Experimental Wind Instruments

Adnan Marquez-Borbon - CCRMA
Rob Ernst - Cogswell Polytechnical College

This project began as a curious exploration of some unusual musical possibilities for acoustic tubes.
It consists of two parts: one being the excitation of multiple tubes by a single source and second,
the seamless introduction of electronic feedback into an acoustic tube.

Multiphonic Bass Clarinet (MBC)


Our Goal was to make a multiphonic clarinet that could be used in a live setting as a droning
instrument. In order to achieve multiple different pitches from a single wind instrument,
we had to use multiple tubes of varying lengths that all protruded from a single mouthpiece.
The goal was to make a drone that had some very interesting beating created by the different
frequencies of each tube, so we cut the tubes according to the desired frequencies.
The other thing we had to think about was that it is going to be played in a live setting,
possibly simultaneously with a keyboard. So the instrument had to be relatively light weight
and compact. Lastly, for amplification as well as greater sonic versatility, we decided to add
small condenser microphones to the ends of each pipe, so that we could plug it into a computer
interface and further manipulate, or just amplify and record the sound.

A cylindrical bore was chosen due to the facility it provides in constructing an instrument.
After researching some of the dimensions of Bass Clarinets it was found that the inner diameter
of the bore is approximately 0.92 inches and that the length of the bore from mouthpiece to bell
lies around 42-43 inches (some sources state variations ranging from 37 to 51.9 in).
The lowest frequency of the instrument is a sounding D2 (73.42 Hz).
Using the formula for tubes closed on one side, fn = (2n-1)(v/4L), it was proceeded to calculate
the length of the tubes based on the lowest frequency desired.


The inner diameter of the PVC pipe is close to our ideal bore diameter and in practice,
the difference in diameter does not affect much of the acoustic characteristics of the tube.

The Build:

The first thing we had to do was to figure out how long to cut the piping to get our selected
frequencies. The frequencies we were shooting for were: 73.42 Hz (D2), 110.0 Hz (A2),
146.8 Hz (D3), and the last tube we wanted to be able to slide from 130.8 to 167.8 Hz (C3-E3).
In order to determine the length needed for each frequency we used the formula: Fn = (2n-1) v/4L
With this formula we determined that the proper lengths for those frequencies were:
1.17 m, 0.779 m, 0.584 m, and 0.656 m. But we could not just cut our pipes to this length,
because we also needed to take into account the distance from the mouthpiece to the beginning
of the sounding pipes. This distance would act as part of the pipe and change the frequency.
So when we subtracted that distance our final numbers were 0.93 m, 0.619 m, 0.424 m, and 0.416 m.
Then we measured our long pipe and started cutting. While we were at it we also cut some 1 1/2 in
pieces of pipe to mate our angled connectors near the mouthpiece.

After cleaning up the edges all we had to do was simply put it together. We attached the mouthpiece
and gave it a try. The first noticeable problem was that it was very hard to make all four pipes
to sound at once. It takes a very focused breath and if you stray a little from it the instrument
can squeak or only sound some of the pipes. After a little bit of practice it started
to sound pretty good.

pipespipes connectorconnectors

detaildetail pitchestube pitches

assemblyassembly MBCfinished MBC

micmicrophone closeupmicrophone closeup

The last thing to do is to build four microphones for each tube and mount them with zip ties.
I built a simple program in Pd and ChucK , as well as a graphic interface in Processing to use with
the instrument in a live performance setting. You can find them here: MBC patches

Here's what the instrument sounds like:
MBC with freeverb
MBC with feedback delay network


Feedback Bass Clarinet (FB2C)


To incorporate an electronic feedback system into an acoustic instrument.


The Build:

Make one microphone of the previously mentioned type, and make the mini speaker connection
by directly connected to a 1/4" mono plug (speaker (-) to ground and (+) to signal).

This instrument is very straight-forward in its construction, however the speaker must be
fixed into the T coupling in such a way that air leakage is avoided. Otherwise, the instrument will
"speak" at the length were there leak lies in (which should be a higher pitch than desired).

The mini-speaker is mounted on the inside of the cap. The inside diameter had to be increased with
a dremel tool because the speaker was a bit too big. Then we drill and hole at the back of the cap
so the speaker cables can go out. Once we have everything set, we hot glue the inside of the cap
opening and mount the speaker. Finally, with the cables poking out of the back of the cap, we hot glued
the hole and cables so we can have a seal and avoid any air leakage.

Finally, the driver is mounted on the side opposite to the tube so that the sound coming from it is
has a direct path to the microphone. The mouthpiece will be put on the mid coupling. The arrangement
is a bit odd, but otherwise getting feedback to occur will be a bit more difficult.

driver-capdriver and cap driverdriver

cap topcap top view FB2Cfinished FB2C

A simple program in Pd and ChucK was written to be used with the instrument.
Here they are: FB2C
NOTE: After realizing that having the feedback loop go through the audio
interface produced some bad sounding digital clipping and that the
gain control was a bit difficult to control, an analog
mixer along with a SP-303 sampler we used instead.
An auxiliary output of the mixer is sent to the input of the sampler,
and the output of the sampler is sent to the aux return bus of the mixer.
Finally, one of the main outputs (or control room out) is sent to the driver
in the tube. Appropriate gain staging is done in order to achieve feedback.
Just make sure that the signal is not "on the red" if you plan on recording,
otherwise the signal will clip and it will not sound good at all.
Once we have feedback, we can put any effect we desire in the sampler.
In this case delay and pitch shifting tend to work quite well.
Other routing schemes were tested, but in the end the result was the same.

Hear the instrument: WARNING! IT CAN GET VERY LOUD
FB2C (6' pipe @ ~47 Hz)
FB2C (16' pipe @ ~17.6 Hz)



Last modified: 1 December 2008