| Home | | | Projects | | | RealAudio™ | | | About me |
While doing research looking for new speakers to use in my music/video system, I ran across a link to Siegfried Linkwitz's web page.
I was already familiar with some of Mr. Linkwitz's work as the co-inventor of the crossover topology which came to be known as Linkwitz-Riley, as well as his work with Audio Artistry. You can refer to the Audio Artistry web page for some reprinted reviews, as well as the review at Stereophile Magazine.
Mr. Linkwitz has been kind enough to publish much of what he has learned about loudspeaker development, specifically relating to dipole loudspeaker performance and active crossovers, on his web page.
The Phoenix system consists of three parts:
The main panel of the Phoenix system is a mid-tweeter-mid arrangement with an open baffle. Since the tweeter has an enclosed back, this results in a dipole radiation in the lower midrange, a directional radiation in the treble, with the change from dipole to directional smoothed a bit by a bit of poly fill in the back of the baffle (and in my case, grill cloth which covers the back side of the main panel).
The only thing that I did differently in my implementation of the Phoenix was due to my lack of a table saw. In order to avoid having to cut long straight edges with a guide board and a hand circular saw, I used pre-cut MDF shelving. I have to admit that this was only a semi-successful experiment. It did avoid the problem of cutting straight edges, but the MDF is not quite as strong and non-resonant as I had hoped. If I ever rebuild the main baffles, I will probably use the baltic plywood that Mr. Linkwitz recommends.
Following are some pictures of the progress of my Phoenix system. Each small picture links to a page with a larger version of the same picture. Just click anywhere in the small picture to go to the larger version.
This is a picture of the MDF shelving after cutting to length, and routing
the first insets for the midrange drivers.
The sequence for cutting the mounting holes is to first cut the "rebates," then cut the smaller diameter hole all the way through. A circle jig from Jasper Audio is invaluable for cutting the holes. I used a model 200 which I bought from my local Woodcraft store. They mostly sell mail order, so check out the link if you need some high quality wood working equipment. I have to admit that they are usually a little too upscale for my budget, so I get most of the equipment at Home Depot, and just go to Woodcraft for the specialty supplies I can't get elsewhere.
This is a front view of the main panel after cutting the driver mounting
holes, and gluing the side pieces and braces.
Another slightly experimental aspect of the assembly is relying on glue only to attach the sides and internal braces to the front panel (no screws). On solid wood or even plywood I would not be very worried, because the glue joint is as strong as the wood, but I'm not sure yet how well the MDF will hold up. It is strong when pressure is applied in some directions, but does not have much strength in other directions. I'm afraid one of the weak directions is torsional force on the edge, so I'll have to see if the joints hold up long term, or if the MDF cracks from the weight.
This is a view of the rear of a panel, showing the braces which are installed
just above and below the tweeter cutout.
This is a view of the main panels and the rear supporting spine from all
three sides. This is just to give an idea of the proportions (depth vs.
width).
This is a rear view of one panel next to the front view of a second. The rear
cavity has been painted black at this point.
This is a view of the front of all three panels after being laminated
with Wilsonart high density plastic
laminate (but before the driver cutouts have been cut throught the laminate).
I found a laminate which matched fairly closely the surface of
my Mitsubishi television, so I decided to use that as a front surface finish
(since these will eventually end up in a front/center/right configuration
around the television).
This is a completed main panel awaiting the drivers.
The white plaster column behind the panel conceals a surround
speaker.
Not obvious in this
picture is the hook-and-loop fastener along each edge to hold the grill cloth
in place.
The grill cloth will cover the rear of the speaker, not the front.
That is partially for aesthetic reasons, and partially because the Phoenix
is designed to smoothly transition from dipole radiation in the bass to
directional in the treble, and the grill cloth (along with a bit of poly-fill)
will help smoothly roll off the high frequencies coming off the back of the
midrange driver.
What isn't visible in the previous picture is that the cable connection to
the main panel is through a Neutrik
Speakon® connector. The picture to the right is a close-up photo of
the cable and connector for the tweeter and mid-range drivers. The picture
isn't very impressive, but if you've ever dealt with having to make sure
four separate wires get connected to the proper terminal without miswiring,
you can appreciate the convenience of the Speakon connectors.
The advantages of the Speakon®
connector are that the connector locks in place, so there is no danger of
the cables being accidentally pulled loose, and the connector is 4 pole,
so that the tweeter and midrange channels are always connected correctly,
there is never a danger of accidentally connecting the low frequency amp
to the tweeter (which could damage the tweeter pretty quickly).
The connectors used are
NLMP4R
on the speakers, and
NLMP4
on the amps.
I decided on some
Sanus stands from
Best Buy for the main panels. I had originally considered stands which
hung the main panels from cables, as suggested by Mr. Linkwitz as a way to
reduce structrural vibration transmission, but found the main panels too
heavy to make that practical.
The panels sit nicely on the stands, although perhaps a little precariously
given how much of the weight is at the top of the panel (because of the top
mid-woofer).
The dipole woofer is described very adequately on the Phoenix woofer page at Linkwitzlab.com, so I will direct you there for all the technical details.
This is just to give a better idea of how you go from pieces of wood
to the woofer shown on the Linkwitzlabs page. Here are all the cut pieces
lined up on the floor. For reference, the large pieces are 14"x19"
and 16"x19".
Here are the top and inner pieces clamped and glued.
This is just a view from a different angle of the woofer pieces
clamped for gluing.
This is the woofer with top and baffles glued together and drivers installed.
The only step left is to screw the base plate onto the assembled woofer.
Grill cloth can wrap completly around the assembled woofer to hide the
baffle openings if desired.
Dipole speakers require compensation for the bass roll-off caused by the open baffle. The Phoenix system incorporates that compensation into an active crossover which also handles dividing the low, middle, and high frequencies to feed to the woofer, midrange, and tweeter drivers respectively. Note that this is done at line level, not speaker level, so the system requires separate amplification for each driver.
The crossovers and amplifiers for each channel are mounted together, with
amplification provided by
LM3886 power op-amps. The IC's are mounted on
model PM21 PCB's from
Marchand Electronics. This picture
is the crossover PCB, amplifier PCB, and power supply tacked together
temporarily on a piece of plywood.
The crossover circuits were modified by the addition of a balanced input circuit using a simple to construct instrumentation amplifier buffer. The RFI filter RC networks were not installed on the Phoenix board and were instead built as input connector filters placed directly behind the XLR input connectors. The ceramic chip capacitors specified in the Phoenix PCB bill of materials were replaced with surface mount capacitors soldered directly to a solid ground plane, with a brass sheet soldered to the ground plane and bent to form part of the mounting bracket for the filter PCB. That mounting arrangement gives a low inductance path from the filter capacitors back to the chassis shield. I found that for optimum RF rejection I had to use ferrite beads rather than just bare wires to connect the XLR connector to the input of the RC filter. I used the RFI tester I built to check the susceptibility of the input to RFI demodulation.
For a more thorough discussion on the benefits of balanced impedance connections, see the writings of Bill Whitlock and the late Deane Jensen at Jensen Transformers, the Clean Audio Interface Guide from Benchmark Media, and the June 1995 issue of the Journal of the Audio Engineering Society.
The crossovers and amplifiers are mounted in rack mount enclosures from
Middle Atlantic Products.
Markertek carries their enclosures,
search for "CH-3" in the search engine on the home page. Markertek is
also a good source for the
Neutrik XLR and Speakon® connectors
used as input and output connectors.
Having finished the enclosure assembly, I would probably opt for enclosures
from Hammond or Gray Enclosures over the Middle Atlantic enclosures. The
MA boxes were about $25 cheaper per enclosure, but were more diffucult to
assemble than I would have liked.
These surround speakers are not technically part of the Phoenix project, but I'm including them here since they are also another Linkwitz design. Mr. Linkwitz has a nice description of his surround sound configuration on his web site, including a link to the construction details if you would like to build your own (as I did).
They are diffuse surround speakers with limited high frequency response. They would work well for the original Dolby recommendations for matrixed surround, and also work well in most uses for discrete surround as well (e.g. Dolby Digital). They will probably not be a good match for surround music which places direct instruments in the surrounds, but they work well for ambience, and are a fair match for reproducing the surround array effect from a movie theater.
This is the surround speaker enclosure, made from 4" PVC pipe, mounted
on a wooden base. An end cap forms the bottom of the enclosure, and an
adapter flares to a 6" opening to accept the surround driver.
My wife apparently has different tastes or a different level of tolerance
than Mrs. Linkwitz, so I was requested to find an alternate look for the
surrounds. This is a decorative plaster pillar, with holes bored in the top
and bottom so that the PVC pipe speaker enclosure can fit inside.
This is a close-up of the driver mounted on top of the PVC pipe, which is
installed in the decorative plaster pillar.
This is a view of the surround speakers installed in my living room (looking
from the TV to the back of the room).
You may notice that the left surround is not placed symmetrically with respect
to the couch, due to the fireplace hearth in the way (out of frame in this
picture). I haven't decided what to do about this yet, or how noticeable it is.
My only options are to move the couch slightly, which causes other problems
in the room, or move the surround up onto the hearth.
If you have the Adobe
Acrobat reader (or equivalent, such as Ghostview on Unix or Linux), you
can take a look at how I plan to configure my Phoenix system here:
plan view of system
elevation view of system
The linked picture is a photo of my Phoenix system installed as the
left/center/right in a 5.1-style video and music system. The center
channel does not have the dipole woofer because of space constraints.
The surround channels are handled by the single driver speakers described
above.
Page last updated: 16 July 2002
| Home | | | Projects | | | RealAudio™ | | | About me |