3-way with tilted front baffle

Hi,

A while ago I made a discussion about similiar project, but back then I was thinking about OB design. After finding out more about ob speakers and realizing how hard it is to get it work, I decided to postpone that idea and go with sealed/reflex boxes for this project.

The drivers I bought for this project:

ScanSpeak D2010
Seas MCA12RC
Peerless 830868

And this is what I've been planning:

Layout

Or at least this is the latest layout as I've been trying to tweak it for few weeks now. I've tried many different tilt angles and now it's 15 degrees.

At the moment the crossover has lowpass for the woofer, bandpass for the midrange and a lpad and highpass for the tweeter. All 2nd order electrically. The midrange polarity is inverted.

This is how the responses look like:

http://koti.mbnet.fi/twisted-/3way/frequency_responses_01.gif

I made the measurements by myself in a "infinite" baffle (and added near field measurements), then added baffle diffraction to the midrange and woofer. After that I substracted -6d from the woofer and midrange so the levels matched with the tweeter. The woofer will be placed in a reflex box but as the box doesn't exist, the response is lacking the port output. The midrange will be placed in a closed box.
I didn't mount the woofer flange to the measurement baffle, as I thought it would be best for the relative acoustic offsets.

Now the thing I've been using most of my time is to get the phase responses to match at xo points.

This is how the responses looks like without the relative differences in the tilted layout (just for a comparison):

Phase responses
Red is tweeter, green is midrange and pink is woofer.

And here is the system phase responses with relative offsets included:
System phase responses

XO points are 450Hz and 3250Hz.

And here are my questions :

- The 15degree tilt sounds quite a lot to me, is this a reasonable amount of tilt? Has anyone of you made anything similiar and how did it turn out?

- I've been using 300cm for the mic distance in the system frequency simulations as I though this would be pretty close to the real listening distance. Is this the right way to do it? Changing the mic distance obviously has an effect to the relative phase responses.

- The bass output at ~150hz is about 2dB above the midrange level and I've been wondering if it's going to make the sound too boomy. If the sound turns out unsatisfactory, can I low down the woofer with series resistor just like a tweeter or is there any kind of problem with this (or alternatively using l-pad)?

- How does the phase responses look like to you? Whatever I've tried to do, the responses at 450Hz xo point hasn't been significantly improved. Is this good enough?

-Have I done the layout in Z direction correctly? The tweeter is 300cm away from the mic, but the midrange and woofer is more closer. Now I tried to put the woofer to value 0, the midrange to 10,30cm and tweeter to 13,7cm. The relationships are equal to this , but now all of the drivers are more than 300cm away from the mic and the phases look like this:

System Phase responses 2
Again, red is tweeter, green is midrange and pink is woofer.

Which is the correct way? Can any of the drivers be closer to the mic than the defined mic distance?

And here's two questions that aren't exactly related to this particular project, but if someone wants to clarify these things to me it would be appreciated.

- There are lots of commercial 3-way designs where the bass driver is located to the side of the cabinet like this. How the xo point phase can be matched in this kind of situation and how low the xo point has to be this to work properly? I'm guessing the flaw of this design is that the woofer is quite a lot more far away from the listening point than the mid/tweeter.

- Does the location of the reflex port have any real effect to the driver/port output combined response? I am guessing there is somekind of phase difference between these two outputs? Does it make a difference how far away from the driver the port output is? (for example in the front or the back of the baffle)
 
I believe the lower the crossover frequency, phase alignment becomes less important. Whether your phase mis-alignment being out at 450Hz by approx 55 degrees is "good enough" I'm not sure. Reading John Kreskovsky's phase paper might help understand.

Overall I would say your phase tracking is bad. The mid/tweeter is out by 120 degrees. Inverting the tweeter will make this 60 degrees - but still bad.

What is the off-axis performance like? specifically what is the upper rolloff like on the midrange? You might want to bring the tweeter / mid XO down a bit if the MCA12 is beaming (ie. more than 3dB down) at 3250Hz

You mentioned you are using a 300cm (ie. 3 metre) mic distance? Is this correct? If so - then that's great if you can gate out room relections as you'll be able to get good point summation of the drivers (being a 3 way - you need more distance between baffle and mic) and you'll be able to measure the lower xo point farfield for xo simulation.

What is your system impedance like? Do you have any nasty phase angles in the bass region and low impedance troughs?

You might need to post your xo schematic and impedance curves so we can be of further help.

You might need to go for asymetric slopes. (ie. 3rd electrical on the midrange HP to better integrate with the woofer and poss. same on the tweeter HP to integrate better with the mid).

David.
 
One more comment - side firing woofers are only a good idea for producing omnidirectional bass (ie. below 100Hz). Anything above this you'll get severe rolloff and poor summing with the midrange. With your drivers, side-firing is not an option.

If you want a cabinet with a leaner profile - try a trapezoidal or angled front baffle (ie. non-parallel sides). However this would be harder to cut / build since you are going for a sloped baffle anyway.

David.
 
Dave Bullet said:
I believe the lower the crossover frequency, phase alignment becomes less important. Whether your phase mis-alignment being out at 450Hz by approx 55 degrees is "good enough" I'm not sure. Reading John Kreskovsky's phase paper might help understand.

Overall I would say your phase tracking is bad. The mid/tweeter is out by 120 degrees. Inverting the tweeter will make this 60 degrees - but still bad.

I believe you misunderstood my first post. This is how the simulated phase responses look at 300cm, red is tweeter, green is midrange and pink is woofer.
At the mid/tweeter XO point (approx 3,2KHz) the red and green lines are aligned, at mid/woofer XO point which is around 450Hz, there is about 15 degree difference in green/pink curves.

Here is the other simulation I mentioned, the relationships to each other are same as in the first one, but now the mic is set up in the SoundEasy so, that the bass is exactly 300cm away from the mic, and the other two drivers are more far away. In This first simulation the tweeter is set to 300cm which again equals to the defined mic distance, and the mid/bass drivers are closer than 300cm. I don't know which one is the correct way to make the simulations.


You mentioned you are using a 300cm (ie. 3 metre) mic distance? Is this correct? If so - then that's great if you can gate out room relections as you'll be able to get good point summation of the drivers (being a 3 way - you need more distance between baffle and mic) and you'll be able to measure the lower xo point farfield for xo simulation.

I only use 300cm distance for the mic in the SoundEasy simulations, as I thought this would represent more closer to the real listening distance. As I am simulating a tiltel front baffle, moving the mic closer or more far away affects to the distance relationships to each driver, and that way will give different phase curves. Of course when the speaker is ready, the listening distance/height will vary, but I though it would be a good idea to design it so, that the optimal listening point would be 3 meters away from the speaker, and at the height of the tweeter.

I measured all the drivers in this test baffle at 1m, just to get the phase differences or acoustic center differences and infinite baffle frequency responses for the SoundEasy simulations.

Measuring the final speaker at 3m would be a problem, but I could make the measurements from 1m and make the simulations temporarily at 1m in SoundEasy, and if I get a good match I know all the simulations are made correctly, right?
 
Hi,

Tilting the baffle can be circumvented by using some assymetry in
the mid / treble crossover.

As said phase tracking at bass / mid c/o is less important as it does
not cause assymetric off axis lobing, mild phase differences can be
used to affect the way the drivers sum.

http://www.troelsgravesen.dk/3WClassic.htm

Has the bass / mid inverted but mid / treble the same .......

:)/sreten.
 
sreten said:
Hi,

Tilting the baffle can be circumvented by using some assymetry in
the mid / treble crossover.

As said phase tracking at bass / mid c/o is less important as it does
not cause assymetric off axis lobing, mild phase differences can be
used to affect the way the drivers sum.

http://www.troelsgravesen.dk/3WClassic.htm

Has the bass / mid inverted but mid / treble the same .......

:)/sreten.


Yes, that would be one approach to the problem. Do you suggest that I would make a straight standing front baffle and use assymmetric crossover, or just reducing the tilt and make a assymmetric xo? (of course this should be simulated to see if it works)

In my previous 2-way speaker I did use assymmetric crossover to get the phases align, but I though I would try something different for this project, which is tilted front baffle. Just another approach to fix the problem, right? I guess both of these have their good and bad points.
 
I believe you misunderstood my first post. This is how the simulated phase responses look at 300cm, red is tweeter, green is midrange and pink is woofer.

Yup - sorry, I must have mis-read your post.

However, good phase tracking through the crossover means more than just phase aligned at the precise Fc.

For example - your woofer / midrange phase is 30 degrees out (and growing) at approx 800Hz where the woofer is only -15dB down. Would this nor smear the midrange? With 2nd order acoustic slopes this is no trivial task to get right through the crossover. (Personally, I struggle with 4th order :rolleyes: )

Same goes for the mid / tweeter xo - quite a bit of divergence above the Fc and happens quite quickly (from about 4KHz).

PS: How does system impedance look?

Cheers,
David.
 

djarchow

Member
2007-04-05 7:23 pm
I am a bit unclear as to why you measured the drivers infinite baffle and not in the real cabinet? Are you doing that to figure out the tilt needed to align the drivers before building the actual cabinet?

If so, here is how I did it for my transient perfect design. I measured the drivers on a flat baffle on each drivers axis at 1 meter from the baffle surface. Once I had these measurements I designed my XOs to match my target slopes as closely as possible as far into the stop band as possible. At some point though, due to the inherant bandpass nature of all real drivers, the drivers acoustic response with the XO began to roll off faster than my targets. Of course in the system screen they didn't integrate properly because of the phase shift due to the acoustic center offset mismatch and the eventual faster rolloffs than my targets. Mind you though, that as an example, with 2nd order 2.k kHz XO points, I still matched my woofer target slope well past 10 Khz.

Then I checked in the system screen how much additional phase shift I needed to align the phase of the two drivers at Fc at my design point (3 meters from the tweeter axis). I added this much additional offset/delay in SoundEasy to the driver layouts and with a little playing with the distances got the driver offsets correct to give me flat frequency response, proper phase tracking between the drivers and good transient response at my design point.

Then I translated these offsets into a new cabinet design which offset the drivers properly. I made a new test cabinet this time with a stepped baffle and remeasured the drivers exactly as I did the first time around. Then all I had to do was make any tweaks to my XO design to deal with the diffraction effects of the stepped cabinet.

In your case you would do basically the same thing, except design a cabinet with a tilted front baffle to match the amount of offsets you determined in the system screen. The only problem you may run into is that you will most likely need a different amount of offset for the mid tweeter than you will for the mid/woofer which may necessitate a stepped cabinet instead of just a tilted baffle. Of course the mid woofer XO is impacted less by driver offset errors due to the lower frequency.

I hope this helps.

Regards,

Dennis
 
Dave Bullet said:
However, good phase tracking through the crossover means more than just phase aligned at the precise Fc.

Sure, your right about that.

For example - your woofer / midrange phase is 30 degrees out (and growing) at approx 800Hz where the woofer is only -15dB down. Would this nor smear the midrange? With 2nd order acoustic slopes this is no trivial task to get right through the crossover. (Personally, I struggle with 4th order :rolleyes: )

Same goes for the mid / tweeter xo - quite a bit of divergence above the Fc and happens quite quickly (from about 4KHz).

Now this is the thing I've been struggling with, I just can't get a good phase tracking near the XO points.

And here is what I mean, and don't fully understand:

Comparison

For this simulation I put the woofer and midrange to excatly same point, right front of the mic.
RED = Woofer at Z 0cm
GREEN = midrange at Z 0cm
PINK = midrange at Z 2,5cm
BLUE = midrange at Z 5cm

Now as I move the midrange more far away from the mic, the phase is getting closer at the XO point 450Hz, but above that, the phase curves cross and the difference is getting more and more bigger the more far away I put the driver.
What am I doing wrong here? I believe I made the HB transform correctly, as this wasn't the firts time I've done it. Can I somehow get the curves line up correctly throughout the crossover point area?

PS: How does system impedance look?

Here is the impedance curve as it looks at this moment. The impedance peak is at 30Hz 52ohm.
 
djarchow said:
I am a bit unclear as to why you measured the drivers infinite baffle and not in the real cabinet? Are you doing that to figure out the tilt needed to align the drivers before building the actual cabinet?

Exactly. After I've decided how to place the drivers, I can either build a prototype cabinet as I did last time, or if I'm 100% sure about the cabinet costruction I can make the real cabinet. After the cabinet is built, I can make new measurement right from the cabinet and then design the crossovers in SoundEasy.

The only problem you may run into is that you will most likely need a different amount of offset for the mid tweeter than you will for the mid/woofer which may necessitate a stepped cabinet instead of just a tilted baffle. Of course the mid woofer XO is impacted less by driver offset errors due to the lower frequency.

Yes, this might end up being a problem.
Although I might end up chancing the midrange driver, as the sensitivity is quite poor, and the situation might change after that.

I hope this helps.

Every comment is welcome :)
 

djarchow

Member
2007-04-05 7:23 pm
As I mentioned there are a couple things impacting your phase. One is the distance in your simulation between the driver and the mic posiition. The other is your acoustic response and how closely it matches the target response. If you match your target spl response far enough into the stop band, yoour phase for that driver will match the target phase very close as well. Then all you have to do is adjust your offsets enough to bring the phase into alignment for the two drivers. But if you aren't matching your targets close enough, you will never get good phase match over a wide range (though you can get it at fc)

I would like to see your spl and phase response for each driver/XO (especially your mid and tweeter) in the XO screen along with a plot of the target spl and phase response.

Regards,

Dennis