I have been following a process involving several steps to export a complex shape from inkscape SVG to a dxf file and later import the 2D dxf into openscad and then do a linear extrude to get a 3D model.
I was wondering if there is a shorter or direct way of doing this, like getting the data points from the inkscape vector graphic and then using the polygon primitive in OpensCAD to create a polygon that I can extrude. I tried opening the svg in a text editor like Gedit, copying just the data for the polygon, using an online svg to WKT converter and finally copying the converted points into OpensCAD and creating a points variable with that data. This worked as a proof of concept but the origin was far off and I couldnt figure out how to easily fix that in the conversion process. It was also too much of a process to follow. As I was messing around, I stumbled upon an inkscape plugin for doing the conversion in a straight forward way, exporting actual OpensCAD code with the origin variable defined. This is awesome. you can even launch OpensCAD with the exported file to preview from the plugin’s widget.
Installing The Extension
Cd into your sources directory and clone the repository:
cd src && git clone https://github.com/fablabnbg/inkscape-paths2openscad.git
I installed the latest version of android studio after upgrading my laptop to Ubuntu 18.04. There are subtle changes that you have to make to the old projects after importing them into the new android studio to make gradle sync successful and build dependencies satisfied.
In the project under Gradle Scripts, open the build.gradle(Project:name) file. In the buildscript section, make sure you have included the Google Maven repositories, i.e google().
buildscript {
repositories {
google()
jcenter()
}
dependencies {
classpath 'com.android.tools.build:gradle:3.1.2'
// NOTE: Do not place your application dependencies here; they belong
// in the individual module build.gradle files
}
}
You may also include the same repository under the allprojects entry:
I found myself faced with the need to install KiCAD from source after upgrading my laptop to Ubuntu 18.04 and installing KiCAD 4.07 which is the stable version from the Ubuntu repositories but it had graphics issues. The cursor painting was was leaving a trail on eeschema making it unusable. I wanted to install from source as a last resort after trying every intel graphics and X11 hack I could find to no avail, if there was even the slightest chance to fix the bug it would include at least building the right packages from the source.
installing the dependencies
There are some dependencies that have been outlined on the github page, check and install for your distro. This list may not be exhaustive here as I may have had some of them already installed.
I also built ngspice from source following my earlier tutorial in which I fixed a bug. libngspice doesn’t really install well from Ubuntu repositories either.
Get the Source
Get the source code by clicking the KiCAD launchpad download link. This gets you the source for version 4.0.7. save the file somewhere in your computer [I always use the directory ~/src] and extract the .tar.xz file, cd into the folder.
cd ~/src/kicad-4.0.7/
create a build directory:
mkdir build && cd build
We need to Install without scripting to avoid wxpython3.0 wxpython2.8 wars: I couldn’t install wxpython2.8 in Ubuntu 18.04 .
You have to initialize bzr first or the build command will complain. I have no idea what that is, I only used git but I think it’s some version control tool like git. Anyway the command is:
bzr whoami "First Last<first.last@domain.abc>"
The option for switching off scripting is: T-DKICAD_SCRIPTING_WXPYTHON=OFF. Run cmake to generate the makefiles.
cmake -DKICAD_SCRIPTING_WXPYTHON=OFF ../
I ran make -j5 followed by sudo make install, I realized it builds everything twice, takes long. If I were you I would just run:
First things first, I assume you have downloaded your raspbian OS and burnt it on to an SD card that is booting properly on your pi zero W.
Then you have setup the pi to connect either to your local network or to a hotspot through your phone or PC. Basically you have internet access on your pi and you can ssh into it.
Wayne Stambaugh made a presentation of the new features coming to KiCAD version 5 at FOSDEM and it was amazing.
So how do you get it running on Linux right now? I am using Ubuntu for work and the easiest way is to use a ppa. If you head to the KiCAD Ubuntu download webpage you will find the instructions on how to install from a ppa. If you are already using the stable version and Ubuntu has not released the new version yet, you can install the nightly development build via another ppa. The instructions to upgrade to the nightly are:
Add the nightly ppa[please confirm this ppa from the site as it may change] to your sources with the command:
After that, you will have the latest version of KiCAD.
I have tested it and it opens the projects created by the old version pretty well, with a few library swaps. It backs up the old project without any changes in the same folder, just in case you want to roll back to the stable version.
One of the reasons I decided to risk and upgrade was because of the embedded simulator. I could already export netlists from KiCAD schematic editor and use ngspice to do AC and DC simulations and draw graphs pretty well, but an embedded simulator would be even better for my students as they do not like working in the terminal that much. I quickly tested it with a voltage divider and it works OOTB.
Simulation Program with Integrated Circuit Emphasis. NGSPICE is an open source mixed-signal circuit simulator. It is freely available for use in Linux and Windows. It is recommended to use Linux for NGSPICE. NGSPICE requires you to describe your circuit as a netlist. A netlist is defined as a set of circuit components and their interconnections.
Why install from source?
Because the most recent code has some nasty bug fixed. If not, you can fix the bugs in the code yourself, build and install the freshly built binaries.
What bugs?
Well, there is this one bug I found, when you plot multiple lines like in transistor output characteristics, you get the lines joining back to the origin after the last point as seen below.
When you make a schematic in KiCAD, you place components and then interconnect them with wires by clicking on the green place wire icon and clicking on the pins of the components you want to connect, one after the other. Pretty simple. When you export a spice netlist, the default setting is that the wires are given numbers as their names. In fact you can specify this setting in the export netlist dialog. It is however easier in spice simulation, especially if you are using ngspice in a Linux terminal, if the wires had meaningful names.
voltage divider schematic
The resulting netlist is as follows:
* Sheet Name: /
R2 1 3 470
R1 1 2 1k
v1 2 3 DC 12
.control
op
tran 0.5s 1s
tf v(vout,0) v1
print all
.endc
.end
This is a brief step by step instructional tutorial to compile openscad from the latest sources hosted on github. If you want to install the ‘stable’ version for your linux distribution, then you should head to the openscad download page and grab the binary instead.
Why build from source?
I have been learning and teaching how to create mechanical 3D CAD models in openSCAD and export stl files for a while now. In this time, I have come across a number of bugs and potential feature I would like to request. One of those features is the partial rotate_extrude which is in the documentation but doesn’t work in the stable Ubuntu 16.04 binary. A quick search on the internet took me to the github repository for openscad where these bugs have already been fixed and the changes committed and merged, and the features implemented and incorporated in the latest sources. Instead of wait for my distribution to build updated binaries, I decided to take the shortcut and uninstall the official apt binaries to install from source. Continue reading “Building and installing openSCAD from source in Linux”
I read a blogpost on mcuoneclipse about MCUXpresso installation and I downloaded the installation files for MCUXpresso from NXP and headed for the terminal to install, only there was an error about some P&E probe udev permissions file failing to be overwritten during the installation process. I found the answer in the support forums and here are the steps to follow:
press CTL+ALT+T to open a terminal, change to the directory where the file is and run the binary file with –keep option. It will fail but leave the extracted folder with files behind for the next step
cd Downloads && sudo ./mcuxpressoide-10.0.0_344.x86_64.deb.bin --keep
If the installation succeeded, there is no need to bother much. If it failed the .deb package will be kept. Get into the install folder and open the install script in a text editor
cd install && sudo gedit install.sh
Add the –force-overwrite option to the dpkg command by editing line 35 in the file, save and close gedit
The mcuoneclipse post already has all the info on downloading the SDK and adding them to MCUXpresso, so just head there for that. No need to replicate everything… Just make sure you select Linux as your OS and MCUXpresso as your IDE when creating an SDK configuration.
There are situations where what you need is an API of all the low level register access definitions and functions/Macros or whatever you want to call them, routines even… and you want to write your own code the best way you know how, whether using an infinite while loop in main or using an embedded operating system like FreeRTOS. The main focus at the beginning of writing embedded software is figuring out the memory mapping, startup and hardware drivers. Once you have those in place and test them roughly, you want to write application code. The tools provided by manufacturer as an IDE provide you with a way to create a new project for a specific micro-controller, putting together the MCU header file, the linker script, the startup script and potentially the peripheral registers declarations or drivers.
For ARM®, there is of course the Cortex® Microcontroller Software Interface Standard (CMSIS) that helps abstract hardware access. Its implementation provides a standard interface for the core of the micro-controller, the ARM CPU and peripherals. It has been extended beyond the core to interfaces like CMSIS-DAP debug and later to others including Drivers, DSP, RTOS-API, Software Packs and SVDs which are System View Descriptions for peripherals e.t.c. This has kind of standardized software interfaces make porting code to new microcontrollers easier and encourage code re-use. It is, I believe the reason why Mbed, FreeRTOS and other operating systems have been ported to ARM microcontrollers and support for different families of these microcontrollers added so fast. It has also allowed GUI tools developers add more features to their software suites to better help the developer debug hardware. Stuff like peripheral register views were not available in eclipse when I started writing tutorials on how to install vanilla eclipse, then add hardware debug plugins. USBDM is one debug software that I loved to learn about. GNU ARM Eclipse Plugins by Liviu Ionescu is another success story for me. I loved that at a time when I could not afford to procure the then CodeRed suite license, I could still own my hardware by adopting purely open-source tools like OpenOCD, USBDM, GNU-ARM toolchain and Eclipse IDE. Continue reading “Dynamics of Embedded Systems Development for ARM based Microcontrollers”
Recent Comments