Baffle step correction circuit - how to design?

[mr_push_pull]

My question is: how to design the baffle step correction circuit from the graphs in the Baffle Diffraction Simulator? I have used Passive Crossover Designer (http://www.pvconsultants.com/audio/crossover/pcdc.htm) to simulate the results of the BSC I designed but I don't know if my strategy of choosing the correction curve is correct. I have used 100 Hz as minimum attenuation point (because that is the the more attenuated frequency by the baffle diffraction that I can see in the graph!) and 1000 Hz as maximum attenuation, at that frequency is where the highest peak is found. I believe that's not 100% correct, the values suggested by "The Edge" differ by large amounts.
And one thing I don't understand in BDS: why isn't the response below 100 Hz shown? Doesn't it count in designing the correction circuit?
And the last question: what program do you recommend using for baffle diffraction simulation: the Excel Baffle Diffraction Simulator or "The Edge" The results given by these two programs are pretty different, especially for high frequencies.

 

[Mark25]

here's a basic one by Rod Elliot of ESP, change the values to suit your needs, it's line level though:


http://sound.westhost.com/bafflestep.htm

 

[mr_push_pull]

Thank you for that link, but that page explains active equalization, which I don't preffer, because as long as the baffle step is a phenomenon that occurs at the speaker, the BSC circuit should also be part of the speaker, not something belonging to the amp (anyway, active correction can be achieved with a multiband equalizer, isn't it)?

 

[Mark25]

Thank you for that link, but that page explains active equalization, which I don't preffer, because as long as the baffle step is a phenomenon that occurs at the speaker, the BSC circuit should also be part of the speaker, not something belonging to the amp

so you would passivly EQ a subwoofer then?

(anyway, active correction can be achieved with a multiband equalizer, isn't it)?

if....
you can stand the SQ loss
one of the equalizer f's coincides with your BS freq'


then yes

 

[Jennice]

Now I'm curious!

If Baffle Step correction is such a hassle for DIY'ers to implement at the speaker end (after the power amp), then how does every manufacturer do it?
After all, a speaker manufacturer can't send a line-level filter with the box and hope that everybody has a seperate pre- and opwer-amp!?

Jennice

 

[Svante]

My question is: how to design the baffle step correction circuit from the graphs in the Baffle Diffraction Simulator? I have used Passive Crossover Designer (http://www.pvconsultants.com/audio/crossover/pcdc.htm) to simulate the results of the BSC I designed but I don't know if my strategy of choosing the correction curve is correct. I have used 100 Hz as minimum attenuation point (because that is the the more attenuated frequency by the baffle diffraction that I can see in the graph!) and 1000 Hz as maximum attenuation, at that frequency is where the highest peak is found. I believe that's not 100% correct, the values suggested by "The Edge" differ by large amounts.
And one thing I don't understand in BDS: why isn't the response below 100 Hz shown? Doesn't it count in designing the correction circuit?
And the last question: what program do you recommend using for baffle diffraction simulation: the Excel Baffle Diffraction Simulator or "The Edge" The results given by these two programs are pretty different, especially for high frequencies.

Edge calculates the compensation circuit from the actual baffle step response as in the simulation. It assumes that there is a full 6 dB step, the load is resistive (ie R1 really is a resistor). In Edge, the compensation is always a smooth transition consisting of a single pole and zero. The frequencies are found from the +2 dB frequency of the simulated response.

There are a few reasons why this compensation is not an ultimate or exact solution. First, the load in the passive filter case is not a resistance, but the loudspeaker and other filter components. Second, most baffle geometries does not produce a smooth baffle step, so there will be a residual ripple even after compensation.

Also, another reason that Edge may produce different results than other software is that there is a rule of thumb that circulates on the web, which is based on the early work of Olson. He demonstrated the baffle step for a sphere, consecutively others have assumed that this is applicable to a baffle of the same width. In my view, this is an assumption that leads to errors in the compensation, and this would lead to a difference between Edge and others (that use this assumption). I don't think BDS falls in this cathegory, however.

HTH
/Svante

 

[rabbitz]

I find it easier to use a 0.5 woofer for BSC at the back of the box as suggested by planet10. Works a treat but a bit more expensive as you need additional woofers. You do gain SPL and less components in the crossover.

Do a search and there's some info in the Wiki.

 

[Mark25]

Now I'm curious!

If Baffle Step correction is such a hassle for DIY'ers to implement at the speaker end (after the power amp), then how does every manufacturer do it?
After all, a speaker manufacturer can't send a line-level filter with the box and hope that everybody has a seperate pre- and opwer-amp!?

Jennice

the answer in a lot of case's is that they don't! There are many ways to achieve BSC though;

a .5 way bass driver, with low pass at the BS
a seperate bass driver on the rear of the box, a la sonus faber electa amator.......just a few......

 

[mr_push_pull]

Re: Re: Baffle step correction circuit - how to design?

Originally posted by Svante There are a few reasons why this compensation is not an ultimate or exact solution. First, the load in the passive filter case is not a resistance, but the loudspeaker and other filter components.[/B]

But if one uses impedance equalization?

So, how do I design the BSC? I mean, how do I derive the values of the inductor and the resistor (or the capacitor and the resistor, depending on what type of correction circuit I use), by looking at the diffraction "response" (if I can call it like that)?

 

[BillFitzmaurice]

The first thing you do is ignore trying to accurately predict the baffle step; you must actually measure your woofer response. There are too many variables to arrive at a correct solution otherwise.

Once you have plotted the actual woofer response you'll be able to see the frequency where the response begins to rise. At that frequency you calculate from your usual source (http://www.lalena.com/audio/calculator/xover/ if you don't have another favorite) the value for a first order low pass filter; that will compensate for the rising response.

Having done that you then add additional stages at your crossover frequency to achieve the desired final acoustic and electrical slopes. To do so involves repeated measuring of response and substitution of component values until the desired result is attained.

This, by the way, is the process used by most professional crossover designers, including the best there is in the business, Joe D'Appolito.

 

[Jennice]

the answer in a lot of case's is that they don't! There are many ways to achieve BSC though;

a .5 way bass driver, with low pass at the BS
a seperate bass driver on the rear of the box, a la sonus faber electa amator.......just a few......

... and this .5 woofer adds, what the baffle step removes at the low end (in terms of output)?

 

[rabbitz]

Yep Jennice.

Just a bit of mucking around getting the xo frequency right. I start off with the calculated BS F3, and then arm myself with a range of inductors. Just uses a simple 1st order to roll off the 0.5 woofer. The last one I did worked out the best rolling off just a bit higher than the BS F3.

Some info: http://www.diyaudio.com/wiki/index.php?page=Baffle-step+Physics

This is only for a speaker in free space. If you have a speaker against a wall or very close to it, you don't get the baffle step loss and therefore no BSC required.

 

[Jennice]

rabbitz,

Your icon doesn't make it that easi to take you seriously, you know...?

anyways... Is that why some speaker manufacturers suggest a certain distance to rear wall, to get the low frequency gain from the reflection?

If so, I'm beginning to see a faint candle above the head..

Jennice

 

[Svante]

Re: Re: Re: Baffle step correction circuit - how to design?

But if one uses impedance equalization?

So, how do I design the BSC? I mean, how do I derive the values of the inductor and the resistor (or the capacitor and the resistor, depending on what type of correction circuit I use), by looking at the diffraction "response" (if I can call it like that)?

If you manage to compensate the impedance such that the impedance is resistive, then you are OK. However, a simple zobel is not always sufficient (but sometimes it is).

Edge can suggest component value in the compensation designer, enter the value for R1 and press "suggest f1&f2". Values for R2 and L are then dispalyed.

 

[Jennice]

Svante,

I noticed your link to the XDir tool. Is that vertical or horizontal directivity?
It it usual with a phase shift for the woofer, since the tool allows for that (or just a part of the cross-over)?

Jennice

 

[rabbitz]

Jennice, you still not taking us rabbitz seriously? :>)

It comes down to whether the speaker is in 4PI or 2PI space. With 4PI you get baffle step and reduced boundary reinforcement but with 2PI you don't get baffle step and you have the assistance of boundary reinforcement. Different designs for both types as it's not hard to have a speaker in 2PI that sounds boomy...... has to be designed for that location. You might demo a speaker at a dealer and it appears bass shy (set up in 4PI) but set it up at home closer to the wall and it sounds right.

To me if it's within 75mm of the back wall that close enough to 2Pi for me (I'm probably wrong) and then a transition to 4PI from there.

 

[Jennice]

Rabbitz,

2PI... 4PI...!?

What the heck are you referring to? I just thought I was starting to undertand this Baffle Step phenomenon, and here you rabbitz go, sending me back to the land of the un-knowing!

Please feel free to enlighten me when you find the time...

Jennice

 

[Svante]

Svante,

I noticed your link to the XDir tool. Is that vertical or horizontal directivity?
It it usual with a phase shift for the woofer, since the tool allows for that (or just a part of the cross-over)?

Jennice

In my mind, it is vertical directivity from two point sources above each other. But you are free to put the point sources beside each other instead, and then it represents the horizontal directivity. It might have been been unpedagogical of me to name the two point sources "woofer" and "tweeter", since they are nothing but point sources. However, it has to do with that I wrote this little program to test the directivity around the crossover frequency of a two-way system, and point sources is a crude approximation of the drivers. In real life, the directivity of the drivers themselves, will affect the overall directivity too.

The reason for the phase shifts is the crossover. You might be familiar that a "schoolbook" Butterworth (odd order) or Linkwitz-riley (even order) filter has a phase shift of n*90 degrees between the branches (for all frequencies, actually). So, a first order filter has the woofer lagging 90 degrees, a second order; 180 degrees etc.

 

[rabbitz]

Sooooooo Sorry Jennice.

Didn't mean to confuse you. PI is TT as in Circle Area = TTr².

Easier to think of it as:
4PI is a full sphere in open space with sound radiating in all directions.
2PI is half a sphere against a wall with sound radiating outwards.

Hope that makes it clearer. It was much clearer in my head before I typed it.

 

[Jennice]

Svante,

I think I'm beginning to understand the idea of your little program. Should the distance between the drivers be from centre to centre, or edge-edge, since the radiating field isn't only from the centre of the membrane?


Rabbitz,

Please confirm if I got your description right:

4Pi space (the Pi as in 3.14....., and calcutated in radians) describes a speaker which is located away from walls and radiates in all directions (4Pi). Thus, the sound radiated backwards is "lost".
2Pi space then describes a speaker close to a rear wall, where the sound - which would normally radiate backwards - is reflected by the wall, and send forward.

The baffle step then represents the marginal frequency (transition range), where the driver unit changes from "seeing" the speaker baffle as the reflecting surface (2Pi), to seeing nothing (4Pi).

Did I get this right?

If so, Rod Elliot has a BS frequency calculation that gives different results (metric SI units) than others in here calculate. What's your formula and tool.