DiyMobileAudio.com Car Stereo Forum banner

1 - 16 of 16 Posts

·
Registered
Joined
·
1,310 Posts
Discussion Starter #1
Ok, so waveguides are new to me and I'm planning to use one in my new build for a center channel. I will be using two Aura Whispers in a pod similar to the one Patrick Bateman (member not serial killer) made in the "Anyone Used One Tweeter" thread. I have attached a picture below for an example, but the drivers will be rounded over as well once installed.

My question is that if you use traditional center to center spacing, my first null will be at 766hz, which isn't very good. I got that number by using (speed of sound)/(center to center spacing)/4. 13,000/4.5/4=766.

So, I'm wondering if the waveguide changes this. Since each driver is in it's own "waveguide" and uses the center piece of pvc as a shared waveguide, if the ctc spacing is actually less. I would be great if it is true and I could geat the comb filtering pushed up higher and that's what I'm wondering...

Will my comb filter occur at 766hz or at some point higher because of the waveguide?

 

·
Registered
Joined
·
1,310 Posts
Discussion Starter #3 (Edited)
Why? That "waveguide" wouldn't be effective on anything but treble. And it's a horrible design for a "waveguide."
I know you have a great deal more knowledge than me on this subject...so just a bit more help would be appreciated.

The "waveguide" should have a diameter of around 5", left to right, which I figured would help with directivity down to around 2700hz (you're right about treble there). The glass and dash would create the up/down portion of the "waveguide". I thought if directivity started at 2700hz and the center to center spacing was near zero (thanks to the two waveguides "touching" at the center pvc-- hich pushed comb-filtering up to around 3,400hz), I would actually have eliminated most the comb-filtering issues. I figured that would be the case since the comb-filtering would occur AFTER the horn was already creating a forward lobe.

Then I started wondering if that wasn't the case and I would actually have come filtering start MUCH lower since the "waveguide" doesn't help with CTC spacing and the drivers are now 5" or so apart. This is why I created the thread.

If the latter is the case, would I be better to put the drivers as close together as possible and then put the PVC on each side? This would push the comb-filtering up higher and still help with directivity control in the treble correct?

Here is a better picture of what my setup will ultimately look like (courtesy of Patrick Bateman):

 

·
Registered
Joined
·
1,310 Posts
Discussion Starter #4
Another question.

If this setup is truly "bad" as is, is it really any worse than what I would get firing up from the dash off the glass? I measured last night and it looks like the closest I can get is about 4" away. To me, that doesn't make much of a difference from my 5" CTC spacing with this setup as far as comb-filtering goes, but I get no benefit of better directivity in the treble.

Again, I'm early in the process of this build which is why I asked the question, it's a lot easier to fix now than down the road when things are integrated/installed.
 

·
Registered
Joined
·
1,486 Posts
Build it and measure it. I am a big believer in waveguides. Particularly for a center channel. I was always skeptical about the way Patrick did his round overs but didn't have the time to test the effects. I would build a second wave guide with the center piece of PVC removed and body filer smoothing the transition from the MDF to PVC. I think the second waveguide would outperform the current design.
 

·
Registered
Joined
·
1,205 Posts
I think I am going to have to agree that the short rounded over sections will probably not be effective at lower frequencies.

This ESP article seems to fit the problem and provide information towards dealing with it:
Practical DIY Waveguides - Part 1

I will mention, however, that having tried to tune an MS-8 with speakers directly on axis... good luck. That isn't the device's strong suit. It much prefers correcting speakers that are off axis. It has the same types of problems with horns. It is my theory that it is related to the device's inability to detect the reflected response and seperate it from the direct response. The result is that the head-bobbing that we generally use to steer the image during an MS-8 tune becomes ineffective.

I would highly suggest trying a quick tune using just the speakers on the dash to see if this is something you can live with before investing too much time here. The MS-8 prefers strong, full-bandwidth capabilities in its center channel speaker, and I am not certain that you will be pleased with its treatment of the Auras.

-Todd
 

·
Registered
Joined
·
1,310 Posts
Discussion Starter #7
Thanks guys, I appreciate the help. I'll respond accordingly and try and give a bit additional information as well.

Build it and measure it. I would build a second wave guide with the center piece of PVC removed and body filer smoothing the transition from the MDF to PVC. I think the second waveguide would outperform the current design.
Since the current setup seems to be non-ideal, I will probably just scrap it before going any further. I believe I will create a setup that puts the drivers closer together but can still be aimed on-axis (like currently designed) and off-axis (firing into the glass). Essentially making a baffle that has the drivers side by side, with removable "guides" on each side. I will test and measure both of those and see which direction seems to be better. If I like on-axis better, I will add more baffles with a 15-degree slope so I have more of a true waveguide and the transition to the pvc is eliminated. If I like the off-axis better, I will just use the baffle with a basic 1/2" roundover to the recessed drivers (they will be attached to a 1" thick piece of mdf (1/2"+1/2").

This ESP article seems to fit the problem and provide information towards dealing with it:
Practical DIY Waveguides - Part 1

I will mention, however, that having tried to tune an MS-8 with speakers directly on axis... good luck. That isn't the device's strong suit. It much prefers correcting speakers that are off axis. It has the same types of problems with horns. It is my theory that it is related to the device's inability to detect the reflected response and seperate it from the direct response. The result is that the head-bobbing that we generally use to steer the image during an MS-8 tune becomes ineffective.

I would highly suggest trying a quick tune using just the speakers on the dash to see if this is something you can live with before investing too much time here. The MS-8 prefers strong, full-bandwidth capabilities in its center channel speaker, and I am not certain that you will be pleased with its treatment of the Auras.

-Todd
I will check out the provided link when I have the chance...thank you.

As for the MS8, I appreciate you bringing it up because I hadn't thought of that. In my personal car, every speaker in the system is nearly on-axis and that MS8 has done well with it (mids in kicks, tweets in pillars, center in open DIN slot), but I believe that is because I'm not running any of the speakers to the point where they are beaming. If I take what Andy W. has stated correctly, if you play a driver in the range where directivity doesn't matter, it SHOULD be the same on-axis and off-axis and therefore axis doesn't matter. This is why he suggests left and right mids in the doors (or in the kicks off-axis), because this will help give the widest stage for the listener and directivity doesn't matter at that point so long as you cross them low enough.

I do appreciate the advice and it's why I'll approach the next baffle with the ability to test it both ways before moving forward.

To answer a few more questions about the intentions of the build:

-every other driver will be using the glass off the dash (maybe it IS best to do the same with the whispers)
-the whispers will not cover full-range, lows will be covered by a JBL 500GTi, the mids will take over somewhere between 350hz and say 700hz

Here are a few more pictures of the original planned layout. The 4" drivers in the corner will be coaxial/braxial and about 2" off the glass. The center channel midrange in this design is within 1" from forming a straight line across the dash for l-c-r. The center midbass is a bit futher forward, but I felt it was less important that the midrange from an alignment setup.



 

·
Premium Member
Joined
·
6,035 Posts
Yes, a waveguide changes comb filtering.

The easiest way for me to describe what a waveguide is doing is that it simply makes a small piston behave like a larger piston.

For instance, take a 3" driver with extended response, and high pass it at 1500hz. It will sound like a tweeter. Now rotate the driver to the right or to the left. You should notice that the high frequencies disappear.

Now repeat the same experiment with a 3/4" tweeter. When you turn the tweeter, it will basically sound the same.

That's what a waveguide does. It lets you attenuate the off-axis sound.

Of course, it's easy to visualize this with light. The headlight of a car is much more directional than the lamp on an end table, because the car headlight has a lens.

Also, it's not a coincidence that a car headlight uses a small bulb - that's because we need a proper shape to load the lens. Same thing with sound.

You can simulate all of this with the array simulator from FRD consortium. For instance, if you want to simulate a 2" driver in a 3" waveguide, like the "one tweeter thread", just model the speaker as if it was 3" instead of 2".

This applies to all well behaved pistons also. You can make a 1" tweeter behave like a 3" midrange, or make a 3" woofer behave like a 6" woofer. The trick is that the wavefront has to be well behaved. This usually means that the cone needs to be flat, or you need to be using a compression driver. You can get away with a non-ideal piston if it's small. It's not perfect, but the problems go higher and higher in frequency as the piston gets smaller. That's why I used 2" drivers in my waveguide there instead of 3" or 4". If I was willing to pony up the big bucks for the TangBand flat piston woofers I could have used a larger waveguide. (Or just use a compression driver.)
 

·
Registered
Joined
·
1,205 Posts
Thanks for chiming in, Patrick. I always appreciate your well worded input to perplexing problems involving horns. It never fails to teach me something new.

In my head, horns work by bending (bowing) the wavefront. In my mental model, the sound 'drags' along the surface of the horn and 'accelerates' at its center. The end result is a manipulation of the shape and velocity of the wave as it leaves the mouth of the horn. As frequency changes, so does the degree to which the size and shape of the throat controls the degree of 'bend' applied to the sound wave. I relate the size of the horn's mouth to the 1/4 wavelength of the frequency it has an effect on. The radius of the throat relates to the directivity of the wavefront. At a certain point the wavefront may be significantly 'larger' than the horn can effectively control, and from that point downward the horn becomes useless with respect to controlling directivity.

Does this mental model approximate the way sound behaves or am I completely off base?
 

·
Premium Member
Joined
·
6,035 Posts
Thanks for chiming in, Patrick. I always appreciate your well worded input to perplexing problems involving horns. It never fails to teach me something new.

In my head, horns work by bending (bowing) the wavefront. In my mental model, the sound 'drags' along the surface of the horn and 'accelerates' at its center. The end result is a manipulation of the shape and velocity of the wave as it leaves the mouth of the horn. As frequency changes, so does the degree to which the size and shape of the throat controls the degree of 'bend' applied to the sound wave. I relate the size of the horn's mouth to the 1/4 wavelength of the frequency it has an effect on. The radius of the throat relates to the directivity of the wavefront. At a certain point the wavefront may be significantly 'larger' than the horn can effectively control, and from that point downward the horn becomes useless with respect to controlling directivity.

Does this mental model approximate the way sound behaves or am I completely off base?
I think that's description, but please take what I say with a grain of salt. I'm just a dopey computer programmer; guys like Geddes have studied audio their whole life. So they could speak to this authoritatively.

The easiest way to visualize directivity is to play with the simulators that are in hornresp. (Which is free.)

As I see it, the waveguide acts as a lens, and it works down to a frequency equal to the diameter. For instance, 1000hz is 13.5" long, so if you want a waveguide that works to 1khz, it needs to be 13.5" in diameter.

The angle of the walls controls the width of the pattern, simple as that.

If you do the math, you'll notice it doesn't take much of a waveguide to go down to 4khz; less than 3.5" in diameter.

Also, waveguides greatly simplify crossovers. This reason alone may be a great reason to use them. When both of your drivers have the same directivity, crossovers are huuuuugely simplified, because the off axis energy is the same for both. (In other words, when you cross over from a 1" tweeter to a 6.5" woofer, the crossover is a miserable p.i.t.a. because the off-axis energy is completely different than the on-axis energy. That's because the tweeter has more energy off-axis than the woofer does. A waveguide fixes that.

Have you ever tried to tweak the crossover in a car stereo which had a large midbass and a dome tweeter? And you found that you could get it to sound right on SOME songs, but not on others? And the soundstage would wander all over, or simply was all bunched up on the driver's side of the car? A lot of that is due to directivity. The woofer is very directional, due to it's large size. And you're closer on axis to one woofer than the other. Exacerbating that is the fact that the tweeter is NOT directional, because it's so small. Ideally you'd want both to have the same directivity. (It doesn't even have to be narrow; it just has to the same directivity. You may have noticed that guys like Linkwitz have screwed around with both; the Orion is narrow directivity and the Pluto is very wide directivity.)

As a bonus, it also reduces excursion, which makes it easier to get away with simple crossovers.

The more I screw around with waveguides, the more I think that JBL has the right idea. Instead of using the big old-school horns like they did in the 90s, JBL uses a small shallow elliptical oblate spheroidal waveguide, and they match the directivity of the tweeter to the woofer.
 

·
Registered
Joined
·
1,310 Posts
Discussion Starter #11
Yes, a waveguide changes comb filtering.

The easiest way for me to describe what a waveguide is doing is that it simply makes a small piston behave like a larger piston.

For instance, take a 3" driver with extended response, and high pass it at 1500hz. It will sound like a tweeter. Now rotate the driver to the right or to the left. You should notice that the high frequencies disappear.

Now repeat the same experiment with a 3/4" tweeter. When you turn the tweeter, it will basically sound the same.

That's what a waveguide does. It lets you attenuate the off-axis sound.

Of course, it's easy to visualize this with light. The headlight of a car is much more directional than the lamp on an end table, because the car headlight has a lens.

Also, it's not a coincidence that a car headlight uses a small bulb - that's because we need a proper shape to load the lens. Same thing with sound.

You can simulate all of this with the array simulator from FRD consortium. For instance, if you want to simulate a 2" driver in a 3" waveguide, like the "one tweeter thread", just model the speaker as if it was 3" instead of 2".

This applies to all well behaved pistons also. You can make a 1" tweeter behave like a 3" midrange, or make a 3" woofer behave like a 6" woofer. The trick is that the wavefront has to be well behaved. This usually means that the cone needs to be flat, or you need to be using a compression driver. You can get away with a non-ideal piston if it's small. It's not perfect, but the problems go higher and higher in frequency as the piston gets smaller. That's why I used 2" drivers in my waveguide there instead of 3" or 4". If I was willing to pony up the big bucks for the TangBand flat piston woofers I could have used a larger waveguide. (Or just use a compression driver.)
PB,

I'm planning on using the setup above as a center channel with the MS8 for midrange up. After doing a bit more research, I think I'm going to remove the center PVC and put the drivers as close together as possible.

Please let me know if my logic makes sense:

With the spacing I have right now, I should get my first "big" null around 750hz (which is right in the middle of the vocal range and not very good). If I put the drivers together, that null bumps to around 2200hz. This is right in the area most people, including me, make cuts (per the equal loudness contour) and therefore I can cut down to the null I have reducing it's effect.

I can still keep the "waveguide" so my directivity is narrowed above 3khz which rids the setup of many of the nasty reflections it would normally suffer.

I'm also thinking about getting rid of the PVC completely from the face. I think putting the PVC on the side, or using a good gradual roundover with a router, will help with diffraction at the edges of the baffle without hornloading part of the midrange.
 

·
Premium Member
Joined
·
6,035 Posts
PB,

I'm planning on using the setup above as a center channel with the MS8 for midrange up. After doing a bit more research, I think I'm going to remove the center PVC and put the drivers as close together as possible.

Please let me know if my logic makes sense:

With the spacing I have right now, I should get my first "big" null around 750hz (which is right in the middle of the vocal range and not very good). If I put the drivers together, that null bumps to around 2200hz. This is right in the area most people, including me, make cuts (per the equal loudness contour) and therefore I can cut down to the null I have reducing it's effect.

I can still keep the "waveguide" so my directivity is narrowed above 3khz which rids the setup of many of the nasty reflections it would normally suffer.

I'm also thinking about getting rid of the PVC completely from the face. I think putting the PVC on the side, or using a good gradual roundover with a router, will help with diffraction at the edges of the baffle without hornloading part of the midrange.
Actually the trick with the narrow tweeters only works at high frequencies. You wouldn't want your midranges close together. The only exception to this is if you could put them right in front of you, a la an Ambiophonic setup.

Check out the Opsodis paper - it has all the angles and spacing and crossover points.

Also, after listening to an Ambio setup at home for a few months, I went back to 'conventional stereo.' It's not that Ambio doesn't work. In fact, it does. The problem was that there's really no real soundstage on 90% of the recordings out there. There are a few exceptions to the rule, and they can be breathaking. But most of the music out there is basically mono with a few studio tricks thrown in for good measure. And ambio set ups sound really processed if you're not in the sweet spot.

So I just went back to plain ol' stereo.

I'll bet the ambio stuff would be mind boggling for movies or videogames, because they tend to have a lot more going on in the soundstage.
 

·
Moderator
Joined
·
18,618 Posts
awesome thread.

Let me step in and ask a (likely) OT question...
How is it that horns don't cause issues via comb filtering when the horn itself seems like it would create reflections? Is this all in how you align the driver to the throat (ie: driver and throat diameter same size), and do you use absorption in the horn?
 

·
Registered
Joined
·
1,310 Posts
Discussion Starter #14
Actually the trick with the narrow tweeters only works at high frequencies. You wouldn't want your midranges close together. The only exception to this is if you could put them right in front of you, a la an Ambiophonic setup.

Check out the Opsodis paper - it has all the angles and spacing and crossover points.
I think based on this response that we aren't on the same page with my intentions. Basically, I'm using this as a center channel and not left/right. The reason I'm using two drivers is to gain output based on how bad the efficiency is on the Whispers (both by doubling drivers and dropping the impedance).

So I'm essentially trying to create a good array for the lower octaves (why the drivers need to be close together) and a waveguide for the upper ones (to reduce reflections).

If that approach makes sense, I'd appreciate you going back and looking at my last post. I believe that by putting the drivers beside each other, I push the first null up to around 2khz. By putting them in a "waveguide", I narrow the directivity from around 3khz+ which helps reduce nasty reflections.
 

·
Registered
Joined
·
1,310 Posts
Discussion Starter #15
awesome thread.

Let me step in and ask a (likely) OT question...
How is it that horns don't cause issues via comb filtering when the horn itself seems like it would create reflections? Is this all in how you align the driver to the throat (ie: driver and throat diameter same size), and do you use absorption in the horn?
No problem going a little OT man. I think it's a good question and I'm curious about the answer myself.
 

·
Premium Member
Joined
·
6,035 Posts
I think based on this response that we aren't on the same page with my intentions. Basically, I'm using this as a center channel and not left/right. The reason I'm using two drivers is to gain output based on how bad the efficiency is on the Whispers (both by doubling drivers and dropping the impedance).

So I'm essentially trying to create a good array for the lower octaves (why the drivers need to be close together) and a waveguide for the upper ones (to reduce reflections).

If that approach makes sense, I'd appreciate you going back and looking at my last post. I believe that by putting the drivers beside each other, I push the first null up to around 2khz. By putting them in a "waveguide", I narrow the directivity from around 3khz+ which helps reduce nasty reflections.
LOL it might help if I actually read the first post :p

If you want to eliminate the comb filtering, that's a piece of cake. Here's what you do:

Step 1: measure the center to center spacing of your two drivers
Step 2: Divide 13500 by the number from Step 1
Step 3: Run one driver full range. Low pass the other driver at a frequency which is a multiple of the number from step 2. The lower that multiple is, the less comb filtering you'll get.

Here's an example.

You've got a pair of 2" AuraSound Whispers. You space them two inches apart. 13500 divided by 2 is 6750hz. You run one driver full range, and the second driver is low passed at 6750hz.

Simple enough? The filter is there to eliminate the comb filtering. Below 6750hz, the drivers are within one wavelength of each other, so comb filtering isn't (much) of an issue.

But if you want to improve things, you can drop that multiple, to reduce comb filtering even FURTHER. For instance, run one full range, and the other low passed at 3375hz, or even 1688hz.

You can get really wacky with this, and do a 3-way, a 4-way, hell go for the gusto and do a ten-way!




And this is exactly how the Quad ESLs and the Unity horn and the Synergy horn work!

Note that the Quad uses a delay network, because the drivers on the outside are further away from the listener than the drivers in the center of the array. The Unity horn doesn't need that, because it *physically* delays the drivers. (I can explain how, probably off topic for this thread.)

But anyways, as long as the array isn't crazy large, you probably don't even need a delay for the drivers on the edges of the array.

Tom Danley talks about how this stuff works here:

why not the perfect reproduction?

The company I worked for was a NASA contractor, my job was working on sound sources for acoustic levitation. My Boss was an acoustician who had come from England after the war. He had been part of the team at Mullard that developed sonar transducers during the war etc.

Anyway, he was a hifi buff and loved speakers too and after he found I built my own electrostatic speakers, it wasn’t long before he asked me to fix his ESL-63’s. By fix, he meant neutering the spark gap limiters that interrupted his listening.

I took a deep breath and said sure.

These were different than any of my flat panel speakers AND they did a cool thing, it “sounded like” the origin of the sound was right behind the actual speaker.

The confusing part for me was why, with my large panels, the sound was somewhere off in front but this was different and the answer was slowly clear once the grill was removed, THIS was no ordinary speaker.


This speaker produces / radiates a simple spherical segment, a part of a sphere and so your ear hear that as a source somewhere behind the actual driver. Thinking of your clear film experiment, consider that if you were behind the speaker, the sound radiates from your mouth in a forward in all directions, a sphere, spherical radiation.
 
1 - 16 of 16 Posts
Top