Experiments with a CNC router - sound hole rosettes

Initial trial using Deskproto CAM software was to cut a .jpeg image in shallow relief as a Bitmap project using the Wizard based interface. I was curious to see how the machining worked out with a view to eventually piercing right through a work piece with the cutter to make a rosette.

Two test runs were made the first in 3mm clear Plexiglas and the second in 3mm thick figured maple. The work blanks were screwed to wooden blocks clamped to the router bed and cut with a solid carbide V engraving bit supplied with the router kit. Maximum depth of cut (negative Z direction) was 2.5 mm with the tip of the cutter zeroed on the upper surface of the work piece.

The software calculates depth of cut for black as zero depth and for white as maximum depth. A colour image is converted by the software to grey scale the depth of cut depends then upon the density of grey for each pixel (ranging from minimum black to maximum white).

The attached portrait photograph was used for the test - well lit with no shadows that would distort calculation of depth of cut. Not a rosette but bear with me!!

The test in clear Plexiglas machined well without problem. The carved image (a.k.a Lithophane) can not be seen by back lighting but is revealed in incident lighting by brushing a translucent white liquid over the machined surface. The fluid fills the deep cut areas in white.

The trial with figured maple, back lit was also successful.

So much for the 'artistic' stuff. Next to try to machine a rosette as a Bitmap project

Bitmap machining is a fairly slow process the cutter moving side to side across the work piece only advancing a fraction of a millimetre with each pass of the cutter. To avoid spending too much machining time in the trials, therefore, relatively small sound hole pattern examples have been selected for processing.

For the first trial a Nahat Eyun design has been used. The rosette will be cut in 2D so from a photo of an original Eyun the image was converted into a plain black and white .jpeg image using Photoshop editing software. So in the Bitmap process pure black represents zero cutting depth and white the set maximum depth.

This example will be a challenge to cut because of the fine detail and type of cutter used but the image may be made thicker with smoother edges in places and larger in diameter to compensate if necessary. The diameter of this first test Eyun is about 5 cm

The cutters supplied with the router kit are V shaped solid carbide engraving bits with a 20° included angle. This means that machined edges of a work piece will slope at a 10° angle and will not be perfectly vertical.

Furthermore the cutters, being only half round in section, cut with greater dynamic forces (vibration) than a more delicate (and costly) parallel sided spiral cutter might. See the attached image of the irregular 'straight line' cut on the Lithophane example in relatively hard acrylic plastic (Plexiglas) previously posted. Part of this irregularity was due to slack in one of the linear slide bearings that has now been replaced.

Mounting a 2mm thick work piece securely for machining so that it may also be easily released on completion without damage presents a bit of a challenge requiring some experimentation.

To protect the router bed plate from damage and for convenience of set up and removal, each work piece is mounted on a scrap wooden block clamped to the bed plate. The clamps provided with the router kit are not very good so a T-slot clamp and a couple of T-slot low profile work stops from Lee Valley work nicely instead.

There are a number of alternative solutions to mounting the work piece on the block. If the machining operation pierces through the work piece to greater than full thickness then the work piece must be uniformly mounted with adhesive so that it doesn't fall apart during machining. One method is to first cover the underside of the work piece with painter's masking tape (weak adhesive so easier to release) and then stick the work piece directly to the mounting block with either diluted hide glue or alternatively double sided adhesive tape.

Alternatively if the work piece is not machined quite to full depth leaving a small thickness to be removed by sanding after removal from the mounting block then the work piece may be simply mounted directly to the block with four screws, one in each corner. This method has been successfully tried with the Lithophane examples previously posted but not so far for a rosette.

The attached image shows an MDF mounting block clamped to the bed plate with a completed work piece held in place with masking tape and diluted hide glue. MDF (medium density fibreboard) material has been found to be too soft for this application resulting in a rough, poorly defined machined edge as the cutter breaks through the underside of the work piece.

The first test with the Nahat eyun design, undertaken a few months ago, was to machine an African ebony work piece 1.3 mm thick using the Bitmap process. As the cutter depth of cut was set deeper than the work piece thickness the underside of the work piece was covered with masking tape and then glued to an MDF mounting board with water soluble fish glue. Upon completion of the machining operation the work piece was removed from the mounting block with a thin metal spatula.

Apart from a couple of small missing pieces the rosette was complete although the edges are pretty rough looking. This may be because the tool path was not set fine enough but also because the masking tape and support block material combination was too soft to support the cutter breaking through the work piece for a clean cut at full depth? Anyway a good start that can only be improved through experience.

An alternative method for cutting a rosette with DeskProto software (free version) is to use basic Vector machining where the machine path for the cutter follows the lines and arcs of a 2D drawing or image. This is a much faster machining operation than the previously reported Bitmap machining (where the tool path zigzags across the work piece advancing only a fraction of a mm at each pass). It should also produce smoother edges than achieved so far with Bitmap machining. The free version of DeskProto only allows a single tool pass (several passes may be used in the paid versions allowing roughing and finishing cuts for thick materials) but for rosettes only 2mm thick a single pass may be satisfactory?

The Vector operation starts with a DXF (Drawing eXchange Format) file that may be converted directly from a JPEG image file through free online conversion sites.

Several test runs with different materials were undertaken earlier in the year before the project was put on hold. Securing each work piece to the mounting block was the main difficulty - diluted fish glue worked for machining but separating the rosette from the mounting block (using moisture and heat to soften the glue) without damaging the fine detail was difficult and no satisfactory versions of the Eyun rosette were cut - even when the rosette diameter was increased to 6 cm.
The fine detail of the Eyun rosette is a challenge to cut in wood when the cutter is set to cut through the full depth of the work piece. More uniform materials such as plastic or soft metal may be more successful such as this attempt in soft aluminium cut to almost full depth.
Note that all of these earlier attempts used the same V engraving cutters and were made before the loose linear bearing was replaced - so some tool vibration may be evident causing rough cutting. Also when cutting through full depth the work piece should be mounted on a block of the same material as the work piece to minimise tool vibration. Alternatively the work piece may be cut to slightly less than full depth and then the thickness reduced by sanding - after the machining operation - to reveal the open areas of the rosette.

As an alternative design to the delicate Nahat Eyun trial cuts were made using the central part of an original 16th C lute rosette by Giovanni Hieber - a more 'robust' design that might better withstand the cutting and handling forces without breaking(?)

A hand drawn copy of the original rosette was made and a JPEG image prepared and converted to a DXF file extension for Vector machining. The hand drawn rosette design has imperfections that the software will copy which may or may not be a good thing - rougher machined edges but representing more closely a hand cut rosette? Smooth lines in the image may be achieved if necessary by processing in free CAD software such as Inkscape - a lot of extra work so avoided for these preliminary trials.



Trials were made using spruce sound board material about 2mm thick and glued to a mounting block with diluted fish glue and a paper backing for additional reinforcement - the cutter being set to cut full depth of the work piece. The results were pretty rough looking - again due to some cutter vibration and perhaps material non-uniformity due to hard grain in the spruce. Separation from the mounting block without damage was again difficult.



This summarises the results of trials completed earlier this year. Picking up this project and starting again the next trials will be Vector machining in a close grained hardwood (boxwood) of the Eyun design, 6 cm diameter and 2.5 mm thick. Depth of cut will be 2.4 mm and after machining the underside of the work piece will be sanded to open up and reveal the rosette design. This way the work piece may be simply fixed to the mounting block with four screws allowing easy release without damage.

The CNC 3018 Pro router is designed to work with a laser engraving head in place of the router motor/spindle. I originally did not intend to try laser engraving/cutting due to the hazards involved but have now decided to give it a try to complete my cnc machining education.
I have purchased the SainSmart 5.5Watt Laser Module Kit designed to fit the CNC 3018 Pro router and am currently testing the installation to ensure it works as intended so that it does not have to be returned. If OK then laser cutting of rosettes will be reported on this forum as a separate topic.

The kit comes with the laser head and control modules easily installed with wiring and mounting hardware and includes the all important safety goggles. As laser burning produces fumes the router and computer were moved into my small workshop that has a fume extraction fan and is otherwise enclosed for safety.

Control firmware for the router is open source LaserGRBL that also converts any image to Gcode for controlling the laser head movement. Like the DeskProto firmware used for routing operations previously posted LaserGRBL can engrave photographic images (similar to Bitmap router operation) and also engrave or cut to line drawings in Vector mode.

As a preliminary functional test the .jpeg image of the B/W Eyun rosette previously posted was first converted to a line drawing using Photoshop software. This was then loaded into Laser GRBL for auto conversion to GCode. The laser beam was focused onto a piece of plywood and a test burn initiated. I need to have some practice at getting optimum focus to a fine point but this first engraved example with a relatively coarse focus setting (it took 9 minutes to engrave) is the outcome. The laser may repeat the same path if required to cut completely through the work piece (presumably setting the depth of burn a little deeper for each path?). Further experimentation will be required to achieve optimum settings but this is a start.

The laser head will now be replaced by the router spindle to continue with trial vector machining of rosettes.