Electromagnetic harp development

in Blog

17 March 2014

Augmenting a harp with electromagnets to play the strings without touching them.

A collaboration with Emily Hall and David Sheppard. I am building the electronics for this project. The aim is to be able to make 19 different strings vibrate independently without visible human intervention (computer-controlled, in other words).

Here is an overview of the electronics development process from start to finish.

Update January 2014

First experiments with electromagnetic transducers similar to the kind found in the popular EBow designed for guitars.

Finding suitable electromagnet actuators

These can be made by removing the diaphragm and end cap from cheap buzzer elements. I tried about a dozen different buzzers and eventually found one that worked better than the others. It was actuated as if it were a speaker – by connecting it to a power amplifier that is in turn connected to a sound source – a laptop in this case. I used LM386 power amplifier ICs because they are apparently similar to what the original Ebow uses.

Testing actuators on a guitar

This initially looked like it would work on an electric guitar with standard gauge strings (11s). However, the electromagnets need to be very close to the string to get it vibrating initially, but once vibrating, the electromagnet is so close that the string hits it and buzzes. Many hours were spent trying to mount the buzzer at the ideal distance from the string, but gave up in the end. It’s easy to do it handheld because you naturally position the electromagnet the correct distance from the string and adjust as the string vibrates, but doing this in a fixed way is nearly impossible.

Update February 2014

Testing the modified electromagnetic buzzers on a real harp.

The first tests with the guitar gave me some hope that the harp with its much more resonant sound board would allow the electromagnets to be run at a lower power to prevent the strings hitting them.

First try on a harp with modified electromagnet buzzers and LM386 amplifiers

Unfortunately, the metal wound strings on the harp seemed far less responsive to the electromagnets. They did react but only slightly. Interestingly, string tension seemed to play a big role. Generally we found that higher tension strings reacted more than floppy ones at the same pitch.

We tried multiple transducers per string, positioning them at various points a long the string, grouping them in one area of the string and other configurations, but found the effect of adding further transducers to be minimal. They were still incredibly sensitive to distance from the string. So much so that we had serious doubts that this would be a suitable system in a concert hall environment where the harp would be moved around and knocked by passing performers. Much more powerful electromagnets were needed.

Update March 2014

Stronger electromagnets were sourced from China and a basic 4-channel driver circuit was put together. The new system was ten times more powerful than the original. The full size pedal harp was replaced by a more portable lever harp that was strung with electric guitar and electric bass strings.

New electromagnets and prototype driver circuit

The difference was immediately obvious. The magnets could be held more than 5mm from the strings and still cause them to vibrate loudly. The magnets got only slightly warm unlike the modified buzzers that got very hot in the center.

Emily posted a quick video of this in action here.

This version does not have dynamic control of the electromagnets – the ‘volume’ of the strings cannot be changed. The current circuit simply uses an amplified half-waved rectified audio signal to switch high power transistors on/off that in turn pulse the electromagnets on/off at the desired frequency.

Update late-March 2014

A new version has been designed, PCBs and components ordered. Each unit will have four analog audio inputs with four corresponding electromagnet outputs. Five of these units will stack on top of each other to provide twenty channels. This new version takes any audio signal and converts it into a format that can properly modulate the electromagnet strength enabling full frequency and amplitude control. Strings (and their harmonics) can now be actuated with any amplitude envelope.

Update early-April 2014

The new design has been built up and tested. A stack of five PCBs gives twenty audio inputs powering twenty electromagnet outputs. Power is shared between the boards requiring just one DC power adapter for all PCBs and electromagnets. Pluggable terminal blocks allow different electromagnets to be interchanged quickly. Each audio input has an LED that lights up when the input signal reaches -3dB of the available headroom.

New prototype for driving twenty electromagnets, with level control

A Max patch was written to provide 20 audio outputs to 20 electromagnets. The signal going to each magnet has the following controls: fine tune (+/- quartertone in case a string goes out of tune), harmonic select (play a harmonic on a string) and overall level. A global trim control calibrates the audio output levels to the electromagnet driver board audio inputs. MIDI note on/off controls the velocity sensitive oscillators and MIDI continuous controller messages control the above parameters. Multiple sets of parameter values can be stored and recalled. The system can be played directly using a MIDI keyboard, or can be designed/recorded using a MIDI sequencer and then played back through this software.

Software to drive 20 electromagnets with a MIDI keyboard

Back to Blog