Dipoles for Dummies

[augerpro]

Hi all. I wanted to start a thread to hopefully start a discussion around the finer points of Dipole design since a number of us in this forum are starting to build some. Some background links that I've found invaluable:

http://www.linkwitzlab.com/

http://www.geocities.com/kreskovs/John1.html

http://www.musicanddesign.com/tech.html

Now a couple questions:
1) What are the major resonances that limit how high a U frame can go? Linkwitz mentions there will be a resonance at a frequency F where the distance from the cone to the rear opening is a quarter wavelength of F. Is this the major resonance?

2) Using Boxnotes I wondered if this resonance could be shown. I think "Driver to rear wall" is it, but it's not exactly the same value, perhaps because of the way it calculates the cone position with the cone depth value. Now because there is still an impedance change at the rear opening, there still be a reflection. So I'm thinking that all the resonances in boxnotes are still valid to take into account. What do you guys think?

3) Floor reflection. There is some info on this at John Kreskovsky site. Apparently the dipole roll off is not as bad as theory says (which assumes free space) and the floor will cause the woofer to radiate in to 2pi space progressively as the frequency lowers. John mentions generally where this change occurs according to woofer placement, but no info on magnitude. I found one post at another forum that said where the distance from the woofer to the floor is 1/2 wavelength of a frequency F, the critical frequency, the amplitude will be up 3dB over freespace calculation, and 6dB up one octave below that. So if the distance from woofer (center?) to floor is 20" then the response is up 3dB at 337hz and 6dB up at 168.5hz. What do you guys think? Granted we could say the same thing about any typical box speaker, so all things being equal, the dipole may still sound bass shy relative to the box speaker.

4)Mid/woofer crossover point considering baffle width. This mystifies me more than it probably should, and will probably be clearer after I do some real measurements, but what advice can some one give on this? How do you deal with the dipole peak? Do you keep the crossover above the baffle step? AND the dipole peak (tough to do that one)?

5)Power response. John K has some excellent info on this at his site. Basically just because a dipole can offer an even polar response, does not mean that power response will be even. Lots to dig into on this one.

6)Good sim programs?
-Edge is cool, but can't do U frames or angled, folded back wings without screwing up the baffle step. I suppose if you only look at bass response you could model fairly well with it. Doesn't include teh floor.
-John K has a cool program in his ABCDipole guide that considers driver Q and Fs, does H, U, and flat baffle, and directivity, but nothing on sensitivity or baffle diffraction.
-Soundeasy does just H or U, no flat baffle. Can do sensitivity, directivity, and baffle diffraction. The results don't agree real close with John K's so I don't know what to think. And I can't reduce a U baffle to a flat baffle correctly, so I'm thinking Bohdan needs to look at this model.
-I tried xlbaffle but I'm just not seeing where this an improvement over any of the others.

7)Linkwitz mentions that the U baffle will operate *I believe* lower that predicted, due to end correction like you see of a port in a vented box.

I hope all of you find this baffle and we can get some of these "details" put in to one thread for future use.

[Dennis H]

The takeaway I got from SL's writing is the theoretical models can't really predict everything that's going on (e.g. the FR of the back side of the driver with its basket and magnet). Mock up a baffle with cardboard or whatever and the real drivers and measure the results -- faster and more accurate than running a bunch of sims that may or may not have anything to do with reality.

[Saurav]

6)Good sim programs?

There's a U/H frame MathCAD worksheet on the FRD Consortium page. I think MJK and John K. are listed as co-authors/developers. The free MathCAD 'viewer' required to use it is no longer available from the manufacturer, but I found a copy pretty easily through a search.

xlbaffle - My take-away from reading about it is that it gets you into the ballpark, nothing more. It probably doesn't do anything that all the others put together don't do. I haven't tried the ABC guide, and I don't have SoundEasy. Do either include the room gain?

One more program that I found useful is Linkwitz's spl_max1 spreadsheet.

[dlneubec]

Brandon,

I'm not as far as you into the dipole research yet, and my project will have a dipole mid only, but I've had many of these same questions and thoughts pass through my feeble mind. 8O I hope some of the OB/dipole experts chime in here.

Too bad we're not trying to design the same type of speaker or we could somehow pool resources! It might be less expensive that way to experiment.

[augerpro]

Mock up a baffle with cardboard or whatever and the real drivers and measure the results -- faster and more accurate than running a bunch of sims that may or may not have anything to do with reality.

Yeah, I'll be making a mock up tomorrow. I'll do various measurments and two lengths of U frame so I can hopefully extrapolate how dimensions in between behave and see what the reality is. Of course it's only the reality for that mockup. It would be nice to get some correlation through various theories so I don't have to do everything empirically all the time.

[augerpro]

There's a U/H frame MathCAD worksheet on the FRD Consortium page. I think MJK and John K. are listed as co-authors/developers. The free MathCAD 'viewer' required to use it is no longer available from the manufacturer, but I found a copy pretty easily through a search.

My impression was that you licensed for the worksheet for a year after you paid a fee? That kind of kept me away. Although if it's worth it I'll pay. ABCDipole was money well spent.

I haven't tried the ABC guide, and I don't have SoundEasy. Do either include the room gain?

Nope

[augerpro]

Brandon,

I'm not as far as you into the dipole research yet, and my project will have a dipole mid only, but I've had many of these same questions and thoughts pass through my feeble mind. 8O I hope some of the OB/dipole experts chime in here.

Too bad we're not trying to design the same type of speaker or we could somehow pool resources! It might be less expensive that way to experiment.

Tell me about it. Trying to justify spending $600 for two woofers (B&C and Lambda) just for testing is kind of hard to swallow. I sure would like to see some tests like I do on the Lambda and B&C mid for comparison. And even the PHL since it seems to be decent performer, unlike the typical hifi HE drivers. If you have any good ideas let me know, I'd love to get some measurements of these interesting drivers done for the community.

[Saurav]

Oh, one more useful link:

http://www.quarter-wave.com/OBs/OB_Theory.html

Going through his site, it looks like there are other OB-related MathCAD worksheets available too.

http://www.quarter-wave.com/Models/MathCad_Models.html

Again, no idea how much this overlaps with ABCDipole or any of the other tools.

I'd been thinking of buying some of these tools. Does it make sense for me to buy MJK's worksheets? Do you already have ABCDipole? Maybe we could enter the same numbers into both tools, compare their results, and also compare how well they correlate with your measurements. I have an ECM8000 and Speaker Workshop, but I'm a complete novice when it comes to measuring anything, so I won't be much help there. I also have an 8 month old daughter, so my free time is a little limited

Mid/woofer crossover point considering baffle width.

One hint I got from MJK's papers is to 'underlap' the XO, i.e. set the low-pass lower and the high-pass higher. You can use this to bring the peak down, and 'extend' the response of both drivers in either direction, and get better integration.

This is probably easier to do with a passive XO. I'm trying something like this with my CX3400 by using the high and low outputs for the upper panel and woofer cabinet (which allows me to set different XO points). But that means I need a 3-way XO to get 2-way output.

Edit: Responses to your other posts...

My impression was that you licensed for the worksheet for a year after you paid a fee?

From his website, yes. I found that one worksheet on the FRD Consortium page, so I assumed he'd made that available for free. I could be wrong. $25 for a year is fine with me, I don't do this anywhere near as often as some of you guys

I have a pair of Audax PR170M0s. I know they're not on the list of drivers you're considering, and as Jeff pointed out in the other thread, the 0.5mm Xmax really limits what you can do. But if you're interested in measuring them, I could send them to you. I might end up using them in my next set of speakers, so I have a vested interest in seeing them tested

[augerpro]

Again, no idea how much this overlaps with ABCDipole or any of the other tools.

I'd been thinking of buying some of these tools. Does it make sense for me to buy MJK's worksheets? Do you already have ABCDipole? Maybe we could enter the same numbers into both tools, compare their results, and also compare how well they correlate with your measurements.

I have ABCDipole. It's very good. I believe MJK's spreadsheet is newer, because John K. was helping MJK develop some of it in thread I was reading. I know they were comparing results to see if there was an inconsistency. If I had to choose I would pick ABCDipole simply because John K is the dipole guru and has an incredible amount of info on his site.

One hint I got from MJK's papers is to 'underlap' the XO, i.e. set the low-pass lower and the high-pass higher. You can use this to bring the peak down, and 'extend' the response of both drivers in either direction, and get better integration.

I've considered this. The only thing that makes me leary is the phase agreement which could make for some unpredictable lobing off axis.Although if you can get phase quadrature (like BW3) it will have even power response, not sure what that implies about lobing though.

[augerpro]

I have a pair of Audax PR170M0s. I know they're not on the list of drivers you're considering, and as Jeff pointed out in the other thread, the 0.5mm Xmax really limits what you can do. But if you're interested in measuring them, I could send them to you. I might end up using them in my next set of speakers, so I have a vested interest in seeing them tested

I see them mentioned all the time, so I'd be game to measure them. Won't be until my next round of testing which will probably be in June. I'll PM you when I'm getting ready to test.

[Saurav]

I have ABCDipole. It's very good.

OK, I'll probably pick that up then. At some point I'll have to do some of my own work, since I can only piggy-back on your design work so far

I see them mentioned all the time

Yep. Although if you test them, the resulting intersection of the "measure and simulate" world with the "just listen to it" world may cause a rift in the space-time continuum.

[Deward Hastings]

What Dennis said.

It's pretty amazing that Linkwitz not only pretty much invented the "dynamic driver dipole" as a genre, but managed the old fashioned way (pencil and paper and testing and listening) to find the "sweet spot" for so many elements of the design. ABCDipole does a pretty good job of modeling some of the design details . . . easier perhaps than working through the process as laid out at Linkwitzlab.com . . . and will both make it easier for and provide reassurance to designers looking to use other drivers and who want to try variations on the Linkwitz design. It's a good "what if" tool that will keep you from making a lot of false steps and outright mistakes and will get you as close to a good design as a tape measure and ten minutes with an ORION.

It's pretty clear that ORION performance, or near to it, can now be accomplished with significantly less expensive drivers than those Linkwitz used (drivers that were not available 10 years ago), and there remains the possibility that better drivers will come along someday too. There's always room for improvement.

And nobody seems to have worked out the details of making it all work on a flat baffle . . . yet.

[dlneubec]

Tell me about it. Trying to justify spending $600 for two woofers (B&C and Lambda) just for testing is kind of hard to swallow. I sure would like to see some tests like I do on the Lambda and B&C mid for comparison. And even the PHL since it seems to be decent performer, unlike the typical hifi HE drivers. If you have any good ideas let me know, I'd love to get some measurements of these interesting drivers done for the community.

I'm trying to get to a decision on what drivers to use. For example, the Lambda TD12H or the Lambda TD12X for the woofer? The PHL 1120 or the B&C 6mnd44 for the mids? Not to mention what high sensitivity tweeter to use.

Jeff said he is going to run his xmax simulation on the 6mdn44 to see how it would look in his design as compared to the PHL. If it compares well, I may pick one of them up to test.

[Martijn_H]

The scanspeak Iluminator serie looks interesting for dipole mid, because of their open structure. I'm using Accuton C90 mid with the Mundorf 2340 as a dipole tweeter. Lots of people use a single dome as a tweeter, but the principle of a dipole tweeter is preferable.

When you got the possibility to make an active filter use a very very shallow baffle for the mid and tweeter. This gives you a better of ax response.

Martijn

[John_E_Janowitz]

I have a couple older links to some stuff that Larry Selmer did. He had corresponded quite a bit with Linkwitz back in the day regarding dipole bass. He did lots of work regarding dipole bass. You'll probably find the following page quite interesting:

http://www.geocities.com/larryselmer/DRL.htm

I haven't talked with Larry for a couple years now. I'll see if I can contact him.

John

[augerpro]

Martijn> the Illuminator definitely shows some promise for OB. Don't tell Jed that though or he'll scrap his Tombstones and spend another mortgage payment for new drivers Is that the AMT tweeter? Very interesting...I bet that is a nice combo. I do think teh mirror image concept of radiation is worth striving for, tweeters make that difficult.

John> thank you for the links, I'll be sure to check them out.

[JimS]

Brandon - I'm sure you've already checked it out, but I'd highly recommend Jon's initial Arvo post and the follow-up Isiris post as required reading - it's a lot of pages to wade through and no models/tools, but some very good design decision info buried in there!!

https://www.htguide.com/forum/showthread.php?t=7644

https://www.htguide.com/forum/showthread.php?t=23598&highlight=isiris

Jon appears to be on a "work-related holiday" from the forum right now, but CraigJ and ThomasW have built and extensively listened to both u-baffle and flat baffle versions (although with very different drivers) so maybe they'll chime in with some impressions . . .

I've been listening to the "classic" bi-amped Arvo's (RS28A, dual M8As, dual RS315HFs - see last page in the Arvo thread for details) for a few months now and all I can say is :E :T :E :T ;x(

I'm planning on posting a longer review on my build thread when I get a bit more time, but initial thoughts are:

1) Soundstage is much wider and more enveloping than the previous monos and very, very clear (My RS28As appear to have no issues crossing at 1250Hz with this filter - apparently Jon knows what he's talking about :W )

2) I noticed the largest difference in the bass. Upright just sounds "right", woody, etc. and lower end of electric becomes really clear in most mixes. YMMV in a different room (and yes they are placement sensitive - didn't really sing until almost 4ft from rear wall) - however, my experience leads me to wonder why so many recently are focusing on dipole midrange with monopole bass if they've got the wall space behind!! Of course, I'm a drummer and occasionally tend to notice what the "bass drum accompaniment" is doing :P

3) These were a really, really "cheap" build for the end result. I'll try to post a full BOM in the build thread, but my best guess is that total cost was around $1,100 :E (driver cost at sale prices was ~$650 at the time and I stocked up on copper before the price increases, but it's likely you could still get it in <$1,500). I'm sure that newer/better/$pendier drivers could improve performance, but this build would be a great place to start for many folks interested in full-range dipoles.

4) No measurements to back this up ops: just lots of years of playing live music, but I'll try to borrow some gear over the summer to see what's going on objectively. If anyone is in the DC area and interested, drop me a PM. Depending on if/when DIY DC is scheduled this spring, I might be able to arrange to show them there.

Best,
Jim

[Martijn_H]

Martijn> the Illuminator definitely shows some promise for OB. Don't tell Jed that though or he'll scrap his Tombstones and spend another mortgage payment for new drivers Is that the AMT tweeter? Very interesting...I bet that is a nice combo. I do think teh mirror image concept of radiation is worth striving for, tweeters make that difficult.

John> thank you for the links, I'll be sure to check them out.

Yes that is the AMT https://www.htguide.com/forum/showth...&highlight=AMT


I think you can make a very good dipole with cheap drivers like the Neo3 pdr dayton mid and low. With a shallow baffle you can get a better of ax response but it is unhandy for those who want to work passive, because the mid will fall of to high on the low side. Maybe a compromise between a higher xo for the mid and woof can be choosen then. When you keep the baffle for the tweeter and mid very thin you can keep resonances on the backside at the mid tempered. In the netherlands we use more visaton and peerless drivers because dayton is much more expensive here.

Martijn

[CraigJ]

And nobody seems to have worked out the details of making it all work on a flat baffle . . . yet.

Deward,
I don't think that's quite true; see Isiris Junior thread. IMHO, Jon Marsh has made "it all work on a flat baffle" quite nicely. Infact, flat has worked out so well, that I've combined it with a 45" ribbon, a half dozen RS180s, and a XJ-15. Hank, don't wait any longer, the Saint Saens Junior is currently drying, and waiting for it's wings.

Craig

[Saurav]

I went ahead and ordered both ABCDipole as well as the MJK worksheets. I'm playing with some of the worksheets now, the ABCDipole software ships on a CD.

One worksheet in particular seems very useful for what I'm trying to do. You can see a snapshot of it here:

http://www.quarter-wave.com/Models/OB_3_Drivers_2_25_08.pdf

This assumes a flat baffle with 4 corners, 2 woofers, and a fullrange driver (which I'm using as the midrange, since I won't have a dipole tweeter, and I don't think dipole effects matter up at the tweeter frequency). It lets you enter coordinates for the baffle corners, driver positions, and also has inputs for distance to rear wall and distance off the floor.

The only annoying thing for me is that I'm using the free MathCAD 'viewer', which doesn't let me save anything. So each time I restart, I have to enter the data again. I think it lets you export the responses as FRD files though, so you can import them into Speaker Workshop etc.

I'm shooting for a fairly high XO point (300-400Hz) between the woofers and the midrange, so I think my best bet is to try and make this work on a flat baffle. I'm also starting with roughly the same shape/size/driver positions as the Isiris Junior (thanks for that pointer).

One question for those who've built flat-baffle speakers - how 'tippy' are they? I'll have a toddler in the house soon, and I'm concerned about the speakers being easy to tip over. With 2 10" woofers close to the floor I don't think it'll be too bad, but it's still not as bottom-heavy as a cabinet, right. What has been your experience with that?

[Deward Hastings]

> I won't have a dipole tweeter, and I don't think dipole effects matter up at the tweeter frequency

The “dipole effect” is primarily one of shaping radiation pattern, and once the tweeter(s) start beaming dipole cancellation goes away. It may go away for practical purposes even lower than that because of the (almost always) differing shapes of the front and rear of the MT panel. Keeping rear radiation consistent with front in FR and total energy is another issue (and the one that seems to be entertaining Linkwitz these days) . . . but the rear radiation is heard entirely as reflected, so it’s not clear how closely it must match the direct, and questions of phase become moot. At what frequency phase and amplitude matching of the rear radiation cease to matter is not clear . . . Linkwitz previously assumed that the rear didn’t need to be there at all over 1500 Hz. That’s clearly wrong, but the degree of matching necessary is still open . . .

> I'm shooting for a fairly high XO point (300-400Hz) between the woofers and the midrange, so I think my best bet is to try and make this work on a flat baffle.

It seems almost impossible to avoid cavity resonances in low frequency “H” (or “U”, although it’s not clear why one would) baffles, so that’s almost a necessary choice (for that high a crossover). The necessary flat baffle size becomes imposing, though, which raises questions about what benefits and whether the benefits of that high a W/M crossover are worth it.

> I'm concerned about the speakers being easy to tip over

You’ll need substantial bracing at the bottom to stop panel vibration from the woofers anyway, enough to remove any tipping worries. Then come the worries about tripping over the braces . . .

[Saurav]

Thanks for the feedback.

The necessary flat baffle size becomes imposing, though, which raises questions about what benefits and whether the benefits of that high a W/M crossover are worth it.

That's driven by my desire to continue to use my current midrange driver, and having to work around it's low Xmax (yes, I know, why am I trying a dipole speaker with this driver ).

And have you seen Jon's Isiris Junior design? I think it's a 45" tall baffle, 24" at the base narrowing to 16" at the top (going from memory here). That's not a whole lot bigger than my current baffles (42x16). And he uses 2x12" woofers, I'm hoping I can get something smaller to work with 2x10s.

I get your point though, I could use a different midrange driver that would allow a lower XO point, which gives me more woofer choices, allows me to use a U/H frame, and will probably work better on a smaller baffle.

Keeping rear radiation consistent with front in FR and total energy is another issue (and the one that seems to be entertaining Linkwitz these days)

Yep, the rear-firing tweeter... maybe for the next project. I'm already in way over my head with what I'm trying to do here

You’ll need substantial bracing at the bottom to stop panel vibration from the woofers anyway

Define 'substantial' I've seen some speakers with no bracing, I've seen some with a 'center' brace that basically goes around the drivers and attaches to the baffle (and the base). My woodworking skills are extremely poor, but I think I can manage something like that.

[augerpro]

In playing around with ABCDipole I modeled a U frame with 20cm depth versus a flat baffle of 40cm width. The flat baffle was down about 3dB under the dipole peak but what found most interesting was the behavior around and above dipole peak. Where the U frame had a large peak followed by a good size dip, the flat baffle had much smoother, wider peak, not very high (more like a typical baffle step) and was very smooth after the peak. This response would be MUCH easier to shape and cross to mid. Jon could probably shed some light on this if he were around...

Cut up some wood yesterday for a flat baffle and two U frames.


U Frame sim:




Flat baffle sim:

[CraigJ]

Hi Brandon,

I started with Type 2 and Type 3 Arvos, but ultimately decided the flat Isiris "sounded" the best to my ears (the least "box" like of the three). MDF is cheap and easy to play with, and allows for an excellent learning experience.





Craig

[Saurav]

Everyone should post pictures of the backs of their speakers Thanks, that was very helpful. Do you know the purpose of the semi-circular 'top plate' in the bracing? I assume it somehow separates the rear wave of the woofers from the midrange? I saw that in the plans, and couldn't understand what it was for.

Also, compared to these baffles:

https://www.htguide.com/forum/showpost.php?p=298015&postcount=18

Do you think the bracing is required all the way up to the MT section of the baffle? Did you find the top half of the Avro baffles vibrating? I was wondering if an Isiris-style brace, but extending up to the height of the top woofer, would be adequate. The woofer frequencies are probably less sensitive to reflections from surfaces of that size.

And finally, the Isiris brace - that's not used for magnet mounting the drivers, is it? I notice screws in the front, and it doesn't look like it's touching the backs of the magnets.

And maybe I should buy a circular saw before I attempt this. I'm trying to imagine cutting these shapes with my jigsaw, and that won't be much fun

[augerpro]

Thanks for pics Craig. How different would you describe the bass between these? What are the woofers crossed at? Any EQ?

Also the flat baffle still has a bit of U frame thing going on, I thought the picture I saw of Jon's was completely flat? EDIT: I was thinking of the Isiris with the completely flat baffle. Although I see saurav posted a version you had with a truly flat baffle?

[Saurav]

Although I see saurav posted a version you had with a truly flat baffle?

I think that is Craig's Isiris, while the 2 he posted are 2 versions of his Avro's. At least, that's what it seems like to me, from going through several long threads. The Avro's in their various incarnations all had some form of U-frame. The Isiris is Jon's fully OB design.

[CraigJ]

Sorry, I should have marked the pictures better. The first two pictures were of the Arvo Part Types 2 & 3, which did not have completely flat baffles. The picture below, is the Isiris Junior, which Jon designed as a flat baffle. Later, the Isiris was modified for the DDS ENG1 + BMS4540.

Brandon, the woofers are the RS315 HFs, crossed at 352hz with eq. from a dcx. Saurav, I can't answer the bracing height question because I don't know what your are going to make your baffles out of. I believe the Types 2 & 3 were made out of 1.5" mdf and because of the small u, did not require additional support. My third Arvo (Arvo/Isiris Junior) , which you referenced in your post, was made of 1.5" baltic birch laminated with .25" hdf on both sides; it vibrated a bit and needed more support than the cut up 2" x 8" that I had used. My final Isiris Junior is made of 4 layers of .25" hdf and .25" Corian, is very dense and doesn't require it's large brace. Regarding children, my 6 and 9 year old haven't knocked them over, yet. JimS, glad you like you Arvos. Jon has done an excellent job designing them, too bad he can't be home long enough to finish them

Have fun and good luck,



Craig

[Saurav]

Ah, so that was a 3rd Avro, not the Isiris. I have a hard time keeping all this straight, since I wasn't following those threads when they were active.

I'm not sure what I'm going to use for a baffle yet. I'll probably use BB ply, and this was an idea I got from a forum post:

http://www.ikea.com/us/en/catalog/products/10057760

I'll probably have to figure out the bracing requirements once I've built something.

[Dennis H]

Saurav, you probably won't need as much bracing as the Arvo if you use regular woofers instead of subwoofer drivers, only running them down to 80 Hz. The lower Mms and shorter stroke mean there won't be nearly as much vibration from the woofers. The RS270 has about 1/2 the stroke and 1/4 the moving mass of the RSS315 so it will only vibrate about 1/8 as much when pushed to its limits.

[Saurav]

I played around with MJK's worksheet this evening, since I don't have ABCDipole yet.

First, the baffle and predicted response in EDGE. I'm assuming a 6.5" mid, and 2x10" woofers. The baffle is 40" tall, tapering from 16" to 12" wide.

Woofers:

Images not available

Midrange:

Images not available


That's the extent of what I know how to do in EDGE. Then I went to Martin's worksheet, and set it up as close to EDGE as possible. Drivers only, no crossovers, no floor, no rear wall.

MJK Worksheet (Edit: Hmm, I just noticed my upper woofer height is wrong. Anyway...):

Images not available

Combined woofer response (it also plots the 2 woofers individually):

Images not available

Midrange response:

Images not available

Note that the vertical dB scale is very different from EDGE. Once you account for that, the two sims are pretty close. So far so good.

Next I put the baffle on the floor (and selected 'carpet').

Combined woofer response:

Images not available

Midrange response:

Images not available

I tried adding in the back wall, but that introduced so much comb filtering that it was hard to read the graphs.

Added LR4 filters: HP @ 700Hz, LP @ 500Hz, out of phase (just some numbers I dropped in by eyeballing the graphs).

Combined woofer response:

Images not available

Midrange response:

Images not available

Combined response:

Images not available

So, first off - is there anything grossly wrong with my approach so far? If I have some fundamentals wrong, I should fix those before I start fine-tuning anything.

I know this will take more shaping than just LR4 filters. Also, the level mismatch shouldn't really matter for me since I'll be using an active XO. And I don't think I should have to flip the phase between the two, so I'll have to play around with that (I had a pretty deep null right at 500Hz with them in phase).

I can see one advantage to ABCDipole - I think it takes the midrange driver's FRD file as input. With this worksheet, I'd have to export these curves and then pull them into Speaker Workshop. Which is probably what I'll be doing anyway, at least when it comes to modeling crossovers.

I don't have the woofers so I can't measure a test baffle to see how well these correlate. Hopefully Brandon will have some data on that.

[augerpro]

Well I hope to get some measurements next week. If I don't I have a wedding then a vacation in the Mediterranean so any work will be stalled until about the 2nd week of June.

Interesting plots. So the last set includes a floor bounce? Interesting. I would have thought the mid would be beaming enough that it wouldn't be so uneven-and that model does appear to use driver directivity judging by the first set of plots.

[Saurav]

Are you the groom? If so, congratulations

One more thing - I just noticed that I had the EDGE woofer plot use a 10m listener distance, while everything else uses 2m. If I enter 2m into EDGE, the woofer plot shows the same 2.5kHz null as the worksheet. I'm not sure how much that affects the floor bounce calculations.

I would have thought the mid would be beaming enough that it wouldn't be so uneven-and that model does appear to use driver directivity judging by the first set of plots.

Yeah... I don't know enough to have an opinion about that. My current speakers have a 6.5" midrange on an OB, similar baffle size and driver placement. I was messing around in Speaker Workshop and trying to measure it in my room, from my listening position. All my attempts had a dip at 1.5kHz. I knew it was a reflection from somewhere (which meant I should have used a shorter window), but I was interested in the mid-tweeter crossover, so I just ignored it. I guess it could have been the floor bounce. Or it could have been something completely unrelated.

Edit: 2nd order filters, same frequencies, in phase:

Images not available

This makes sense now, I probably had a ~6th order acoustic slope with a 4th order XO, this is closer to 4th order acoustic. Which means I probably won't be able to use my Behringer CX2300/3400 with this, since they have fixed LR4 slopes.

Anyone know of a good active XO kit? Well, there's always Marchand. I'd built my own on a project board before I got the Behringers, and that was really unwieldy.

And it looks like an ~80Hz cross to a sub should be doable.

So the last set includes a floor bounce?

The 2nd and 3rd sets both have the floor bounce, the 3rd set has the LR4 filters as well.

[Martijn_H]

> I won't have a dipole tweeter, and I don't think dipole effects matter up at the tweeter frequency
.

Well, maybe you don't want to use it because you think instead of try!

The dipole tweeter on a very narrow baffle can give you dipole behaviour over 2,5Khz. It gives the image more depth, so on every dipolebaffle I prefer dipole high. I like the neo3pdr dipole more than the scanspeak ringrad. :B

[Saurav]

Well, maybe you don't want to use it because you think instead of try!

LOL. OK, point taken, I should have worded that better. I won't have a rear-facing or dipole tweeter, and I don't think the shape/size of my baffle will significantly affect the radiation pattern of my monopole tweeter.

Maybe in the next project

And I want 95+dB sensitivity, so I'd need multiple Neo3pdr's.

[JoshK]

And I want 95+dB sensitivity, so I'd need multiple Neo3pdr's.

Planars and ribbons don't generally couple that way, meaning you don't get the added sensitivity with adding more of them.

[Saurav]

The ripples in the midrange response are influenced by the 2m listening distance.

Images not available

Same thing at 1m:

Images not available

It's flattened out the woofer curves too.

What's the right way to handle this? I think I want to keep the floor bounce effects in the picture, at least for the woofers. For the midrange... at those frequencies wouldn't you be using a gated measurement to get rid of the floor reflections? I probably shouldn't be worrying about this level of detail in a simulation.

[Deward Hastings]

> I probably shouldn't be worrying about this level of detail in a simulation.

There's many a slip twixt the sim and the . . . well, you know . . .

Things like the floor, and the walls, and what *really* happens when you build what the sim says is "perfect". There's too many variables, and none of them "independent". So you compromise *everything* and just hope the thing will fly . . . oh, wait, that's airplanes . . .

But the sims *are* getting better, and more detailed, and they *do* quickly weed out the really bad ideas. Still, no sim can tell you whether you're willing to put up with baffles shaped like a pear that are a three feet wide at the base. Or trapezoids 5 ft. tall. Especially when a "H" frame works just as well below 150 Hz.

Looking at real in-room response is a nightmare. At low frequencies there's the room modes and position dependent nulls. Then there's the nulls from the box being close to but not flush with the wall behind it . . . especially bad with "surround" speakers that are typically hung *on* the wall and comb like there's no tomorrow, but all too common in front, too, given where L and R are usually placed.. And then there's the floor . . .

[CraigJ]

Still, no sim can tell you whether you're willing to put up with baffles shaped like a pear that are a three feet wide at the base. Or trapezoids 5 ft. tall. . . .

I resemble that comment; introducing my 6 ft. tall version of the Saint Saens Junior. I guess that's one reason I like working with inexpensive hdf/mdf, if you don't like it, just glue some wood together and route again....it's all fun. :banana:



Craig

[Saurav]

But the sims *are* getting better, and more detailed, and they *do* quickly weed out the really bad ideas.

That's all I'm hoping for - that when I spend the money on drivers, I would have selected something that has a decent chance of working.

[Deward Hastings]

I lived with MG-IIIa's for several years . . . when I was single. And I'd still hold out for "big" if it was *necessary* for the sound. But despite being only two inches thick the Maggies were visually . . . significant . . . in my otherwise rather spacious 16x24 listening/living room. So "size" is another of the many variables in play, and the "eye opener" for me was the realization that all the dipole goodness could be had in a speaker no more imposing than "ordinary" towers or monitors-on-stands. And that it didn't entail *any* sonic compromises (except, perhaps, maximum SPL, with which I am not as enamored as some, not having a night club to fill).

But I'm 110% down with "it's all fun". I'm long since past realizing "good enough" in my system, but I continue to "tinker", putting, for example, time and money into improving "surround" speakers that don't make a darned bit of difference "just because". And while I have given up thinking I could make a speaker much better than ORION I sure haven't stopped wondering about making one almost as good a whole lot cheaper. I almost wish I had a reason to. An all-Dayton "orionette" (RS28, RS225, RSS265) would come in at a fraction of the cost of the original . . . the MT panel in particular 20% the cost (in drivers). And designing (said MT panel) from scratch with a rear tweeter in mind opens up the possibility of actual improvement over the tacked-on rear tweeter of ORION+. A very slightly larger "H" frame to house RSS315 wouldn't save any money, but would give a little more "oomph" at the bottom. Knowing that a lot of people have thought these same thoughts leaves me wondering why we don't have examples up in "Mission Accomplished" . . .

[CraigJ]

Hi Deward,
I'm heading off to Milwaukee, so don't have much time. I'm sure you know that Monte Kay has done the all Dayton "orionette". If I remember correctly, Monte felt that it was very, very close to the Orion; which says a lot for the Daytons drivers.
Regarding the "visually significant" comment, I understand completely, that's why I only look at my speakers from an angle, makes them look thinner .
Craig

[Deward Hastings]

Yes, except he didn't use the RS225, but a pair of 7" instead. Is the tweeter a RS28?

[john k...]

In playing around with ABCDipole I modeled a U frame with 20cm depth versus a flat baffle of 40cm width. The flat baffle was down about 3dB under the dipole peak but what found most interesting was the behavior around and above dipole peak. Where the U frame had a large peak followed by a good size dip, the flat baffle had much smoother, wider peak, not very high (more like a typical baffle step) and was very smooth after the peak. This response would be MUCH easier to shape and cross to mid. Jon could probably shed some light on this if he were around...

Cut up some wood yesterday for a flat baffle and two U frames.


U Frame sim:

[snip]


Flat baffle sim:

[snip]

I don't read this board very often but I happen to see this today and wanted to make a couple of point about the use of ABC. Regading the H and U frame simulations please remember that these are bascially tools to help evaluate the potential of H and U frame woofer systems. They do not include the effects of caviety resonances since these are expected to either be damped (by adding appropriate damping materials to the cavietys) or, in the case of an undamped H frame, be well below the intended crossover point to the mid. In the U frame they must be damped for correct U-frame operation, as discussed in the design guide that comes with the spread sheet. For the H no damping is required assuming the x-o is suitable below the dipole peak. Additionally, the deep dip you see in the simulations for the U-frame will be significiantly reduced when damping the rear caviety, as required. In any event, as I indicated above, the H and U simulations are really designed for woofer performance evaluation with an intended woofer to mid x-o point at lease 1/2 octave below the response peak.

Anyone using ABC should feel free to contact me if they have any questions or comments.

[augerpro]

Thanks John for the comments! I do have a question about U frame baffle size I hope you will answer here for he benefit of others reading: you mention that it should be just big enough to fit the woofer. Is this to keep resonances from dropping down into the woofers passband, or osme other reason? MJK mentions that efficiency of the U frame goes up if you increase the baffle size, what are the downsides?

BTW I have a post over at the SE user forum about a U frame sim that I haven't gotten any help on, could you take a look at it?