Hi everyone,
Something I'm a little curious about is stepped baffle speakers. These are speakers where the tweeter (and sometimes more) is recessed from the woofer in an otherwise conventional design. That is, not in a wave guide or horn. I often see these in two way designs, Troels Gravesen has a couple of such kits as well, like this one:
When I first saw them I thought "oh, well this must be to create a perfect impulse response, like Thiel" but these speakers often lack this particular feature. So, if not for this, I wondered "well, maybe it's so they can use a certain order filter more easily" or "maybe it's so the off-axis response is consistent?"
So I wonder, at a very high level, what is the process to think about designing a speaker with a stepped baffle?
1. Decide stepped baffles are cool
2. Build the cabinet
3. build the crossover
or is it more:
1. Decide to use (as an example) 4th order LR crossover
2. Pick drivers
3. Find acoustic offsets
4. Design cabinet for ideal phase matching
Or is it more about using 2nd order filters AND keeping the drivers in positive polarity??
Thank you for any insights.
Something I'm a little curious about is stepped baffle speakers. These are speakers where the tweeter (and sometimes more) is recessed from the woofer in an otherwise conventional design. That is, not in a wave guide or horn. I often see these in two way designs, Troels Gravesen has a couple of such kits as well, like this one:
When I first saw them I thought "oh, well this must be to create a perfect impulse response, like Thiel" but these speakers often lack this particular feature. So, if not for this, I wondered "well, maybe it's so they can use a certain order filter more easily" or "maybe it's so the off-axis response is consistent?"
So I wonder, at a very high level, what is the process to think about designing a speaker with a stepped baffle?
1. Decide stepped baffles are cool
2. Build the cabinet
3. build the crossover
or is it more:
1. Decide to use (as an example) 4th order LR crossover
2. Pick drivers
3. Find acoustic offsets
4. Design cabinet for ideal phase matching
Or is it more about using 2nd order filters AND keeping the drivers in positive polarity??
Thank you for any insights.
Last edited:
AFAIK it is simply to align the drivers' so-called acoustic centers (point source origin) to be of equal distance to the listener's ears. Offset detetmined first, everything else follow. The main benefit is a more "holographic" 3D sound beyond what can be achieved by merely aligning phase over the crossover region.
(For example, SB Acoustics published their Satori TW and MW offset to be 32mm; XO frequency and design then had to be made for it.)
(For example, SB Acoustics published their Satori TW and MW offset to be 32mm; XO frequency and design then had to be made for it.)
Last edited:
He writes :
So, it's to keep both drivers in positive polarity, despite probably not achieving an ideal impulse response?
So, it's to keep both drivers in positive polarity, despite probably not achieving an ideal impulse response?
You are correct, it shouldn't force a stepped baffle.....in most cases.
Folks can get obsessed with a single measurement at a particular point in space. In that case, a designer can justify this approach and reconcile the unintended consequences and construction complications.
If using lower-order filters you can find yourself making all sorts of physical modifications to make the scheme work. 🙂
The Dunlavy speakers pop to mind as the best example of this.
Dave.
@Davey
I thought Dunlavy attempted a phase perfect design, and the low-order filters were then required, and then the stepped baffle was also forced? I'm very curious about the order in which these designs are thought about.
BTW, I think stepped baffle speakers are very cool looking. Maybe the earliest I remember is a powered Revox 3-way.
I thought Dunlavy attempted a phase perfect design, and the low-order filters were then required, and then the stepped baffle was also forced? I'm very curious about the order in which these designs are thought about.
BTW, I think stepped baffle speakers are very cool looking. Maybe the earliest I remember is a powered Revox 3-way.
@eriksquires Yeah, I think they look sorta cool too.
It's the chicken or the egg design approach.....depending upon how you look at it. The Dunlavy speakers are only "phase-perfect" at a single point in space. Move the microphone up/down and you have a real mess.
Suffice it say, stepped-baffles have trade-offs just like nearly everything else in speaker design.
I hesitate to elaborate further because one of the moderators will come in here and generate a straw man argument from my posts and tell me I'm wrong about something.
Kinda tiring.
Dave.
It's the chicken or the egg design approach.....depending upon how you look at it. The Dunlavy speakers are only "phase-perfect" at a single point in space. Move the microphone up/down and you have a real mess.
Suffice it say, stepped-baffles have trade-offs just like nearly everything else in speaker design.
I hesitate to elaborate further because one of the moderators will come in here and generate a straw man argument from my posts and tell me I'm wrong about something.
Kinda tiring.
Dave.
I doubt that the hard 90 degree step is a good solution though. It will create bad reflections/diffraction effects.
Why not use a waveguide and have additional benefits (less low frequency distorsion, more even energy response)?
You ask for the process for a recessed tweeter design? Fine. As a free diy-er, you might happily test-build three flavors of tweeter recessing:
1. Tweeter recessed by Zero == Plane baffle.
2. Tweeter recessed by the stepped baffle approach, like the way in your initial foto.
3. Tweeter recessed by some kind of a continously altering recessing profile like a bowed structure or like in a waveguide or so.
And then ... perform a wavelet analysis, and analyze it focusing on the diffraction artefacts matching the wave propagation times from the driver centers to hit the recessing baffle structures. I would not be astonished if the clear looser was ... the stepped baffle as the common enough practice.
Philosophic Morale: Common enough practices might not always be best practices.
Last edited:
@Daihedz I don't want to try to agree on what's best, nor do I want to publish a comparison of the performance of various designs. I just was curious what the driving decisions were in the designer's process. Like asking, "why did you do a 2-way speaker?" Answer: Well, it had to be small but have some bass.
So it sounds like the process is something like this:
1. Pick drivers
2. Select crossover based on driver overlap, power handling and matching off-axis dispersion over the crossover points
3. Pick stepped or not based on need to keep positive driver alignment
So it sounds like the process is something like this:
1. Pick drivers
2. Select crossover based on driver overlap, power handling and matching off-axis dispersion over the crossover points
3. Pick stepped or not based on need to keep positive driver alignment
Have a look at the impulse response of a Cabasse Sampan 311, certainly a well-known tweeter-recessed 3-ways in Europe of the 70 - 80. This speaker has a tweeter which is recessed by some 18mm. 18mm equals 50 microseconds of acoustical time lag:
The graph shows the step response of one of my cabasse sampan 311 for 2ms after onset of the (initially negative) step. Highlighted are 50 microseconds. Placing the tweeter on a plane baffle would shift the tweeter peak forward (= to the left) in time by these 50 microseconds.
So, does this recessing really improve the step response? Not in my oppinion.
The graph shows the step response of one of my cabasse sampan 311 for 2ms after onset of the (initially negative) step. Highlighted are 50 microseconds. Placing the tweeter on a plane baffle would shift the tweeter peak forward (= to the left) in time by these 50 microseconds.
So, does this recessing really improve the step response? Not in my oppinion.
Attachments
Last edited:
An Addendum and a Nota Bene:
Moving two point sources by 50microseconds from one to another however has a notable effect in the frequency range:
So by recessing you may tweak the frequency response for best possible linearity. You may perform this excercise rather along with a microphone than with a ruler.
In accordance to the thread title (hopefully this time) this could be a part of the process: Measure the frequency response. If there is a suspicious bump a bit around the xover frequency (up to a range of max. +-1 octave), you may try to notch it out by recessing the tweeter, thus introducing a matching amount of tweeter delay. While doing so, watch for the behavior of the harmonic notches at higher frequencies then. And be aware that for notching out a bump within a system with a xover of 2.5kHz, you would have to recess the tweeter by up to 5cm or worst-case even a bit more. This would be impractical. So this technique might be useful only for systems with quite highish x-over frequencies between mid's and tweeters.
Moving two point sources by 50microseconds from one to another however has a notable effect in the frequency range:
So by recessing you may tweak the frequency response for best possible linearity. You may perform this excercise rather along with a microphone than with a ruler.
In accordance to the thread title (hopefully this time) this could be a part of the process: Measure the frequency response. If there is a suspicious bump a bit around the xover frequency (up to a range of max. +-1 octave), you may try to notch it out by recessing the tweeter, thus introducing a matching amount of tweeter delay. While doing so, watch for the behavior of the harmonic notches at higher frequencies then. And be aware that for notching out a bump within a system with a xover of 2.5kHz, you would have to recess the tweeter by up to 5cm or worst-case even a bit more. This would be impractical. So this technique might be useful only for systems with quite highish x-over frequencies between mid's and tweeters.
Last edited:
Sorry, I forgot, and it's past edit time - same goes for the reversed situation, of course: If you have a suspicious bump or notch a bit around the xover frequency ...
I don't want to get anyone bogged down in dry pdf reports, but the first one by Jeff Bagby is quite short, and the other Nugget Speaker is a German 2-way design from 20 years ago and discusses all kinds of cabinet and XO stuff (in stodgy German-English translation)...
So maybe I should ask this another way. Lets say I find a couple of unremarkable drivers (dome tweeter, cone mid-woofer) I want to make into a 2-way box. Normally my process is to measure the drivers in the box and then pick a crossover, in these cases often a 3rd order LP + 2nd order HP with an inverted tweeter often work out really well in terms of phase matching across the pass band due to the early arrival of the tweeter.
The often-accepted limitations here are usually that the speaker sounds much worse above tweeter axis.
In this case, due to excess late night partying I decide no, I'm not going to let the box dictate my crossover. I'm going to insist on using a 4th order (electroacoustic), symmetrical crossover, no matter what, so I have better vertical response profile, and I want to keep my drivers in positive polarity relative to the input.
In my second case, I'm probably going to be forced to recess the tweeter, no?
The often-accepted limitations here are usually that the speaker sounds much worse above tweeter axis.
In this case, due to excess late night partying I decide no, I'm not going to let the box dictate my crossover. I'm going to insist on using a 4th order (electroacoustic), symmetrical crossover, no matter what, so I have better vertical response profile, and I want to keep my drivers in positive polarity relative to the input.
In my second case, I'm probably going to be forced to recess the tweeter, no?