Fireplace mantle built in centre speaker build - Crossover help

So I've been wanting to make a wall mounted centre speaker for a while now and come up with what seemed like a good selection of speakers via playing with VirtuixCAD2 and SpeakerBoxLite.com.

Taking performance, size and price into consideration, I came up with an MTM design that has the following:

2 x Peerless SDS-P830656 5.25" 8 ohm woofers.
1 x Peerless XT25BG60-04 1" 4 ohm tweeter.

I came up with an enclosure volume of 12.8 litres and then modelled this enclosure design on TinkerCad:

Mantle Centre Speaker TinkerCad.png


I built the unit and took measurements with a boom arm and Umik-1 following the guide at AudioScienceReview here and came up with some reasonable looking graphs which follow the manufacturers anechoic graphs close enough to look legit to my untrained eye.

Mantle Centre Speaker.jpg


Now comes the tricky part. I seem to have a ~3khz dip in the tweeter response and I'm trying to design the crossover on VirtuixCAD2 and would like some help.
I measured each of the 3 drivers with the Dayton Audio DATS v3 to get accurate t/s specs and then input that data into the software. I think I've got it fairly flat (as best I can) but I'm not sure if the phase response is going to cause issues or if the resistance is too low (going down to about 3.1 ohms).
I'd love some feedback to assist with the crossover design as I want this speaker to sound great!

Mantle Centre Speaker VirtuixCAD2.png


For some context it's going to be powered by a Denon AVR-X3300W in a 5.1.2 speaker layout and will be mounted to the wall just above the fireplace alcove.
 
A few observations/questions:

1) Your 2 midbasses can be put in parallel and driven off one crossover leg. That will cut your crossover complexity down significantly.
2) Are you asking how to fix the dip at 3 kHz in the tweeter response with the crossover? If so, that's not the way things would typically be done. The dip may be an artifact from the geometry around the tweeter (the molding above being a likely culprit, or the step below). This is likely to change with vertical angle and when you place the mantle against the wall. I can give more suggestions on how to try to figure out what's going on, but they may not be relevant when you get it against the wall (or a temporary test baffle that approximates a wall if you'd rather do that).
3) Have you measured the tweeter and midranges together and tried to get Z offset correct or done something else to handle that in the simulation? If you haven't, your phase information isn't likely to be correct. I can explain this further, but it's kind of involved and don't want to waste the space if you already know it.
4) Given the sensitivity of the tweeter and midbasses, the padding you have on the midbasses seems suspect (the 2.7 ohms).
 
You have choosen good drivers and build an interesting construction. You seem new to constructing complete speakers. Maybe let me explain what I see, even as this includes some points of disagreement.

First, an MTM configuration in horiziontal orientation inverts the advantage of the vertical MTM build.
If only one person listening is sitting in the middle and does not move the head up and down, it may not matter. I know this is often seen in commercial products, as the customers wishes are more important for sales than sound and physics.
A horizontal D'Appolito is none, it is all about the vertical orientation. So a 3rd order x-over is not needed.

**If you don't like the sound changing when you move just a little on the sofa, you may later only use one woofer and leave the other one for optics. Some well done center speaker that look MTM use one woofer just as a low range improver. Also called a 2 1/2 way system.
For now, better first build a good 2-way. A 3-way speaker adds 300% complication.**

Second, in the simulation, you build two identical x-over legs for two identical drivers. This is a waste of work and money. You parallel the two and use one x-over in front of them.
Next, the resistors in series with the drivers are not needed and should be avoided. You may throttle a tweeter, but not a woofer.
A simulation is stupid, if you feed it wrong data the output is wrong too. I honestly can not see much sense in yours. Maybe your measurements have some problems.

3rd Your simulated frequency response seems quite unrealistic, as the woofer is louder than the tweeter. The SPL of the two woofer in parallel (85.7dB+6dB=91.7dB) should nearly match the tweeter (92.3dB). A difference of 0.6dB is not really audible and may even add some detail to the sound.

The chassis you use are quite simple to combine, a 2nd order x-over should do and sound better. Any extra part in the x-over adds something you may not want.
Please parallel the two woofer and measure again. This will change your simulation!

Then you should simply build a basic x-over, cleanly separating woofer and tweeter at the x-over point. 2.0 kHz are best for this combination. Crossing at 5kHz makes no sense and will sound much worse.
These drivers perform very well and need no exotic x-over. They are so "school book" I could even give you values for the x-over, but I don't want to start a discussion about that. In this (quite rare) case only 2 caps 2 inductors may do the x-over. A most simple 12dB x-over.

If you find out, later, that my first point is right, you may go for 2 1/2 way. You have to add a large coil in front of the second woofer, but this will change the values of the other components too. Also you will need a resistor network to reduce the tweeter level. So a complete rebuild with other parts.

Last, you have a nice AVR with the fantastic Audyssey MultEQ XT32 system. Do not try to use a lot of parts in the x-over to flaten the response to the last dB. Let Audyessey do the hard work, a DSP can do it much better than any passive part.
 
A few observations/questions:

1) Your 2 midbasses can be put in parallel and driven off one crossover leg. That will cut your crossover complexity down significantly.
2) Are you asking how to fix the dip at 3 kHz in the tweeter response with the crossover? If so, that's not the way things would typically be done. The dip may be an artifact from the geometry around the tweeter (the molding above being a likely culprit, or the step below). This is likely to change with vertical angle and when you place the mantle against the wall. I can give more suggestions on how to try to figure out what's going on, but they may not be relevant when you get it against the wall (or a temporary test baffle that approximates a wall if you'd rather do that).
3) Have you measured the tweeter and midranges together and tried to get Z offset correct or done something else to handle that in the simulation? If you haven't, your phase information isn't likely to be correct. I can explain this further, but it's kind of involved and don't want to waste the space if you already know it.
4) Given the sensitivity of the tweeter and midbasses, the padding you have on the midbasses seems suspect (the 2.7 ohms).
Thanks for the feedback.
I've put the tweeter at 0mm of z Offset as it's flush with the baffle (unless I need to account for the dust cap?) and I've put the woofers as 10mm z Offset due to the flush mount but cone being concave.

I didn't think I should take measurements with the baffle/enclosure mounted to the wall due to the reflections I'd be measuring and not the speakers performance. I expect my avr will correct the reflections etc to the best of it's ability.

I've made a new crossover design based on the feedback.
How does this look? Anything else I should do?
 

Attachments

  • Mantle-Centre-Speaker-Virtuix-CAD2-rev2.png
    Mantle-Centre-Speaker-Virtuix-CAD2-rev2.png
    169 KB · Views: 28
You have choosen good drivers and build an interesting construction. You seem new to constructing complete speakers. Maybe let me explain what I see, even as this includes some points of disagreement.

First, an MTM configuration in horiziontal orientation inverts the advantage of the vertical MTM build.
If only one person listening is sitting in the middle and does not move the head up and down, it may not matter. I know this is often seen in commercial products, as the customers wishes are more important for sales than sound and physics.
A horizontal D'Appolito is none, it is all about the vertical orientation. So a 3rd order x-over is not needed.

**If you don't like the sound changing when you move just a little on the sofa, you may later only use one woofer and leave the other one for optics. Some well done center speaker that look MTM use one woofer just as a low range improver. Also called a 2 1/2 way system.
For now, better first build a good 2-way. A 3-way speaker adds 300% complication.**

Second, in the simulation, you build two identical x-over legs for two identical drivers. This is a waste of work and money. You parallel the two and use one x-over in front of them.
Next, the resistors in series with the drivers are not needed and should be avoided. You may throttle a tweeter, but not a woofer.
A simulation is stupid, if you feed it wrong data the output is wrong too. I honestly can not see much sense in yours. Maybe your measurements have some problems.

3rd Your simulated frequency response seems quite unrealistic, as the woofer is louder than the tweeter. The SPL of the two woofer in parallel (85.7dB+6dB=91.7dB) should nearly match the tweeter (92.3dB). A difference of 0.6dB is not really audible and may even add some detail to the sound.

The chassis you use are quite simple to combine, a 2nd order x-over should do and sound better. Any extra part in the x-over adds something you may not want.
Please parallel the two woofer and measure again. This will change your simulation!

Then you should simply build a basic x-over, cleanly separating woofer and tweeter at the x-over point. 2.0 kHz are best for this combination. Crossing at 5kHz makes no sense and will sound much worse.
These drivers perform very well and need no exotic x-over. They are so "school book" I could even give you values for the x-over, but I don't want to start a discussion about that. In this (quite rare) case only 2 caps 2 inductors may do the x-over. A most simple 12dB x-over.

If you find out, later, that my first point is right, you may go for 2 1/2 way. You have to add a large coil in front of the second woofer, but this will change the values of the other components too. Also you will need a resistor network to reduce the tweeter level. So a complete rebuild with other parts.

Last, you have a nice AVR with the fantastic Audyssey MultEQ XT32 system. Do not try to use a lot of parts in the x-over to flaten the response to the last dB. Let Audyessey do the hard work, a DSP can do it much better than any passive part.
I followed the audiosciencereview.com guide on taking gated measurements but it's my first time so maybe I did something wrong?

I setup my umik-1 on a boom stand and set my speaker cabinet up on an adjustable hieght pa speaker stand in the middle of my lounge (4m x 3.5m x 2.85m with a round bay window on one side). I did several measurements and found the best height for the speaker and the best distance for the microphone.

I did consider doing a 2.5 way design but thought for simplicity I would do a 2 way design to begin with.
I'm aware that a horizontal MTM speaker design isn't ideal but I thought that given the relatively small driver size, the dual bass drivers would allow lower distortion.

I've done some of the recommendations you've mentioned and changed my design.

Any more feedback would be great, thanks!
 

Attachments

  • Mantle-Centre-Speaker-Virtuix-CAD2-rev2.png
    Mantle-Centre-Speaker-Virtuix-CAD2-rev2.png
    169 KB · Views: 23
Your woofers are in series in the new layout. They should be in parallel. The level is probably going to look wrong compared to the tweeter, but as discussed above, there's probably something going on in your measurement that's skewing things.

The acoustic center is typically close to the front of the voice coil (where it attaches to the cone in a woofer). I would expect something greater than 10 mm for how you have things laid out, but I could be wrong. This can be determined experimentally by leaving the speaker and microphone in the exact same locations, then taking a frequency response measurement of the tweeter on its own, the tweeter and woofer together, then the woofer on its own. You then load the individual driver responses into the model for each driver. Then load the combined response as a reference file. In the model you also have to have the physical setup correct for the individual drivers and the microphone. From there you can adjust the Z offset in the model until the predicted response matches the measured response for the woofer and tweeter combined.

If you don't want to get that involved in the modeling/measuring, you can also just build the crossover and tweak it on the fly. People have different approaches to things.

The higher the crossover frequency, the more critical Z offset becomes if you want an accurate model.
 
Last edited:
Member
Joined 2010
Paid Member
But keep in mind that the sideways MTM is a creation of the marketing departments and is not a good configuration for any loudspeaker.

Can’t help with the XO, i’d just use a single FR driver.

dave

Hmm... the MTM configurations that sound good tend to have at least two tweeters in a vertical array between the woofers. That ensures good vertical dispersion.... As a matter of fact, I got one.. a PSB Stratus C6i. Unfortunately, the marketing department tends to overshadow the engineering department in such things.

I've always wondered about a ribbon driver in a sideways MTM configuration? Maybe the OP could change his tweeter to something that doesn't launch a hemispherical wavefront, that will mitigate the effects of the mounting around it.

OP? It's a nice project, but did you think of just buying your way out with a Maggie center channel speaker? It's shallow, curved and designed to be mounted on the wall.

https://magnepan.com/products/magnepan-ccr?variant=42607785541868

Have fun.
 
Last edited:
Member
Joined 2010
Paid Member
From a vertical perspective, combing for a center channel HT speaker is not an issue. Such speakers are set up, angled, to aim at your seated ears, because we sit to watch movies. We want a narrow vertical beam and a wide horizontal beam. What we're looking for is reasonably "tall" waveform devoid of perturbations by stuff above and below the drivers. In a "normal" configuration such drivers will have a flat screen right above and God Knows what below, so some means of minimizing the interference from those surfaces is important.

As your drawing above shows. such a pair launches a broader "vertical" wavefront at the listening position. Indeed, your drawing proves why such a configuration is perfect for a HT center channel speaker.

The acoustic effects of diffractions from surfaces above and below the drivers will be different, so hopefully they will "spread" out at the listening point.
For music, and L/R speakers, things are different...

...

For the OP, though, I don't think he has much room to add a second tweeter, so perhaps playing with a ribbon or a line driver might make more sense? But I wonder if the ribbon will have sufficient wide dispersion?

...
 
Last edited:
I appreciate that there's potential issues with the MTM speaker design but I've been using various models with this design for years now and most sound great to my ears.

This is an interesting article on the topic HERE

Any critique on the new crossover?

I was told that I should be in series to bring the woofers down to match the tweeter and also so I lower distortion vs parallel. Does this sound correct?
 
Member
Joined 2010
Paid Member
That's a good write up... which emphasizes the importance of avoiding lateral lobing.

Recall that for a HT center channel we want very wide horizontal dispersion. People will sit on couches and chairs around the TV and those angles can easily be 35 degrees! Unless you have a High Falutin' HT room... It's important that people watching TV at a an angle also hear the full sound.
 
I was told that I should be in series to bring the woofers down to match the tweeter and also so I lower distortion vs parallel. Does this sound correct?
Going by the spec sheets, no.

The woofers are 85.7 dB sensitivity. The tweeter is 92.3 dB. As Turbowatch2 pointed out earlier, when the two woofers are put in parallel, you basically just get up to the sensitivity of the tweeter. In practice, the tweeter may still need a little attenuation to produce a more relaxed sound. Though if you like what's often called a more revealing sound, you may like the tweeter without any attenuation.