How to CNC machine a Violin
Using CAD and CNC for a new type of violin: the 'Dutch Violin'The design of classical violins has remained virtually unchanged during the past 400 years. From the moment he started to make violins, Dutch industrial designer Frank van der Horst has been fascinated by this lack of development. Many questions have come up, like "How are the vibrations of the violin’s strings transformed into sound?", "Does a violin need to be symmetrical?", "Would a five-string violin offer advantages?", “How can I get an electric violin to play with an acoustic sound?”.
Some of the violins built by Frank van der Horst in the past 50 years.
Over the past 50 years he has experimented with innovative concepts, the first being an asymmetrical violin, built in 1966. About this violin Franks says: "It looked nice, I thought, but it didn't sound that great. It took me more than forty years to find out why".
In the meantime, CAD/CAM techniques had been developed. Could they help to solve this problem? Is it possible to make a good, and perhaps an excellent, violin with fewer parts and less effort? As good as an classical violin, but different. An instrument to extend the range of violin players, not just a replacement. Franks continues: "This question is crucial for the kind of violin I had in mind".
The first new model, built in 1966, which is asymmetrical.
Using CAD, CAM and CNC technology offers a violin maker many advantages:
- It is possible to change any design variable when searching for the optimum design
- CNC-made parts are more or less (as wood is not constant) identical, and thus interchangeable
- CNC makes the design process more flexible (for example, you could make a mirror violin)
- CNC files make it possible to produce anywhere in the world, using uniform digital data and locally sourced materials
It is asymmetrical, and it converts the movement of the strings into sound more effectively than the classical design.
Cross section drawing of the Dutch violin. It shows how the vibrations of the strings (on top of the bridge) are transferred to the violin's resonance-box.
As shown in the drawing above, one leg of the bridge is connected to the bass bar mounted under the top plate. The other leg of the bridge is connected to the sound post, which rests on the bass bar mounted on the backside plate of the violin.
The sound post passes through a hole in the top plate. The back and top plate vibrate in opposite directions. As a result, these vibrations enhance one another, providing a greater effective vibrating surface area. This means that the top plate does not have to vibrate as hard as in the classical design.
It goes without saying that the sound this instrument produces differs from that of a classical violin.
Design of the Dutch violin in CAD (top plate and back plate transparent).
The image above shows the violin in the 3ds Max CAD software by Autodesk. The top plate, the back plate and half the neck are drawn as wireframes, making them transparent, the other parts are rendered as wood. Except of course the violin's strings. A CAD program with powerful surface modelling options is needed to create the correct curvature for both plates.
Toolpaths in DeskProto: the back plate and two halves of the neck/pegbox/scroll part.
All wooden parts of the violin are CNC machined using DeskProto. All parts have been separately exported as STL files, and correctly orientated to be machined.
The violin's neck is a special part as it contains area's where the cutter cannot reach (undercuts) when machining from two sides. Frank has solved this by machining two separate halves: see the screenshot on the right and the photo further on this page. These will later be glued together to create the neck. An alternative would have been to machine one block of wood from four sides (indexed machining), however using two parts proved to be easier. Second advantage is that now a standard thickness board material could be used.
The BTZ PF 1000-P machine, milling the bottom plate for a new violin.
As you can see a relatively simple 3-axis machine is used: a PF 1000-P by BZT. Franks has added a custom fixture to this machine that allows him to easily machine a part from two sides. He calls it the "Static 4th axis", see the images below.
Back plate and top plate are made of spruce, cedar or any other lightweight tonewood. For tailpiece and fingerboard in many violins ebony is used, however this may in fact be any dense and strong wood with a very fine grain. The neck mostly is done in maple wood, this may as well be mahogany, cherry or (fine) oak.
First the fixed center, next the movable center, both showing the horizontal and vertical metal "rails".
The static 4th axis is a simple concept: it is a rotation axis, however without CNC controller rotation.
On the machine's working table two tailstocks (fixturing helps) are mounted: one fixed and one movable along the X axis. Each tailstock features two metal rails: horizontal (long) at Z=0, and vertical (short) at Y=0, together forming a T-shape.
Frank says: "I used the mill to mill a flat surface and two groves in its bed. An old blade from a handsaw was used to form the T shapes. The movable center is kept in line with two pieces of perspex. The slit in the middle is there to fix the movable center with a piece of wood a bolt and a wing nut."
See the photo's above. To use these tailstocks the block of material needs to be prepared:
"I use a band-saw to saw crosses in the end-grain sides of my work piece, about 8 mm deep. The T-shaped rails on the tailstocks fit tight in the crosses of the block, so as a result my work piece is clamped and aligned. I can easily flip a work piece 180° or (for small blocks) turn it just 90º."
Machining the neck from two sides: side one being machined, next side two after machining.
The left image above again shows the horizontal rails on the movable tailstock. The image of the right shows that the toolpaths for the second side have been accurately positioned: perfectly aligned with the toolpaths for side one. As you can see from the holes for the tuning pegs: this part is the neck (or more accurately the peg-box), though it still has to be split in two halves and assembled.
About his fixturing system Frank concludes: "It really does not look like much, just a first attempt, but still it works very well. I could have made a full cross instead of a T, but I chose for the most simple solution. The clamping also works well for large blocks because of the long horizontal part of the T."
The CNC machined wooden parts for a Dutch violin, before assembling.
In the image above the resulting wooden parts for the violin are shown, after finishing and after a first layer of varnish. From left to right: Neck/pegbox/scroll (one combined part), tailpiece, bridge, top plate (with hole), fingerboard. Not shown are the ribs, the bass bars and the sound post.
Below the tailpiece and the fingerboard you can see a block like the one used to create that part. Just as large as needed. Frank tells that he uses much 'recycled' wood and leftover pieces that otherwise would have become waste material.
A Dutch violin, made in 2014.
The Dutch violin shows a clear difference with the classic model. No corners in the outline of the instrument (the shape is in fact more 'guitar-like') and one round hole instead of two 'f-holes'. The shape of the scroll (which is the ornament on top of the peg-box) is different as well.
Current status of this ongoing project is that the even the latest violin still is a prototype. Neverhtless the sound and handling of the instrument are now up to standard.
As a next step (May 2017) Frank decided to release the design files as open source: they can be downloaded from his website Dutch Violin. Franks states: "You are free to use the files as long as you call the resulting instrument a Dutch Violin". And of course he is looking forward to hear about any results!