You are here: Foswiki>FLAO Web>UaO>UAOSwTestLog (15 Apr 2016, LucaFini)Edit Attach

UAO Software Test Log

The software tests for UAO started on April 4th. Here follows a log of the work done to test the FLAO software.

1st Week: 4-8 April 2016

Resume: The low level tests performing during this week were successful, except for an issue with the new BCU firmware which is being examined by Microgate. All the other problems encountered were depending on OS configuration and were easily solved. During the tests a small number of code bugs were discovered and fixed. During the tests we realized that we had not addressed some aspects regarding the GUI support for the removal of the failed actuator. We obviously provide the support in the AdSec GUI, but it might not be suitable for the night operation with the T.O.. The issue needs some feedback from LBTO people.

Day 1

  1. OS configuration and FLAO Software (Supervisor) installation.
    This step was concentrated in the few hours preceeding allocated daytime. We followed the procedure as described in the related document (see: Installation of FLAO Supervisor) so that it was updated in the process. To be noted:
    1. the OS as installed on server was lacking many development tools (compilers, subversion and the like); this is probably due to the selection of installation flavour. It is suggested to install the systems as "development workstation" so that most of the required tools are selected.
    2. During the installation we could spot out a few items regarding the configuration of server which needed to be fixed (additional disk mount and NIC configuration). This was promptly solved by LBTO people on their next daytime.
  2. Preliminary tests of the AdSec
    1. We could start the AdSec related processes and some engineering GUIs and could upload the new firmware on the AdSec BCU's

Day 2

  1. AdSec Tests.
    Many low level control routines were tested against the new firmware developed for UAO project. This allowed to spot out and fix a few software bugs and to upgrade some configuration files.
  2. WFS Tests.
    We tested all the low level operations on the WFS subsystem successfully. During the process we could remove a few bugs and do some improvements to the code. In particular we moved some hard coded parameters into configuration files, to improve the consistency of the programs. We discovered also a few issues remaining in the OS configuration which were not addressed in the configuration document.

Day 3

  1. AO Arbitrator test.
    We tested thoroughly the functionality of the new Ao Arbitrator. This is most relevant because the AO arbitrator has been completely rewritten. We were able to verify the communication between the AO arbitrator and the two lower level arbitrators for the AdSec and the WFS. Some basic commands (those not requiring setting the AdSec) were also tested completely. We also tested the new FLAOGUI, developed for the UAO project, after solving a remaining OS configuration software: the QT library was apparently installed in a non standard path.
  2. AdSec tests.
    We continued the debugging of the AdSec firmware. In the process we tested the proper functioning of the AdSec status GUI's.

During today's work we could spot out and fix a couple of bugs related to porting to 64bit of the FLAO software.

We also discovered a problem with one of the NIC (slow speed). The LBTO people solved the issue which was due to a bad ethernet cable.

Day 4

  1. Arbitrator test.
    We continued the testing of the arbitrators exercising more commands. We performed successfully all tests not requiring some light.
  2. AdSec tests.
    We could set the mirror using the new procedures and firmware. Unfortunately we also discovered an issue in the firmware which needed support from Microgate. Some answer on this is expected for monday 4/11. The issue is not a stopper for the programmed tests because we have a workaround, but must be solved before testing on sky.

Day 5

  1. Arbitrators.
    We started testing the new FSM on real hardware. The test exercised both "automatic" and "intervention" modes up to the "ReadyForCloop" state. All commands were tested successfully in dark conditions (that is, without a real source), except for the CenterPupils command that requires a source.
  2. AdSec tests.
    We started testing the new procedure to remove failing actuators. Some issues have been found and addressed. This will require more tests next week.

2nd Week: 11-15 April 2016

Day 1

  1. Diagnostic process.
    Some modifications are required in the MirrorDiagnostic code. This will be done in the next few days and may require some additions daytime before the night tests for final verification.
  2. Arbitrators.
    Small modification to the AO Arbitrator state machine needed to manage properly the close loop command both in automatic and intervention mode.
  3. AdSec tests.
    According to the results of tests the actuator failure recovery procedure was modified and tested. The new actuator failure detection procedure in the firmware was successfully tested. The proper management of coil current saturation was verified. The failed actuator removal procedure was upgraded. Some bugs discovered and fixed.

Day 2

  1. Diagnostic process. Still working on some internal details.
  2. Arbitrators.
    We started testing the "intervention" mode procedure with the light (unfortunately the light arrived somewhat late and we had only about 45 minutes before EOB). We could fix some bugs remaining in the AO arbitrator and we arrived close to closing the loop.
  3. AdSec.
    We tested the failed actuator removal procedure in close loop (with zero gain). The procedure seem OK and we only need to test it with real light and some gain to verify if there are dynamic effects preventing the operation.

Day 3

  1. Diagnostic process.
    The Mirror Diagnostic process was debugged and tested. It now properly cooperates in the failing actuator management procedure. See note below for a brief description
  2. AdSec.
    We tested the AO loop with the ARGOS source light. We were able to close the loop with up to 400 modes. In the meanwhile we tested the detection and removal of actuator when the AO loop is closed. We succeded in remove the actuator whithout harming the AO loop. Some more tests will be needed to assess the robustness of the process. We added a small panel for actuator failure notification and removal.

Note: The management of actuator failures are now managed by the AdSec firmware from the point of view of mirror safety. The FLAO Supervisor has the task of notifying and keep track of failed actuator so that a few failures do not make the AdSec unusable.

The failing actuator detection and removal procedure is a as follows:
  1. The FastDiagnostic process detects a failed actuator based on larger errors on position or current. I many cases this will not cause the intervention of the emergency procedure in the AdSec firmware.
  2. When failure is detected on an actuator the AdSec arbitrator is notified.
  3. The AdSec arbitrator propagates the notification to other interested clients (it will ultimately arrive to AOS, too).
  4. A panel is shown in the AdSec engineering GUI from which a "remove" command can be sent.
  5. After a remove command the AdSec arbitrator modifies the error thresholds of the arbitrator so that it is no longer detected as failed and resets the "actuator failed" flag in the FastDiagnostic.

Day 4

  1. AdSec.
    Solved a problem which prevented the flat shape to be reloaded after an actuator failure. The shell setting procedure has been revised.
  2. Ao Arbitrator.
    We repeated the "manual intervention" procedure exercising many cases in different conditions. The process is slow because at each repetition we must wait the time needed for the real devices to settle. We could test various combinations of sub steps exercising several use cases and we are now able to close the loop repeatably with the artificial light source. A few bugs were discovered and removed in the process. We could also test some error conditions (e.g.: skip-frame events) and verify that they are properly managed and notified.

Day 5

  1. AdSec.
    We have tested more thoroughly the actuator removal procedure at loop closed. We could make it more reliable with a slight modification of the sequence of commands.
  2. AoArbitrator.
    We completed the tests of the "intervention mode" commands exercising them in various conditions. We repeated the generation of errors to verify the proper management in all the states and respect to all the subcommands.

Topic revision: r10 - 15 Apr 2016, LucaFini
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback