Speaker design, X-Over, and MTM Questions - Please help a noob!

[sbolin]

Hi!

A couple of questions from a noob designer for the speaker design gurus here - hope you can help me understand a couple concepts.
-I have read several places that the baffle affects the x-over (or usually that the crossover has been designed to take the baffle into account). What does this mean? What is done to the crossover (specifically) to account for the baffle affects?
-I have also read a little about the MTM configuration, and how it is supposed to use a odd-order crossover, usually a 3rd order. Why is that? Will it not work with a 2nd or 4th? What about a 1st order?
- Also regarding the MTM, what is the benefit to this configuration? I have seen many projects use it, but have not found any really good explaination.
- When one speaks of 'tweeking' a x-over, what exactly does that mean? Does it mean using a standard 'text book' cross over, and slightly varying some of the component values? Or actually changing the topology?
- Finally (for now!), can someone expain what the Qes, Qms, and Qts is physically? What does it mean if the Qms (or Qes) is high or low? How does this affect the speaker performance?

Thanks very much for your help!

-SB

[BillFitzmaurice]

There is a baffle step in a drivers response. At 1 wavelength across the baffle the radiation characteristic is hemispherical, changing to omnidirectional as the wavelength exceeds the baffle width and wraps around it; when it gets to fully omnidirectional the on axis response is down 6dB, so to compensate one has the first inductor of the midbass low pass filter kick in not at the nominal crossover frequency but at the frequency where the baffle step is fully in place; this way as the on-axis response is climbing above that frequency that inductor compensates for the rising on-axis response.

MTMs will work with any order. 1st order doesn't offer much driver protection and allows a lot of driver passband overlap, so it is seldom used. Joe D'Appolito (the inventor of the MTM) prefers using a 2nd order electrical configured for a total of 4th order when combined with the driver acoustical rolloffs.

MTM has a better polar response pattern, maximizing horizontal and minimizing vertical dispersion, while using two woofers helps get their SPL closer to that of the tweeter, as the woofers are losing SPL due to the baffle step and tweeters are generally more sensitive to begin with.

Tweaking means changing anything or everything to get better response.

Qes is the electrical losses of the driver, Qms is mechanical losses, Qts is total losses. Simply put the lower the Qts the higher the efficiency, usually. But also lower Qts gives less bass response, usually. A full understanding of Q factors is too much for a simple answer.

[sbolin]

Originally posted by BillFitzmaurice
There is a baffle step in a drivers response. At 1 wavelength across the baffle the radiation characteristic is hemispherical, changing to omnidirectional as the wavelength exceeds the baffle width and wraps around it; when it gets to fully omnidirectional the on axis response is down 6dB, so to compensate one has the first inductor of the midbass low pass filter kick in not at the nominal crossover frequency but at the frequency where the baffle step is fully in place; this way as the on-axis response is climbing above that frequency that inductor compensates for the rising on-axis response.

MTMs will work with any order. 1st order doesn't offer much driver protection and allows a lot of driver passband overlap, so it is seldom used. Joe D'Appolito (the inventor of the MTM) prefers using a 2nd order electrical configured for a total of 4th order when combined with the driver acoustical rolloffs.

MTM has a better polar response pattern, maximizing horizontal and minimizing vertical dispersion, while using two woofers helps get their SPL closer to that of the tweeter, as the woofers are losing SPL due to the baffle step and tweeters are generally more sensitive to begin with.

Tweaking means changing anything or everything to get better response.

Qes is the electrical losses of the driver, Qms is mechanical losses, Qts is total losses. Simply put the lower the Qts the higher the efficiency, usually. But also lower Qts gives less bass response, usually. A full understanding of Q factors is too much for a simple answer.

First, thank you very much for your anwser, it helps me a lot...but of course, now I have a few more questions!

For the baffle step, it would help me if I can get a real life example. So if I have a two-way system with a 2nd order crossover of 2,500 Hz, in a box with a 225mm baffle width (about 9 inches), this correponds to a frequency of about 1,500 Hz. So does that mean that that the inductor value for the low pass changes to what it would be for a 1,500 Hz crossover, but the capacitor stays the same? I am really confused on this.

Next, if I have some kind of impedience compensation circuit in parallel with the crossover, do I treat these as totally separate? Or (in the case of a zobel on the low pass filter) would the capacitor in the zobel need to be added to the capacitor in the crossover (since they operate in parallel)? Same with a notch filter, I guess.

Back to the MTM, if I have two 8 ohm woofer's operating in parallel, the net resistance will be 4 ohms. Does this mean the tweeter must also be 4 ohms or 8 ohms is ok. If 8 ohms, do I need to put a resistor in parallel with the tweeter to make it an equivalent of 4 ohms?

I really appreciate your time in anwsering my questions, I am very interested in this.

[piro]

I am also trying to get an answer aboud baffle step compensation.
How did you get 1.5KHz from 225 mm?
and this is an aritcle I found that I hope will help me too.

[BillFitzmaurice]

So does that mean that that the inductor value for the low pass changes to what it would be for a 1,500 Hz crossover, but the capacitor stays the same?

Yes, that's pretty much correct, so that when you hit the intended corner frequency you go from a 6dB to a 12dB rolloff, or more if required. You have to actually measure the woofer response to see at what frequency the SPL starts to rise.

if I have some kind of impedience compensation circuit in parallel with the crossover, do I treat these as totally separate?

Yes, but don't automatically use one until you're sure that it's necessary.

if I have two 8 ohm woofer's operating in parallel, the net resistance will be 4 ohms. Does this mean the tweeter must also be 4 ohms or 8 ohms is ok.

Go 8 ohms with the tweeter; the lower impedance of the woofers raises their voltage SPL closer to that of the tweeter, which is usually going to have a higher SPL than the woofer section. You only use resistors if in testing you find that the tweeter level has to be padded down.

[planet10]

Originally posted by sbolin
baffle affects

This article outlines ways of dealing with bafflestep.

http://www.t-linespeakers.org/tech/b...intro-bds.html

If i've got 2 drivers i like putting the 2nd one on the back.

dave

[BillFitzmaurice]

So does that mean that that the inductor value for the low pass changes to what it would be for a 1,500 Hz crossover, but the capacitor stays the same?

Actually in this case 1.5kHz is where the baffle step is completely realized; the rise in response more likely starts at around 400Hz. To tame the rising SPL above that an inductor value between 1 and 2 mH is likely for a 4 ohm load.

[eric180db]

Its also important to realize that compensating for the baffle step isnt for every one. this only provides optimal results in anechoic testing. Many well built systems dont compensate for this and here is why. If the listening room requires the speakers to be near the back wall or worse in the corner then the reflection will keep the response step from happening completely. That means if someone is using, say, bookshelf speakers near a wall without sound absorbsion then compensating for the step will actualy yeald a boost at the step frequency. So it is important to know the acoustice of the room also. An extreme example is in wall speakers. There is never a step compensation for these because what is the step (a 16 foot wall?)

[sbolin]

Originally posted by piro
I am also trying to get an answer aboud baffle step compensation.
How did you get 1.5KHz from 225 mm?
and this is an aritcle I found that I hope will help me too.

Piro- Thanks for the info, it helped quite a bit. I also found the True Audio website helpful too. The 1,500 Hz came from a single cycle at 225mm (345m/s / .225m). This is not correct, I know now.

Originally posted by planet10


This article outlines ways of dealing with bafflestep.

http://www.t-linespeakers.org/tech/b...intro-bds.html

If i've got 2 drivers i like putting the 2nd one on the back.

dave

Well, two speakers is an expensive solution to the problem.

Originally posted by BillFitzmaurice


Actually in this case 1.5kHz is where the baffle step is completely realized; the rise in response more likely starts at around 400Hz. To tame the rising SPL above that an inductor value between 1 and 2 mH is likely for a 4 ohm load.

Yes, from the above reading, I understand the baffle response more. I thought it was a singular affect at a frequency corresponding to the baffle width, but that was totally wrong, it is actually an attenuation of all low frequencies below (about) the baffle width. From the True Audio Website, they have a simple L-R circuit (the L and R in parallel) in series with the speaker (I assume that it is before any other crossover components), and for a 0.25m baffle (similar to what we have been discussing) they came up with a 1mH inductor. This is similar to what you say, so I believe this is the same thing. I understand that this simple attenuates the high frequencies above the desired baffle frequency.

Originally posted by eric180db
Its also important to realize that compensating for the baffle step isnt for every one. this only provides optimal results in anechoic testing. Many well built systems dont compensate for this and here is why. If the listening room requires the speakers to be near the back wall or worse in the corner then the reflection will keep the response step from happening completely. That means if someone is using, say, bookshelf speakers near a wall without sound absorbsion then compensating for the step will actualy yeald a boost at the step frequency. So it is important to know the acoustice of the room also. An extreme example is in wall speakers. There is never a step compensation for these because what is the step (a 16 foot wall?)

I am sure this is true, but I don't really agree with it, but only for two reasons. First, in my opionion it is best to try to design a speaker that by itself has a smooth frequency response, and does not rely on room affects. Second, from a design standpoint it doesn't seem like a good idea to depend on something you can't control, such as room placement. What if the person using the speaker moves to a new place, and moves the speaker, or the room size is very different than you expect, or can't place it near a wall, etc etc. I can control (more or less) the speaker's frequency response, but I can't control the room it is put in. Anyway, I agree mostly with what you say, and am sure many speaker designers follow this advise. For my own speakers, I will consider this, as the fewer elements in the crossover the better, plus I know I have designed the speaker in such a way that it is required to place it near a wall. But in general, I don't know about this...

[piro]

what drivers do you use?