Test Summary
Part I: Failed Integrated Hardware/Software Simulation (8:00am - 1:00pm)
Jamey, Sarah and Andrew met early in the morning (8:00am) and tried to get to do an integrated hardware/software flight simulation. By using the Launch Tower Computer (LTC) as a high speed network to CAN bus converter, we thought we could replace our sensors with "virtual sensors" from Dave's discrete event simulator running on Jamey's laptop. We chose to use the real APS, Recovery and ATV nodes, and simulated GPS, pressure and IMU.
Amazingly, Jamey got all the connections just right: messages from the simulated nodes on Jamey's laptop made it onto the CAN bus and CAN messages from real hardware got injected into the simulator and rocketview. It was cool.
Well, it was cool until we realized that the LTC's Intel 82527 couldn't read "RTR" messages because it's a completely screwed up chip, and that basically killed the test right there. Sigh.
Part II: Succesfull System Test in Gabriel Park (2:00pm - 9:00pm)
Eric and Frank showed up, and we packed everything into Andrew and Frank's vans. The failed simulation put us behind a bit, so we didn't get to Gabriel Park until 2:00pm (sorry Josh!).
We first set up launch control (LC) at the top of the hill (about 300 ft south on SW 37th from the corner of SW Vermont and SW 37th), and shockingly, started following the checklist to set the rocket up. It was... a new experience ;) And as such, we found a lot that we wanted to change. Eric dutifully updated a local copy of the CurrentCheckList page, which helped us really nail down exactly what we needed to do. We even worked out a few more pieces of the checklist we didn't realize was missing, like what happens if we need to restart the launch while on the pad.
We turned on the Flight Computer (FC), prepped it for flight, and after a bit of hassling, got it ready for it's "flight". Frank, a newly re-arrived Nathan, Bart and Andrew drove down and set up the Launch Tower Computer (LTC) this time on it's tripod. We then set up the rocket, and then Jamey, Kieith, Josh and Eric went at it.
I believe we tried launching something like a dozen times over the course of the afternoon; it was a pretty intense work day. Here are some highlights:
- We updated the kernel and CAN driver on the LTC (it now uses the ADEOS kernel and appropriate "uncanny" CAN driver code.
- During one launch attempt, the "transition from armed to rocketready" command got dropped between the field server and the FC. The FC thus never asserted the rocketready signal which then turns on the the rocketready launch interlock relay, so despite going through the whole launch sequence, the rocket never "launched". This is exactly what should have happened if we lost that packet, which was cool. We restarted the launch, and as expected, it worked fine.
- We caught a bug which might have fired the recovery system right after leaving the tower. Whoops. Since we disabled coast detect, the apogee code was running right after boost, so it's possible that due to sensor filtering on the pressure and GPS we might have gotten an apogee detect. Our fix was to check the IMU for low levels acceleration (non-boost, and non-high speed deceleration) before we start checking for apogee.
- We experienced the behavior from the 12V-to-19V shore power DC-DC converter where it jammed the ATV and wifi signals. Turning off shore power solved that problem. And since we can do that an arbitrarily long time before launch, this is not a critical issue.
- We observed the flight computers interacting with the recovery's pyro timers and it did what we expected.
- There is one last non-critical but important issue left: why the last bits of code we added today didn't work as expected on the IMU_COAST bit. Luckily the timers fire as expected, so this is not a big deal, but we'll fix it "offline" either on Tuesday or Wednesday.
Finally, as it was starting to get really dark around 8:30pm, we quit for the day.
Thanks to everyone who helped out at the test today, especially to:
- Burt and Pam Triplett (Josh's parents) for bringing us water and pizza(!). That was great.
- Frank, for running the LTC by himself at the end of the day while Andrew, Jamey and Eric worked on testing at the LC.
Where
We met first at Andrew's workshop in SE Portland to pick up the tents, rocket, etc, and then headed out to Gabriel Park in SW Portland. We were in the very northeast corner of the park, just south of the corner of SW 37th Ave and SW Vermont St (google map).
What
Full PSAS system test. By the checklist. Goals:
- Integrated hardware/simulator test to try post-launch detect sequencer code: FAILED (not critical).
- Change lv2_node_id.h
- Recompile fcfifo & tools/ccpsim
- regen CanBusIds.java
- recompile rocketview
- install new run_thredas on rocket
- socat on ltc
- link fc-sim.sh to ltc
- point rocketview at real rocket & simulated tower
- hook avionics CAN bus to LTC CAN bus (REQUIRES NEW CABLE?)
- watch rrr light during launch
- pull umbilical manually
- Make sure we don't fall behind on the network if the disk is not full: OK
- Test the ATV power output (on some kind of pole?) to get away from nearfield: NOT DONE (not critical)
- Run through the checklist a bzillion time, by the checklist. Did we mention the checklist? DONE
- Modify the checklist as necessary to reflect changes in software/hardware since February DONE