Build Thread -- Newbie modeling an hframe - open back mid

[brianpowers27]

I am totally new to the open baffle modeling process. I have gained a very rudimentry understand about what needs to happen by reading from John K, Martin King, Linkqitz, etc...

I am setting out to build a first attempt at an open baffle. My budget is tight and I want to go all passive. I really like the idea of minimizing radiated energy into the room via di-pole cancellation.


I own the following drivers for one pair of speakers:

2x alesis 5.25 http://www.parts-express.com/pe/sho...tnumber=299-160 (Nearfield FRD Attached.)

4x no-name 12" woofers (qts = 1.2) http://www.parts-express.com/pe/sho...tnumber=292-320

4xDayton ND20fb4

I really would like to find the best tool for the modeling process. AJINFLA explained that the X-Baffle and the Response Modeler do not account for the H-Frame specific properties. I assume the H-frame has some level of pressurization that is independent from flat baffle designs(Pressure cavity?). I believe that both programs do account for the di-pole peak. THe response modeler 2.1 was just released today with this mod. (Thanks Jeff!)

I was thinking a nice starting point would be: (MDF)
--woofer baffle of 14"x30"x.75"
--two wings 8"x30"x.75".
--The top and bottom of the hfram would be covered 15.5"x8"x.75"
--Open Baffle for mid 12"x24"x.75". REsts in the middle of the h-frame
--Tweeter (Dayton ND20fb 3500hz) at 16" on open baffle (Rear firing tweeter is mounted on additional back baffle piece.) wired in series with rear firing out of phase.
--Midrange at 12" on open baffle
--30degree angle brace from h-fram rear to open baffle rear.(pine 1"x4")

[ThomasW]

Here's a radical thought....

Build a "H" for the 12"s, your dimensions look good to start. I'd make the front baffle out of doubled 3/4" MDF.

Next experiment with different length sides. Glue 1"X2" or 2"X2" blocks on the sides of the front baffle. Screw the different length sides to these blocks for a quick and easy way to change the setup.

Now listen/measure to what changing the depth of the sides does to the FR.

MDF is cheap and there's no substitute for some hands on experience.

After that play with the mid-woofers in the same manner.

IMO you'll actually learn more from this exercise than mastering all the modeling programs available.

Obviously just my $0.02....

[brianpowers27]

That is a good idea. I could create two wings for each side. I could start with the wings fully overlapped and expand them until I find a good match.

The biggest question in my mind is not how the wings will change the sound but how I will reach the best integration between the two drivers. I have considered shooting for a 300hz 2nd order xo point. It seems as if the mids will be happy to being a nice 6db rolloff with a 12" baffle. I have done some preliminary modling to find this information. I will next do a proof of concept.

[Evil Twin]

Linkwitz's site covers calculations regarding H frames and their additional resonance effects-

This page should be helpful....


LInkwitz on H frame and U frame baffles

[Saurav]

I found MJK's MathCAD worksheets pretty accurate when I was modeling / measuring my OB woofer. I'd assume his H/U frame worksheets would be similarly accurate. Unfortunately, he may have pulled them from his website. It used to be a $25 purchase for a year's license, and you had a bit of a learning curve with the free MathCAD Explorer or whatever the tool was called, but after that, it was very useful.

Not sure that really helps I think his site has a white paper comparing H/U/OB woofer setups, you may find some guidelines or starting points there. But you've probably gone through that already.

[brianpowers27]

Linkwitz's site covers calculations regarding H frames and their additional resonance effects-

This page should be helpful....


LInkwitz on H frame and U frame baffles

This is a great page, I have looked over it before. I gained my understanding of bi-pole vs cardoid here.

I read somewhere and I am having trouble remembering where... I believe there is a correlation between a 1/4 wavelength and the depth of the h-frame. I would like to understand a little more about this.

It is my understanding that a hfram is essentially a open baffle with a longer path length. The u-frame creates the asymmetrical cardoid response...

[Paul Ebert]

Interesting project, Brian. Too bad the Alesis drivers are out of stock. I wonder if they'll get any more in.

[brianpowers27]

Interesting project, Brian. Too bad the Alesis drivers are out of stock. I wonder if they'll get any more in.

I wouldn't lose any sleep over this driver selection. I am only using it because this is what I have on hand. I am guessing that the tbw5-704 would make an equal if not better replacement. Since I am crossing high, a smooth poly cone fits the bill. The peerless xls are also on closeout.

[brianpowers27]

I found MJK's MathCAD worksheets pretty accurate when I was modeling / measuring my OB woofer. I'd assume his H/U frame worksheets would be similarly accurate. Unfortunately, he may have pulled them from his website. It used to be a $25 purchase for a year's license, and you had a bit of a learning curve with the free MathCAD Explorer or whatever the tool was called, but after that, it was very useful.

It appears the Mathcad has gone into obsolescence :cry:

[JonMarsh]

This is a great page, I have looked over it before. I gained my understanding of bi-pole vs cardoid here.

I read somewhere and I am having trouble remembering where... I believe there is a correlation between a 1/4 wavelength and the depth of the h-frame. I would like to understand a little more about this.

It is my understanding that a hfram is essentially a open baffle with a longer path length. The u-frame creates the assymetrical cardoid response...

There are additional resonance effects caused by the H structure, compared with a flat baffle; not as severe as the W baffle like the PHoenix dipole woofer, but something to take into account; for that reason, it's better to use an H section below 125 Hz typically, IIRC. With a crossover in that range, not so much to worry about, but still some peaking in response.

Sometimes, you know, it's easier to make sawdust and measure the results than to worry too much about modeling everything in advance.

Last question, why H baffles? If you're using a 5" midrange with dual 12's, I'd suggest working towards something like this, using Xbaffle or one of the other tools.....



Jamo R 907

[brianpowers27]

Last question, why H baffles? If you're using a 5" midrange with dual 12's, I'd suggest working towards something like this, using Xbaffle or one of the other tools.....



Jamo R 907

I figured that an h-baffle would give a little extra support to the bass section, due to the added baffle width. I could seek other ways to get the width, or just keep it fairly skinny. I also figured the H part of the baffle would provide some structural integrity. I will consider what it would mean to make it flat...

[Rudolf]

Brian,
with those 12" woofers (qts = 1.2) you could perfectly follow MJKs H-baffle proposals for the Eminence Alpha 15.

His Comparison of the Bass Performance of Passive Open Baffle, U
Frame, and H Frame Speakers should answer some of your questions as well as his Goldwood projects. There are no other resonances to consider than the 1/4 wavelength of the H-baffle.

[brianpowers27]

Rudolf,

Thank you for that information. I have skimmed many articles but I don't always retain the details. I remember reading about the 1/4 wavelength now that you have pointed me to the article by King. I will need to look at my application and review those notes before I do a mock up.

[JonMarsh]

Brian, to relate this to your application, and one I did a few years back (Arvo Part), the resonance from the H-Frame (a relatively shallow one in the first version, a tapered one in the second) led me to choose a crossover point no higher than 180 Hz passive to the midrange unit from two 10's or two 12's (depending on the version). The rest of the baffle was a flat baffle, designed with an asymmetric layout for flattest on axis response; in retrospect, I would do that differently, with a symmetric layout, and use the crossover to compensate for the other effects; then the forward power response would be more uniform.

On a largish baffle, a 150-180 Hz crossover shouldn't be hard for the 5". OTOH, it may be running out of steam- use Linkwitz's SPL Max spreadsheet to get an idea of how low it can go with a dipole of a specific size, for the given driver Xmax.

OTOH, what about your 12's? Besides the H baffle resonance, what is their inductance like? Many of the high Qts low cost drivers don't have very low inductance, and the rise in midrange distortion due to inductivity modulation suggests not using them much above 100 Hz. For that reason, it may be worthwhile to consider a lower Q, more modern driver, and use a simple passive inline equalizer between your preamp and power amp, as I did in one version of the Arvo (you should be able to search the past threads for info, if interested). Or PM me if you want an example schematic.

[Saurav]

There's also the rule of thumb of keeping the baffle width no more than 2.2x the cone diameter. As far as I understand it, this helps keep the power response smooth - the baffle provides dipole loading up to a higher frequency, above which the driver's increasing directivity takes over. Of course, this reduces the midrange driver's low frequency output, so you either need to move the crossover point higher (which requires a better woofer and will eat into c-t-c spacing between the woofer and mid), use EQ to boost the midrange low end (requires a high xmax driver), ... or ignore this guideline and go with a wider baffle.

[Dennis H]

Rudolf has a good thread with sims showing why narrow baffles are best for dipoles if you want even power response. Short version, the dipole peak happens where the baffle width equals a wavelength and you get funny off-axis response if you cross much higher than that. Above the dipole peak, the off-axis response can be louder than the on-axis while at lower frequencies the off axis will be quieter. I really like his design proposal for a 3-way that tapers from wide at the bottom down to as narrow as you can make it at the top -- almost a triangle.

http://www.audiocircle.com/circles/index.php?topic=64048.0

[brianpowers27]

Rudolf has a good thread with sims showing why narrow baffles are best for dipoles if you want even power response. Short version, the dipole peak happens where the baffle width equals a wavelength and you get funny off-axis response if you cross much higher than that. I really like his design proposal for a 3-way that tapers from wide at the bottom down to as narrow as you can get it at the top.

http://www.audiocircle.com/circles/i...?topic=64048.0

This is a good idea. I saw this one earlier today.

http://www.audiocircle.com/circles/index.php?topic=49940.20

Image not available

[Dennis H]

I'd do something similar except with a flat baffle, at least for the mid and tweeter. U and H frames are for low frequencies and all they really do is make it look smaller from the front.

[brianpowers27]

I'd do something similar except with a flat baffle, at least for the mid and tweeter. U and H frames are for low frequencies and all they really do is make it look smaller from the front.

Perhaps the main baffle could be a pyramid with all drivers attached. The wings could extend in a pyramid like shape to about where the midrange starts.

I would like to model something like this in Sketchup, when I get home tonight.

[Saurav]

My attempt (and it was a challenge building even this):

Image not available

I've switched to a different tweeter with a smaller waveguide, so if/when I re-build the baffles, I'll probably make the upper baffle a little narrower.

The lower baffle isn't wide enough to get flat response to the 80Hz XO to the sub. I'm using an active XO and a driver with tons of available Xmax (AE OB12), so I solved this problem by throwing in EQ (a shelving lowpass filter) to get the response I wanted. This may not be practical with other drivers and/or a passive XO.

[Dennis H]

Brian, like Thomas and Jon said, you'll probably need to experiment with the wings to get them right. Cardboard wings taped onto the baffle will let you make quick and easy changes.

[brianpowers27]

Brian, like Thomas and Jon said, you'll probably need to experiment with the wings to get them right. Cardboard wings taped onto the baffle will let you make quick and easy changes.

WHy Didn't I think of that? Cardboard is even better!

[brianpowers27]

I've switched to a different tweeter with a smaller waveguide, so if/when I re-build the baffles, I'll probably make the upper baffle a little narrower.

The lower baffle isn't wide enough to get flat response to the 80Hz XO to the sub. I'm using an active XO and a driver with tons of available Xmax (AE OB12), so I solved this problem by throwing in EQ (a shelving lowpass filter) to get the response I wanted. This may not be practical with other drivers and/or a passive XO.

It seems that the waveguide would help to align the AC. The stepped woofer also helps.

What woofer did you use? QTS? Is it a pure open baffle, or a U?

[Saurav]

The old waveguide put the tweeter AC about 3" behind the midrange, and that was beyond my crossover skills. The new waveguide is a lot smaller / shallower (and loses pattern control about an octave higher), but the tweeter is roughly the same distance from the baffle as the midrange. I'm waiting for good weather to take it outside and measure, but I'm hoping the impulses will be much closer together now.

The woofer is an AE OB12. You won't find it on their website, but there's a thread on here where it's discussed. It's basically built from the IB12/15 parts, with a different dustcap and I think a different cone, or different treatment on the cone, or something.

I went with a flat baffle, there are no wings on the back.

[brianpowers27]

Brian, to relate this to your application, and one I did a few years back (Arvo Part), the resonance from the H-Frame (a relatively shallow one in the first version, a tapered one in the second) led me to choose a crossover point no higher than 180 Hz passive to the midrange unit from two 10's or two 12's (depending on the version). The rest of the baffle was a flat baffle, designed with an asymmetric layout for flattest on axis response; in retrospect, I would do that differently, with a symmetric layout, and use the crossover to compensate for the other effects; then the forward power response would be more uniform.

I assume when you say other effects, you mean diffraction?

On a largish baffle, a 150-180 Hz crossover shouldn't be hard for the 5". OTOH, it may be running out of steam- use Linkwitz's SPL Max spreadsheet to get an idea of how low it can go with a dipole of a specific size, for the given driver Xmax.

I may consider a different point. At this point I am just guessing on the figures...

OTOH, what about your 12's? Besides the H baffle resonance, what is their inductance like? Many of the high Qts low cost drivers don't have very low inductance, and the rise in midrange distortion due to inductivity modulation suggests not using them much above 100 Hz. For that reason, it may be worthwhile to consider a lower Q, more modern driver, and use a simple passive inline equalizer between your preamp and power amp, as I did in one version of the Arvo (you should be able to search the past threads for info, if interested). Or PM me if you want an example schematic.

They are about the cheapest 12s one can buy. I imagine that the total project cost (drivers, xo, wood) will cost under $150. I am on a tight budget and just wanted something to play around with. If I like the way this turns out I will consider using some nice drivers. I can imagine using something like the Goldwood 18" PA or a more expensive Dayton 15" IB

Thanks for the offer on the inline equalizer. That may not work too well given the digital connections I have between source and amp. I would consider other types of passive eq (BSC,etc..)

[Evil Twin]

WHy Didn't I think of that? Cardboard is even better!

I suggest a sturdier compromise, something like 1/4" MDF from Home Depot. Easy to cut, but won't flap in the breeze, so to speak.

[brianpowers27]

Here is what I modeled last night in 3d. Nothing built yet. I tried other variations with smaller wings and had a hard time making it look right to my eye. I may try modeling again tonight.

Height = 48"
Top width = 6"
Bottom = 18" wide + wings
Wings = 9.5"

[JonMarsh]

Think about this thing, of using an H Frame across the whole bandwidth- you've just created a major diffraction generator for the midrange and tweeter drivers that will create a very high level (not attenuated) reflection that will add with the primary sound with a time delay and create comb filtering in the frequency response. You sure you really want to do that?

[Rudolf]

Jon is right with the reflections around the tweeeter and mid driver. You should restrict wings to the woofer area. What it really means to follow that "baffle width no more than 2.2x the cone diameter" rule is shown in my personal testbed. It looks like this:

Image not available

The tweeter on top of the 3" fullranger is hardly visible. At this time it is obviously not made for good looks, but for functionality only.

And this is a proposal by "ultrachrome":

Image not available

I am not sure whether you are doing right with starting into dipoles with a 3-way system combining flat baffle and H-frame. There are no tools to simulate it and without measurement you might have a hard time to control the development process.

[brianpowers27]

Think about this thing, of using an H Frame across the whole bandwidth- you've just created a major diffraction generator for the midrange and tweeter drivers that will create a very high level (not attenuated) reflection that will add with the primary sound with a time delay and create comb filtering in the frequency response. You sure you really want to do that?

You bring up some excellent points. The overall appearance is important so I am going to work toward a compromise between form and function. I will go back to the 3d drawing board later tonight.

[brianpowers27]

Image not available

The tweeter on top of the 3" fullranger is hardly visible. At this time it is obviously not made for good looks, but for functionality only.

And this is a proposal by "ultrachrome":

Image not available

I actually like the way these look. I think that if I reshape the top/sides a little and keep the general concept I may find something that satisfies my aesthetic needs.

I am not sure whether you are doing right with starting into dipoles with a 3-way system combining flat baffle and H-frame. There are no tools to simulate it and without measurement you might have a hard time to control the development process.

Duly noted. I am a glutton for punishment. I have a fully calibrated measurement system and plan to use it. This is also not my first 3 way xo build.

[brianpowers27]

Question: When you say that the baffle width shouldn't exceed 2.2x cone diameter, what are you measuring as width?

[JonMarsh]

You should use the radiating diaphragm diameter.

[brianpowers27]

You should use the radiating diaphragm diameter.

When you say baffle width, I assume that you mean the total distance for one side of the baffle.

Baffle width + 2x depth of wings...

[Rudolf]

I have a fully calibrated measurement system and plan to use it. This is also not my first 3 way xo build.

Sorry, I did not realize that. My apology. ops:

When you say that the baffle width shouldn't exceed 2.2x cone diameter, what are you measuring as width?

May be I should explain the concept. The "rules" will be explained by the way.

My design is along the philosophy of Linkwitz and Kreskovsky. All dipole drivers should operate only up to the first dipole peak of their baffle and then be crossed to the next driver on a narrower (part of the) baffle. The crossover will be easier if the first dipole null is attenuated in some way. Therefore the driver should become directive above the dipole peak. This condition is met if the effective baffle width does not exceed ~ 2x effective cone diameter. "Effective" in both cases means the "working" dimension, which will be different from the physical dimension.

The working width of the baffle is the wavelength of it´s dipole peak at the driver position. You can find it quite exactly by simulation with EDGE. In almost every case the effective width of the baffle will be larger than its physical width. Kreskovsky explains it in some more depth in his Design Considerations.
Following this strategy all drivers will operate as true dipoles with a (rather) uniform polar radiation pattern resulting in constant directivity in the horizontal plane.

The down side of this approach: All drivers operate for a large part in the 6 dB/oct dipole loss area and have to be EQed 9-15 dB along their individual passband (for a 3-way-system). The gap between necessary power and limited diameter makes heavy demands most notably on the midrange driver.
Mounting plates of dome tweeters should be no wider than 5-6 cm and the tweeter depth should not exceed 1.5-2 cm to fulfill the baffle width condition up to 5 kHz. Planars will brake the rules in every case. Edit: I have been proved that the last sentence (about planars) is wrong. Please forget about it.

It is almost inevitable that the tweeter (pair) has to operate down to almost 1500 Hz. Otherwise the midrange driver can´t exceed 3" which makes it practically unuseable below 500 Hz.

You should know about these grueling demands before investing in this kind of uncompromised dipole. I should mention that even Linkwitz and Kreskovsky had to make compromises in the mid-to-tweeter area with their designs.

Edit: Forgot to mention that this design is strictly symmetrical. No moving of any drivers to the side of the baffle.

[Evil Twin]

Baffle width is total width of baffle- like diameter of circle, versus radius of circle. Plus 2X depth of wings.

One of the desirable targets with any speaker design is to design both for flat axial response and similar power response- while the former is straightforward regardless of other choices, the latter requires more thought and planning, and considering the behavior of the drivers as well as their behavior on the intended baffle or the specific enclosure. This is an area where matters require more effort, as a dipole format complicates the radiation pattern as a function of frequency.

[brianpowers27]

The down side of this approach: All drivers operate for a large part in the 6 dB/oct dipole loss area and have to be EQed 9-15 dB along their individual passband (for a 3-way-system). The gap between necessary power and limited diameter makes heavy demands most notably on the midrange driver.
Mounting plates of dome tweeters should be no wider than 5-6 cm and the tweeter depth should not exceed 1.5-2 cm to fulfill the baffle width condition up to 5 kHz. Planars will brake the rules in every case.

It is almost inevitable that the tweeter (pair) has to operate down to almost 1500 Hz. Otherwise the midrange driver can´t exceed 3" which makes it practically unuseable below 500 Hz.

This may be an area where this design deviates from the ideal Dipole. I am set on using these specific drivers and the tweeter is unlikely to cross well beloew 3k. The tweeter's dome diameter is .75". (1.9cm) It sounds as if I will likely lose some of the true dipole's uniformity.

I may try a few different configurations. I like the idea of using a h-frame for the woofers and keeping the top open. (Similar to the Orions) I do like the tapering idea but would like to find a way to taper with a nicer look. It seems as though the offset between the woofer and the mid would be slightly less critical than the offset from the mid to the tweeter.

I will model some of my ideas in Sketchup tonight.

[brianpowers27]

I assume that the effective working baffle width is the same as effective in the equation below.

U_and_H_Frames.pdf
Leffective = 7.5” + 0.6 x reffective
Leffective = 7.5” + 0.6 x 9.0”
Leffective = 12.9” = 0.328 m
f1/4 = c / (4 x Leffective)
f1/4 = 344 m/sec / (4 x 0.328 m)
f1/4 = 262 Hz

In practicality one could use the h-fram for the woofer to gain a slight overall width advantage compared to the open baffle. The h-frame would also carry the h-frame transmission line peak.

Maybe I could substitute an m-frame for the h-frame...

http://www.diyaudio.com/forums/showthread/t-95185.html

[brianpowers27]

THe thought process is starting to come together. I understand now the tradeoff between efficiency and q when using a flat vs h-frame. The m-frame seems unknown at this point. I blieve that I might use something like this.

14" wide with 2" wings from the floor to the top of the upper woofer. The baffle will taper to the top with the mid and the woofer centered.

The tiny wings should strike a nice compromise between the h-frame and the open baffle. The resonance peak will be near 520hz, which is way outside of the xo range.

[Dennis H]

In practicality one could use the h-fram for the woofer to gain a slight overall width advantage compared to the open baffle. The h-frame would also carry the h-frame transmission line peak.

I was just fiddling with adjusting the t-line peak with wing size. It works out that making the wings approx. 1/6 baffle width for a U frame or 1/3 baffle width for an H-frame will make the t-line peak fall on the first dipole null, an octave above the dipole peak. It seems like the help in killing that null could be useful to extend the usable frequency range or at least to make XO design easier. Of course, that's just another rule of thumb and modeling in Martin's worksheet and building the real thing are the next steps

[Rudolf]

Brian,
Dennis is making a good point. Probably you should take a look at the difference between effective baffle width and effective TML length (Leffective).

Image not available

Leffective (straight blue line) is measured from the baffle plain of the H frame to the effective front or back end of the H frame (blue circle line).

The effective baffle width is measured from the center of the driver on the rear of the baffle plain to the front of the baffle plain (but not the center of the driver again!!).

Leffective x 4 is the wavelength of the resonance of the H frame cavity, the effective baffle width x 2 is the wavelength of the dipole peak frequency.

You can easily imagine how both length differ with different H frame geometries.

[brianpowers27]

I was just fiddling with adjusting the t-line peak with wing size. It works out that making the wings approx. 1/6 baffle width for a U frame or 1/3 baffle width for an H-frame will make the t-line peak fall on the first dipole null, an octave above the dipole peak. It seems like the help in killing that null could be useful to extend the usable frequency range or at least to make XO design easier.

I was assuming a baffle width (13")and a baffle height of (26").

Please correct me if I am wrong...
Leffective = 2” + 0.6 x reffective
Leffective = 2” + 0.6 x sqrt((13"x26")/3.14)
Leffective = 2" + (.6 * 10.37)
f1/4 = c / (4 x 8.22)
f1/4 = 13550 in/sec / (4 x 8.22 m)
f1/4 = 411Hz

It doesn't seem as though one could construct any ratio between width and depth without accounting for height. I assume that your ratio was based upon a sqaure ratio of width/height.

Of course, that's just another rule of thumb and modeling in Martin's worksheet and building the real thing are the next steps

I don't have access to Mathcad or his worksheets.

[brianpowers27]

Here is one mock up.

[Dennis H]

Rudolf, I'm not sure I buy that definition of Leffective for defining the TL peak, at least not for very shallow wings. If the baffle were 24" wide and the wings were 1", Leffective would be a bit over 12", but we know it should be much closer to 1" than 12".

Brian, Google Mathcad Explorer. It's a free viewer. The worksheet is on the FRD page.

[brianpowers27]

Here is version 2. For scale, my coffee table subwoofer (26" height is right next to the design.)