Andy Cavatorta


The Irvine is a new type of electronic musical instrument commissioned by AVL to commemorate the 75th birthday of their CEO — prolific inventor and engineer Herr Doktor Professor Helmut List. It creates music using gallium phosphate crystals, one of his favorite AVL inventions. These crystals do not occur naturally on our planet. The secret process for growing them was invented at AVL after 20 years of research. This exotic piezocrystal can operate in incredibly high temperatures and pressures and has dozens of industrial applications. And now one musical application.

an early breakthrough
an early breakthrough
growing crystals from seeds
growing crystals from seeds
a beautiful specimen
a beautiful specimen

The Irvine is part of a re-imagined history of electronic music. If electronic instruments found a home in the symphony orchestra in the early 20th Century, how would they have evolved differently and what would we demand of them?

Rather than cheap plastic consumer electronics that we are expected to upgrade every year, they might be finely crafted instruments the reward long relationships and intimate familiarity. Rather than controling processes that crank out automatic music, performers would need to craft every nuance with their hands, ears, technique, and musicality. Advanced players may develop unique and recognizable voices.

The Irvine is intended to play as expressively as singing, with continuous and discrete expression in pitch, dynamics, and timbre.

Studio Diary for The Irvine

We started at the edge of the dark forest of possibilites.

Every year, a new batch of MIDI controllers appears, attempting to improve on the keyboard by adding more expressive or intuitive features. Many of these features are fine ideas. But I've never heard any of these instruments used to make music I enjoy. So making improvements is more challenging than it looks

Prototype 1

I always start with the sound. It shapes everything that follows. We were seeking an honest way to make music — good music that makes you feel — with the direct sounds of the crystals. No gimmicks or fakery. No designs that insert the crystals as a superficial afterthought. The moving music should come from the voices of the crystals, controlled by the performer.

I was hoping to cut custom-shaped crystal oscillators under a microscope from blocks of raw crystal. I'd been reading up on the physics of crystals and how different shapes and sizes resonate at different frequencies. The most interesting part is how cutting the crystal along different axes totally changes its behavior.

Our hook-up for these rare crystals came through Piezocryst, an epiphite of AVL focusing on applications for piezoelectric crystals. I was disappointed to find they don't release raw crystals because their crytstal-growing method is still secret and larger crystals can be used as seed crystals to grow more. What they could offer us was their 6MHz crystal resonators product.

We were grateful to be able to get any of these rare crystals. But 6,000,000Hz is far above the human hearing range of 40 to 16,000Hz. So we had a lot of things that didn't fit together.

As always, we start by prototyping. With the gallium phosphate resonators, we created a Colpitts oscillator — a basic crystal oscillator circuit. It created a stable sine wave at 6MHz. So we had a singing crystal but nothing like audible sound.

small crystal tuning fork
Simple Colpitts circuit
quartz resonator from Piezocryst

Once we had one oscillator, we created a second one that sings at an ever-so-slightly different frequency. When we mix their two signals together, we get a more complex signal with lumpy beats that have a frequency equal to the difference of the two source frequencies. In physics this adding of waves is called superposition. In radio it's called superheterodyne.

We could already see the signals on the oscilloscope. By using superposition, we could now hear the raw oscillations of two crystals together. We began pushing the crystals into uncharted states to see what distinctive voices might result. We tried high and low voltages, adding physical pressures and vibrations. The video below shows the effects of lequid nitrogen on the crystals. On the oscilloscope's bottom trace, we can see the signal bloom into multiple frequencies as its temperature drops.

Prototype 2

So much of this is not about technology or even music. It's about our miraculous human hands and what can do with them.

We started with continuous pitch control, for sliding between pitches. Drawing inspiration from the Ondes Martenot and similar instruments, we suspended a loop of cable in four pulleys to create the 'transport', which can slide smoothly left and right. We connected this to the performer's hand using a 'yoke' that slips around the performer's finger. You can see João testing this mechanism above.

Then we needed to play some music to see how this feels. So we added some sensors to precisely measure the movement of the transport and some software to turn transport position into MIDI messages.

Next, we added the pitch keys — copper strips that we can trigger with just a touch. This enables the performer to instanly zero in on exact pitches while sliding up and down with the transport.

This simple first prototype is a test platform for many questions. What is the best spacing between the keys? How does it feel to play the transport and touch keys at once? Do the capacitive sensors behave correctly for this application?

It's good to build prototypes before you know what you're doing. Because they can answer questions you hadn't yet thought to ask. But so many questions could not be answered until we could play music.

Prototype 3

Prototype 3 could play music! So we could start to feel what it's like to play it. It had a pitch transport, pitch keys, and three voice keys that set the breathing, or volumes, of the three voices. It also had a small, outboard box with sliders to control the timbres of the three voices.

And we had to finish it all in a sprint so it could be ready in a few weeks when the composers Tom Huber and Sergio Minutillo arrived to test it.

We designed and tested a variety of yokes for the transport. And we also tested the the hand positions, the range of the voice keys, and the shapes and spacing of the pitch keys.

testing hand-yokes
testing scale with human hands
tested on humans

Finally, we developed all of the electronics and software.

João's custom circuit boards
the voice key mechanism
last-minute soldering

We finally got over the finish line just as the composers were landing in NYC to meet us.

Our first round of playing was half music and half live troubleshooting.

The most exciting part of this process was watching new techniques being invented. Our composers Tom and Sergio brought it back to Munich to keep composing, learning to play, and inventing new techniques.

Prototype 4

Prototype 4 was brutal. We used this period to test and weed out many of the most ambitious ideas. In particular, solid glass circuit boards and creating different musical pitches by electrically pulling the crystals into different states.

The glass circuit board idea was gloriously impractical. We would use 15mm thick plates of borosilicate glass as the board and copper foil as the traces. I had done this with acrylic at MIT with mixed, if beautiful, results.

Test circuits for pulling crystals into tune
glass with circuit traces
lighting the glass circuit boards

Version 4 concept
Version 4 concept
All maquettes together like a turducken

The glass would have weighed over 250kg, requiring a much bulkier support structure. And I was having prophetic nightmares about troubleshooting circuit boards that cannot be modified without a diamond drill. We will make this beautiful idea a reality someday. But we'll have to factor it into the budget and schedule from the beginning.

And the gallium phosphate crystals could not be budged far from their natural 6MHz frequencies. 6Mhz. FFS. Human hearing tops out around 16kHz. What are we supposed to turn the genuine sounds of the crystals into music?

The continuous pitch controller felt wrong and the way it tethered the performer's hand was annoying.

Version 4 was where many of my design fantasies got crushed. Again, this is why we prototype.

Prototype 5

Prototype 5 was our breakthrough! This was the first version that embodied the original vision and had all of the musical features.

It featured rich timbral control for each of the three voices. A new continuous pitch controller that glides freely under the performer's hand. An internal tube amplifier for each of the voices. A system for recording and playing loops of music, enabling the performer to build up layers in a composition. And a vast and complex electronic inner life featuring ten small computers translating between the digital and analog components.

But the breakthrough was in how we generated the voices. Each of the crystals was only approximately ~6MHz. Some were higher or lower. We paired high and low crystals whose frequencies were different by about 100KHz. Those were each placed into Colpitts oscillators and their output signals were multipled to produce a superposition signal equal to the difference.

100KHz is still far above the human hearing range. To make specific audible signals, we combine the 100KHz crystal signals with digitally-generated signals to create a new superposition in the audible range. The natural analog vibrations propagate all the way from the crystals to the amplifier speaker to the listener's ear.

We needed a much more sophisticated structure to contain all of the new features. More space for so many parts. A stand and pedals. And a first attempt to separate the sensitive acoustic electronics from the brutal electronics like power transformers and tube amps.

2D exterior design
3D exterior design

We started with a design that should be easy to take apart and reassemble. Because we would be doing a lot of that. But as we continued to add more features and change the design, it became an unwieldy beast.

We had found and solved lots of small wiring problems. But we had a general problem that could not be solved. Every wire is also an unintended antenna, interacting with all of the other fields and signals pulsing through this instrument. Some short cables grew to a meter in length, stretched when the instrument was open and falling in chaotic piles when it was reassembled. Isolation became impossible and the instrument hummed and throbbed with mysterious noises as if it was haunted.

It looked so easy at first
Jesse is only smiling because he is always smiling.

The photo below shows the new brass and copper pitch keys, the drawbars used to set the timbres, and the six glass globe that protect the crystals from breeze and changing temperatures. Underneath are some of the dozens of sketches and diagrams we draw with each version. And also the custom circuit boards with slide potentiometers.

Here is Merche Blasco with one of our first tests of the system that controls the timbre of each voice.

When our composers Tom, Sergio, and now Christof Kosel, arrived from Germany, Jesse and I had been awake for 36 hours frantically troubleshooting mysterious electrical noises and software glitches. I'm aware this is part of a time-honored tradition of demoing research for clients. But it's still embarrassing every time.

It was also not the last time.

The delays were forgotten quickly when the composers finally got to explore the new instrument. It was amazing to see and hear them continuing to discover and invent new techniques and sounds.

We also encountered tuning problems for the first time. The natural resonance of each crystal was slowly drifting with changes in the temperature and air around the crystals. The crystals are very stable. But even a tiny inaccuracy of ±0.00003% was glaringly audible.

Prototype 6

I promised myself that Prototype 6 would introduce no new features. It was time to focus on solving the many challenges we'd encountered already.

The biggest challenge was something so prosaic — cabling. I thought of this project as transforming signals, coursing data, how it meets the hands, the sounds and how they feel. Boring things like cabling were an afterthought.

But it's a huge part of the project. We have five types of electrical power, 16 ports of Ethernet, data and signals from over a hundred different sensors, the many stages of the musical signal path. We spent at least a hundred person-hours making and installing custom cables for Prototype 5. They are the electromagnetic achilles heel of the whole project. If they are not solving problems, they are creating them.

We made a full-scale model of the new case to help us find a sane and organized way to lay out hundreds of cables.

You can see my optimism in this early video report about our progress.

And just like the case for prototype 5, it all started out promisingly. But our code name for Prototype 6 could have been Slippery Slope.

We were already adding new features such as staccato/legato and hold buttons that changed how the voice keys worked. And an automatic system to measure and compensate for the drifting frequencies of the crystals.

The most interesting new feature we could not resist adding was vowel formants. I wanted this instrument to play as freely as singing. The performer can freely mix pitch, dynamics and timbre. But we could make this even more like singing words by enabling the performer to assign vowel sounds to each of the three voices.

In human speech, vowel sounds are made by subtracting certain bands of frequencies from our harmonically rich human voices. The Irvine's pure sine-wave tones are harmonically sparse. So we added a tube distortion unit to each voice to fill out the frequency spectrum. Then we could make vowel sounds by filtering out specific frequency ranges.

In the midst of adding all of these features something was causing circuit boards and computers to overload and burn out. And we could not find the source. Each new casualty disproved whatever tenuous hypothesis we were clinging to at the time. We ripped out most of the electrical lines, changed power supplies and network topologies. And everything would be fine. Until it wasn't.

Our simple wiring layout was becoming a haunted electromagnetic jungle again. Behold the carnage.

When our composers came again to visit, they again found us at the end of another all-nighter. Shambling, amiable zombies. But the Irvine was playing more solidly and more beautifully than ever. So we were finally getting close.

the jungle
final tests
first playing with Tom

Each of these prototypes produces a number of collateral designs and technologies. Prototype 6 included a new type of speaker with a diaphragm of sitka spruce tonewood — the type of wood used to make a guitar or cello.

Some composers add amplified instruments to a traditional orchestra. It can produce interesting effects but the timbres of the amplified speakers never truly meld with the wooden and brass instruments. This tonewood speaker is an attempt to solve that problem.

The tonewood colors and limits the sound in exactly the ways audiophile strive to escape. But for our purposes, the tonewood plane of these speakers may vibrate in the same modes and ranges as the face of any string instrument. So the ranges of timbres produces can meld smoothly.

Initial experiments with the tonewood speaker were very promising. But with our endless, hectic pace of prototyping, it was set aside for a period when there is more time.

tonewood speaker driver
tonewood speaker front
tonewood speaker detail

Prototype 7

We had one last shot at getting everything right before sending this to Graz for a performance. And as we do, we redesigned and rebuilt every single piece from scratch. Because it was possible for every piece to be perfect.

We remade all of the pitch keys to be wider and to match the size of huan finger pads. And to have their pitches stamped on them clearly.

The transport handle is new and based on many iterations of feedback from the composers.

The pedals are rebuilt to be easier to play.

The new structure is a steel skeleton instead of a wooden box. And on top of that is a sculpted exterior we call the baroque meteorite.

The pitch keys were cut, shaped, and polished by hand. And stamped with letters using a manual 12-ton press. The video below shows Maria slicing brass into segments using the mighty Famco 612 horizontal band saw. These segments were later shaped by hand into keys.

slicing brass plates into strips using a router
new key slugs, rounded and polished
pitch keys with stamped letters

The new fingerboard was created from solid maple and a 3D model and cut using a CNC router.

Making seamless joints
Yvette is skeptical about this process
The keys fit tightly

The new front panel, including the transport and voice keys, was all cut from thick aluminum stock and shaped by hand.

New voice keys
Voice Keys with Springs
Gently shaped thick aluminum

Pitch keys from behind, plus the transport mechanism looking legit.

The designs of physical musical instruments have such fixed expectations. It's hard to create a new instrument that looks like it's from the present moment. They tend to look like the past. Or to look like futurism by rejecting the past. And nothing looks older than yesterday's futurism.

The final design for the Irvine uses classic Austrian ornimentation in a completely unfamiliar material - a solid block of carved aluminum. This is the baroque meteorite design. It started with a rough clay model which was physically disected and measured by Jeannette Subero and turned into an ornate 3D model.

rough model
dissection and measurement
3D model

Jeannette slathering details onto the 3D model.

We spent days preparing to cut our big slabs of solid aluminum.

practicing the details in wood (4 hours)
practicing smooth contours in wood (4 hours)
practicing smooth contours in aluminum (6 hours)

The final assembly included a new steel frame, new circuit boards and cabling, and our carved aluminum top. The breakdown of our shop's CNC router put a stop to all of our precision metalwork. So we couldn't create the ornately carved aluminum top just yet. But a friendly production shop helped us out by using their CNC router to create a temporary cosmetic cover out of high-tech modeling foam.

The circuit layout is finally not a shit show. Or at least less of one. 5th time is the charm.

Yvette, assembling the new steel skeleton

Skeleton and nervous system

All hands on deck for the final fitting and assembly.

Clemens in Graz, adding the final lettering.

Labels. Now anyone can play it. JK it's still hard.

Prototype 8

We have one final prototype left to build in this series. Prototype 7 is complete and performs well. Prototype 8 will incorporate every good feature we've explored on this journey.

This will include complex vowels for each voice, the a new baroque meteorite top carved from solid aluminum, a parametric aluminum basket to cover the underside, fully redesigned circuit boards, tonewood speakers, and a small bust of Dr. Krempl hidden inside among the wires and crystals.

Watch for more developments in 2018!

parametric basket
parametric basket
parametric basket