Member
Joined 2003
For the past while I have been researching different speaker arrangements for an affordable HT center channel that overcomes the major drawback in most commercially available center channels: the horizontal MTM.
When an MTM is vertical, off-axis nulls will occur at at +45 degrees, and -45 degrees on the vertical plane, at the crossover frequency. In most listening conditions, these nulls would be aimed at the floor and ceiling and so are not a problem. However, when we make a horizontal MTM for a center channel, these nulls are now at 45 degrees left and right, rather than up and down, meaning horizontal off-axis performance suffers greatly, and moreover the crossover frequency where the response suffers is usually the vocal range of the frequency response where we want the center channel to perform best.
So, how can we make a horizontal speaker without this problem? There are a few options:
-A 3-way with drivers oriented such as Zaph's center channel here. Here, the crossover point between the horizontally arranged woofers and the midrange is low enough that off-axis performance doesn't suffer. The midrange and tweeter are located vertically so that off-axis performance suffers only in the vertical plane. However, moving to a 3-way can add a lot of cost to the speaker.
-A WMTMW. Again, not very cost effective, but in this method we have the ability to move the crossover point of the MTM higher to not be in the vocal range, or at least reduce the effects to the vocal range.
-A 3-way with a coaxial mid-tweeter, such as the Kef Uni-Q. There aren't a lot of good coax drivers available to choose from.
-design a horizontal MTM using a wide-range driver instead of a tweeter. Here, we can use a wide-range driver crossed over low enough so that off-axis response is not a problem. This is the same idea as the Zaph 3-way above, but using a wide-range driver in place of the midwoofer and tweeter. Here, we can keep the driver quantity low, and crossover part count low as well.
I chose the last option, as a cost effective solution is what I am after. So, what drivers to choose? The biggest hurdle here is finding a good wide-range driver that works well from about 500Hz to 20kHz, and is sensitive enough to mate with a couple 6.5" woofers. The sensitivity is the biggest one, as most 3" drivers have very low sensitivity of ~80dB@1W/1m, and I can't use multiple without feeling the need for a tweeter to cover the high end. However, I did find the Fostex FF85K which is 88dB@1W/1m and is highly regarded by the full-range crowd.
88dB is great, and leaves a lot of options for the woofers. The woofers will be crossed over low, so only need to be about 88dB sensitive, and perform well up to about 1.2kHz. No problem, right? I originally had started this design around the Fountek FW168 for it's cost/performance ratio, but then moved to the Peerless 830883 when I found a few cheap ones on the meat-swap (formerly known as the Trading Post). I can post the design for the Fountek as well if there is any interest.
So here's the design. I was a little worried about the wonky response of the FF85K in the datasheet, particularly the rising response in the top octave, and the 4-5dB dip in the midrange. I could not find any 3rd party measurements for this driver to confirm the frequency response, so I left a few tweakable options in the crossover:
-looking at the transfer function, there is a hump in the midrange for the FF85K to compensate for the dip in it's response. This hump can be raised and lowered by adjusting the 2.4 Ohm resistor in series with the 0.62mH inductor.
-the dip in the transfer function in the top octave can be decreased by unwinding the 0.1mH inductor in series with the driver a couple turns at a time. Unwinding this inductor will raise the top octave response.
What we have here is a horizontal MTM where a wide-range driver is used in place of the tweeter. Component count has not increased in comparison to a traditional MTM center channel, and we have overcome horizontal off-axis problems!
Peerless 830883 and FF85K Center Channel:
IMAGE: Simulated Frequency response, phase, transfer function, and impedance
IMAGE: Crossover Schematic
IMAGE: Cabinet Drawing
Notes:
-This is a modeled response using the FRD tools, not measured.
-The 830883 frequency response is traced from Zaph's measurements, and the impedance is traced from the datasheet.
-The FF85K frequency response is traced from the datasheet, and the impedance is traced from Planet10's measurements in the Tysen thread.
-All inductors are Solen 16AWG solid wire air core, except the 2.25mH which is a Sledgehammer 15AWG steel laminate from Madisound.
-if using a different 0.62mH inductor, note the change in DCR and adjust the series resistor accordingly.
-Cabinet is ~30 Litres, tuned to 42Hz. A 3" straight port fits nicely in the middle on the back of the cabinet. If you wish to build this speaker, I recommend downloading Unibox, and choosing your desired port tuning.
-I have not built this speaker yet. I will post any changes I make once I have built and listened to it.
When an MTM is vertical, off-axis nulls will occur at at +45 degrees, and -45 degrees on the vertical plane, at the crossover frequency. In most listening conditions, these nulls would be aimed at the floor and ceiling and so are not a problem. However, when we make a horizontal MTM for a center channel, these nulls are now at 45 degrees left and right, rather than up and down, meaning horizontal off-axis performance suffers greatly, and moreover the crossover frequency where the response suffers is usually the vocal range of the frequency response where we want the center channel to perform best.
So, how can we make a horizontal speaker without this problem? There are a few options:
-A 3-way with drivers oriented such as Zaph's center channel here. Here, the crossover point between the horizontally arranged woofers and the midrange is low enough that off-axis performance doesn't suffer. The midrange and tweeter are located vertically so that off-axis performance suffers only in the vertical plane. However, moving to a 3-way can add a lot of cost to the speaker.
-A WMTMW. Again, not very cost effective, but in this method we have the ability to move the crossover point of the MTM higher to not be in the vocal range, or at least reduce the effects to the vocal range.
-A 3-way with a coaxial mid-tweeter, such as the Kef Uni-Q. There aren't a lot of good coax drivers available to choose from.
-design a horizontal MTM using a wide-range driver instead of a tweeter. Here, we can use a wide-range driver crossed over low enough so that off-axis response is not a problem. This is the same idea as the Zaph 3-way above, but using a wide-range driver in place of the midwoofer and tweeter. Here, we can keep the driver quantity low, and crossover part count low as well.
I chose the last option, as a cost effective solution is what I am after. So, what drivers to choose? The biggest hurdle here is finding a good wide-range driver that works well from about 500Hz to 20kHz, and is sensitive enough to mate with a couple 6.5" woofers. The sensitivity is the biggest one, as most 3" drivers have very low sensitivity of ~80dB@1W/1m, and I can't use multiple without feeling the need for a tweeter to cover the high end. However, I did find the Fostex FF85K which is 88dB@1W/1m and is highly regarded by the full-range crowd.
88dB is great, and leaves a lot of options for the woofers. The woofers will be crossed over low, so only need to be about 88dB sensitive, and perform well up to about 1.2kHz. No problem, right? I originally had started this design around the Fountek FW168 for it's cost/performance ratio, but then moved to the Peerless 830883 when I found a few cheap ones on the meat-swap (formerly known as the Trading Post). I can post the design for the Fountek as well if there is any interest.
So here's the design. I was a little worried about the wonky response of the FF85K in the datasheet, particularly the rising response in the top octave, and the 4-5dB dip in the midrange. I could not find any 3rd party measurements for this driver to confirm the frequency response, so I left a few tweakable options in the crossover:
-looking at the transfer function, there is a hump in the midrange for the FF85K to compensate for the dip in it's response. This hump can be raised and lowered by adjusting the 2.4 Ohm resistor in series with the 0.62mH inductor.
-the dip in the transfer function in the top octave can be decreased by unwinding the 0.1mH inductor in series with the driver a couple turns at a time. Unwinding this inductor will raise the top octave response.
What we have here is a horizontal MTM where a wide-range driver is used in place of the tweeter. Component count has not increased in comparison to a traditional MTM center channel, and we have overcome horizontal off-axis problems!
Peerless 830883 and FF85K Center Channel:
IMAGE: Simulated Frequency response, phase, transfer function, and impedance
IMAGE: Crossover Schematic
IMAGE: Cabinet Drawing
Notes:
-This is a modeled response using the FRD tools, not measured.
-The 830883 frequency response is traced from Zaph's measurements, and the impedance is traced from the datasheet.
-The FF85K frequency response is traced from the datasheet, and the impedance is traced from Planet10's measurements in the Tysen thread.
-All inductors are Solen 16AWG solid wire air core, except the 2.25mH which is a Sledgehammer 15AWG steel laminate from Madisound.
-if using a different 0.62mH inductor, note the change in DCR and adjust the series resistor accordingly.
-Cabinet is ~30 Litres, tuned to 42Hz. A 3" straight port fits nicely in the middle on the back of the cabinet. If you wish to build this speaker, I recommend downloading Unibox, and choosing your desired port tuning.
-I have not built this speaker yet. I will post any changes I make once I have built and listened to it.
Member
Joined 2003
Member
Joined 2003
I'm a bit surprised in the lack of interest this has gathered. Total build cost ends up being $250-$300 using FW168, add ~$50 on that for 830883. I can post the Fountek FW168 crossover as well if there is interest. This speaker can obviously be tipped upright and used as a regular speaker too...
My speakers arrived yesterday, great success! However, I am a little disappointed in the FF85K. The one I received had excessive black goop between the aluminum cap and the cone, and was not very evenly distributed. I'll take a photo later tonight when I'm not at work.
My speakers arrived yesterday, great success! However, I am a little disappointed in the FF85K. The one I received had excessive black goop between the aluminum cap and the cone, and was not very evenly distributed. I'll take a photo later tonight when I'm not at work.
Member
Joined 2003
That looks quite different from the published one:
http://www.fostexinternational.com/docs/speaker_comp/pdf/ff85k.pdf
http://www.fostexinternational.com/docs/speaker_comp/pdf/ff85k.pdf
I'm actually working on an identical project, using a Markaudio Alpair5 for the fullrange, and two Dayton DA175s as the woofers.
I have it assembled in a prototype box and have done some initial listening using the variable crossovers in my bench amp. Later this week I'll be taking acoustic and impedance measurements in the box to finalize the crossover.
So far, however, it sounds pretty good. Hard to tell in the garage, terrible acoustics. The Alpair5's +10k hz rise is pretty harsh, in that environment anyway. Definitely helps to stand well back from the speaker, which will be the case in the home theater these are being designed for, but I might tame it a bit with a notch filter.
Just as you I'm suprised noone else on this board has really tried this, yet. Or at least I haven't stumbled across that particular post
An externally hosted image should be here but it was not working when we last tested it.
I have it assembled in a prototype box and have done some initial listening using the variable crossovers in my bench amp. Later this week I'll be taking acoustic and impedance measurements in the box to finalize the crossover.
So far, however, it sounds pretty good. Hard to tell in the garage, terrible acoustics. The Alpair5's +10k hz rise is pretty harsh, in that environment anyway. Definitely helps to stand well back from the speaker, which will be the case in the home theater these are being designed for, but I might tame it a bit with a notch filter.
Just as you I'm suprised noone else on this board has really tried this, yet. Or at least I haven't stumbled across that particular post
Member
Joined 2003
Nice to see! Regarding your speaker, you will probably have smoother response by moving the Alpair driver off-center a bit, and round-over the baffle edge. The round-over is probably out of the question for your cabinet, as you have screws lining the edge...A cheap metal cone driver is a good option for bass duty, since the crossover should be low enough that you won't have to worry about the cone breakup.
If all goes as planned I'll be building my cabinet early August during my vacation time.
I think that for the most part this type of speaker has limited options for wide-range driver choice, as the sensitivity has to be high enough to work with the bass drivers. A regular 2-way might have some more flexibility as the sensitivity will be lower. Another option would be to use an active crossover, in which case sensitivity matching is not a concern, just need a large amp to make loud happen.
If all goes as planned I'll be building my cabinet early August during my vacation time.
I think that for the most part this type of speaker has limited options for wide-range driver choice, as the sensitivity has to be high enough to work with the bass drivers. A regular 2-way might have some more flexibility as the sensitivity will be lower. Another option would be to use an active crossover, in which case sensitivity matching is not a concern, just need a large amp to make loud happen.
this is just a prototype box, with no concerns put into aesthetics, only build speed 
Attached is a pic of a previous finished single-driver I did. (the blemish on the edge is just a little glue, it came off). These will be of a similar design, when finished.
On these I'm going to experiment with a "staged" roundover, with a truncated roundover like on the speaker in the attached picture around most of the speaker, and then transition to a full flush half-round near the mid-tweet.
As far as sensitivity goes, the DA175 and the Alpair5 both have a rated sensitivity of 85 SPL 1W/1M, but the DA175 is an 8 ohm speaker where the Alpair5 is a 4 ohm, so the amp delivers more power to the Alpair, relatively acting as if it were more sensitive. The overall speaker won't, unfortunately, be really sensitive, but it should be able to handle a decent amount of juice.
Attached is a pic of a previous finished single-driver I did. (the blemish on the edge is just a little glue, it came off). These will be of a similar design, when finished.
On these I'm going to experiment with a "staged" roundover, with a truncated roundover like on the speaker in the attached picture around most of the speaker, and then transition to a full flush half-round near the mid-tweet.
As far as sensitivity goes, the DA175 and the Alpair5 both have a rated sensitivity of 85 SPL 1W/1M, but the DA175 is an 8 ohm speaker where the Alpair5 is a 4 ohm, so the amp delivers more power to the Alpair, relatively acting as if it were more sensitive. The overall speaker won't, unfortunately, be really sensitive, but it should be able to handle a decent amount of juice.
An externally hosted image should be here but it was not working when we last tested it.
Member
Joined 2003
DcibeL said:
The black stuff is the glue used to attach the dustcap... for the last year & a half the execution of this assembly step has seemed to be a bit random... i changed part of my EnABL recipe on the FE126/127 to accomodate the difference... i've not seen an FF85 with that much thou.
dave
I don't think this is universally true. When D'Appolito first introduced his version of an MTM arrangement (AES paper) it included a specific crossover configuration that, together with the driver arrangement, yielded much more uniform polar response. I'm certainly not suggesting that the project that is the focus of this thread isn't worthwhile--I'm looking forward to reading about its successful development and completion. I'm just suggesting that a close look at D'Appolito's work on MTMs might yield some design ideas that will be helpful as you try to achieve your goal of more uniform polar response.
Few
There are other benefits of using a full-range as a tweeter, of course, as well. It's these other reasons that made me decide to go with this configuration, especially as I'm not an extremely experienced designer.
1) Woofer choice - You can choose woofers based solely on their performance in the bass range, without having to worry about cone breakup and the like. Finding a woofer that is great in both the low end and midrange generally results in an expensive driver. This configuration allowed me to choose a relatively inexpensive woofer that performs great down low, without having to worry about its inadiquacies
2) crossover design - since there is a HUGE overlap between the tweeters resonance (125hz for the A5) and the woofer's breakup, I have a huge range of crossover points available that require no compensation for driver behavior
3) It's different and it looks cool?
1) Woofer choice - You can choose woofers based solely on their performance in the bass range, without having to worry about cone breakup and the like. Finding a woofer that is great in both the low end and midrange generally results in an expensive driver. This configuration allowed me to choose a relatively inexpensive woofer that performs great down low, without having to worry about its inadiquacies
2) crossover design - since there is a HUGE overlap between the tweeters resonance (125hz for the A5) and the woofer's breakup, I have a huge range of crossover points available that require no compensation for driver behavior
3) It's different and it looks cool?
Hi Tresch,
I agree. There's more to the project than the radiation pattern, and it's worth pursuing for lots of reasons. My motivation for pointing out D'Appolito's work was that posts have a way of taking on lives of their own and I didn't want other readers to overlook some of the MTM subtleties. I certainly don't want to discourage anyone from pursuing projects! I share the interest in finding ways to lower the crossover point to minimize the discontinuity between the tweeter (or wide range tweeter-substitute) and the woofers and I'm looking forward to seeing how the project develops.
Few
I agree. There's more to the project than the radiation pattern, and it's worth pursuing for lots of reasons. My motivation for pointing out D'Appolito's work was that posts have a way of taking on lives of their own and I didn't want other readers to overlook some of the MTM subtleties. I certainly don't want to discourage anyone from pursuing projects! I share the interest in finding ways to lower the crossover point to minimize the discontinuity between the tweeter (or wide range tweeter-substitute) and the woofers and I'm looking forward to seeing how the project develops.
Few
To put some details on the D'Appolito configuration:
Interdriver spacing (acoustic centers) of 2/3 of the wavelength at the crossover frequency.
Locked 90 deg phase difference between woofers and tweeter, which can be achieved (for example, but not restricted to) with odd-order Butterworth acoustic x-over characteristics. With 1st order, the off-axis errors are in a -3dB band, with 3rd order it's a bit more, peaking at -9dB at about 30 degrees. But no deep nulls like that we get with a Linkwitz-Riley (which is 0 deg locked phase). To the latter the OP was referring to, me thinks.
- Klaus
Interdriver spacing (acoustic centers) of 2/3 of the wavelength at the crossover frequency.
Locked 90 deg phase difference between woofers and tweeter, which can be achieved (for example, but not restricted to) with odd-order Butterworth acoustic x-over characteristics. With 1st order, the off-axis errors are in a -3dB band, with 3rd order it's a bit more, peaking at -9dB at about 30 degrees. But no deep nulls like that we get with a Linkwitz-Riley (which is 0 deg locked phase). To the latter the OP was referring to, me thinks.
- Klaus
There's a really simple solution to this problem. Here's what you do:
Does that make sense?
I personally am not a fan of using a full range in lieu of a tweeter, as it reduces your efficiency, your extension, and your power handling. I think the best solution is to use an extremely small center-to-center spacing, or use a tweeter with gobs of excursion.
For further reading, download "Application of Linear Phase Digital Crossover Filters to Pair Wise Symmetric Multi Way Loudspeakers part 2", by DB Keele.
- The directivity of your speakers will narrow to 83.6 degrees when the distance between the woofers is one half wavelength. Therefore, do not run your woofers above this frequency.
- For example, if you are using four inch woofers and the center-to-center spacing is seven inches, the beamwidth will collapse to 83.6 degrees at 964hz. (speed of sound / 7 inches / 2) Therefore, don't run your woofers above this frequency.
- Once you do the math, it becomes clear that horizontal arrays are practical, but you need an extremely small center-to-center spacing, or a very robust tweeter.
Does that make sense?
I personally am not a fan of using a full range in lieu of a tweeter, as it reduces your efficiency, your extension, and your power handling. I think the best solution is to use an extremely small center-to-center spacing, or use a tweeter with gobs of excursion.
For further reading, download "Application of Linear Phase Digital Crossover Filters to Pair Wise Symmetric Multi Way Loudspeakers part 2", by DB Keele.
Yes, with convolution filters sort of "everything" is possible, since one can adjust all driver's magnitudes and phases independently within reasonable bounds... and yield a pretty much optimized system in various regard (especiallly tailoring off-axis response to the needs). I'm about to follow that route, but for many DIYers that is not an easy option.
- Klaus
- Klaus
Member
Joined 2003
Well I am currently looking at this project being completed (apart from the cabinet finish) at the end of August. I may have to order another FF85K and hope for the best, as the extra goop on the one I have is sure to cause some trouble, or at least change the drivers' response from the manufacturer's datasheet. I expect that sensitivity and high frequency response will suffer. I'll try it out though, and hopefully won't have to go that route.
Member
Joined 2003
My cabinet arrived, but the builder screwed up the baffle so terribly much it's unreal. I am very upset about it, the builder has come highly recommended from various other projects on the interweb.
The cabinet arrived with the baffle glued on, so I can't remove it easilty and put a different one on. I'm waiting to hear back from him, hopefully he will wish to rectify the issue.
The cabinet arrived with the baffle glued on, so I can't remove it easilty and put a different one on. I'm waiting to hear back from him, hopefully he will wish to rectify the issue.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.