Transition current hardware platform to Vivado

Description of the problem

Right now we are using XPS as a platform to generate our hardware output file, this software is annoying and tedious to use.

Proposed solution

We are planning on changing this platform to Vivado(another xilinx tool that is more modernised and supported)

The 18-19 team has been recommended to essentially start this process over. That will be my job, this will entail starting from the current stable release.

assigned to @jtalbert and unassigned @ktrost

Components to Build in Vivado

• PWM_Recorder
  • Recorder with shift-register debouncing
  • Recorder with Debounce counter module (compare with above)
  • IP Core
• PWM Signal Out With Killswitch
  • PWM Generator
  • IP Core

I've moved the old branch for this project to match master, all the old commits are there, just undone by a merge.

I got a project working (hello world code runs) with a recorder block included in the design. I'll start building the unit test branches (one per sensor/output) today.

I have scripts for building the projects and software test applications for the PWM units. The tests are currently extremely thin, but are able to read and generate PWM signals successfully.

I have integrated a PWM recorder test into the CI framework. This required mounting the Xilinx folder on the VM. There is a document in our google drive with the login info (not public for security reasons). Any pull request will now run the HW simulation tests automatically.

Remaining Tasks:

• Create Simulation Testbench for PWM_Generate
• Create Full System hardware design
• Test quad_app on the hardware platform

The PWM_Generate kernel module has a testbench that passes.

I have created the XSDK project for the new platform, swapped the Zybo on quad 01, and run some tests.

• IMU: Working Correctly
• Optical Flow: Probably working correctly (it's a little hard to test with the USB cable, but it gives +x for one direction and -x for another.
• LiDAR: Working correctly
• PWM Capture: I think working correctly. Throttle definitely works, but the controller is low on battery so identifying the other channels is not going quickly.
• PWM output: Working.

Current Work Items:

• Get batteries for the controller
• Do a no-motor test run
  • Flat bench tests
    • Motor PWM goes to 1 ms on start-up (controller in killed mode)
    • Motor PWM increases slightly when controller set to on
    • Motor PWM increases proportionally with throttle stick
  • Basic Stabilization/IMU Tests
    • Motor PWM varies with quad angle
    • Motor PWM counteracts quad angle
• Do a flight test
  • Report Results
  • Fix the problems

The hardware interfaces seem to work correctly. The current issue is that the controls are not functioning on this quad. I am going to attempt to revert the quad_app software to the golden build and see if that helps.

It didn't.

Additionally, I'm not sure how to connect the quad to the ground station to get the control graph. @bertucci, can you meet up Tuesday morning?

Manual flight works. Now to do an autonomous flight.

created branch 41-transition-current-hardware-platform-to-vivado

mentioned in merge request !41 (closed)

The autonomous flight eventually worked. I am going to wait and merge this as part of the PCB transition as there are several hardware changes involved with that. This should be mergable again in a couple weeks or less.

Thus far, the PCB is working. For some reason, the I2C controller is reading a NACK from the IMU when the line is clearly pulled low on an oscilloscope. The line does return high for a little bit after when the IMU releases it. Everything on the scope looks like a valid transaction, and the IMU correctly ACKs its address, and not other addresses. I have not yet verified the clock speed (thought about it after I left the lab).

So, I got it working by forcing a full system reset in XSDK. I am evaluating how this applies to SD-card boots.

It looks like things work fine with the SD card.

• Power Working (5V to Zybo from battery via regulator)
• Motor Control Working (PWM)
• IMU Working
  • Looks like an I2C controller cofig issue (I don't know why we haven't seen it before).
  • I'm going to try reverting to the MPU-9150 to see if that made a difference (I don't think it's supposed to).
• WiFi Working
  • UART to/from WiFi Working, I can't test the actual WiFi right now.
• LIDAR working
• Optical Flow Working
  • I don't have a great way to evaluate the accuracy of the sensing, but the communication with it is working.
• RC Receiver Working (order might be reversed)

I need to re-orient the IMU to software.

I have the quad communicating with the ground station again. This weekend/early next week, I will need to publish an errata sheet for the PCB (some pairs of wires need to be flipped).

mentioned in merge request !43 (merged)

closed via merge request !43 (merged)

mentioned in commit c888421d

removed In Progress label