66 • February 2010 • Lighting&Sound America

TECHNICAL FOCUS: LIGHTING / RIGGING

It’s only a chandelier, isn’t it? Just afew lights. What could be so techno-logically interesting about that? Well,that’s what I was thinking almost twoyears ago, when I first heard aboutthe proposed chandelier at theWinspear Opera House in Dallas. (Seepage 46 for the feature story.)However, once I grasped the scale ofthe proposal, it became clear that thiswasn’t your ordinary, everyday chan-delier, if there is such a thing. It allstarted with an artist’s concept picturefrom the project’s architect, Foster +Partners, and the architectural lightingconsultant, Claude R. Engle.This showed a 40' diameter ethere-

al construction with seemingly unsup-ported bars of light hanging in aroughly pyramidical shape from themain auditorium ceiling. Discussionrevealed that the intent was to have

over 300 illuminated bars, which wouldreach down 50' in front of the two topbalcony tiers in a three-dimensionalstructure to provide a centerpiece forthe room while the audience wasbeing seated before the performance.At show time, these bars would riseuntil the ends were flush with the ceil-ing, where they would form a star fieldbefore finally extinguishing at curtainup. The bars themselves needed to beilluminated as evenly as possible,should match the warm white of theexisting incandescent house lighting,and complement the white gold leafon the balcony fronts. The chandelierwas a key part of the room’s artisticdesign and would be the center ofattention—both literally and metaphor-ically—of the audience before the cur-tain went up. Okay, that was the con-cept, but how do you actually build it?

The architects had considered a fiber-optic solution but had not been able tofind a way to raise and lower it 50'while keeping the suspension invisibleto the audience and solely illuminatingthe final 6' portion of each fiber.J.R. Clancy was already heavily

involved in the Winspear installation.As there was a significant riggingimplication to making this chandelierraise and lower safely and quietly overthe heads of an audience, the compa-ny was also invited to bid for the chan-delier addition, which left the projectmanager, Robert Degenkolb, the taskof coordinating an engineering con-cept that would meet both the artisticand engineering requirements. Thechandelier is primarily an artistic instal-lation, with the illumination for theroom coming from conventional lumi-naires, so it needed an engineeringdesign that wouldn’t interfere with theaesthetic concept and was essentiallyinvisible without compromising safetyand reliability. At that stage, J.R.Clancy brought me in to assist withdevelopment of the illuminated rods; itturned into a joint project, with Clancytaking the lead and providing all of thehoists, control, structural support, rig-ging, and suspension while MikeWood Consulting and Scott Ingham, ofIngham Designs, provided the illumi-nated rods, drive electronics, data dis-tribution, programming, and control.The first task for the team was to

develop the rods themselves, asthose would be the only portion theaudience would see, and, once theywere working, everything else in theengineering design would follow fromthat decision. The team experimented

The Retractable LEDChandelierBy: Mike Wood

Engineering a centerpiece attraction atthe Winspear Opera House

Figure 1: Concept. (Photograph courtesy of Foster + Partners) AllphotographsbyMikeWoodexceptwhere

indicated

Copyright Lighting&Sound America February 2010 http://www.lightingandsoundamerica.com/LSA.html

with various materials and illuminationsources before ending up with anacrylic rod and a custom four-channelRGBW LED illuminator that was opti-cally coupled to the top of the rod.We tried many different types of sur-face finishes to ensure even illumina-tion but, in the end, simplicity wasbest, and a polished cast acrylic rodgave the solid bar of light that wasneeded. Cast acrylic performs muchbetter than an extruded version forlight transmission in this kind of appli-cation; the cast material is easier tomachine and form and doesn’t havethe entrained air bubbles and imper-fections present in the die extrudedproduct. Those imperfections glowlike little stars within the rod; they’revery pretty, but not what was wantedhere. A custom 6'-long cast rod, withthe ends polished optically flat, wasselected. For illumination, four stan-dard 3W RGBW LED dies were pack-aged with a custom lens so that theexit angle matched as closely as pos-sible the TIR (total internal reflection)angle of the acrylic; therefore, as littlelight as possible was wasted in cou-pling to the top end. The use of TIR ina solid rod is optically identical to thefiber-optic solution that the lightingconsultant first envisaged; it’s justthat it is 318 separate 6'-long rigidfibers rather than flexible bundles.Although the main desire of the light-ing consultant and architects was forwhite light, as previously mentioned,

the white had to be controllable toget the right color temperature andwarmth to match the color scheme ofthe room, so an RGBW array gavethe flexibility needed.Figure 2 shows an early trial with

various patterns and surface finishes.The plain polished rod (third from theright) performed the best and met thearchitect’s and lighting consultant’srequirements.Once a basic rod shape and style

had been agreed upon, the projectcould proceed to a mock-up of the

entire chandelier—or at least half of it!J.R. Clancy took over a local theatre inSyracuse for a day and constructed afull-scale model using black cord andwhite rope of the same diameter asthe acrylic rods. Some effective use ofthe stage lighting installation complet-ed the job, as shown in Figure 3.With this 1:1 model to approve the

overall scale and, in particular, thediameter and length of the illuminatedrods, the architects and ownerssigned off on the project, and we hadthe green light to proceed to adetailed design.Although we were using 3W LEDs,

we wanted to run them at very conser-vative levels—the installation needs tolast at least 10 years with daily usage,and the key to getting a good life fromLEDs is good thermal management.Each acrylic rod was capped with a

head, which formed the enclosure forthe LEDs and their associated drivers,and also provided secure mechanicalsuspension for the rod. The head wasalso the heat sink for the LEDs; thethermal design, using heat flow analy-sis, was specified to allow at least50% headroom on the expectedpower usage. The final head designwas an interesting collaborationbetween the mechanical engineers,whose main concern was the safetyaspects of hanging a 6' rod over theaudience; the electronics engineers,with their requirements for heat flow,EMI shielding and connectivity; andthe understanding by all that this wasin view of the audience and thereforehad to fit the aesthetics of the design.It seemed like the simplest componentin the installation—just an aluminumextrusion—but it turned out to be oneof the most complicated and criticalparts of the design.Nothing on this project was sim-

ple—not even the cabling. Each rodhead was suspended from a singlecustom cable, which contained ashielded twisted pair for data, a pairof wires for 24V DC, and three syn-thetic fiber suspension members. Thewhole assembly was enclosed in amatte black sheath to produce a finalcable that was just 0.25'' in diameter.This cable had to pass over sheavesand blocks onto a motor drum hoistwithout breaking, so ultra-flex insulat-ed conductor cables of a type morecommonly seen in moving lights wereused for the construction. Beforecommitting to the production run,Clancy ran both proof load testingand an accelerated simulated life test,confirming that we could successfullypass data and power through thesystem over the life of the cable withthe specified connected load. Thehead and rod only weighs about 5lbs;however, we had a concern that therecould be problems if the rods everknocked and entangled with eachother, so minimum 8:1 safety factorswere used in the development of thesuspension system.Now to the hoists: There are 318

rods in total, and the ideal situationFigure 2: Rod testing.

Figure 3: Early trial using white ropes as aproof of concept and scale.(Photograph: Robert Degenkolb)

68 • February 2010 • Lighting&Sound America

TECHNICAL FOCUS: LIGHTING / RIGGING

would be if there was one hoist perrod. However, cost put that possibilityout of the running very quickly, and asolution using fewer hoists had to be

found. In the final installation, the 318rods are distributed over 44 liftingmechanisms, each of which controlsup to eight rods. In the chandelier’s

operational shape, those 318 rods areat widely different height bands, buteach hoist has to be connected toeight rods, which are all at the sameposition. To make this trickier, theoverall conical shape of the chandeliermeans that rods at the same heightare never adjacent and the ceiling itselfisn’t flat; instead, it is both domed andat an angle. To accommodate theserequirements, the hoist cable routing inthe grid from each hoist is incrediblycomplex. A three-dimensional layouthad to be prepared (Fig. 4) so thatnone of the cables interfered with eachother as they passed over or undermultiple other cables on their routefrom hoist drum to loft block.Figure 5 shows how that translated

to reality in the installation. After a lotof work by skilled installers, it endedup being remarkably tidy.Also visible in Figure 5 are the tops

of tubes in the grid that provide guidesand protection for the rods when theyare parked in the void between theplaster ceiling of the auditorium andthe grid. Figure 6 shows those tubesas seen from the surrounding catwalks.The tubes also contain twin stacks ofbrushes, which serve to both keep therods clean and to provide a light block.As mentioned above, each hoist

raises and lowers up to eight rods, andevery LED in each of those eight rodscan be individually controlled from datasent down the cable. We had concernsabout hanging 318 long antennas in anopera house and so wanted to mini-mize that data; we also wanted everyrod and its circuitry to be completelyidentical and interchangeable. The bestsolution was to have an intelligent datadistributor on every hoist, which took inDMX512 data, then formatted it andrebroadcast it down the eight cables,with each rod only receiving its owndata. This has the added benefit thatthe heads are all being fed with homeruns, so no addressing or networking isneeded. The net result is a vast reduc-tion in the amount of data passingdown the cables, which mitigates therisk of electromagnetic noise causinginterference with the sound systems.That, coupled with the silent operation

Figure 4: Hoist and cables grid layout. (Drawing courtesy of J.R. Clancy)

Figure 5: A (not so) tangled web.

of the hoists, meant that we passedthe scrutiny of the audio consultants—pretty important in an opera house! Tofurther reduce data flow, each rod hasa processor and does all its own cross-fading, interpolating, and smoothing ofthe incoming eight-bit data to an inter-nal 16-bit and adding pseudo thermaldelay so that the output matches anincandescent dimming curve. Key tomaintaining the incandescent illusion isavoiding the final “blink” as LEDs turnoff, so a great deal of care was takenwith the final 10% of dimming. This isparticularly important with house lightsas, by their nature, they nearly alwaysturn off into a blackout where theslightest imperfection is apparent.The data distributor is mounted on

the rotating portion of the hoist, withpower and data fed to it through adata-quality commutator. The softwarein the data distributor and headswatch for any glitches in the data feedthrough the commutator and hide it sothe output remains smooth. We werehelped in this by the application—withnormal entertainment lighting, speedof response is critical but, in this case,immediate response can take secondplace to smooth operation. The cross-fade speed is provided as a program-mable operational parameter so, if theWinspear staff wants to change thefade rate through the ETC Mosaic sys-tem, they can.Figure 7 shows some of the

installed J.R. Clancy hoists suspendedfrom a Unistrut grid that is attached tothe underside of the building structure.At the front of each hoist is a mountingframe diverter assembly that directseach lift line to the ceiling tubes. Thediverters incorporate a spring-loadedoverload assembly that stops the indi-vidual hoist in the event a lift linebecomes snagged during operation.The hoists are controlled by a JR

Clancy SceneControl 500 through awired touch-screen pendant interface.The pendant plug-in stations are locat-ed so that the operator has a goodview of the moving rods at all timesand allows for individual, group, andpreset control of the rod positions.Interestingly, the lighting control room

is not normally one of those stations,as it has no view of the chandelier!The pendant also provides a wired E-stop connection to the system andallows for password-protected adjust-ment of the variable speed hoist drivesto reconfigure the system.Control of the 1,590 channels driv-

ing the 318 illuminator heads (eachuses five control channels—four colorsand a timing channel) is providedthrough two linked ETC Mosaic showcontrol systems which, in turn, aretriggered from the ETC UnisonParadigm and AMX touch panels dis-tributed throughout the building.Programming made full use ofMosaic’s ability to map an array ofchannels to a matrix which dramatical-ly simplified the process. Programmingeach of those 1,590 channels by handwould have been painful. Instead, wewere able to overlay moving patterns

and cloud imagery to give a constantlyshifting dynamic to the chandelier.Figure 8 shows the final chandelier.

You can’t tell from the photo, ofcourse, but the programming is quitesubtle; each rod is continuously shift-ing its color and brightness on eitherside of the defined warm white by asmall amount, and no channel is everstationary. This movement is notenough to be overt, but it gives thewhole thing the kind of life that you getwith a candle flame chandelier. Wealso programmed in red-shift as thechandelier dims so that, again, itmatches the color shift in the incan-descent house lights on the balconyfronts. These fades are timed to matchthe one-minute movement of thechandelier as it flies out so that, whenthe rods reach their top preset withtheir bottom ends level with the ceil-ing, the rods have dimmed down to a

Figure 6: Guide tubes between grid andplaster ceiling.

Figure 7: Hoists.

Figure 8: Finished chandelier.

TECHNICAL FOCUS: LIGHTING / RIGGING

very warm star field glimmer with asuperimposed twinkle (Fig. 9).The Winspear Opera House is now

open, and the chandelier runs everynight in front of an audience with no

idea of the engineering complexitiesneeded to perform such a simpleeffect. Instead, it’s an art installationthat, I hope, compares well with theoriginal artist’s concept of Figure 1.

That’s as it should be—we are in thebusiness of theatricality and illusion,after all.The installation was supervised by

Robert J. Degenkolb, of JR Clancy,and Mike Wood and Scott Ingham, ofMike Wood Consulting and InghamDesigns, LLC, respectively, and wasperformed by Joel Svoboda, KenEggers, Charlie Hulme, Byron Payne,and the International Association ofIron Workers Local 263 (Dallas/Fort Worth).

Mike Wood provides technical andintellectual property consulting servic-es to the entertainment technologyindustry. He can be reached atMike@mikewoodconsulting.com.

Figure 9 - Star field.