Sometimes, you just have to make your own reeds.
Traditionally, reed production starts with a piece of cane, which is then split into strips, trimmed, and gouged to about the right thickness. This is where we start:
This is then profiled, meaning that the cane is scraped from the bark side on a profiler to form the blades. After that, or sometimes before that, the basic reed shape is made by cutting the sides using a shaping tool. This is what one half of the double reed looks like after profiling and shaping:
The characteristics and sound of the reed can be very much affected by any part of the geometry, and the shape of the blades especially is very sensitive, so even differences in thickess of 0.05 mm can have a significant effect on the response, intonation, timbre, etc. It's helpful to look through a wetted reed blade to see how the thickness varies:
Generally, the reed blade is thicker in the middle, thinner towards the edges and the tip. The thicker part of the blade just below the tip is called the heart, and the thicker middle part down the length is called the spine. The thin corners are called the wings, and the straight portion with the bark is called the throat.
The actual product of profiling and shaping looks like this, with the blades joined tip to tip:
You can also buy blanks that look like this. It can save you a lot of time, if you are able to work with the provided shape to get the results that you want.
It can then be either folded in half or cut in half. Then the blades together are formed into a more rounded tube around a mandrel, tied with wire, and bound with string or something. The seam is often sealed with Duco cement or something similar. This is a typical finished reed:
Final touches are made to the shape of the blade ("finishing") by scraping with a blade to get it just right.
As I'm sure you can tell, this is actually quite a fine and delicate process and can take quite some time. And then usually you have to make a few before you get one that works right. Over the years, then, various tools have been made and refined to make it easier and to make more reproducible results.
Many, many years ago, I decided to make a reed profiler. A scraper.
Making reeds is a very touchy process. Small changes in the cane, and characteristics of the cane itself, produce different results. Having a consistent, symmetric profile before finishing both results in less work and better quality, and that's the main job of the profiler.
This scaper was an experiment in geometry. The blade on this scraper slides precisely to and fro. The vertical position of the blade is adjusted, much like in a wood planer. The profile is created because the reed is on a rotating cylinder (also called an easel) which is set at a slight angle to the plane of travel of the blade. In addition, the easel rotates eccentrically; the axis of the easel near the tip end is adjusted further away from the axis of rotation. This all results in a smooth, gradual thinning towards the edges and the tip, while maintaining a nice heart. At least in theory.
I got the precision rod from an old printer. Most of the rest, including the bushings that slide along the rod, came from 13mm aluminum plate from some old unidentifiable piece of junk in our garage. The rod was press-fit into the end plates, with the help of the different coefficients of thermal expansion. (I stuck it in the oven.)
The easel was aluminum rod (tube, actually), turned on the lathe. The ends were press-fit into two gymbals. Finally the whole thing was mounted on a piece of poplar.
The angle of the blade could be adjusted by loosening the big bolt, and the blade slid up and down on precision keyways. The entire platform could be disassembled easily for cleaning.
It was crude compared to the commercial types, but it sort of worked. It still required a lot of manual labor and a lot of fine, finicky adjustments.
This is now ancient history. I didn't end up making many reeds with it.
There were some obvious problems with that first profiler. The whole process of scraping was rather slow, and I'm lazy. So I decided to try to take this process to another level.
In recent years, I built a CNC mill/router using some linear motors I got through salvage. It occurred to me eventually that I should try making reeds with it.
I generally use OnShape and Fusion 360, which are parametric CAD applications, to design models and generate gcode. Parametric CAD is nicely suited for this. It allows for dimensions (or other parameters) to be defined in a table as constants or expressions, and adjustments to those values can be automatically applied to the model. In this case, the reed dimensions are all defined in a table, including coarse measurements like lengths and widths as well as fine measurements including the thickness of the cane at various points. These values can also be used to make graphs of the reed profile using sketches.
If anyone knows a better way to make little graphs like this in OnShape, please let me know!
You might think that a precision CAM process could do this pretty easily, but it's not that simple. There are a number of complications:
1. The setup needs to be precise, and getting it just right can be tricky.The usual digital caliper is very handy. I also have this dial indicator for testing lenses, and it's perfect for measuring the cane thickness. My eyes are old, and a loupe is really helpful, along with a bright light.
In my experience, it's best to use a square end mill. A ball end mill needs more Z clearance when doing the side edges, and so it's easy to accidentally cut into the fixture. The square end mill also seems to make a smoother finish.
The fixture on my CNC table is a 28mm OD aluminum tube mounted on two V blocks. Y is on the long axis.
It is very carefully aligned with all the axes, and the Z height is also very carefully set. Getting Z just right can take some trial and error. Centering on X deserves special care, because, as it's a cylinder, any error in the X will result in terrible error in the Z along the sides. An edge finder works pretty well for that. Y is a lot easier to do, and it doesn't have to be perfect. The ends of the cane should be at +/- 60 mm, so it's just centered with that in mind.
Although the cylinder fixture is centered, it still needs to be rotated so that the cane itself is centered on X. I just measure the width, move the tool to half that plus half the tool diameter in X, then rotate it until the cane just touches the tool. That should be good enough.
The 28mm diameter seems to fit the Bonazza cane that I have been using. It might be necessary, eventually, to use cylinders of different diameters (and CAM setups to match). But see below for a trick that might help with that!
The reed stock itself is a strip of gouged (not shaped) cane, and it is held in place with double-sided carpet tape and a spring clamp. I've been machining one side, with the other side clamped down, switching back and forth between coarse and finish passes. The idea is that the clamp ensures that the cane is firmly down against the fixture, especially near the middle, and doing the full coarse passes first means the cane will be more compliant for the finishing passes. Clamping it too tightly, though, will compress it too much.
Selecting the right type of carpet tape can be helpful. It should be thin, consistent in thickness (preferably without thick fibers), strong enough to hold the stock in place, and also release without too much trouble (probably after soaking in water). It's important that it releases reasonably easily, as the cane will be extremely thin and fragile. I've been using this tape ????, which is a bit thick, but it can be reused many times. Just wipe it off under water, and it's plenty sticky again after it dries.
The X and Y motors on this CNC are pretty precise servos with an encoder accurate to 5 microns, and those axes suffer virtually 0 backlash. The Z axis is also a servo, but it is a lead screw type and the encoder is on the motor shaft, not the linear slide, so there is some backlash. Fortunately, this can be compensated by using all downward milling.
It should be possible to do this on machines with more backlash, but that will complicate everything. GRBL Advanced, which is what I use, will supposedly compensate for backlash, but I haven't tried it yet. Some CAM programs can also compensate for it when producing the G-code. If you want to try this on your CNC system, just make sure you account for backlash in some way.
It would probably be much easier and more reliable to use a rotating axis, if you have one. I'll probably make one eventually.
I used OnShape to create a reed profile in a public document. You are free and welcome to copy it and do your own experimentation. (I'll put up the link to the public document when I'm satisfied with the profile.) Unfortunately, OnShape does not have a free CAM option that is up to the task. At least not that I am aware of. Fusion 360, however, has excellent CAM support. I really prefer OnShape for modeling and Fusion for CAM.
To get from model to G-code, I export the OnShape part as STEP and import it into a new document in Fusion. It must be saved to work with the next step. I have a separate document with the fixture and stock bodies defined. At the beginning of that docuement's history, I have a derived part inserted, which is - you guessed it - the reed body from the previously saved document. When trying out a different reed model, I simply insert the new derived part into the beginning of the history and remove the old one. There is very little dependent on that geometry, and it's easy to fix the couple of exceptions that arise. The CAM also needs to be regenerated.
In the CAM settings, I include the fixture body and the stock body and set the origin to the surface of the fixture, in the middle. Thus my Z origin is at the surface of the fixture and my X and Y origins are in the middle of the fixture, stock, and reed model.
I use parallel 3D milling, downward direction only. For the coarse pass, I just do the blade surfaces. For the finish pass, I do both the blades and the throat, removing the top layer of bark on the throat. Removing the bark makes forming the tube much easier with no cracking, opens the throat more, and also makes it fit the bocal dimensions better. As a bonus, since it is cut all the way around, shaping the reed is easily done with just a sharp knife. No shaping tool is needed!
Be careful that your tool does not try to cut out the sides of the reed and down into the fixture. It will happen with some parallel milling options and with adapative milling in Fusion. Just do the top surfaces only.
The cane blank needs to fit well onto the fixture.
I've been trying something out to help with this. As a side effect, it also compresses the grain, which might make it more consistent and stable. It can reduce the thickness by as much as 0.2mm, so the compression is real. It does add a couple of days to the prep time, though.To start, soak it for many hours, until it sinks. Then heat it up in a pot and boil it. While it's still hot, clamp it between two pieces of tubing that are the same diameter as the fixture. Then put it in a toaster oven at ~400 F for a few minutes. That's right, cook it. Purists will shudder at this, but it doesn't damage the cane if done with care. When cooking it, make sure it doesn't completely dry out, or it may crack!
Then just let it cool and dry while clamped. After all this, the cane should, at least for a while, match the fixture. It might still not be a perfect match, but it's fine as long as some light clamping will hold it down evenly.
After the mill has done its work, remove the tube from the table, with the cane, and immerse it in water. With the right kind of carpet tape, the cane will release from the tube easily after soaking for 10 minutes or so.
Since it is milled all the way around and thin on the edges, it can be cut to shape with a sharp knife or razor, without needing a shaping tool. At this stage, it can also be cut down the middle to separate the blades, although some people prefer to fold it.
Now here's a good tip. (A soldering iron tip.) The next step is to form the tube. This is sometimes done by steaming the cane and carefully shaping it around a mandrel. Some can do it without steaming. Either way, that causes cracks for me around 80-100% of the time. Instead, mount a mandrel onto a soldering iron.
To get started, after completely saturating the cane in water (until it sinks), press the hot mandrel down on the throat, rolling it back and forth to evenly heat the cane, and fold up the edges with pliers. (Do this on a heat-resistant surface! Do not spoil your kitchen counter or start a fire!)
After starting it that way, put the blades together and roll O-rings onto it. Or fix it in some other way. Then add the wires. Soak the cane again, and then just press it onto the hot mandrel, squeezing it with pliers to round it out, adjusting the wires along the way.
The soldering iron (hot mandrel) method works pretty well. The cane must be heated evenly, or it might just bend in the middle, exerting a lot of stress there and causing a crack. This is especially likely if you have left the bark on. It must not be allowed to dry out, so keep soaking it.
Finally, the reed assembly can be finished with sealing and wrapping, in whatever way you prefer. Then it needs to be trimmed and tested! Sometimes I don't even need to do any further finishing.
If you have removed the bark, you might wish to use a coat of cement or nail polish to stiffen the throat up again. I haven't explored this at all, but it could in theory improve vibration of the blades.