[joshsehn]

I thought I should share my experience with Open Baffle project.

When I first learned about Open Baffle speakers I was really suspicious about how that could work... so I just did a few simple tests starting with a single 6.5" driver and 3/4" tweeter. I was blown away by the clarity of the sound but it was seriously lacking bottom end of course...

I decided to go further with the project because I like to learn and I had some confidence that I could build something that sounds nicer that what I could afford to purchase. Given my goals, I feel the speaker is tremendous success... I had lots of fun and learned alot during the build. Thus far it is without a doubt the best sounding speaker I've ever built.
 
Its a fully active system with 75wpc for each driver (except 1" and 2" drivers are driven in parallel)
Per side these are the drivers:
2 @ 10" Peerless (Same sub as Orion) in a similar H frame running upto 105Hz
2 @ 6.5" Peerless (105 - 1.65khz / Open back of course) 
2 @ 2" Dayton Dome (1.65khz - 4.6khz / 1 front, 1 back)
2 @ 1" Peerless Dome (4.6khz up / 1 front, 1 back)

Here is what it looks like now:


The 10" drivers are in a 24" H Frame (similar to SL's Orion). I used two of the 6.5" drivers instead of the single 8" driver for the lower midrange because I thought this might alleviate cone breakup when approaching 1.5khz on a larger 8" driver. I'm using both the 1" and 2" domes on both front and back.

I just got some testing software so I'm not really sure of the optimum/standard testing so I would be grateful if someone would tell me the best/standard test methods or recommended settings.

Nonetheless, here is some measurement data.

This is an outdoor measurement at 2m (2ms window - no smoothing)


In room response at listening position (200ms)

Note: I'm also using a 15" woofer below 50Hz to help smoothen the bottom end response and provide extra kick 

Waterfall plot at listening position. (200ms)

[versus rider]

nice work josh. How do you arrive at the dimensions for an H frame. The reason I ask is that I like my open baffle bass courtesy of Eminence Alpha 15's but am now embarking on a front horn project that will keep the open baffle bass. The H frame would allow me somewhere to mount the front horn, something that is going to be problematic with the baffle.

[jimdgoulding]

Geez  

[matevana]

Good job!  I run my peerless woofers up to 334Hz, crossed at 24dB/octave. This greatly influences the power curve and provides a presence that the 6.5" drivers simply can't offer. It gives that chest thumping "whack" that I was missing when I crossed over at 100Hz initially. You may wanna give that a try as you continue to experiment. Also, the curve seems rather flat for such a wide array of drivers/alignments. I'm guessing you're using something like a Behringer DEQ or DCX to address dipole cancellation and smooth-out the response? I realize those were measured outdoors, but do you run a house curve inside with your dipoles? I have a lot of experience dialing-in a favorable house curve... which seems to make a huge difference in how dipoles present themselves. 

I am running the same Peerless 10" drivers, along with an 8" pro sound woofer and a Morel MDT-20 tweeter mounted coaxially, in a single point-source alignment. The system is true dipole up to about 6.5k.  I experimented with a 2nd rear firing tweeter but had too many problems combating reflections and image smearing. Did you have a chance to A/B your system with and without the rear firing tweeter? Curious what you thought.     

[joshsehn]

How do you arrive at the dimensions for an H frame. The reason I ask is that I like my open baffle bass courtesy of Eminence Alpha 15's but am now embarking on a front horn project that will keep the open baffle bass.

I just built as per lovely examples and work of others (read Sigfried Linkwitz's Orion).

A good rule of thumb I estimate would be to keep the operating frequency BELOW the dipole frequency. In my case, a 24" H frame offers a dipole frequency of about 140Hz and thus I'm running the woofers up to 100Hz. If the woofers run right up into the dipole frequency, you could have trouble dealing with a large peak. I guess you would have to base your decision on the frequency range you hope to achieve with your woofers. Keep in mind you loose 6db below the dipole frequency, so if you use a small baffle which allow you to operate upto 300hz, you probably will not have enough displacement to provide adequate SPL in the lowest octave.

I run my peerless woofers up to 334Hz, crossed at 24dB/octave

I presume you are not using an H frame so your dipole frequency is probably way higher (ie. 400Hz ?)   As discussed above, I'm using a  24" H frame so I don't think I can run much above 100hz without creating a HUGE hump in the response. With the DUAL 6.5" drivers I've got plenty of output down to 100Hz (I.e. flat response)

 

Also, the curve seems rather flat for such a wide array of drivers/alignments.

Yes, I'm using a DCX to provide for the crossover requirements as well as the 6db/octave dipole slope. I believe the time alignment really helps to give flat response. (as per outdoor measurements).

but do you run a house curve inside with your dipoles?

yes, I use DEQ to provide some smoothing of the room response but I really haven't changed much from the outdoor measurement. Most all of any additional manipulation is mainly used to help reduce 2 room modes (-8db at 63Hz and -9db @ 90Hz) corresponding with the dimensions of my room.  I've played around with a "house curve" as you say, to try and deal with reflections/cancellation of the rear output of the dipole but can't say that has been really helpful for me and doesn't really seem to solve the problem, at least in my experience.

Did you have a chance to A/B your system with and without the rear firing tweeter?

No I haven't A/B test to eliminate the rear firing tweeter. I was thinking about reducing the level of the rear tweeter and I've played around by covering it up but that didn't seem to make as big a difference as I had anticipated. I DID however recently incorporate some sound absorbing material on the wall behind the speaker and it made a BIG difference. Much more refined "image" now. (Or less "smearing" as you said) I can't say I have much desire to change the imaging by replacing the rear tweeter any longer. I'll try and publish some polar response data here.

8" pro sound woofer and a Morel MDT-20 tweeter mounted coaxially, in a single point-source alignment.

I've been thinking about that kind of idea as well... inspired by the "SuperV" I've seen elsewhere on this forum.

The system is true dipole up to about 6.5k.

Can you share any pictures or measurement data?

[joshsehn]

here is a polar response plot... I'm missing the 90Degree measurement however. 


I did try getting some measurements at extreme angles but I had difficulty getting the microphone to pickup adequate response at 90degrees.... 

[Angaria]

Gotta love systems like this.... Seems like you could achieve more consistent directivity around 1khz by narrowing the baffle?

[djn]

here is a polar response plot... I'm missing the 90Degree measurement however. 


I did try getting some measurements at extreme angles but I had difficulty getting the microphone to pickup adequate response at 90degrees.... 

Very nice plot Josh.  I'd like to hear those.

[matevana]

Here's a few quick pics.

Type: OB dipole up to approx 5.5k Hz. Biamped helper woofer, actively crossed and roll-off compensated. Coaxially mounted pro audio mid w/traditional 1" fabric dome tweeter. Acoustically dampened and isolated panels w/closed cell suspension. Visco-elastic feet.

Helper Woofer: Peerless 10" Nomex XXLS woofer. 334Hz lowpass, 24dB/octave.

Mid: Eminence Acoustinator 8" poly mid range, coax 1 1/8" 18TPI threaded, allowed to roll-off naturally.

Tweeter: Morel MDT-20 mounted coaxially, 1.8mf single cap, 6dB per octave.

Initial crossover points surprisingly good. Will spend some time tweaking. The drivers produce some respectable SPL's w/o strain. These were initially designed with many genres in mind. The acoustinator CX2008 mid is a bit dark, but the warm character of the Morel MD-20 crossed low balances nicely and yields a realistic presentation. The voicing is very pleasing to me. For HT, I rely on a seperate sub from about 60Hz down. For music, I listen w/o any sub. The dipole bass, even with only 1 Peerless driver and no H-Frame, is very realistic to me. These are supprisingly good when listening from CD Transport->Behringer SRC2496 ->Behringer DEQ2496 -> Crown XLS1000 (lows with built in crossover) & Dayton DTA-100 T-amp (highs). 

[joshsehn]

Seems like you could achieve more consistent directivity around 1khz by narrowing the baffle?

I believe narrowing the baffle could help... but I'm really thinking that the next step I should take is to get rid of the baffle entirely (I want to keep some H-Frames for bass of course).

[joshsehn]

Here's a few quick pics.

Very nice... I like the nice small and simple configuration. I'm also interested in the point source coax idea.

Do you have any measurement data?

I used Fuzzy Measure Pro... Testing by ear and using the simple pink noise and a RTA was simply not adequate. I'm sooo happy that I bought the testing equipment because I was able to make big improvements to the setup much quicker with much more confidence. Highly recommend the testing equipment. BEST INVESTMENT EVER.   

[matevana]

Very nice... I like the nice small and simple configuration. I'm also interested in the point source coax idea.

Do you have any measurement data?

I used Fuzzy Measure Pro... Testing by ear and using the simple pink noise and a RTA was simply not adequate. I'm sooo happy that I bought the testing equipment because I was able to make big improvements to the setup much quicker with much more confidence. Highly recommend the testing equipment. BEST INVESTMENT EVER.   

Not familiar with Fuzzy Measure Pro. Sounds interesting though. I only use the RTA on the DEQ along with a calibrated measurement mic. I try to get the in-room response flat (+/- 1dB) from around 60Hz up through 15k Hz.  Then I dial in a house curve based on my musician's ear  which usually consists of a cut to the last ten or so sliders by about 1.5dB and any other (annoying) room node that might be present. With dipoles its also helpful to set the RTA reading to mid or slow, since you want to take into consideration reflections off the rear wall. It seems that too much of the rear wave is ignored if the meter reads too quickly.

SL had an interesting evolution with and w/o the rear firing tweeter on his Orion. Initially he had no rear firing tweeter, then he added one with no attenuation, then reduced the output by a few dB, and now I believe it is back to being flat. I have spent many hours observing the rear wave and I believe there is no clear answer here. Allowing the highest frequencies to be truly dipole in nature seems to produce a more natural, reverberant sound. The down side is that the stereo image can tend to wander and be smeared with a loss of focus. My current rig is only dipole till about 6.5k, and combined with the single point source alignment, produces a very realistic image.       

[joshsehn]

Not familiar with Fuzzy Measure Pro. Sounds interesting though.

http://www.supermegaultragroovy.com/products/FuzzMeasure/

I only use the RTA on the DEQ along with a calibrated measurement mic. I try to get the in-room response flat (+/- 1dB) from around 60Hz up through 15k Hz.

That was also my approach... before I got fuzzmeasure. I still use some attenuation of significant room modes at very low frequencies and a few other relatively small changes to help minimize room effects.

But what I have found, now that I have better measurement capability, was that I am able to set the system response MUCH more accurately. I was completely surprised by the new measurements compared to the old pink noise. (not to mention I don't have to listen to pink noise for hours try to figure things out) 

I feel that I was only able to do this accurately since I was able to measure independently of the response of the room. The result are quite pronounced and reveals a much more realistic and musical presentation.

IMO, the DCX pink noise averaging gives the room way to much control over the measurements and I found this to be very misleading (mostly due to the cancellation effects of room reflections). Hence, I spend alot of time trying to "Fix" the room response using the DEQ measurement but never felt I was able to achieve the best possible performance.

The measurement software was VERY revealing and helped me make changes that I otherwise would never have been able to see.

SL had an interesting evolution with and w/o the rear firing tweeter on his Orion. Initially he had no rear firing tweeter, then he added one with no attenuation, then reduced the output by a few dB, and now I believe it is back to being flat. I have spent many hours observing the rear wave and I believe there is no clear answer here. Allowing the highest frequencies to be truly dipole in nature seems to produce a more natural, reverberant sound.

I studied SL's work a lot and I must say it was most enlightening and really very helpful.

Since I purchased measurement gear, I was much more able to attempt to emulate his recent change (Orion rev 3.2) with regard to the overall sloping of the response. You will see in my plot that I've mimicked the down slope fairly closely and included a 3.5kHz BBC dip. I am very happy with the changes and the musical presentation is MUCH better.

The down side is that the stereo image can tend to wander and be smeared with a loss of focus. My current rig is only dipole till about 6.5k, and combined with the single point source alignment, produces a very realistic image.

I agree about the tendency of the image to wander/smear to some degree depending largely on distance from the rear wall. Due in part to the newly acquired measurement capability, I found that if I put some wall treatment behind the speaker the stereo degradation problem was substantially reduced and thereafter the stereo image has become MUCH more focused. This wall treatment also helped reduce the negative cancellation of the rear wave by a few db.

[matevana]

http://www.supermegaultragroovy.com/products/FuzzMeasure/

That was also my approach... before I got fuzzmeasure. I still use some attenuation of significant room modes at very low frequencies and a few other relatively small changes to help minimize room effects.

But what I have found, now that I have better measurement capability, was that I am able to set the system response MUCH more accurately. I was completely surprised by the new measurements compared to the old pink noise. (not to mention I don't have to listen to pink noise for hours try to figure things out) 

Hey Josh,

Thanks for the link on FuzzMeasure.  Too bad they don't make it for a real OS like Linux    Maybe one day I'll break down and buy a Mac.

All good points btw. I think I have experienced everything you have commented on at some point in the tweaking process. I really must say that I am enjoying my setup right now, in its more simple config as pictured above.

I'd like to bring up the 10" peerless x/o points once more. I played a lot with this last night.  I too used to cross them at around 100Hz to the midwoofer. I really believe they can benefit from a higher crossover, with a steep slope for many reasons. First of all, the Peerless XLS (and XXLS) are one of the few subs that can be crossed fairly high. They exhibit lower harmonic and tall order distortion in the 50-300 Hz range then even two of the 6.5" drivers that you use, covering the same range. i.e, they can play louder and stay clean. They also move a lot more air due to their increased surface area and xmax and are more efficient. Lastly, they seem to have a lot more "presence" in this pass band then the two woofers would offer.

Since you have a DCX, it would be fairly easy to test this theory. To be fair, the relative loudness (including room gain and the benefit of your HFrame) should be the same when testing. Only provide power to the drivers in question. Listen to a variety of material first through the XLS and limit the passband to 100-300Hz, to be fair to your 6.5" drivers. Then do the same for your 6.5" drivers at the same relative loudness. In my experiemnts, I was convinced that the XLS easily out performed, allthough I was testing using a slightly different pair of mid woofers. I'm really interested in your findings here.

Also, just curious what you used to treat the rear walls. I'm guessing some type of foam?

Ed   

[joshsehn]

I'd like to bring up the 10" peerless x/o points once more. I played a lot with this last night.  I too used to cross them at around 100Hz to the midwoofer. I really believe they can benefit from a higher crossover, with a steep slope for many reasons. First of all, the Peerless XLS (and XXLS) are one of the few subs that can be crossed fairly high. They exhibit lower harmonic and tall order distortion in the 50-300 Hz range then even two of the 6.5" drivers that you use, covering the same range. i.e, they can play louder and stay clean. They also move a lot more air due to their increased surface area and xmax and are more efficient. Lastly, they seem to have a lot more "presence" in this pass band then the two woofers would offer.

Absolutely agree with everything you've brought up... but in my case the problem seems due to the configuration of my H-frame. I think they are too efficient for me to cross them much higher and they easily start to overpower the rest of the system. I suspect I would have very big hump in response above 100Hz with massive peak around 150.

Since you have a DCX, it would be fairly easy to test this theory. To be fair, the relative loudness (including room gain and the benefit of your HFrame) should be the same when testing. Only provide power to the drivers in question. Listen to a variety of material first through the XLS and limit the passband to 100-300Hz, to be fair to your 6.5" drivers. Then do the same for your 6.5" drivers at the same relative loudness. In my experiemnts, I was convinced that the XLS easily out performed, allthough I was testing using a slightly different pair of mid woofers. I'm really interested in your findings here.

With my existing configuration I haven't been able to push the 6.5" drivers to any significant excursion levels (i.e. the point where they are moving very much). But sure, I can give it a go and publish a plot later tonight.

Also, just curious what you used to treat the rear walls. I'm guessing some type of foam?

I used a few acoustic ceiling tiles. Very cheap.

[matevana]

Absolutely agree with everything you've brought up... but in my case the problem seems due to the configuration of my H-frame. I think they are too efficient for me to cross them much higher and they easily start to overpower the rest of the system. I suspect I would have very big hump in response above 100Hz with massive peak around 150.

With my existing configuration I haven't been able to push the 6.5" drivers to any significant excursion levels (i.e. the point where they are moving very much). But sure, I can give it a go and publish a plot later tonight.

I used a few acoustic ceiling tiles. Very cheap.

Just for grins and giggles, i wonder how your same configuration would fair in a flat panel instead of H-Frame, since you dont seem to need the output down low. Would look pretty cool too and reduce the footprint. Ceiling tile is a great idea btw!  And looks a lot better than convoluted foam.   

[joshsehn]

Since you have a DCX, it would be fairly easy to test this theory. To be fair, the relative loudness (including room gain and the benefit of your HFrame) should be the same when testing. Only provide power to the drivers in question. Listen to a variety of material first through the XLS and limit the passband to 100-300Hz

OK... just for you  I performed the following test:

• I set the Dual 10" to have no crossover
• I turned off the dipole EQ.
• I ran a sweep from 20Hz to 500Hz.

Here are the results. (I'm using 7ms window so the slope is nice and visible. Any larger window and we get big distortion due to reflections because I did this all indoors).



You can see the output is quite low at lowest octave and gathers ALOT of volume by 200Hz.

I put the regular filter I use on the DCX back in place (-15db at 96Hz with 12db/octave HP. -15db is the max cut/boost on the DCX) and ran another sweep from 20-500Hz and here is the result of that.



Then I applied the standard crossover settings I'm using and here is that result. (This time with 14ms window so you can see the crossover cutoff better)

Just for grins and giggles, i wonder how your same configuration would fair in a flat panel instead of H-Frame, since you dont seem to need the output down low

But I do need the output down low  and moving to a smaller dipole (e.g. 15" flat panel for instance) I would end up loosing a bunch more output at 20Hz because the dipole frequency would move up an octave.

Perhaps I should clarify, now that I've actually measured while I'm writing this I realize I was mistaken early as these are NOT exactly a proper 24" H frame! 

The frame is really only 16" total front to back with the woofer on a panel in the middle which makes it closer to 30" total baffle length. So it makes sense why my dipole frequency is closer to 200hz rather than what I quoted/guesstimated earlier. Regardless, the foot print isn't actually that big. It's almost the same size at my Paradigm monitor 11s front to back but it is a little wider. I haven't really listened to the Paradigms in some time because these ones are so much more pleasing.   

[matevana]

OK... just for you  I performed the following test:

• I set the Dual 10" to have no crossover
• I turned off the dipole EQ.
• I ran a sweep from 20Hz to 500Hz.

Wow, awesome testing there! That makes a lot of sense. Part of my assumption in suggesting a flat panel was in thinking you werent too concerned with the output at 20Hz, since you're also using a separate sub. By shifting the XLS's band higher (and effectively moving everything else up a notch) I thought you might be able to take advantage of having each driver produce sound only in their ideal range , i.e., lowest possible distortion, etc.

True that the footprint isn't bad, even with an H-Frame. I'm partial to the look of the new Jamo dipole series and also the look on people's faces when they ask "Where's the box?" or "I thought drivers had to be enclosed to get any amount of bass".

It's also interesting that with the advent of products like miniDSP, it easier for mass manufacturers to consider producing "out-of-the-box" dipoles that people can simply plug-in and enjoy. Then again, to me, trying to incorporate every last tweak is half the fun.

My latest experiment involves A/B'ing my DEQ with a (dare I say it) Rane 31 band old school analog EQ that I picked up on ebay for $35. I use the RTA on the Behringer for measurement only, and then duplicate the settings on the Rane. Of course you have to live with something for a while, but my initial impression is very favorable. I'm still using the DAC in my Behringer SRC2496, but there is something to be said for analog equalization ahead of the amps.

Thanks for all the time you took to test that out. You gotta admit this stuff is addictive!          

[mcgsxr]

I will admit to fooling around with an analog EQ for my subs, and it has been very helpful in boosting the lower frequencies.

Clearly not as good as a digital one, but as I already own it, I will play around with it.

[joshsehn]

Part of my assumption in suggesting a flat panel was in thinking you werent too concerned with the output at 20Hz, since you're also using a separate sub. 

I really like full range.   

My separate subwoofer is only playing upto 50Hz with LR24 LP and it is set at a relatively low volume. It just provides that little extra boost at the lowest octave and overlaps the frequencies played by the HFrame woofers which set to playing down to 20Hz with a 6db/octave HP below that. The effect of the 6db/HP is that the 20-30Hz is also reduced a bit. The 15" subwoofer is positioned close to the listener and seems to have the effect of distributing the room modes and helping level the response.

The net result are; stronger base, flatter response all the way down to 20Hz and the ability to play to much higher SPL with very little distortion. It provides bass output you can really feel when the program material calls for it but still not overpowering at all. Meanwhile, thinner recordings with less bass still sound like they should.

By shifting the XLS's band higher (and effectively moving everything else up a notch) I thought you might be able to take advantage of having each driver produce sound only in their ideal range , i.e., lowest possible distortion, etc.

IMO, the xover frequencies I've chosen accomplish exactly that: Each driver producing sound in only their ideal range with lowest possible distortion.

My latest experiment involves A/B'ing my DEQ with a (dare I say it) Rane 31 band old school analog EQ that I picked up on ebay for $35

A good analog EQ should sound very nice.... you may have trouble doing the the fancy crossover stuff required for a dipole but with a microphone and some test gear you should be able to get very good results in many other setups.

It's also interesting that with the advent of products like miniDSP, it easier for mass manufacturers to consider producing "out-of-the-box" dipoles that people can simply plug-in and enjoy.

Absolutely. My limited understanding of the MiniDSP is that you could accomplish most everything required for a great dipole setup.

For commercial deployment as the DCX is likely not optimal for retail / non-hobbyist as you suggest and other technical reasons i.e. the filters don't work as good as they should and don't have enough cut/gain. Ultimately a retail solution should have a proper preamp and of course I would prefer phase linear crossovers and EQ. I should investigate MiniDSP further.