"Hi Dave, Thinking about it a bit more, I suspect the sparse but contiguous pulses will excite the resonance better than equi-spaced sparse pulses, as it will contain the resonant frequency rather than a sub-harmonic of it. Cheers, Matt."
Yes indeed! This project involved some fantastic discussions.
I was approached in April 2014 by Sam Collins, technical director of creative production agency Forma, who invited me to discuss building an entirely new musical instrument. The instrument / installation was the idea of artist and composer Ed Carter, who wished to make a number of tuned steel plates resonate using electromagnets. The large "bell plates" would be made by master gong-maker Matt Nolan.
After some experiments to make sure that the system would work in principle, a number of prototypes were built and tested in Forma's basement studio. It's rare for a project to have such a thorough R&D phase, but this one was very luxurious. The experimental sessions allowed problems to be ironed out, suggestions to be explored and improvements to be incorporated. It also allowed the four of us to play around with magnets and make noise, which is of course a great way to spend time.
Note that the plates weren't physically struck to produce tones - the electromagnets caused the plates to vibrate by attracting the steel and then relaxing many times per second.
We found that fixing anything across two points on the bell plate would significantly dampen the sound, but at the same time the electromagnet had to be accurately and firmly held around 1mm from the steel.
Mounting the magnet on a separate stand invariably resulted in a sudden clang as the magnet attached itself, and so another approach was needed.
The solution was to use an arm fixed to a single point on the plate, which would help to keep any damping effects to a minimum.
The hanging points of the bell plates were positioned in the areas that vibrate the least, so that the sound is disrupted as little as possible. Matt has a special rig that vibrates a plate at a particular frequency. Sand is poured on top, which then accumulates at the areas which vibrate the least. These nodal points are marked and drilled for use as hanging points. Matt really knows his stuff - it was great learning from him. This photo (c) Matt:
The driver circuitry consisted of a rapidly switching FET, the frequency of which could be set via a rotary encoder. The resonant frequency of the bell plate could be found by sweeping the dial until it hit the "sweet spot", at which the plate sang out loudly in quite an eerie manner.
Pressing the rotary encoder toggled fine / coarse tuning, and pressing a button saved the frequency to the Arduino's EEPROM memory. A second button could be used to store the harmonic frequency of the plate, i.e. a second tone one octave above the main resonant frequency.
These stored frequencies could the be recalled by a MIDI input to the Arduino (surprisingly easy to set up).
The final bell plate resonator control system consisted of a MIDI file player and 10 oscillators.
The 20 frequencies (10 main and 10 harmonic) could then be triggered via MIDI - from a keyboard, laptop, or from files on an SD card in the built in player.
As the control unit contained circuitry repeated 10 times, it was worth getting some custom PCBs manufactured.
[2019 update: I originally used Fritzing for this, but now there are far quicker and cheaper options.]
Each board had a microcontroller socketed in the centre, which responds to its designated MIDI notes and controls the magnets accordingly.
The next phase of the project involved the creation of some interesting lighting rigs. Ed wanted to create an effect involving moving shadows as his composition was played on the bell plates, and decided to use three motorised swinging arms to achieve this. The rigs were to be MIDI controlled, so that the movement and fading of the lights could be choreographed via Ed's sequencing software.
The central aluminium boss for each arm was drilled using a jig made for a previous project - basically a drill held horizontally in a vertical slide milling attachment, with the bull wheel on the lathe being used as a dividing head.
Brackets for the nema 34 stepper motors. No limit switches, but a hefty physical stop just in case an unexpected cormorant landed on the arm and caused it to lose its position.
Motor with boss and arm attached. The arm was 1.5m long in total, and travelled 180° back & forth.
More boxes of electronics, including a stepper motor driver, 60V DC power supply, 12V DC power supply, 230V AC dimmer system and the control board. The microcontroller accepted MIDI signals via an optocoupler and changed the velocity of different notes into light fading speed and motor speed. The light fading commands were sent to digital potentiometer with 8-bit resolution, which in turn controlled the 0-5V needed to control the dimmer unit. The dimmer unit could handle mains up to 750W. The motor commands would generate the appropriate pulses for the stepper driver, and included ramping to make the movement smoother.
Testing in the orchard...
Once the electronics and hardware had been delivered, Ed wrote a fantastic composition for the bell plates.
The plates were then set up for a very atmospheric two week installation in St. John's church, Seaham.
The concept of this project was linked to the 1962 George Elmy lifeboat disaster, with various themes and compositional elements taken from details of the event.
(following 2 photos: Colin Davison / Rosella Studios)
On November 15th 2014 the bell plate system was set up on Seaham marina, for an incredible live performance accompanied by Durham Miners' Association Brass Band.
The PCB design has since been used by Johnny, Will and Timo for other projects involving ferrous metal objects oscillating at their resonant frequencies. Let me know if you'd like a system to do this - I have PCBs in stock.