"The EBow is a battery-powered electronic device for playing the electric guitar. The EBow uses a pickup – inductive string driver – feedback circuit, including a sensor coil, driver coil, and amplifier, to induce forced string vibrations. The EBow is monophonic, and drives one string at a time, producing a sound reminiscent of using a bow on the strings." (Wikipedia)

The idea of the MultiBow, which I'm currently trying to realize, builds on this concept. It's essentially six EBows firmly installed under the strings. The allready installed hex-pickup serves as sensor coils and the driver coils are integrated with small class-d amplifiers into another pickup looking package. So the six signals from the pickup are going throug The Teensy-audio-board I use. It has 8 output-channels and I do only need one or two for the main output. One idea was to have additional discrete outputs for each string to process them externally. But using them as part of the MultiBow does make more sense to me, as it opens up the possebillity to modify the feedback signal to alter the effect on the strings (see moog guitar).

Here is a little demo video to proof the concept:

The digital path is currently just dry. But it affects the signal by delaying it a few milliseconds. That doesn't seem to be much, but I fear the so introduced phase shift reduces the efficency of the sustaining effect. The high power consumption (about 10W when all strings are in action) in comparison to another approach I tried, seems to confirm that.

There are two ways I think of to enhance efficency:

1. Compensate the phase shift by delaying the the signal a certain amount of time. The exact timing could be calculated by taking the latency and the signal frequency into account. But I currentliy don't have an frequency detecting algorithm as the one in the Teensy example patches doesn't work with the short sample buffer of just 16 samples I've set. Also, it would be quite cpu expensive to apply this to all six strings simultaneously.
2. Use a waveshaper to modify the waveform. My train of thought: If we put a bipolar sinosoidal wafeform into a coil, it changes the magnetic field direction within the phase, so a magnet would be pulled and pushed. But a string isn't a magnet. Regardless of the direction of the magnetic field, the string is pulled in both cases. So eliminating the postive or negative portion of the signal may avoid the effect of pulling the string while it' swinging in the other direction.

Do have further ideas or advice on this? Let me know in the comments!

Some pictures: