"There are billions of precious things in the universe that we have no time to study"

- Karlheinz Stockhausen

Stockhausen's Himmels Tur (Heaven's Door)

The first Heavens Door was designed by Stockhausen in collaboration with a German cabinetmaker. The instrument is a directly struck Idiophone and is ostensibly a double door with leaves consisting of rectangular tone panels without resonators. There are twelve panels which, in accordance with the range required by the composition, encompass a minor third. The first iteration of the instrument was rectangular in shape and employed a different method for affixing the tone panels to the door leaf frame. Below is my concept sketch for the construction of a new Heaven's Door which responds to Stockhausen's first conception of the instrument "to resemble an old church door". My design also reevaluates the panel hanging from the perspective of a luthier, employing a method inspired by marimba design. The second iteration of the Door was built for Stuart Gerber, assistant professor and percussion coordinator at Georgia State University in Atlanta had its inaugural performance (and North American debut of the piece) at Spoleto Festival USA in Charleston in June of 2007.

After the instrument's debut I was delighted to be able to write an account of the making of the instrument for publication in the Galpin Society Journal, . Following is my abstract:

"In June 2006 the world was introduced to a new musical instrument conceived by the groundbreaking German composer Karlheinz Stockhausen for the performance of his piece entitled Heaven's Door, the fourth composition in a series titled KLANG/ - Die 24 Stunde des Tages. The present article concerns the documentation of the instrument created for that performance as well as a second iteration of the instrument, which was constructed by the author. This second instrument was modified aesthetically and in its acoustical mechanics under the guidance of Stockhausen's percussionist, Dr. Stuart Gerber, in preparation for the North American debut of the piece in Charleston, South Carolina at the Spoleto USA festival in June, 2007. The article explores the challenges of creating the vertically oriented struck plate idiophone (conceived of by the composer in a dream) and the relationships between composer, performer and instrument-maker in the development of a completely new instrument."

Click here for the full article and here for the color plates.



Wednesday, October 31, 2012

The Door was set to be performed at Lincoln Center in NYC yesterday, but unfortunatly Hurricane Sandy put the skids on and the door is headed back to Atlanta unplayed. A reschedule is in the works. In the mean time, WABE did this article on the piece: WABE on City Cafe and a corresponding Georgia State Article

October4, 2012

In preperation for the New York performance, local station WQXR did this profile, including a posting of a full-length recording of the piece here.

August 27, 2013

Georgia State University Magazine shot a video feature on Heaven's Door. Click here. or here.


Project Log

Friday, January 12, 2007

Commencement of commission for Stockhausen's Heavens Door. I am excited to announce that I have recently undertaken construction of one of these instruments, conceived by Karlheinz Stockhausen for the performance of his piece "Heaven's Door". Above is an early drawing of my design for the door, showing superimposed over the front elevation the geometry and proportion in comparison to Vitruvian Man and the Golden Section.


Project Updates:

Wednesday, January 31, 2007

The drawing is complete. My design shows a vertical emphasis over the first door, derived from the Golden Section and the geometry of a three-center arch.



Friday, February 2, 2007

Material has been obtained for tone panel mock-ups to prove the new panel hanging method. These wenge boards have been planed to thickness. Here are three 15cm wide quartersawn boars laid up together as they will be glued. The relief profile will be cut at the backs of the boards and the ligature hole drilled at each board before they are joined.

A full-scale front elevation was drawn for Stuart to approve the new shape and proportions.


Monday, February 12, 2007

The location of the fundamental nodes of the tone panels is established at 256mm from the centerline. After researching the several equations conceived to determine bar dimension and arch vs. wood density as relates to pitch for the marimba (Fletcher and Rossing, Murray and Greated, etc), I came to appreciate that the proportions of the Heaven's Door tone panels exist outside of the influence of mathematics. Marimba equations assume a certain relationship between length, width and thickness. The Heaven's Door panels must be equal in width (2 panels wide) and roughly equal in height (6 panels high) and of a thickness that can be obtained in solid lumber of the desired specie (1" thick in the case of wenge). The thought of laminating the tone wood was considered and could be explored for future Doors, but the parenthetical improvement in tone from this method must remain outside the scope of this project. In my experiments, an adequate, if not exemplary, tone was achieved with 1" thick material using a simple trial and error method for determining the best node location to enhance the fundamental tone. The panel was placed on top of two moveable, insulated sticks laid on a graduated framework. The reverberation of the fundamental tone was timed with a stopwatch when struck. This node is where the hole for the cable will be placed, as shown in these photos of boards awaiting to be joined into a panel and the template.


Monday, February 19, 2007

Construction of the super-frame has commenced. Quarter-sawn white oak has been obtained for the framework. The following photos show the process of assembling the bottom rail. the super-structure frame is a lamination of three pieces of white oak, approximately 1" in thickness. The ends are offset so as to make a dado-type joint with the adjoining piece. Here we see that both ends of the bottom rail have a central tenon that stand proud of the quartersawn face lamination by the width of the vertical supports, which is 15cm. The edges of the rail will be planed flush after the clamps come off. Each piece will be joined with the next with glue and dowel pins

A finished tenon


Monday, February 19, 2007

Work on the super-frame arch.

The various parts are identified with a lettering system. A1 (shaded here) is the left side front-facing lamination at the base of the arch. Its identical sisters are A2, A3, and A4, A2 and A4 being the rear-facing laminations. B1 is the central lamination at the left side base

Each piece is allocated on the boards. The A and B laminations have the greatest arch and require the widest boards. Here is a 10" wide oak plank for A1.

Here is the same board planed to thickness and surfaced.

This is the poor man's CNC machine. I have made full-scale templates in CADD which are glued to the face of each board. The exact angle is noted and cut on the miter saw. . .

. . .then cut on the bandsaw.

Here is the finished A1 lamination. The ruler and the line drawing beyond indicate the extent of the B1 lamination that will be underneath and form the locking dado with the next 3-lamination assembly of the segmented arch.


Thursday, March 8, 2007

The first assembly of the arch is glued up. This is A1, B1, and A2. Note quartersawn cuts for the A laminations and flat sawn for the B core.


Sunday, March 11, 2007

One of the stiles is glued up.

A detail of the bottom of the stile

The arch components are cut

Two segments of the arch are glued up but not yet glued together. Lamination E is at the apex of the arch.

Detail of the E-D-E lamination

The rail laminations are planed level by hand before feeding into the planer

The arch laminations are made flat with the scraper before final sanding with a palm sander

The framework is starting to rise.

The complete framework is laid out on the floor for final fitting before gluing up. The paper has a full-scale printout of the arch geometry from CADD to compare the work against the drawing, making apparent any need for slight adjustment.

The cheeks of the tenon are scored before gluing to ensure good glue penetration

The top of the arch is assembled. The A/B and C/D assemblies on each side are joined at the apex with the E lamination.

The stretcher rail for the bottom of the frame is temporarily clamped at the bottom of the arch as a quick method to ensure the proper distance is held.

The arch is assembled.


Sunday, March 18, 2007

The arch is worked over with scraper and beltsnader prior to assembly to the rails

The clamping pattern for ensuring the arch and bottom rail snug completely to the stiles (side rails)

The frame is assembled. Work will now commence on the individual door leaf frames.


Wednesday, March 27, 2007

The door leaf frame is laid out. The "A" lamination of the arch still has its template paper on. This is a dry fitting to check the mating of the parts. The door leaves are constructed of two laminations rather than the three of the outer frame. The addition of the tone panels will make the overall assembly close to the thickness of the outer frame.


Monday, April 2, 2007

The same exercise as before with the outer frame is performed for the door leaf frames: Each piece is assigned a letter code. Full size templates for the arch segments are printed full-scale, pasted to the wood and cut on the bandsaw. The straight pieces are planed and chopped to length.

rails for the door leaves. Again, a combination of flat and quartersawn pieces.

Rails glued up

Both door leaves in progress. There will be a great amount of tension on the door leaves from the wire rope assembly of the tone panels. The leaf framework will be further strengthened by the uppermost curved tone panel which will be permanently affixed across the top of the leaf, more in the manner of a soundboard rather than a floating tone bar. This design will counteract door warping.

An in-progress leaf is resting in place inside the outer frame.


Friday, April 6, 2007

The framework is on saw horses for the final surfacing. The door leaves are near completion with the archwork in progress.


Tuesday, April 11, 2007

The archwork for the door leaves is glued up. The hinges have been purchased. There will be three Hager anti-friction medium duty lift-off hinges per side.


Sunday, April 15, 2007

The outer frame gets a final working over with the palm sander before the door leaves are attached

Holes are drilled for the tennon pegs

Tennon pegs are turned from the same white oak of the frame.

One of the door leaves is hung


Sunday, April 22, 2007

The support brackets for the first Door were metal and required through bolts which are visible from the audience side. This is one of several subtle details I felt were too 'Machine Age' for this project and I have endeavored to make the new Door rigid without the use of steel. My bracket design strives for simplicity and minimal exposure. Here is a strut and one of the two dadoed feet. There is a bracket situated behind each of the two stiles of the outer frame, attached with bolts to threaded inserts buried in the framework. There is no visible support from the audience side. For the performance installation, full-height curtains are hung on either side of the door to obscure the brackets, as in the debut performance.

A general view of the mayhem. A half basement is not large enough for this project and the gardening tools. Here the brackets are installed and the finishing has begun on the door leaves. The skeleton for the platform extends in place in front of the door. I have created two large cavities for the resonation of the footwork required for the performance. The platform of the first Door was merely a hinged (for easy transport) piece of plywood. Mine has greater thickness to coincide with the dimension of the bottom rail (absent in the first Door) and has the bonus effect of creating these resonating cavities. The platform will be covered with 2" wide strip tongue and groove white oak flooring.

I hate this stuff. I am begrudgingly filling the grain of the oak for a fine finish. This is a test area. I quickly realized I should be tinting the filler because the final stain is quite dark. This view also shows the bottom cleat of the door leaf. It is screwed on rather than glued so it can be repositioned - or remade as the case may be - to adjust the distance of the bottom anchoring points away from the frame.

Goop in progress and the resulting product. The piece will take at least two more coats of stain before it is dark enough.

Some of the wonderful flake that turns up in the quartersawn oak. This is the rear of a door leaf, which is a combination of quarter and flat sawn. The face of the outer framework is entirely quartersawn and shows the flake throughout. This will certainly reinforce the 'old door' aesthetic Stockhausen conceived for the piece.


Sunday, April 29, 2007

The platform in progress. The decking is standard 2 1/2" wide tongue and groove white oak strip flooring. It will hold up better to the blows of the performance than plywood. However it does make the platform very heavy - almost as heavy as the door frame.

A general view of the progress. The frame and platform are stained but do not have a final finish (the spotlight is casting on the platform). A clamp temporarily holds the platform tight to the frame to prevent tipping forward. The platform will be attached with two bolts from the rear of the door frame.

The swaging is complete. The "Ultra-Tec" fittings have been permanently swaged to the 1/8" 7x19 wire rope with 3/16" o.d. vinyl coating. This is also the platform.


Saturday, May 6, 2007

Progress begins on the tone panels. The first Door had panels of six different varieties of wood, chosen by Stockhausen for their various densities to produce different pitches. The method for achieving pitch for the second Door is to cut relief dados of various widths on the back of each panel, such as the method used in marimba bars. This makes the use of different woods unnecessary, however the second Door will have tone panels made of both wenge (used in the first Door) and sapele (not used in the first Door). Stuart reported that the wenge was the best sounding of the panels of the first Door, so we attempted to make all the panels of our Door with this wood, despite its liabilities of weight and cost. I was lucky to find quite a number of quartersawn wenge boards here in Atlanta, but alas not enough for the entire project. My suppler, who I have worked with for over 15 years, suggested that finding more quartersawn wenge - from any location - in the timeframe of the project was highly unlikely. Rather than compromise and use a less-than-quartersawn cut, the decision was made to move to another wood. Sapele was available quartersawn and has close to the density of wenge. It was an acceptable substitution. The second Door will therefore have six panels of wenge and six of sapele.

Two sapele boards being marked for the hole placement. As previously determined, the first node is at 256mm from the centerline. The board centerline is marked and the template is used to autopunch a set hole

The set hole is circled. The brad point bit is not tall enough to penetrate the board. . .

. . .so an extension bit is used off the drill press to finish the hole. The extension bit will be used again to clean out the hole after the panel is assembled. The 5/16" hole allows easy passage for the smaller of the two swaged cable ends (just over 1/4" diameter)

Rods the same diameter as the hole are pushed through the holes for alignment during assembly.

Two sapele boards laid up. Each panel is between 3-4 boards tall.

After the clamps come off, the seam is planed with the scraper. . .

. . .and fed through the drum sander. This is a wenge tone panel

The first panel is set in place with the wire system. a stick with holes in the appropriate place for the tensioners is temporarily clamped in place. Small rubber washers are set between the bottom of the panel and bottom rail. The tolerance between the panel and the framework is tight, due to the fact that the upper panel is the source of the wires. To keep the panels off of the framework when struck I have inserted two thin oak sticks behind the panel at the node. The dampening affect is negligible.

A detail of the assembly


The newspaper shot


Wednesday, May 9, 2007

A tone panel having the relief cut into the back. A 2" wide by 1/8" deep channel reduces the fundamental by a semitone compared to an unrouted panel. Widening the channel in 3/4" increments reduces the pitch an additional semitone.

The cremone bolt set has arrived. The rods slide on the attached tabs. Decorative covers for the tabs are in the bags. A doorknob is inserted into the box. I'll be trolling the antique stores for that.


Progress on the tone panels. The spandrels (upper curved panels) are in place, two wenge panels are in their correct places on the left. On the right a sapele panel that occurs higher in the order is for now sitting on the bottom. I am experimenting with different stick lengths behind the panels to hold the panels off the frame at the node. At this point the decision was made not to permanently affix the spandrels to the frame in a "soundboard method". To preserve the ability to change out ALL the panels in the future, a new method is needed to hold the panels in place. Removable bolts located at the node location is sufficient to resist the wire rope tension. I believe the framework is stout enough to not require the extra stiffening of a permanently affixed spandrel.

Details of the progress

Detail at the back of a spandrel showing the routed pockets for the tensioners. The swaged wire rope is threaded up through the bottom of the spandrel into the tensioning sleeve which is just barely visible at the bottom of the pocket. A hex allen wrench turns the sleeve and tightens the rope. Two bolts tapped at the fundamental hold the panel to the frame. The bolt is tapped into a threaded insert in the panel to make repeated disassembly possible.

The lowest panel of the right hand leaf is tuned against the left hand leaf.


Sunday, May 20, 2007

The first leaf is hung with all its panels.

The right-hand spandrel panel is being tuned. Because of its asymmetrical shape, this panel's relief must be routed differently from the others. I am experimenting with removing material to balance the mass along an off-center and skewed centerline

The spandrel panels are held by two bolts, on the node centerline, at the top edge. No further support at the bottom is required, however to keep the panel from touching the framework I have set a stick into the frame along the bottom edge and placed two felt pads, again at the node, to insulate the panel. Note also the asymmetrical routing of the back. This is the final week of work on the Door. The instrument will be transported to Stuart's studio at the end of this week.


Tuesday, May 22, 2007

The panel hanging system: The upper panel is in place, held off the frame by its bolts and the stick. The wires are threaded up into the upper panel and the second panel is in place. The rubber washers are in place on the wire to separate the panels and the third panel is being slid into place. . .

. . .now all the panels are in place. The bottom washers are on and the bottom rail is ready to be attached. . .

. . .bottom rail now in place with the washers and lock washers in place. Now the bottom rail is screwed on in seven staggered locations. The discoloration on the rail is the result of the piece being dragged on the floor to get it in place. This will not be seen once the door is assembled.

Both leaves are hung. The panels do not yet align properly because the right-hand leaf is not tensioned. Note outward buckling of the panels on the right edge. The left-hand leaf does not have its panels in the proper sequence. The sapele panels will be stained darker to match the wenge. The desire is to obscure the panel edges so the audience perceives a continuous plane over which different sounds are obtained, heightening a mystery to the door. This idea contributed to the deletion of the routed edges, originally planned for the panels. The routed edges also would hinder the glissandi required for the piece.

This rear shot shows the differences in the panel routing. Stuart and I have designed a pitch range for the instrument of about an octave and a third, with care not to make any of the intervals a true step or half-step and indeed no combinations of pitches to be a perfect third, fifth, octave, etc.


Thursday, May 24, 2007

The panels are tuned and in their proper sequence. The sapele panels must now be stained.

The drawing has been updated with hardware tags and the revisions made to the design including the spandrel hanging method and the support bracket and platform change.


Friday, May 25, 2007

The sapele panels are stained. The gloss is temporary from being wet. They will dull to the level of the unfinished wenge panels

There's an old woodworking saying: you can take wood away but you can't put it back. This is the right-hand spandrel that is currently pitched( (incorrectly) lower than the left-hand panel. It was the first of the two spandrels to be routed and I wound up taking too much material away in an attempt to get the panel to resonate properly. Rather than scrap the panel and make a new one, or worse, repitch the other spandrel and consequentially the other 10 panels lower to compensate, I am scabbing on pieces of wood to the routed areas and removing material at the outer edges in an attempt to raise the pitch. I have also discovered that the panels change their pitch, seemingly arbitrarily, after the routing. (!!!) Having sequentially pitched the panels the first time, I found that after a couple of days of setting that the pitches moved and the panels we no longer sequenced properly. This was disturbing at first, and I don't have an explanation, but thankfully the problem did not represent itself when I made a second pass at them. (I have stamped numerals into the edge of each panel to avoid mistaking one for the other) So I have now pitched all the panels twice. The panels will be hung for the second time in Stuart's studio and we shall see if the wood has truly stabilized at a pitch and if indeed they meet the intervals required by the piece. We are not necessarily matching pitches called for by Stockhausen, but Stuart, having a feel for what the pitches should be for a successful interpretation of the piece, is guiding the work.

This is a rather inauspicious last image for this series, but it is the last night of work, and this is what it looks like in the workshop. The platform is turned up on end to resecure one of the t-nuts that holds the platform to the frame. The outer frame is resting against a floor joist, and along with the brackets, resting on the table in the foreground, has received a top coat of tung oil. The door leaves rest against the table in the foreground and the troublesome spandrel in on horses in the middle with its coat of stain. The other panels are leaning against the far wall. The whole lot is headed out the door tomorrow to Stuart's studio at GSU.


Tuesday, May 29, 2007

To build you must destroy. It is the cycle of life. This is the ceremonial demolition of the basement door to facilitate the removal of the project. Our basement door is 6'-8" tall but fortunately the builder put a transom window over it, later removed and covered over with vinyl siding. We're putting the window back after the project is over.

The door is delivered to a Georgia State rehearsal studio

Some process images of installing the panels

A good closeup shot of a "clip-on" fitting at the bottom cleat.

Complete installation. Next stop, Charleston.


Monday, June 4, 2007

Images from the Spoleto weekend. . .

The setup, showing also the instruments required for Kontakte

Stuart rehearsing the piece.

I took one discrete photo during the performance, without a flash of course, that actually came out pretty good: capturing the moment when the child passes through.


Thursday, November 1, 2007

The door is back in the workshop for some rework. After the Charleston performance it was clear that the bulk of the outer frame would make future performances difficult. Stuart has asked me to make the outer frame disassemble as the first door. Wanting to avoid the metal straps of the first door and keep with the timber-frame aesthetic of my design, I have made pinned mortise and tennon joints.

I am also taking the opportunity to redo the method of attachment of the braces, removing the threaded bolt inserts and adding a dovetailed cleat that the brackets will slide on to attach.

This rework will also include a new method for attaching the upper panels to the door leaf framework as the resonance of these upper panels was subdued in comparison to the others. The experimental nature of the project can not be overstated.


Sunday, January 20, 2008

The pins to secure the arch are in place. The pins are slightly undersized for the hole and provided with string pulls to alleviate sticking.

Showing the new bolt location (old location to the right) where the platform is held to the superstructure. The bolt is received by a permanently fixed heavy-duty threaded plate at the back rail of the platform. Because of the weight of the platform the two rigidly fixed bolt points, movement within the frame when assembled is all but eliminated. The rear facing braces, now sliding on dovetailed cleats after the bolts are inserted, check movement when the doors open.

A series of shots showing the threading of nylon rope through the panels. Following the Charleston performance it became clear that the wire rope system was an over-engineered solution and because the top of the tensioning was occurring in the midst of - instead of at the top of - the spandrel panel that the tension was hindering vibration. By using a nylon rope we are eliminating a proprietary hardware (swaged wire rope with specialized fittings) for a more readily available material and at the same time having the point of security at the very top edge of the panel. The new rope will be tensioned (I anticipate) by using a wood key (slat) to make torsion between the loose rope ends that pass by one another on the back of the framework.


Tuesday, April 22, 2008

The spandral has been sliced so that the upermost portion that contains most of the arch is now the fixed (it is glued to the frame) anchorage and the lower portion of the spandral is free-vibrating. it has a piece of the next lower plate scabbed to it to make a larger rectaliniar area of vibration.

Details showing the new anchorage system. The vertical nodal sticks are in place and allow tie-backs to happen to keep the plates from buckling outward. The tensioned rope is attached at both ends to the single peg. This is a more complicated system than used for the Charleston performance, but it is more intuitive - shedding the proprietary hardware and need for wire rope swaging. It also puts less stress on the framework. Its main advantage is, of course, that it is allowing all the plates to vibrate to the same degree. Also, in the foreground of the long shot, is is the outer framework showing the middle hinge removed and wood being scabbed into the mortise. I have determined the center hinge is not required structurally and the leaves are easier to install without them.