McGraw is a pen plotter at DoES Liverpool, that was converted from a laser engraver.
Because of this former life as a CNC machine, McGraw expects to be fed G-Code files. So if you want it to draw anything, you’ll need to convert your artwork to G-Code format. This is where the scripts in this repo come in!
First, run the script to create a virtualenv and install the required Python packages into it. Open a Terminal, cd to this directory, and run:
script/setup
A little while later, it will have set everything up.
Now activate the virtualenv:
. ./venv/bin/activate
Eventually, we want the code in this repo to work for any SVG or G-Code file. But for now, you are expected to either pass a perfectly-formed G-Code file to ./send-serial.py, or to export some G-Code from Inkscape then fix it up with ./fix-gcode.py and send it with ./send-serial.py.
For example, assuming you have a G-Code file that has been exported from Inkscape (in this case, test/a3-landscape-test.gcode):
cat test/a3-landscape-test.gcode | ./fix-gcode.py | ./send-serial.py
Note: send-serial.py will attempt to guess McGraw’s device path on your system—defaulting to common paths like /dev/tty.usb* and /dev/ttyUSB*—but you may need to provide a path explicitly:
… | ./send-serial.py test/a3-landscape-test.gcode --device /dev/cu.usbserial1
McGraw will now start drawing your file.
You can send one-off G-Code commands to McGraw by piping them into ./send-serial.py, eg:
echo 'M3 S05' | ./send-serial.py # pen up
echo 'M3 S120' | ./send-serial.py # pen down
When fixing up your G-Code, you can provide a custom speed parameter (the default is 800):
… | ./fix-gcode.py --speed 1200 | …
If any of the scripts give you a ModuleNotFoundError when you run them, then you’re probably running them outside the virtualenv. Activate the virtualenv, and re-attempt:
. ./venv/bin/activate
cat test/a3-landscape-test.gcode | ./fix-gcode.py | ./send-serial.py
You can stop send-serial.py partway through a file by pressing Ctrl-C. Note: This will prevent further commands being sent to McGraw, but McGraw will continue whatever commands have already been sent – the only way to prevent that is to cut power to McGraw using the red button on its circuit board.
svg-to-gcode.sh is meant to be a friendly wrapper to a complex vpype command. Vpype is a Python utility creating and modifying vector graphics like SVGs. In case, we’re using the vpype-gwrite plugin to output G-Code. The settings in mcgraw-config.toml tell vpype-gcode to use the specific commands (eg: for pen up, pen down) that McGraw the Pen Plotter expects. But it doesn’t work right now.
svg-to-gcode.py is meant to use the vpype and vpype-gcode Python libraries to read the specified SVG files, optimise them for plotting (combining lines, sorting lines by proximity), and then export them as G-Code files. The settings in mcgraw-config.toml tell vpype-gcode to use the specific commands (eg: for pen up, pen down) that McGraw the Pen Plotter expects. But it doesn’t work right now.
fix-gcode.py automatically performs the replacements that we used to do by hand (see Converting from Inkscape GCodeTools output, below), making G-Code files exported from Inkscape’s built-in GCodeTools “Path to Gcode” extension suitable for drawing on McGraw.
send-serial.py reads G-Code commands from the given file, or from whatever’s piped into it on standard input, establishes a Serial connection with mcGraw, and sends each line of the file to McGraw, printing McGraw’s responses. Since it’s using a “call and response” method (meaning that McGraw has to wait for the next command after finishing the current one, rather than already having the next command in its internal queue) send-serial.py potentially slows down McGraw, especially when drawing lots of small line segments. Perhaps one day we could re-write send-serial.py to use a batched approach.
Jackie provided a Juicy G-Code flavour file, with the following command snippets:
| expression | what it does |
|---|---|
G17 |
select the XY plane (this probably isn’t required, as G17 is the default in grbl) |
G21 |
use millimeter units (this probably isn’t required, as G21 is the default in grbl) |
G90 |
use absolute distances (this probably isn’t required, as G21 is the default in grbl) |
G28 |
rapid move back to home |
G1 F400 |
linear move, at feed rate 400 |
M3 S05 |
set spindle speed to 5 – McGraw interpets this as “pen up” |
G4 P0.3 |
dwell for 0.3 seconds – we use this to avoid “ticks” caused by the pen remaining in contact while the head moves |
| expression | what it does |
|---|---|
M3 S05 |
set spindle speed to 5 – McGraw interpets this as “pen up” |
G4 P0.3 |
dwell for 0.3 seconds – we use this to avoid “ticks” caused by the pen remaining in contact while the head moves |
G1 X.000 Y.000 |
linear move to XY position 0,0 |
| expression | what it does |
|---|---|
M3 S120 |
set spindle speed to 120 – McGraw interpets this as “pen down” |
G4 P0.3 |
dwell for 0.3 seconds – we use this to avoid “ticks” caused by the pen remaining in contact while the head moves |
| expression | what it does |
|---|---|
M3 S05 |
set spindle speed to 5 – McGraw interpets this as “pen up” |
G4 P0.3 |
dwell for 0.3 seconds – we use this to avoid “ticks” caused by the pen remaining in contact while the head moves |
The DoES wiki explains how to replace some expressions generated by GCodeTools, with equivalents that McGraw understands. These are archived here for convenience:
| before | what it does | after | what it does |
|---|---|---|---|
G00 Z5.000000 |
rapid travel to XYZ 0,0,5 |
M03 S05; G04 P0.3 |
set spindle speed to 5 (pen up) then dwell for 0.3 seconds |
Z-0.125000 |
a coordinate 0.125mm below the surface | nothing (so 0) |
|
G01 F100.0(Penetrate) |
linear move at feed rate 100 |
G04 P0.3; M03 S120 |
dwell for 0.3 seconds then set spindle speed to 120 (pen down) |
F400 |
speed 400 |
F800 |
speed 800 |