5. The Development of the Tarhu Cone: first published www.spikefiddle.com 2017
by Peter Biffin
The possibility of using a wooden cone as a stringed instrument “soundboard” first came up when I was working with guitar maker Greg Smallman around 1977. Greg had been making classical guitars and I had been making lutes and experimental instruments. We formed a partnership to make steel string guitars, and began our work together designing a new form of acoustic jazz guitar. Neither of us had made a jazz guitar before and rather than do the rounds looking at what was currently available, we decided to start from scratch and just make something up from first principles.
In exploring the possibilities offered by arched versus flat soundboards, we attempted to define for ourselves what an archetype for a plucked stringed instrument soundboard might be like - to try and find an underlying principle that could guide all other work. Gradually the idea of a cone shape emerged as a good working model – something that was strong but light-weight because of its geometry, and that likewise had an inherent graduation in stiffness from stiff in the middle to more flexible at the edges. I found a book on loudspeaker design, and we immersed ourselves in that for quite awhile. We weren’t planning an instrument with a cone in it that we would make and sell – it was research that didn’t have any particular final form in mind.
We made a variety of wooden cones while getting a feel for the general parameters involved. The cones weren’t ever in a very playable form, just mounted on a workbench with a string tightened by a peg jammed in a hole in the bench. Some of the sounds that came from these cones were absolutely inspiring, and it seemed that the wooden cone as an archetype for a plucked string soundboard was an idea that had a lot of potential.
One of my experimental instruments that had been in development for a few years prior to this period was a fretless banjo – a banjo body with a fretless fingerboard and sympathetic strings. As Greg and I pursued the cone-as-model idea further, the fretless banjo was one of the grateful recipients of these new concepts. Banjo acoustic design, with its very heavy metal ring around the top edge of the body, was ideally suited to experiments that involved a very lightweight, acoustically isolated flat soundboard (ie, the skin). I glued narrow deep struts made of light-weight wood onto the skin in various patterns, so that the overall weight was still very low, but because of the struts the stiffness of the skin was greatly increased and the bridge was capable of driving a much greater surface area. Experiments with this instrument were attempting to emulate a cone’s stiffness and weight relationships, but in a flat plane.
With the introduction of this concept, the attack and decay of the banjo sound suddenly became quite controllable, and the range of possible sounds that could be derived from a banjo body expanded enormously. This same banjo body has been used in many different experiments since then, always with some form of strutting glued onto the skin, and it is still one of my favourite instruments. The banjo experiments showed clearly that strong, rich and varied sounds can come from an acoustically isolated very light-weight soundboard that has a stiff middle and a flexible edge, regardless of whether that soundboard is cone-shaped or flat.
Greg and I explored other concepts encountered in the loudspeaker/speaker box conceptual model, including experiments with different “soundbox” types, such as using large horns instead of bodies etc. These experiments resulted in some very useful test instruments with all components being removable and replaceable. Using these instruments we were able to examine the acoustic function of a guitar’s principal components, and gradually alternative ways of making a guitar body emerged such that it behaved in quite a different manner compared to a traditional guitar. Consequently, the back and sides of our guitars became much heavier and more inert, and the soundboards became lighter and more isolated acoustically from the rest of the structure.
At the end of 1979 I left Australia to continue experimental work in Bombay, and at that time Greg and I dissolved our partnership. In Bombay I worked on North Indian Vina (further developing the techniques for making spherical instrument bodies using strips of wood), experimental forms of tanpura and the Sarangi. I returned from India late in 1980 and began a new series of experiments with various forms of spikefiddle, continuing with the concept of gluing struts onto skin soundboards that had been successful with the banjo. I began with an instrument in the form of the Middle-Eastern kamancha, but did most of the experimental work using the Turkish tanbur (with various modifications to suit my interest in playing music inspired by both Turkey and India). These instruments again used a heavy/inert body coupled with a very light-weight “soundboard” (made from goat skin).
While there was much in the sound of these Eastern bowed instruments that I was drawn to, with their very light-weight skin soundboards, they also had a few inherent design elements that caused problems I couldn’t solve. The soundboards of both Kamancheh and Tanbur have a circular plan view and are flat. This means that either the soundboards have to be very small in order to allow the bow to play near the bridge (as in Kamanchah) or else the bow doesn’t move in an arc, and only one string is bowed (as in the Tanbur). The way in which a cone-shaped soundboard would elevate the bridge up away from the soundboard edges became increasingly attractive – in this configuration, an instrument could have a larger soundboard, but the increased size would not restrict the bow from moving in an arc. It began to feel like “why try and approximate a cone in a flat plane when for a bowed instrument the cone shape itself works to my advantage?”
Over the 10 years or so that I played the strutted-skin tanburs I therefore kept coming back to trying wooden cones in bowed instruments. Unfortunately connecting string and cone in a way that worked effectively for bowing always eluded me. I never made headway at all with cones in bowed instruments until after I started work on the Chinese erhu in the early 1990s.
Tarhus bear virtually no resemblance to the erhu in either physical form or acoustic design (except that they both belong to the generic instrumental form of spikefiddle). The erhu's body is a small cylinder covered at one end by snake skin, which acts as a soundboard. The importance of the erhu for me was the key it provided for understanding how bowed instruments work. The unique feature of the erhu is that the bow moves at right-angles to the plane of the soundboard, rather than the bow moving parallel to the soundboard as it does in the violin family. It is quite an education to begin bowing an erhu in the normal way, with the bow at right angles to the soundboard , and then gradually rotate the body so that finally the bow is moving parallel to the soundboard. Through this progression the sound goes from strong and rich at the beginning to squeaky and almost non-existent at the end. From this experiment it became quite clear that there is a strong relationship between bowing direction and the way in which the string is able to make the soundboard vibrate.
I then looked at the violin family through the lens of these perspectives, and was able to see the function of the violin soundpost in a new light (the soundpost is the little dowel stick that is wedged in between the soundboard and the back of the violin, just behind the bridge - it is known as “the soul of the violin”). The significance of the soundpost is often described in terms of it connecting the soundboard and the back such that an acoustic coupling between these two components is created, and that this coupling is essential for the violin's sound. My erhu experiments showed that by far the more important function of the violin's soundpost is that it acts as a pivot point, and in so doing changes the way vibrations enter the soundboard. The bow moves the string in the plane parallel to the soundboard, but the soundpost/pivot point alters the direction of these vibrations such that they move the soundboard at right angles to the plane of the bow. The enormous change in sound that occurs in a violin when the soundpost is removed doesn't arise because the soundboard/back coupling is lost, but because without the soundpost the vibrations in the string hardly shake the soundboard at all.
It therefore became clear that all I had to do in order to use a cone in a bowed instrument was to come up with a bridge design that transferred vibrations from one plane to another. The spikefiddle concept of having a neck continue right through the instrument to the tailpiece seemed to have the most potential - the through-neck provided a convenient rigid support for the stationary/pivot side of the bridge, with the cone underneath being driven by the moving/active side. These ideas came to fruition in 1995, with the making of the first tarhu.
In themselves, cone-shaped soundboards are reasonably free of vices. Because of their regular geometry, they tend to vibrate in predictable patterns with relatively few peaks and troughs in their response when compared to flat soundboards. While it is easy to see the cone inside the tarhu body as the most important part of the design, the concepts behind the bridge design remain absolutely crucial in achieving a good sound from a bowed cone (along with the details of how the edge of the cone is designed). The most interesting thing to emerge from a reflection on several decades of working with bowed cone instruments is the huge percentage of experimental time that has been devoted to bridge and cone-edge design – much more time than has been given to the design of the cones themselves. If the suspension of the cone is wrong (the edge), or if the way the sound enters the cone is wrong (the bridge), the best cone in the world will make sounds that are completely unusable. In the early years I wasted enormous amounts of time trying to fix sound issues by changing some aspect of the cone’s design, only to discover later that the cone itself had absolutely nothing to do with the problem.
The trend in tarhu cone design over the last 22 years has been towards lower and lower stiffness - cones that don't stand quite as high as the ones before, or cones that have a wider, more flexible edge. In 2015 I began using cones with a curved profile in which the cone and the edge are no longer two separate components. The 40-year journey with wooden cones began by picturing them as a model for stringed instrument soundboards. The current work with low-stiffness curved cones is actually a step in the opposite direction - some of the concepts employed in stringed instrument soundboards are now informing tarhu cone design.
©Peter Biffin 2017