Summary of Switching Signal Tests
Initial Tests with New Manager
Justin and Ryan tried testing an updated version of the manager with additions to the Backends class to support switching in pulsar modes. We were unable to switch the noise diode using VEGAS. At least one problem is that VEGAS is not/cannot respect the requested switching parameters. We requested a switching period of 0.04 s. The actual switching period as determined by VEGAS was 0.04097536 s. These correspond to 62500 and 64024 "granules", where a granule is 64 * fpga_clock. We believe VEGAS is rounding something to be an integer number of HWEXPOSR or SPECTICK, or is failing because the requested switching period is less than the minimum HWEXPOSR allowed in spectral line modes, but we are still investigating the details of this.
We also tried taking a pulsar cal-mode scan with the cal frequency changed to 1/0.04097536 s = 24.40491065850306 but still could not see the cal. So it appears that VEGAS was still not properly switching even at this switching period. However, this could be because of a switching period that fell outside the typically allowed range for spectral line modes.