• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

2.5-way speaker on axis cancellation - why?

pwnz

Member
Joined
Sep 10, 2020
Messages
13
Likes
9
I am currently in the process of activating a pair of old JBL 4622N cinema speakers for home stereo use but I'm facing some strange issues. I got these for almost nothing and I am not yet willing to let 'em go.

index.jpg


These are dual 15 inch woofers + big horn (JBL 2384) units. The original passive design is a classic 2-way with both woofers running in parallel and a crossover at approx. 700 Hz to the horn. This might work well for really large listening distances like you find in a typical theatre but I found very disadvantageous in a home environment, seeing huge vertical cancellations/lobing below 700 Hz.

Now my idea was to go fully active and 2.5-way, meaning the lower woofer will gradually fade out above the lower bass frequencies. The upper one will also do the mid part up to ~ 600 Hz.

So here is what I tried:

First with tons of parametric EQs I linearized all the drivers so they measured flat to at least and octave above / below the crossover. Then I crossed over both woofers at 600 Hz (LR24). The lower one goes through an additional 6 dB/oct first order low pass at 360 Hz and 6 dB attenuation . The upper woofer goes through a corresponding -6 dB shelving low pass so that their output adds to 0 dB. At the crossover frequency the lower woofer is now ~ 10 dB down.

In theory this should look something like this:

crossover.png


Now my problem is that I still see these massive on axis cancellations at ear height a few meters away from the speaker. There is a very wide 7 dB-dip between 100 Hz and 400 Hz. Only if I put some delay on the upper woofer/tweeter this goes away somewhat. Shouldn't there only be 90 degree phase shift between the woofers anyway? I don't see where these enormous phase issues stem from. How is this usually done in practice? I see tons of speakers with multiple bass drivers which should result in massive lobing if my observations are correct.
 

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,227
Likes
17,806
Location
Netherlands
Shouldn't the second woofer be playing much lower to act as a baffle step compensation? 30 inches gives about 152Hz. The -6 dB shelving also doesn't make sense to me. That's what the lower woofer should take care of: compensating the baffle step.
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
2,478
Likes
3,315
Location
Detroit, MI
I agree with @voodooless that I do not understand your proposed filtering. As you mention for a 2.5 way you just roll off your lower woofer early. This is typically done with a 1st order shelving filter on the lower woofer ONLY. A 2.5 way transfer function should look something like the plot below (taken from Zaph's ZRT 2.5 way).

ZRT-2.5way-modeled-transferfunctions.gif


I should also note that a 1st order low pass filter will result in much more phase shift than a 1st order shelving filter. See below for a comparison of a 1st low pass filter @ 360 Hz and a 1st order shelving filter @ 360 Hz. Although they have similar magnitude response in the area of interest the low pass filter has much more phase shift.

Screen Shot 2021-04-07 at 11.05.13 AM.png


The other issue you may be running in to is the spacing between those drivers is quite large. If you are at a relatively close listening position (~2 m) there will be non-trivial relative delays between the drivers which will also cause more phase shift.

Michael
 
OP
P

pwnz

Member
Joined
Sep 10, 2020
Messages
13
Likes
9
Shouldn't the second woofer be playing much lower to act as a baffle step compensation? 30 inches gives about 152Hz. The -6 dB shelving also doesn't make sense to me. That's what the lower woofer should take care of: compensating the baffle step.

That's exactly what I thought, too. Unlike with my smaller speakers I can't measure any meaningful baffle step, though. In my yard I can see a slow 2 dB rise from 100 Hz to 300 Hz but in my listening room the woofer measures flat without any BS compensation! Probably because of adjacent boundaries that act as a virtual baffle expansion at these low frequencies.

I should also note that a 1st order low pass filter will result in much more phase shift than a 1st order shelving filter. See below for a comparison of a 1st low pass filter @ 360 Hz and a 1st order shelving filter @ 360 Hz. Although they have similar magnitude response in the area of interest the low pass filter has much more phase shift.

View attachment 122556

The other issue you may be running in to is the spacing between those drivers is quite large. If you are at a relatively close listening position (~2 m) there will be non-trivial relative delays between the drivers which will also cause more phase shift.

Michael

Thanks, that phase response difference between shelving filter and low pass could explain the issues I'm seeing.

So what I'm wondering now is how one would realize a 2.5-way in an infinite baffle or wall mount speaker where there is no baffle step. Just shelf both of the woofers in opposite directions? Seems so trivial ;-)
 

Zvu

Addicted to Fun and Learning
Joined
May 1, 2020
Messages
830
Likes
1,416
Location
Serbia
You have to make measurements with at least 5ms usable gate.

That famous baffle step frequency equation says that frequency response is -3dB at 150Hz for 76cm wide front baffle. Given that wavelength of 150Hz is huge, loudspeaker behaves as if it is 2Pi as soon as it gets any close to the back wall.

The way you eq-ed them, they just overlap too much. This is what JBL did with DD67000

Screenshot_2021-04-07-22-34-14-081_cn.wps.moffice_eng.jpg
 
Last edited:
OP
P

pwnz

Member
Joined
Sep 10, 2020
Messages
13
Likes
9
I managed to prop one speaker up on the kids' playhouse in my backyard but even with a working 6 ms gate the interesting measurements below 200 Hz are, let's say, inconclusive.

In my room, box about 2 ft away from the rear and side wall and averaging out room modes / SBIR, it looks like there is no baffle step at all. So it seems "baffle step compensating" 2.5 way wouldn't really work here. Might have to low shelf-EQ my own baffle step to incorporate the lower woofer. Or maybe I'll disconnect it entirely. That spares me some hassle :D

The DD67000 approach is interesting. Looks like the lower woofer just goes through a first order low pass at ~120 Hz and that's it. With my much lower crossover point the lower woofer would play quite a bit louder above the steep Xover than the actual midwoofer, including all the nasty resonances. Do you think that would matter? Simulation looks like this. Ignore the bass rolloff below 200 Hz. There's no data there so VituixCAD just extrapolates.

spl.png



I'm strongly leaning towards scrapping the lower woofer. I use a subwoofer array anyway that I cross over at 80 Hz so this might be an entirely pointless exercise. That lower woofer would probably be high passed already before it contributed anything meaningful to the overall bass output.
 

Zvu

Addicted to Fun and Learning
Joined
May 1, 2020
Messages
830
Likes
1,416
Location
Serbia
Use TS and simulated response bellow 200Hz and merge it to farfield. In JBL they use serial coil and serial notch to further shape the response.

I'd put upper woofer in compression and leave lower ported. Use natural slope for upper woofer, and i'd adjust the lower woofer to complement the upper.

Or turn the bass cabinet on its side, make it a clear two way and see/measure what happens. For me this is the fun part of hobby.
 
Last edited:

voodooless

Grand Contributor
Forum Donor
Joined
Jun 16, 2020
Messages
10,227
Likes
17,806
Location
Netherlands
I managed to prop one speaker up on the kids' playhouse in my backyard

I'd love to see some pics of that :eek:

The DD67000 approach is interesting. Looks like the lower woofer just goes through a first order low pass at ~120 Hz and that's it.

Well yes, it's a bit wider than your speaker, so the 120 Hz should just about equate to the baffle step frequency. So it is the baffle step compensation. Since your measurements below 200 Hz are inconclusive you really can't tell if you have a baffle step or not.


Why is the horn so loud? And why does the upper woofer have so much peaking Looks like a high-Q lowpass filter? Or is the dropoff towards the low end just so high? I would definitely filter away the crap beyond 1 kHz on the lower woofer.

I'd probably try the following:
D1: EQ as flat as possible (you will need to compensate for the horn response), LR24 at 700 Hz, and lower gain to suit.
D2 + D3 LR24: lowpass at 650 Hz
D3: additional 1st order low pass at 150hz.

What EQ do you have on the horn? It's fairly flat it seems:
2384 polar.jpg

A shallow shelving filter or a wide PEQ at 1500 Hz should take care of most of it I guess. Your measurement shows a sharp dropoff after 10 Khz? Why is that?
 
Last edited:

puppet

Senior Member
Joined
Dec 23, 2020
Messages
446
Likes
284
Your HP/LP filter response looks more like a Butterworth24 than a LR24.
 
OP
P

pwnz

Member
Joined
Sep 10, 2020
Messages
13
Likes
9
Use TS and simulated response bellow 200Hz and merge it to farfield.

Seems reasonable, thanks.

For me this is the fun part of hobby.

It's all fun and games until the darn box weighs 200 pounds and is the size of a refrigerator :D

Why is the horn so loud? And why does the upper woofer have so much peaking Looks like a high-Q lowpass filter? Or is the dropoff towards the low end just so high? I would definitely filter away the crap beyond 1 kHz on the lower woofer.

That was just a quick simulation where I wanted to show what the overlap of the woofers would look like. Didn't pay much attention to the horn or overall frequency response. If you like I can post individual gated measurements of all drivers in the box under horizontal angles from 0-90. All at a height between upper woofer and horn throat from 1 m away.

Here's what the upper woofer looke like on axis. I agree on removing the resonances for the lower one as well.
woofer.png


I'd probably try the following:
D1: EQ as flat as possible (you will need to compensate for the horn response), LR24 at 700 Hz, and lower gain to suit.
D2 + D3 LR24: lowpass at 650 Hz
D3: additional 1st order low pass at 150hz.

Will give this a try. Thanks to active technology (yay) it's just a few clicks. What a horror it would be to design a passive crossover for this thing :D

What EQ do you have on the horn? It's fairly flat it seems:

Mainly just a low Q PEQ at around 2000 Hz as you suggested.

Here's a measurement of the horn before and after EQ + LR24 high pass:

horn_1.png

horn_2.png


A shallow shelving filter or a wide PEQ at 1500 Hz should take care of most of it I guess. Your measurement shows a sharp dropoff after 10 Khz? Why is that?

I don't know which driver was used for the measurement you posted but my old 4622Ns still have the ancient JBL 2430 with a heavier aluminium diaphgram and lower resonance frequency compared to the newer 2432. Off-axis the drop isn't quite as bad. I might try different drivers in the future.
 
Top Bottom