3-Way/Side-Firing Woofer - Measurements for XO Design

[Robert Dunn]

Ok, guys, I'm sure that I'm over-thinking things but I haven't designed a side-firing woofer before. Hopefully someone can set me straight. Please bear with a rather lengthy explanation as I meander around to the questions.

I'm working on a three-way in a tall box. Basically, it's a 4' tall, 10" wide, 14.5" deep box with the mid located near the top of the box and the woofer on the side roughly half way from the floor. The woofer is ported on the front of the box near the floor. FWIW, the woofer is an SB29NRX75-6 (spec sheet is here).

I have constructed a prototype box, taken measurements, and started working on the XO. Taking the mid and tweeter measurements is easy enough but I'm not sure what woofer measurements to use in the XO design.

The below shows measurements for NF woofer (green), NF port adjusted for relative woofer diameter (mustard), and summed response (blue).



I then adjusted the combined NF down to a theoretical 1 meter SPL shown below (purple). I also took a 1 meter woofer/port measurement on the mid axis (red) primarily to get an idea of the distance/delay (thinking of acoustic centers here) between the mid and woofer. The delay seems to be slightly more than 8".



[Note that all of these are MLS measurements (at 2.834V with a calibrated mic) and actual SPL is, therefore, 3dB higher.]

The low-frequency limit of the FF woofer/port measurement is a bit under 200Hz. The NF curve is theoretically useful below roughly 500Hz (based on D'Appolito's white book if I did the math correctly). The on-mid-axis measurement is, of course, more than 1M from the mic.

Finally, I plan to cross-over to the mid at 250Hz or below. While wavelengths are long here, I'd like to keep things in phase as much as possible.

So here are the questions: Should I design the XO using the NF curves? Or should I take FF on-axis woofer measurements and splice with the NF for XO design? Given that low order filters will allow significant woofer output (side-firing and reflecting off of whatever surfaces are nearby) into, say, the 1kHz+ region, what acoustic order would you target?

I'm sure that more questions will come to mind just as soon as I post this. In the meantime, thanks for any advice.

[rpb]

Re: 3-Way/Side-Firing Woofer - Measurements for XO Design

Ok, guys, I'm sure that I'm over-thinking things but I haven't designed a side-firing woofer before. Hopefully someone can set me straight. Please bear with a rather lengthy explanation as I meander around to the questions.

I'm working on a three-way in a tall box. Basically, it's a 4' tall, 10" wide, 14.5" deep box with the mid located near the top of the box and the woofer on the side roughly half way from the floor. The woofer is ported on the front of the box near the floor. FWIW, the woofer is an SB29NRX75-6 (spec sheet is here).

I have constructed a prototype box, taken measurements, and started working on the XO. Taking the mid and tweeter measurements is easy enough but I'm not sure what woofer measurements to use in the XO design.

The below shows measurements for NF woofer (green), NF port adjusted for relative woofer diameter (mustard), and summed response (blue).

[ATTACH=CONFIG]37226[/ATTACH]

I then adjusted the combined NF down to a theoretical 1 meter SPL shown below (purple). I also took a 1 meter woofer/port measurement on the mid axis (red) primarily to get an idea of the distance/delay (thinking of acoustic centers here) between the mid and woofer. The delay seems to be slightly more than 8".

[ATTACH=CONFIG]37227[/ATTACH]

[Note that all of these are MLS measurements (at 2.834V with a calibrated mic) and actual SPL is, therefore, 3dB higher.]

The low-frequency limit of the FF woofer/port measurement is a bit under 200Hz. The NF curve is theoretically useful below roughly 500Hz (based on D'Appolito's white book if I did the math correctly). The on-mid-axis measurement is, of course, more than 1M from the mic.

Finally, I plan to cross-over to the mid at 250Hz or below. While wavelengths are long here, I'd like to keep things in phase as much as possible.

So here are the questions: Should I design the XO using the NF curves? Or should I take FF on-axis woofer measurements and splice with the NF for XO design? Given that low order filters will allow significant woofer output (side-firing and reflecting off of whatever surfaces are nearby) into, say, the 1kHz+ region, what acoustic order would you target?

I'm sure that more questions will come to mind just as soon as I post this. In the meantime, thanks for any advice.

I've never done a side firing woofer, but my opinion is that second or third order slopes should be used. I think that the integration will be somewhat hit or miss anyway, so I'd want as little overlap as possible, and the woofer restricted from going very high. The filter will probably cause some big issues in the woofer response around 100 hz. I'd minimize that as best possible, and let the blend at 250 be off a bit if that's the way it works out. I think tonal balance should outweigh the blending. You may get lucky, or brilliant, and get everything working perfectly, but If that doesn't happen, I'd minimize the damage of an altered bass response if possible. A slight dip at 250 hz probably won't be noticable. A 2 dB bump at 100 hz might. I don't mean for this post to sound negative, but I realize this will not likely be real easy, so it may mean choosing the lessor of two evils. You may actually need to do this by ear after trying the simulated filter. The 250 range is very room, and room location dependent.

[ReissM]

Re: 3-Way/Side-Firing Woofer - Measurements for XO Design

Looks like a fun project. You're torn between:
1. keeping the woofer and mid close to each other for phase reasons, and
2. putting the woofer close to the floor for increased boundary augmentation (and also to minimized floor bounce issues.)


Hmmm... let me throw some ideas at you. You can dismiss them if you don't like 'em. It might already be too late for my suggestions because you said that you already have a protoype box built.


A. How about lowering the crossover point to 100Hz where the wavelengths are much longer (11.3 feet) and more omnidirectional.

B. Then you can physically locate the woofer lower in the cabinet to minimize the floor bounce problems, while still having the phase integrate reasonably well with the mid. (because of the longer wavelengths). Do some quick trogonometry and convince yourself that the path-length differences (and thus phase) of your woof & mid will be mitigated by going with an 11.3 foot wavelength.


Choosing 100Hz reduces the amount of midrange frequencies coming out of the side firing woofer, (At 250Hz, it may be more difficult to integrate the side firing woofer's output with the midrange driver's output. At 100Hz it will be easier.)

I would suggest a 3rd order or 4th order network on the woofer to simplify the acoustic/phase integration with the midrange driver. You may have to add a notch filter to the woofer's x-over network to make the response behave... but that's a small price to pay to overcome the other issues mentioned above.

Try splicing the NF and FF curves, import them into Jeff's PCD, and design a steep filter that only lets the woofer sing to 100 Hz. Be aware of the filter "Q" during this step... you don't want nasty peaking. (maybe a notch filter will help with this.)

Get the polarity (phase tracking) correct from woofer to mid, then keep the mid's polarity fixed and determine the tweeter's polarity for best integration with the mid. You can run hundreds of trials with Jeff's PCD before you attempt to build the crossover. Best of luck on this design.

[PEB]

Re: 3-Way/Side-Firing Woofer - Measurements for XO Design

Hi Robert. I'm keenly interested in your project, because it mimics what I am going through right now. I can give you advice based on my experience. But first some questions:

Can you provide us with a 1-2M SPL of your midrange in its acoustic cavity? We would want to see what it is willing to do before choosing a cross point frequency. Even NF of the mid will help.

Excellent idea to build a prototype cabinet. That’s what I did.

http://techtalk.parts-express.com/showthread.php?237810-Collapsible-test-box-for-8-quot-woofer

Can you take your proto sub outdoors for ground plane measurements? If so, then measure at 1-2M distance. (Let us know if you need tips on the GP technique.) Take 4 curves, one for each face of the sub. This will give us an idea of the SPL with woofer and port summing, and show the performance for front- or side-firing. If your SW can average curves, then average the 4 curves and that will give us an idea of the performance in your room at the lower frequencies.

Do you have means to modify your test cabinet to move the woofer down low? You really don’t want the port opening too close to the woofer or you may get additional or louder pipe resonances. (A woofer firing 90° across the port opening is the worst case.) If the woofer is down low to the floor, then you would put the port up high. Or you could go to an elbow port like I have. The port opening is now high inside the cabinet, but the elbow exits out low. In this configuration, you can have the elbow exit any panel you want.

Let’s not worry about delay, or anticipate XO slopes just yet. This is the time to experiment with the low-F range.

[Robert Dunn]

Re: 3-Way/Side-Firing Woofer - Measurements for XO Design

Looks like a fun project. You're torn between:
1. keeping the woofer and mid close to each other for phase reasons, and
2. putting the woofer close to the floor for increased boundary augmentation (and also to minimized floor bounce issues.)


Hmmm... let me throw some ideas at you. You can dismiss them if you don't like 'em. It might already be too late for my suggestions because you said that you already have a protoype box built.


A. How about lowering the crossover point to 100Hz where the wavelengths are much longer (11.3 feet) and more omnidirectional.

B. Then you can physically locate the woofer lower in the cabinet to minimize the floor bounce problems, while still having the phase integrate reasonably well with the mid. (because of the longer wavelengths). Do some quick trogonometry and convince yourself that the path-length differences (and thus phase) of your woof & mid will be mitigated by going with an 11.3 foot wavelength.


Choosing 100Hz reduces the amount of midrange frequencies coming out of the side firing woofer, (At 250Hz, it may be more difficult to integrate the side firing woofer's output with the midrange driver's output. At 100Hz it will be easier.)

I would suggest a 3rd order or 4th order network on the woofer to simplify the acoustic/phase integration with the midrange driver. You may have to add a notch filter to the woofer's x-over network to make the response behave... but that's a small price to pay to overcome the other issues mentioned above.

Try splicing the NF and FF curves, import them into Jeff's PCD, and design a steep filter that only lets the woofer sing to 100 Hz. Be aware of the filter "Q" during this step... you don't want nasty peaking. (maybe a notch filter will help with this.)

Get the polarity (phase tracking) correct from woofer to mid, then keep the mid's polarity fixed and determine the tweeter's polarity for best integration with the mid. You can run hundreds of trials with Jeff's PCD before you attempt to build the crossover. Best of luck on this design.

I think you're right, I will need to keep the XO point low. But I also don't want to push it so low that the mid is stressed at higher SPL. 100Hz is probably too low.

I'm sure that I need to splice the NF & FF. The thing that I'm struggling with is, what FF curve to use? If I take the FF from the front on the mid axis, I'm both farther from the woofer and 90 deg off axis. While that seems right for the XO phasing, it seems wrong for the SPL being radiated into the room. If, on the other hand, I take the FF from the side, the curve will have a more accurate SPL in the higher frequencies but I lose the off-axis information that I need to integrate the drivers. Or that's what I keep telling myself.

The solution is, of course, higher order XO where FF accuracy will be less of an issue. OTOH, more expensive parts.

Thanks.

[Robert Dunn]

Re: 3-Way/Side-Firing Woofer - Measurements for XO Design

Hi Robert. I'm keenly interested in your project, because it mimics what I am going through right now. I can give you advice based on my experience. But first some questions:

Can you provide us with a 1-2M SPL of your midrange in its acoustic cavity? We would want to see what it is willing to do before choosing a cross point frequency. Even NF of the mid will help.

The mid 1M curve below However, I did not take a NF measurement. Need to do that and splice to FF for sure since the FF is good only down to around 200Hz. I'll post that, hopefully later today.

Can you take your proto sub outdoors for ground plane measurements? If so, then measure at 1-2M distance. (Let us know if you need tips on the GP technique.) Take 4 curves, one for each face of the sub. This will give us an idea of the SPL with woofer and port summing, and show the performance for front- or side-firing. If your SW can average curves, then average the 4 curves and that will give us an idea of the performance in your room at the lower frequencies.

I wish that I had portable measurement equipment but, no, it's not practical. However, I think that my simulations so far are reasonably accurate. Here is a simulation in free space taken at the mid height, centered on each side.


Alternatively, here is a polar simulation for selected frequencies up to about 5kHz taken on the mid axis (front).



The polar graph suggests that things begin to become directional around the 320Hz curve. Of course, in-room is going to be quite different.

Do you have means to modify your test cabinet to move the woofer down low? You really don’t want the port opening too close to the woofer or you may get additional or louder pipe resonances. (A woofer firing 90° across the port opening is the worst case.) If the woofer is down low to the floor, then you would put the port up high. Or you could go to an elbow port like I have. The port opening is now high inside the cabinet, but the elbow exits out low. In this configuration, you can have the elbow exit any panel you want.

Gee, I hope I don't have to move the woofer. FWIW, the port does have a 90deg bend. The woofer and port NF curves (see first post) do show a summed anomaly around 440Hz but otherwise look ok to me.

Thanks.

[Robert Dunn]

Re: 3-Way/Side-Firing Woofer - Measurements for XO Design

Here are NF/FF tagged curves for all three drivers in box. Woofer curve was taken on the mid axis.



Any thoughts?

[jhollander]

Re: 3-Way/Side-Firing Woofer - Measurements for XO Design

My opinion is to splice the woofer near field with the woofer far field measured on the tweeter axis and skip the offsets if using PCD.

I’m not sure how you could derive a z and x offset if you were to measure the woofer on axis, splice the near field then rotate in PCD, save frd, etc.

I would also consider raising the woofer up and measure outside to see if I could get some measurements below my x-o point.

[Robert Dunn]

Re: 3-Way/Side-Firing Woofer - Measurements for XO Design

My opinion is to splice the woofer near field with the woofer far field measured on the tweeter axis and skip the offsets if using PCD.

That's essentially what I did for the posted curves, only I used the mid instead of the tweeter as the reference.

I’m not sure how you could derive a z and x offset if you were to measure the woofer on axis, splice the near field then rotate in PCD, save frd, etc.

I think something rational could be done, maybe measure at 1m on woofer axis, adjust SPL to account for additional distance of woofer to 1m on mid axis. But, for the moment, I'm going to work with the on-mid-axis measurement, work up an xo, and see how that measures/sounds.

I would also consider raising the woofer up and measure outside to see if I could get some measurements below my x-o point.

I do wish my equipment was portable but it's not. Not to mention that the box weighs 73lbs.

Thanks.

[Robert Dunn]

Re: 3-Way/Side-Firing Woofer - Measurements for XO Design

Sadly, I think this project has reached a full stop. It's time to cut the losses and treat this as a learning experience.

After XO simulations, it's clearly not possible to design a passive XO that will have both reasonably flat SPL through the woofer/mid XO band AND maintain a reasonable minimum impedance. Well, that's not exactly true -- it can be done but the inductors become very large and the capacitor values are simply insane. Driving the woofer with a plate amp appears to be much cheaper than passive XO parts.

Anyway, I think that I now understand why we see so few passive 3-way designs.

Thanks to all for encouragement and suggestions. If anyone wants to see sample XO schematics, SPL or impedance curves, etc., let me know and I'll happily share.

[Face]

Re: 3-Way/Side-Firing Woofer - Measurements for XO Design

A 4ohm woofer would cut down on the inductor size.

[jhollander]

Re: 3-Way/Side-Firing Woofer - Measurements for XO Design

Maybe post some FRDs, ZMAs, and the offsets.

[Robert Dunn]

Re: 3-Way/Side-Firing Woofer - Measurements for XO Design

Maybe post some FRDs, ZMAs, and the offsets.

Sure. I'll have to post the last one in a separate message (limit of 5 attachments).

SPL measurements were at 1m on mid axis. Tweeter xyz = 0, -5.25", +0.35". Woofer = +5", -18", -7.25".


Attachments:
Mid Imp.txt Mid SPL.txt Tweeter Imp.txt Tweeter SPL.txt Woofer Imp.txt

[Robert Dunn]

Re: 3-Way/Side-Firing Woofer - Measurements for XO Design

Last attachment.

Woofer SPL.txt

Would you like to see the xo that I have so far?

[jhollander]

Re: 3-Way/Side-Firing Woofer - Measurements for XO Design

Sure and your sim graph if possible. There are some very talented people here who might have suggestions. I'm not necessarily one of them, but am curious what made you give up on this.