, this can only work in the case of a small sound bar about 2 feets wide maximum.

I am on a soundbar project and the sims i made suggest that 2 tweeters in the middle work better than 2 tweeters at the extreme sides giving narrow but better polar response (for a mono or mid signal). As commented don't believe comb filtering being an issue.