I milled the PR13_B_SPACER_PELV mounting brackets which go on top of each leg.
The brackets are necessary to attach the legs to the chest section where the main computer resides. Currently I am redesigning the chest parts to fit a Odroid-XU3 computer, the main changes are in the position of external USB and Ethernet ports.
The cut files for the mounting bracket are in SourceForge here. In the README file is the list of gcodes files.
For future parts I’ll try to change the cutting strategy for milling. Instead of using several drills and endmills, I’ll try to use a single 1.5mm endmill for all cuts, avoiding tool changes.
I finished assembling the second leg for my DARwIn-OP clone.
I finished writing the gcode cut files for the pelvic mounting brackets (PR13_B_SPACER_PELV) which go between the legs and the chest, and I’ll be cutting them soon.
Meanwhile I was redesigning the chest brackets to accommodate the ODROID-XU ARM computer instead of the original FitPC with an Intel processor. As an alternative, I am also considering to use an ODROID-U3 computer, with has a smaller footprint and a less powerful ARM processor (and consuming less power). The latest ODROID-U3 (Revision 0.5) has an native SPI interface, so which could be used to interface sensors (accelerometer and gyroscope).
But now, Hardkernel is going to release the ODROID-XU3 in August and some info is available. The main differences with the ODROID-XU that I find interesting for my project are:
Faster Cortex-A15 cores (2GHz).
Full octa-core Heterogeneous Multi Processor system.
More energy efficient.
No more LCD connector to get the native I2C #1 bus, but it is available now in the easier to access I/O expansion connector.
Included energy sensors.
Audio-In added (mono only).
OpenCL working in Linux.
Same form factor as the ODROID-XU, so easy replacement.
Anyway, I am scratching my head on how to cut the chest brackets on my Sherline mill, which is too small for these brackets, so I’ll be making them in parts.
I decided to do some electronics and software testing before finishing building the second leg of my DARwIn-OP robot.
I wired my ODROID-XU with the GY-80 10DOF module using the I2C bus on the LCD connector (CON15) and a PCA9306 to adjust communication voltages.
The following is the connection diagram tested:
This 10DOF can be powered from 2V to 5.5V (VCC_IN). Its sensors are internally powered at 3.3V, but according to this documentation its I2C interface is rated at VCC_IN, so the PCA9306 has to translate from ODROID-XU’s 1.8V to VCC_IN (and not to 3.3v).
This 10DOF has 4 I2C devices in the following addresses:
0x1E: 3-axis compass (HMC5883L chip)
0x69: 3-axis gyroscope (L3G4200D chip)
0x53: 3-axis accelerometer (ADXL345 chip)
0x77: pressure and temperature sensors (BMP085 chip)
With this small C program, the accelerometer chip can be query in order to detect it and verify the setup. I am currently testing Ubuntu Server 14.04 off the shelf on the ODROID-XU, which includes the GCC compiler by default.
Now that I have I2C communication running, I followed the chips documentations and some googled examples in order to write some draft C++ classes to read the 10DOF sensors. They are available in this Code Section in the SourceForge repository for cloning the DARwIn-OP Robot.
I could successfully test the 10DOF powering it (and the PCA9306) with 3.3V and 5V, but at 3.3V they draw 0.024A (or 0.079W) and at 5V they draw 0.085A (or 0.425W). So in my final DARwIn-OP design I will most probably include a 3.3V regulator to lower consumption.