GALFA System Code Revisions
v0.5.0 (October 12, 2004):
- Wideband transform is now 512-points. Both 8k- and 512-point PFBs are
still 4x overlap.
- Narrowband now reports 7679 points instead of previous 7935 points. Bin
width is still 871.9Hz, bandwidth is now 6.696MHz reduced from 6.919MHz.
- Shape of the low-pass filter slightly changed. The window function was
offset by half a sample.
- Math in the LPF changed to round coefficient multiplies correctly.
Incorrect rounding was creating a small DC component in the middle of the
narrowband transform. The spur is now gone.
- Added a PFB bypass to both transforms. If the host sets a bit, the
hardware will bypass the frontend filter on the PFB and the transform
becomes an FFT. This is in the gdiag user
interface with the "(" and "
)" keys for PFB and FFT, respectively.
- GALFA now records ADC overflows as an error condition. This is stored in
2-bits of the G_ERR column of the FITS file like the other error
conditions
- GALFA-generated FITS files (GXFITS) now have 512 points for wideband,
7679 points for narrowband.
- GXFITS now reports only one DAC value for each beam/polarity instead of
all 14 values every time.
- New optimization for setting DAC levels, 5x faster than previous method
- Time-domain diagnostics can now trigger on ADC overlap to better
capture RFI-related overflows (-ovftrig
option to gdiag).
- Tweaky new features and bug fixes in graphical diagnostic user
interface. See gdiag --help.
- Socket interface for controlling and getting data from GALFA through a
TCP socket connection.