Summary comments and recommendations made to the NAIC commissioning team on 8 Sep 2004 regarding experience to date with A1946, Hi Jeff, First of all, let me mention that there was good progress made during the last 3 weeks with A1946, and that ALFA is not far from being ready to rock'n roll as far as our group is concerned. Resulting from the A1946 observations carried out so far, I am enclosing a list of recommendations for improvements to the ALFA/WAPPS/CIMA system, that our group would like to see addressed. The recommendations listed below fall into 3 main groups: FITS HEADERS, TIMING ISSUES and CIMA OBSERVING MODE. Several of the recommendations made, you are already familiar with, for we have discussed them in Arecibo over the last few weeks. We add a few that may have more general interest than that of our group, for your consideration. The points with the higher priority, for us, are flagged by an asterisk(*). Please do not hesitate to contact us, if any clarification were needed. Thanks for your help. Cordially, -riccardo for the A1946 crowd __________________________________________________________________________ FITS HEADERS We have recommendations for FITS headers that are immediately relevant to our work, and others of a more general nature. We list the former first, with emphasis on priority for those, from our perspective. It may be necessary to expand the FITS hdr column structure, to include parameters not currently considered. We recommend that if such expansion were to take place, it would do so all at once, so that adaptation of data processing and monitoring programs does not need to be done many times, and there is no proliferation of "mark stages". Additionally, there are a number of columns whose current definition/implementation is unclear to us. We suggest that some concerted dialogue between NAIC staff and the observing teams would quickly lead everyone to a solution of mutual satisfaction. Specific: 1.* Individual RAJ,DECJ positions for each beam in the FITS headers. These should be actual, on-sky positions in J2000. 2.* Doppler correction (heliocentric to topo, as computed and posted in telescope position panel in CR), projected along the line of sight to the source, needs to be added to the header. This should always be recorded even if the topocentric frame with vel=0 is the one adopted. 3. The hdr should contain an identification of the pattern used to obtain the data (e.g. "fixedazdrift", "basketweave", etc.), as well as parameters specific to that pattern (for "fixedazdrift", one would like "requested AZ", "secs", "calmode", "caltype", "async secs", using the names chosen in the CIMA pattern). In order for such an implementation to satisfy all patterns, perhaps one could allocate enough integer/doubles/etc. parameter slots, that would get filled or partially filled by the datataking program, according to each pattern used. 4. RESTFREQ (col. 14) should be the rest frequency of a spectral line not the center frequency of the band. The "CENTER FREQUENCY" of the band is listed as CRVAL1 in Col. 3. It's up to the observer to define RESTFREQ. In our case, for example, it is 1420.4058 MHz. General: 5. CRVAL2 and CRVAL3 are requested positions. Currently they are RA and Dec. Some observers will wish to specify coords in a different system, such as galactic l and b. Note that a parameter specifying the chosen coordinate system should be added too. 6. CDELT1V (col 16):delta vel per channel. This quantity varies across the bandpass; how is it defined? at the rest freq? at Center freq? It is unclear how this can be applied to broad bandwidth setups. (it doesn't bother us, we just wonder what's its use...) 7. It would be very convenient if times could be given more consistently in the same units. Currently, units are a mix of (1) hours since midnight, (2) seconds since midnight, (3) modified Julian date, (4) fractional day of the year. TIMING ISSUES 8.* As reported in previous emails and verbally, mismatches exist between spectra and time stamps: - differrent pols of the same beam occasionally have different time stamps - cal pulses which should be found in the same record for all beams are found in different records in different beams. This needs to be fixed. 9.* The "asynchronous" nature of the cal is highly inconvenient, and produces - in combination with the timing errors mentioned above - erratic calibration. A synchronous cal which fires exactly on the same tic as the start of a data record is highly desirbale. CIMA OBSERVING MODES 10.* The "fixed azimuth drift" mode tracks strips of sky at constant *current* Declination. Since the duration of survey efforts is foreseen to stretch over years, tracks of constant *current* Declination will not line up, leading to misalignments of arcminutes over hours-long drifts, over a few years' baseline. It would be desirable for the "fixed azimuth drift" mode to track a constant *epoch* Declination, e.g. Dec2000, to obviate the problem described. 11. Any minimization of the time lost to data taking by file closure/opening, cal firing, etc. would be beneficial.