ao fits file for pdev spectrometer (cimafits)

last mod: 27nov09

fits file organization.
Fit file binary table
Interpolating positions.
Outstanding issues:
History:
Other pdev Info

Terminology

Pols: The number of polarizations used. It can be 1, 2:polA, polB, or 4: PolA,polB,U,V (stokes).
band: Each spectrometer box can compute two 170 Mhz frequency bands. Each band can have two spectra (polA,polB) or 4 spectra A,B,U,V if stokes.
box: We have 14 spectrometer boxes. Each box can compute two 170 Mhz bands (high and low frequency bands). See specrometer specs for more info.
dump: When taking spectra this is the integration of a single spectra (or group of 4 spectra if stokes).
field: A fits row contains multiple fields. Each field contains information for the time of this row. Some fields are: datapoint, statuspointer, raPos,decPos, etc..
row: Data is written to the fits file 1 row at a time. A row has about 137 fields. One of these fields is a pointer to the spectral data. A second field is a pointer to the status data. When taking data at higher rates each row will contain multiple dumps of spectral/status data. The other fields in the row usually contain values for the first dump of the row.

Fits file organization: (top)

We will have two spectrometers each with 7 spectrometer boxes. This will handle the primary observer and the piggy back experiment.
Each pdev spectrometer box will handle 1 alfa beam, 300 Mhz all pols.

It will output two fits files. 1 file for the upper 170 Mhz band (s1) and 1 file for the lower 170 Mhz band (s0).
Each file will hold all the pols for that frequency band.
When outputing data, the spectra for a single integration (dump) are output (to the data heap) together in the order:

polA,polB,stokesU,stokesV.

The files are named:
nnnn.yyyymmdd.bXsYgZ.sssss.fits etc. p1231.20080311.b0s1g0.00100.fits

nnnn..nn is an identifier for the experiment. it is usually the projectid of the observer.
yyyymmdd . is the date when the file was opened AST.
bX: for alfa this will be b0,b1,b2...b6 corresponding to the alfa beam.
sY: s0 for the 1st frequency sub band of a beam (lower band at IF) , s1 is for the for the 2nd (higher frequency at IF).
gZ:group. g0 is data from the first pdev spectometer (1st 7 boxes), g1 is data from the 2nd 7 boxes.
sssss is a 5 digit sequence number the increments during the day. Each scan starts on a multiple of 100. If multiple files are taken during a scan then then increment by 1. eg startscan: .00100.fits, .00101.fits, .00102.fit, startscan:.00200.fits, .00201.fits.
all files end with .fits.

Primary header

Simple
BITPIX
...
NAXIS
...
END

Not much info, mainly NAXIS, and EXTEND=T --> binary extensio

binary extension header

XTENSION='BINTABLE'
Mandatory:BITPIX,NAXIS ,NAXIS1,NAXIS2,PCOUNT,GCOUNT,TFIELDS
EXTNAME,VER_DATE,THEAP,PCOUNTS
CTYPEn,CUNITn,CRPIXn,CDELTn for the various axis
define TTYPEn,TFORMn,TUNITn,TDISPn for each field of record (about 137 fields).
CALCTL: 0-cal always off, 1-cal always on, 2:winking cal.
WCALON: if winking cal then the number of accumulations for the calon.
WCALOFF: if winking cal then the number of accumulations for the caloff.
WCALPHASE: 0. to 1. . The location for the cal transition in the last accumulation.
PHxxxxxx. The 32 word standard pdev header. Includes :PHMAINID,PHSP1ID,PHADCF,...
PHxxxxxxx. the pdev sp1 header. Contains config info for the spectrometer such as: PHFMTWID,PHFMTTYPE,PHFFTLEN,...

binary table

1 binary row contains about 137 fields. A row can contain contain spectra from 1 or more dumps. The number of dumps in a row will be determined by: output a row once a second or every 20Mb of data (whichever comes first). A row looks like:

dataHeapPtr

StatusHeapPtr

tdim1

tdim2

other 133 fields

dataHeapPtr: points to the spectral data for the dumps of this row.
statusHeapPtr: points to the status info for each spectral dump of this row.
ptrs are 2 4 bytes integers holding: array_len, bytes_offset_start_ofHeap.
tdim1 defines the dimensions of the spectral data as an ascii comma separated string:

nSpcChannels,1 setPerRa,1 setPerDec,nPolsPerDump,NumDumpsInRow

tdim2 defines the dimensions of the pdev status data as an ascii string.

8,1setPerRa,1setPerDec,1setPerAllPols,NumIntegrationsOutput

Heap

The heap is partitioned into two parts. First the status heap that holds the pdev status info for each integrated spectra (10*2 bytes). This area is preallocated to hold all the pdev status info you could have for the maximum allowed size of the binary table.
The pdev status heap is followed by the data heap that holds the spectra. There will be some unused space between the binary table and the heap, and between the status heap and the data heap.

Status heap.

Each entry will hold all of the status info for one binary table row
For a single integration there are int*2 stat[10] data values independant of the number of pols output).
stat[0]:seqNum:. Increments each dump. wraps at 64K
stat[1]:fftAccum: number of ffts accumulated for this dump. it will be different than wall time if blanking or 32bits stokes (which has 1 extra fft time with no accumulation).
stat[2]:calOn: will be 1 if the cal input bit was high for any fft during the dump.
Overflow counts: The following fields count various overflows. They can have values 0..13. The number of errors is 2^stat[n].
stat[3]:adcOverFlow: number of ADC overflows. These are not so bad since they clip.
stat[4]:pfbOverFlow: number of polyphase filterbank overflows. These are overflows in the butterfly stages of the fft. This overflow will mess up the spectra since the value wrap around rather than clip.
stat[5]:satCntVShift: Optional upshift after butterfly stages, before power computation.
stat[6]:satCntAccS2S3: overflows accumulating power for U,V into the 40 bit accumulator.
stat[7]:satCntAccS0S1: overflows accumulating power for polA,B into the 40 bit accumulator.

stat[8]:satCntAshftS2S3: overflows upshifting U,V in the 40 bit accumulator for output packing.

stat[9]:satCntAshftS0S1: overflows upshiftig A,B in the 40 bit accumulator for output packing.
For a description of the pdev processing pipeline see (pdev processing stages.)

Data heap

Each entry will hold all of the spectra for one binary table row.
For a single integration the spectra are output as:

polA,polB,U,V

Example fits file layout:

Suppose we have a fits file with 8192 channels/spectra, polA and polB recorded at .1 second dump time. Assume there are 10 time samples per fits row and lets take 5 seconds of data... The fits file would have:

Primary header
bintable xtension.. extname: 'CIMAFITS'
BINARY TABLE (5 ROWS) (1 second/row and 5 seconds of data)

each row 1 seconds worth of data.
DATA keyword points to data heap with the (8192chan)*2(pols) * 10(tmSamples/row) of (int32 or float) spectra
STAT keyword points to the stat heap with the (10 statAr) * 10tmSamples/rows) of shorts.

Heap

STAT HEAP

(10 StatShorts * 10 tmSamples/row) * 5Rows

space
Data Heap

(8291Chan)*(2Pols)*(10 tmSamples/row) * 5 rows

02jun08: fits table (top)

Here is an updated fits table definition for the ao fits format:
Teminology

Actual. What was measured.
Req: What was requested by the user.
Enc: Encoder value. This is what is read back (or interpolated from a readback).
Band: 1 spectra.
Bin: 1 channel of a spectra.
TimeAxis: We can have multiple dumps in 1 row. These will be pixels along the time axis.Pixel 1 is the first dump of the row.
STAT:

WHERE LOADED in the software bpOut.c:

SCI - loaded in scanInit,
LHS - loaded in ldHdrScan;
LHR - loaded in ldHdrRec
LHRILP: interpLdPos called from ldHdrRec

IMPLEMENTED

D: done/implemented
P: pending implementation.. italics in description may be questions about how to implement.

Times/positions refer to start of spectral integration.

all positions,times are computed from crval5 (gmt secs from midnite)
crval5 =startScan time + nrows*RowTime (nrows counts from 0).
StartScanTime - taken from pdev header start scan time. Loaded when scan started (second from 1970).
The Positions, times are for the start of the time interval not the center.

NAME	TYPE	Units	stat	DESCRIPTION
data	desc		D:LHR	points to the spectra data for this row
stat	desc		D:LHR	points to the spectra status info for this row
tdim1	char[32]		D:SCI	dimensions of spectra: cr1,cr2,cr3,cr4,cr5: chan,ra,dec,pols,numDumps:"8192,1,1,4,10" 8k channels, full stokes, and 10 dumps
tdim2	char[32]		D:SCI	dimensions of status: cr1,cr2,cr3,cr4,cr5: statWds,ra,dec,pols,numdumps:"10,1,1,1,10": 10 statwds/spec. Only 1 set per N pols.
crval1	double	Hz	D:LHR	Center Freq of band (topocentric) (see crpix1 for which pixel it is) From topoa/bfreqhz keyword of pnetctl
cdelt1	double	Hz	D:LHS	Frequency bin interval. Negative number --> increasing pixels give decreasing frequency. Computed as: adcClkHz/fftLen/(decimation)
crpix1	double		D:LHS	Pixel of center frequency crval1. count from 1. Should be the center of the pixel. Computed as: fftlen/2 + 1 Need to check that this computation is correct
crval2	double	deg	D:LHRILP	Actual J2000 Ra this beam on sky. computed as: Readback encoder az,za and interpolate to center of spectra (if multi spectra/row use first spectra) Subtract off model corrections for the encoder position add any alfa beam offsets. Convert az,za to ra,decJ2000
crval3	double	deg	D:LHRILP	Actual J2000 Dec this beam on sky (see above for computation).
crval4	double		D:LHS	Polarization axis. Fits coding is:(1,2,3,4: I,Q,U,V), (-1,-2,-3,-4: LL,RR,RL,LR circulars), (-5,-6,-7,-8:XX,YY,XY,YX linears). We dump all spectra from an integration in a single record (pols are not in separate records). So code crval 4 with multiple digits, one for each spectra. The order should be the order the spectra appear in the record. The values will be: 1: polA+polB summed (stokes I) -56:polA,polB linears. 1234: stokes. The actual data is: let x=polAVoltage , y=polVoltage: 2xx polA 2yy polB 2Re(xy) stokes U (assuming linears) 2Im(xy) stokes V (assuming linears) Pdev can not dump a single polarization in spectral mode. Notes: We call all signals linear...sbn, xband, and lbw with the hybrid are actually circular. Even for the linears what you Really get is what comes out the polA (left) or polB (right) spigots in the control room. Some receivers may not be lined up very well with xx, or yy. For a linear feed the difference of the spectra is Q, For a circular feed the difference of the spectra is V. We don't try to correct for this, You just get the sum, difference, etc... keyword num_pols will also have the number of pols output 1,2,4
crval5	double	sec	P:LHS	Time for start of first pixel in this row recorded as seconds from midnite of date_obs UTC. Note: We can have multiple dumps per row. So this will refer to the first pixel on the time axis. dat_obs gets updated every row. If the scan crosses midnite (UTC) then crval5 will jump by 86400 seconds. This should be the center of the 1st pixel rather than the edge.
cdelt5	double	sec	D:LHS	Spacing between pixels on the time axis (dumps) if multiple dumps per row. This will be wall time (the same value is stored in duration).
azimuth	double	deg	D:LHRILP	Actual azimuth for this beam on the sky (pixel 1 on the time axis). The model corrections have been removed so back computing from az,za should give ra,Dec. Note:Looks like this is the feed azimuth. For the dome, the source azimuth is 180 degrees from this.
elevatio	double	deg	D:LHRILP	Actual elevation for this beam on the sky. (see notefor azimuth).
glon	double	deg	D :LHR	Actual galatic longitude for this beam on the sky
glat	double	deg	D:LHR	Actual galactic latitude for this beam on the sky.
date-obs	char[16]		D:LHR	Date (UTC) for the start of the first pixel of this row. Format: "YYYY-MM-DD". The fits definition says it assumes this is the start of an observation. If not it must be defined in the comment fiield. We are not using it this way. Probably should also be changed to the center of the pixel.
mjd-obs	double	day	D:LHR	Modified Julian Date for spectra (pixel 1 on the time axis). It is update every row. Probably should be switched to center of pixel.
lst	double	sec	D:LHRILP	Local mean sidereal time at start of 1st pixel on the time axis. Computed from mjd. Should be changed to center of pixel.
exposure	double	sec	D:LHS	Integration time for 1st spec on the time axis (if blanking is not enabled). Notes: We can have multiple dumps per row. If blanking was enabled then the exposure time of each dump could be different. So make exposure the time if no blanking was done. Users can check the number of ffts accumulated in each dump for the exact exposure time of each dump. If stokes 32 bits this does not include the 1fft time needed between dumps where no integration is done.
duration	double	sec	D:LHS	Duration for a dump (1st spectra on the time axis). This is the wall time. It will be equal to exposure except when stokes 32 bit mode is used. In this case, the pdev spectrometer does not integrate for 1 fft time. In this case duration will be longer then exposure.
tsys	double	K	P:LHS	Last computed Tsys. Set to 1.0 if noT available.
bandwd	double	Hz	D:LHS	overall bandwidth of spectra. Computed as abs(cdelt1) * number of channels
restfreq	double	Hz	D:LHS	Rest frequency used for doppler correction. Normally it would be the rest frequency of the line of interest. Computed from pnetctl keyword resta/bfreqhz. I should be for the reference pixel.
specsys	char[8]		D:LHS	Velocity frame for crval1. Always "TOPOCENT" for ao data.
				Req_ are values that where requested by the user. They may not be what happenned: eg: req az=100,za=10 but telescope only got to 99.89 and 9.93
req_sys	char[8]		D:LHS	Requested velocity coordinate system. values are: "TOPOCENT","GEOCENTR","BARYCENT","LSRK","UNKNOWN" from pnetctl req_sys keyword
req_vel	double		D:LHS	Requested velocity in req_sys coordinate system. Doppler correcting the restfreq with req_vel should give crval1. From pnetctl req_vel keyword (from cima)
req_vel_unit	char[8]		D:LHS	unit of req_vel. "m/s" or "Z" From pnetctl req_vel_unit keyword (from cima).
req_vel_type	char[8]		D:LHS	values are:"VOPT-F2W","ZOPT-F2W","ZRAD-F2V", or "UNKNOWN" From pnetctl req_vel_type keyword (cima sends).
req_raj	double	deg	D:LHRILP	Requested ra position J2000. Interpolated to the start of the first spectra of the row. If alfa then this is the req_raj for the beam that is used to point the telescope (normaly beam 0). This includes any rates and offsets that had been including in the pointing reqeust. Probably needs to be moved to the center of the spectra. (just update mjd).
req_decj	double	deg	D:LHRILP	Requested dec position J2000. Interpolated to the start of the first spectra of the row. If alfa then this is the req_raj for the beam that is used to point the telescope (normaly beam 0). This includes any rates and offsets that had been included in the pointing reqeust.
	req_xxxc1,c2 These are the indivdual components from the requested tracking command: pnt tr posc1 posc2 -cpos -o offc1 offc2 -coff -r ratec1 ratec2 -crate There is a position , an offset from that position, and a rate. The offset and rate are used when scanning (or doing off source). The 3 of these get combined and then converted to req_raj,decj. These values have not been interpolated to crval5. They normally don't change during a scan (except for maybe a starttime). The offsets, rates can have a starttime (which has not been included here). The starttime is normally the start of the scan (if a timed start was used). req_cs Requested coordinate system coding: "GALACTIC","B1950 ","J2000 ", "BECLIPTI", "JECLIPTI", "CURRENT " (of date), "HA_DEC ", "AZZA_SRC", "AZZA_FEE" (feed azimuth) ,"AZZA_NOM" (no model), "AZZA_GC " (great Circle azimuth).
req_posc1	double	deg	D:LHR	Requested postion (coordinate 1). Usually ra.
req_posc2	double	deg	D:LHR	Requested position (coordinate 2) Usually dec.
req_cspos	char[8]		D:LHR	Coordinate system for position request. (see above for coding)
req_offc1	double	deg	D:LHR	Requested offset (coordinate 1). Usually ra, hour angle or azimuth. The offset are normally fixed for the entire scan so you can use it to see how big the strip was supposed to be (the rate causes the resulant position it to move).
req_offc2	double	deg	D:LHR	Requested offset (coordinate 2). Usually dec, or za.
req_csoff	char[8]		D:LHR	Coordinate system for offsets (see above for coding)
req_pnttime	double	sec	D:LHR	The time stamp in seconds from midnite(UTC) for the pointing info that has not been interpolated to crval5. This include req_posxx,req_offxx,req_ratexx.
req_ratec1	double	deg/sec	D:LHR	Requested rate for first coordinate (usually ra,ha,az) in deg/sec. (per solar second)
req_ratec2	double	deg/sec	D:LHR	Requested rate for 2nd coordinate (usually dec,az) in deg/sec. (per solar second)
req_csrate	char[8]		D:LHR	coordinate system for rates (see above for coding).
req_equinox	double		D:LHR	Equinox for req_csposxx. 1950, 2000. or 0 if current.
rate_dur	double	sec	D:LHR	How long the rate has been applied (at req_pnttime). Units are solar seconds.

enc_time	double	sec	D:LHR	seconds from midnite (utc) when the enc_xxx values were read. Will match crval5.
enc_azimuth	double	deg	D:LHR	Azimuth encoder position (dome side) read from the encoders. The value are interpolated to enc_time (which is the same as crval5). They are the raw values with no model correction applied. For alfa: since there is only 1 azimuth encoder, this is for the alfa beam used for pointing.
enc_elevatio	double	deg	D:LHR	Elevation encoder position read from the enocders. The values are interpolated to enc_time (which is the same as crval5). These are the raw values with no model correction applied. For alfa: since there is only 1 elevatin encoder, this is for the alfa beam used for pointing. Master determines if this is the dome or the carriage house.
enc_altel	double	deg	D:LHR	Alternate elevation encoder position read from the enocders. The values are interpolated to enc_time (which is the same as crval5). These are the raw values with no model correction applied.This is the carriage house (dome) that is not being used to point the telescope.
cur_err	double	deg	D:LHR	sqrt(az_err*sin(za) ^2 + el_err^2) .. This will probably not work very well if we are not in tracking mode. (wapps actually used the agc values that are computed..). May want to switch to used agcerr when in tracking mode (but will need interpolation).
allowed_err	double	deg	D:LHS	Great circle error needed to be "tracking". Typically .00278= 10".
az_err	double	deg	D:LHR	Azimuth position error as: (azEncPos - requestedAzEncPos) interpolated to the start of the first spectra of row. For Alfa it is the error for the beam used for pointing.
el_err	double	deg	D:LHR	Elevation position error as: (elEncPos - requestedElEncPos) interpolated to the start of the first spectra of row. For Alfa it is the error for the beam used for pointing.
model_offaz	double	deg	D:LHR	Pointing model azimuth offset for the actual azimuth position interpolated to the start of the first spectra of the row. For alfa it is the model correction for the beam used for pointing. The value is computed as: ActualAzEncoderVal=ActualAzEncoderNoModel + model_offaz The model_offsets are evaluated at the az,za encoder values from 1 second earlier. This could possibly introduce errors of up to a few asecs.. (see model gradients)
model_offza	double	deg	D:LHR	Pointing model zenith angle offset for the actual zenith angle position interpolated to the start of the first spectra of the row. For alfa it is the model correction of the beam used for pointing. Computed as: ActualZaEncoderVal=ActualZaEncoderNoModel + model_offza See comment for model_offaz
user_offaz	double	deg	P:LHR
user_offza	double	deg	P:LHR
	beam_offaz/za,rfeed_offaz/za, prfeed_offaz/za, beam_offra/dec, alfa_ang are used for alfa.
beam_offaz	double	deg	D:LHS	Azimuth offset from alfa center beam to this beam (great circle) when alfa rotation angle is 0. The value is measured on the sky.
beam_offza	double	deg	D:LHS	Zenith angle offset from alfa center beam to this beam when alfa rotation angle is 0. The value is measured on the sky.
rfeed_offaz	double	deg	D:LHRILP	Rotated feed offset. Azimuth offset from alfa center beam to this beam (great circle) using current alfa rotation angle. The value is measured on the sky. Adding this value to the az,za of the center beam and then back computing from az,za to ra,dec should give you the ra/dec for this beam (except for prfeed_offaz,za). Check that the signs of the offsets are correct 30nov09:finally checked the signs, and they were wrong (moreinfo)
rfeed_offza	double	deg	D:LHRILP	Rotated feed offset. Zenith angle offset from alfa center beam to this beam (great circle) using current alfa rotation angle. The value is measured on the sky. Adding this value to the az,za of the center beam and then back computing from az,za to ra,dec should give you the ra/dec for this beam (except for prfeed_offaz,za). Check that the signs of the offsets are correct 30nov09:finally checked the signs, and they were wrong (moreinfo)
prfeed_offaz	double	deg	D:LHS	The azimuth offset from the beam used for pointing the telescope to the central beam. Only used when doing point source calibrations using a non central beam for pointing. In this case the offset will be -beam_offaz of that beam. Check that the signs of the offsets are correct
prfeed_offza	double	deg	D:LHS	The zenith angle offset from the beam used for pointing the telescope to the central beam. Only used when doing point source calibrations using a non central beam for pointing. In this case the offset will be -beam_offza of that beam. Check that the signs of the offsets are correct
beam_offraj	double	deg	D:LHR	The RA offset to go from the central beam RA to this beams RA. Only used for alfa. Computed as: crval2 - raJCenterPixelActual
beam_offdecj	double	deg	D:LHR	The Dec offset to go from the central beams Dec to this beams Dec. Only used for alfa. Computed as: crval3 - decJCenterPixel
	map_offra,dec,az,za: When a scan in driven (mapping, crosses, etc), you can use map_xxx to find the distance from the current interpolated measured position to the center of the map or strip.
map_offra	double	deg	D:LHR	Distance from current ra position to the center of the map : (crval2 - raMapCen)cos(dec). Mapping routines use the cmd: pnt tr raMapCen decMapCen -o azOff zaOff -r azRate zaRate where raMapCen, decMapCen is the center of the map. Note that the distance is great circle ( cos(dec))
map_offdec	double	deg	D:LHR	Distance from the current dec position to the center of the map: (crval3 - decMapCen). Mapping routines use the cmd: pnt tr raMapCen decMapCen -o azOff zaOff -r azRate zaRate where raMapCen, decMapCen is the center of the map.
map_offaz	double	deg	D:LHR	Azimuth offset of current position to the center of the map (great circle) computed as (azimuth - azMapCenter) * sin(za)
map_offza	double	deg	D:LHR	Za offset of current position to the center of the map (great circle) computed as ((90-elevatio) - zaMapCenter).

alfa_ang	double	deg	D:LHR	The current alfa rotation angle. Positive values are clockwise sitting on top of the alfa rotator feed assembly. The value is not interpolated so if alfa is rotating it may be a little off. Values: -180 to 180.
para_ang
vel_bary	double	m/sec	D:LHRILP	barycenter velocity of the observer projected to the requested raj,decj direction. It is interpolated from the pnt scram heliocentric velocity. These values are computed from the jpl ephemerides once a second. If there is a large pointing error than this would differ from the velocity in the actual direction.
vel_geo	double	m/sec	D:LHRILP	geocentric velocity of the observer projected to the requested raj,decj directions. It is interpolated from the pnt scram geocentric velocity. These values are computed from the jpl ephemerides once a second. If there is a large pointing error than this would differ from the velocity in the actual direction.
vel_obs
backend	char[8]		D:LHS	backend name "pdev"
backendmode	char[24]		D:LHS	mode of observation. value is CHANNELS TYPE, N bits. the value can be: CHANNELS:1-chan, 2-chan, summed, stokes.. 1-chan is polA or polB 2-chan is polA and polB included summed is polA + polB added together stokes: the four stokes parametes. TYPE: spectra,tm-dm note that tm-dm is not yet supported in fits BITS: 8,16,32 .. this is the number of bits on output eg: "2-chan spectra, 32 bits"
caltype	char[8]		D:LHS	type of cal used. Let D1 be caldiode 1 and D2 be cal diode 2. The the values are: "lcorcal": low correlated cal. D1 -> polA and polB "hcorcal":high correlated cal. D1 -> polA and polB "lcal": low uncorrelated cal. D1-> polA, D2->polB "hcal":high uncorrelated cal. D1->polA, D2-> polB "lxcal": low uncorrelated cal: D1->polB, D2->polA "hxcal":high uncorrelated cal. D1->polB, D2->polA "l90cal:low correlated shifted cal. D2->polB,D2+90deg->polA "h90cal:high correlated shifted cal. D1->polB,D2+90deg->polA
obsmode	char[8]		D:LHS	Observing pattern used. Some names are: "ON","ONOFF","DPS","RADECMAP","DECRAMAP","BASKET","FIXEDAZ"
scantype	char[8]		D:LHS	Each observing pattern is made up of one or more scans. scantype names the type of scan within a pattern. Some names are: "ON","OFF","UP","DOWN". Patterns that have a single scantype are usually labeled with "ON".
pattern_id	int*4		D:LHS	unique number identifying pattern. Same format as scanid: Ydddnnnnn where Y is the last digit of the year, ddd is the daynumber of the year (count from 1) and nnnnn increments from 0 each day for each new pattern. eg. an onoff pattern would be 1 pattern and have 4 scans (poson,posOff,calOn,calOff).
scan_id	int*4		D:LHS	unique number identifiying scan. format: Ydddnnnnn where Y is the last digit of the year, ddd is the daynumber of the year (count from 1) and nnnnn increments from 0 each day for each new scan.
recnum	int*4		D:LHR	The dump number withing the scan for first spectra in row. It counts from 1. Each row will increment by the spectra per row. The count is for the entire scan. If there are multiple files for scan, the recnum does not reset at the beginning of the nth file of the scan (n>1).
total_recs				This is supposed to be the total number of records in the scan. It is currently left as 0 since when the data is written , we do not know the final count. It could be switched to the requested number of records/rows per scan.
chan_in	int*4		D:LHS	FFT length used for spectra.
bad_data
plat_powerA	double	dbm	D:LHS	total Power in PolA read by power meter before fiber optic transmitter. Not used by alfa. Value should gt -40dbm across 200 Mhz. This value is only read whenthe power levels are adjusted so it may not be current for this scan.
plat_powerB	double	dbm	D:LHS	total Power in PolA read by power meter before fiber optic transmitter. Not used by alfa. Value should gt -40dbm across 200 Mhz.This value is only read whenthe power levels are adjusted so it may not be current for this scan.
cntrl_powerA	double	dbm	D:LHS	total Power in PolA IF read by power meter after the fiber optic receiver downstairs. Not used by alfa. This value is only read when the power levels are adjusted so it may not be current for this scan.
cntrl_powerB	double	dbm	D:LHS	total Power in PolB read by power meter after the fiber optic receiver downstairs. Not used by alfa. This value is only read when the power levels are adjusted so it may not be current for this scan.
syn1	double	Hz	D:LHS	Synthesizer for first upstaris 1st LO. Values: Usually rfi + IF frequency if high side lo.
syn2	double	Hz	D:LHS	Synthesizer for downstairs mixing. Values: Alfa: this will be the 2nd LO used to complex mix the 250 MHz IF1 to baseband. The same information is stored in bintable header locations PHLO2MX0 and PHLO2MX1. syn2 will have the value that corresponds to this subband. It will normally be 175e6 or 325e6 Single pixel: The LO used for the mixing from IF1 (usually 1-2 Ghz) to IF2 (250 MHz). This can be a high or low side lo. The 2nd IF is then complex mixed to baseband with PHLO2MX0 or PHLO2MX1 (in the bintable header).
tot_power	double			Was the 0 lag from auto correlators. Pdev doesn't use this.
tcal_numCoef
tcal_coefA
tcal_coefB
backendmask	uchar[8]		D:LHS	Array of bitmaps specifying which pdev boxes, subbands, and pols were used for the experiment (at least for this group). It can be used to find all of the files that were used for an observation or preallocate arrays. array of 8 bytes: 1 for each pdev box used. 0 is beam0,mixer 0, 1 is beam1,mixer1...etc 8 bit val. A single byte where the lower nibble (4bits) si the lower subband and the upper nibble is the higher subband. 4 bit nibble: specifies which pols are used: 1 - pol A only or polA and polB summed. 2 - pol B only 3 - polA and polB separate 15- stokes. Values: Alfa: backendmask[0..6] have 00110011 if polA,B, 11111111 if stokes,and 00010001 if summed pols. Use beam to index into backend mask and ifn to decide which subband. Single pixel: depends on which pols and subbands were selected. Use ifn/2 to index into backendmask and ifn%2 to get the subband.
num_beams	char		D:LHS	Number of beams used for the experiment. Values: Alfa: 7 Single pixel=1.
num_ifs	uchar		D:LHS	The number of subbands we have on this beam. Values: Alfa=2 . the low (170 Mhz) subband and the high (170 Mhz) subband Single Pixel. The total number of subands selected on this beam. We can have a maximum of 7 boxes with 2 subbands per box so the maximum would be 14 subbands.
num_pols	uchar		D:LHS	number of pols. This will be the 4th dimension of tdim1:(nchan,nra,ndec,npol,ndumps). Values: 1 for polA,polB,or summed polA+polB 2 for polA and polB recorded separately 4 for stokes.
beam	char		D:LHS	Beam number used for this data. Values: Alfa = 0 .. 6 Single pixel=1
ifn			D:LHS	IF subband number for this data. Values: Alfa: 0 for the low band and 1 for the high band. Single Pixel: Index into backendmask for this subband. ind=ifn/2 is the index into backendmask. ifn% 2 =0 is the lower 4 bits of byte (low band). ifn%2=1 is the higher 4 bits of byte (high band)
pol	char			We store multiple pols in 1 record so this variable is not used.
prfeed	char		D:LHS	alfa beam used to point the telescope. Values: 0..6
input_id			D:LHS	The number of the pdev box for this record. It will be the same value N as the bNs0g0 in the filename. Values: ALFA: it will have the same value as beam 0..6 Single Pixel:It will be the pdev box used for this data. The mapping of synth/mixer numbers to beam is: BEAM syn/mix number bm0 . mix 1 bm1 . mix 5 bm2 . mix 2 bm3 . mix 6 bm4 . mix3 bm5 . mix 7 bm6 . mix4
uppersb	char		D:LHS	If true then the spectra are in increasing frequency order (cdelt1 > 0). If false then the spectra are stored in decreasing frequency order (cdelt1<0). This is probably opposite from the definition used for the wapps.. But an upper side band isnot flipped.. so i'll stick with this definition. This is set correctly for alfa. Single pixel with all of the permutations needs some work.
attn_cor	char		:LHS	Currently not used. should put pdev atten here but then need an array I,Q polA,polB maybe code at 2 bits/digitizer: 0 - min 1 - < 16db 2- >=16 db 3 = max attn. Or we could stick the mixer atten settings in the btable header like the rms's.
master	char		D:LHS	which side of the azimuth is controlling the pointing. Values:0 dome, 1 carriage house.
onsource	char		D:LHR	We are tracking the source within the preset tolerance (True when cur_err <= allowed_err). Allowed_err is usually set to 10 arcSeconds.
blanking
lbwhyb	char		D:LHS	lband wide linear to circular conversion with hybrid. Values: 0 no hybrid, linear pols 1 hybrid in, circular pols
shcl	char		D:LHS	Reciever shutter (all but sband narrow). Values: 0 shutter open, 1 shutter closed.
sbshcl	char		D:LHS	sband narrow shutter. Values: 0 shutter open, 1 shutter closed.
rfnum	char		D:LHS	Receiver number for reciever selected on platform (rf/if switch positions in dome). Values: 1 327, 2 430gr,3 610,5 lbw,7 sbw,8 sbh,9 cb,10 cbh,11 xb, 12 sbn,17 alfa
calrcvmux	char		D:LHS	receiver on the platform that receives the ttl signal to fire the cal. Values:1 327,2 430gr,3 610,5 lbw,7 sbw,8 sbh,9 cb,10 cbhi,11 xb,12 sbn,15 alfa (note alfa is not 17 since the control word is only 4 bits).
zmnormal	char		D:LHS	platform rf transfer switch before the first mixer. Values: 0 - rf comes straight down 1 - polA,polB switched before first mixer.
rfattn	char[2]	db	D:LHS	Platform attenuators before the first mixer. [0] is polA, [1] is polB. Not use for alfa. Values: 0..11 .. 0 is no attenuation.
if1sel	char		D:LHS	Platform IF selector. Determines whichRf->If mixer is used. Values: 1 300If, 2 750If, 3 2-12GhzRf 1500If, 4 10GhzUpconverter, 5 straight thru
ifattn	char	db	D:LHS	Platform IF attenuators after the mixer, before the fiber optic xmter. Not used for alfa. Values: 0..11 db.
fiber	char		D:LHS	select fiber/coax to bring signal down from platform. Values: 1 fiber, 0 coax. Note we can on longer switch to coax.
ac2sw	char		D:LHS	bit map for ac power switch to enable certain functions in dome. 2008 coding; B1=2: vlba cal is turned on (1 MHz comb in IF). Note: is it s bit map so 2,3,7 all have B1 vlba cal turned on.
phbsig	char		D:LHS	The adjustable phase delay in the pol A leg of the 10 GHz upconverter. (at 10 Ghz). Values: 0..127 binary numbers map to 0.. 178.5 degrees of delay.
hybrid	char		D:LHS	The 10 GHz upconverer has a hybrid that can be used to convert from linear to circular. Values: 0 hybrd is no used, 1 hybrid is being used.
phblo	char		D:LHS	10 GHz upconverter IF phase shift (delay) in polB LO going to 10Ghz to 1.5 Ghz. Values: 0..127 map to 0 .. 178.5 degrees in the lo (fixed at 11.5 Ghz).
xfnormal	char		D:LHS	control room (downstairs) polA/polB transfer switch after the fiber optic reciever, before the variable gain. Values: 1 straight thru (norma). 0 switch polA and polB.
ampgain	char*2	db	D:LHS	Control room (downstairs) variable gain after the fiber optic receiver (not used for alfa). [0] is pola, [1] is polB (assumes no polA,B switching). Values:-11 to 30.. -11 is 11 db of attenuation. 30 is 30 db of gain. You need to treat this as a signed char. We may stuff -11 into an unsigned char..see what C does...
noise	char		D:LHS	Control room noise source after the fiber optic receviers, before the variable gain. The noise source can be used instead of the signal from the platform/sky. Values: 0 use IF from platform, 1 use noise source.
inpfrq	char		D:LHS	IF mixing stage selected for the control room (downstairs) IF. Values: 0 750NarrowBand,1 300Mhz IF, 2 750IF,3 1500IF.
mixer	char		D:LHS	mixer system used: 0- 750 filter 1- 1250 filter 2- 1500 filter 3- 1750 filter
vlbainp	char		D:LHS	Which vlba IF frequency is used. Values: 0 750, 1 1500.
syndest	char		D:LHS	Destination for downstairs synthesizers (not used by alfa). Values: 1: 260 to 30Mhz downconverter 2: vlba or sband 3: if2 mixers 4: front panel.
calsrc	char		D:LHS	Where the ttl level to fire the cal comes from. Values: 1: interim correlator 2: 25 Hz hardware tied to the station clock. 3: wapps 4: pdev 5: cal always on 6:cal always off 7: external bnc at cal multiplexor 8: external switch at the cal multiplexor.
cal	char			Is cal turned on. I don't think this value is ever loaded.
vis30mhz	char		D:LHS	source for 30 Mhz going to visitors equipment in screened room. values: 1 dome, 0 carriage house.
blank430	char		D:LHS	The 430 Mhz blanking (1st lo, rf, and 260 to 30 Mhz downconversion). Values: 1 430 blanking is enabled, 2 430 Mhz blanking is not enabled.

Interpolating positions/velocities (top)

The subroutine call sequence to interpolate positions is in ldHdrRecs() of bpOut.c.

ldHdrRecs() - called on each data record to fill in header info for this record.

interpLdPos() - interpolate scram info to current time stamp then load in header.

interpPosScrm(): Loop calling interpPntScrm, interpAgcScrm till we get both interpolations done or we timeout. The interpolation wants to bracket the requested time with scramBlocks. Note that interpPntScrm, interpAgcScrm are in libbp.a

interpPntScrm: interpolate scram pnt block info.

interpolate:lmstRd, req_raJRd, req_decJRd, req_azEncRd, req_zaEncRd, modelCorEncAzZa, velHelo, velGeo. Load debug info for two points used

interpAgcScrm: interpolate scram agc block info using two data blocks.

Interpolate: act_azEncRd,act_zaEncRd,act_altZaEncRd

load: lst,vel_bary,vel_geo,req_raj,req_decj,az_err,el_err,model_offaz/za.
azzaToRaDec: Back compute from az,za (with model removed and alfa offsets included) to raJ,decJ actual positions.
Load: azimuth,elevatio, crval2,crval3

Outstanding issues: (top)

To implement or fix: (as of 02JUN08)

tsys: probably won't implement..
user_offaz,za
total_recs
tot_power: probably leave out since no 0 lag
tcal_numcoef,tcal_coefA,tcal_coefB
pol: probably not going to be implemented since we store multiple pols per row.. this specifies which pol.
attn_cor: probably need to store pdev attn info.. but no space for all 4 values .. probably
cal: should be set to 1 if cal was on during row

03jun08:

If the cals are being driven by an external source (cima, wapps) the cal keywords in the bintable header: calctl do not tell us what is happening.
We need to define the numbering scheme for the version key in the bintable header. cimafits is at 2.0. We are different than cimafits.

22may08:BeamNum/sbcNum:

The downstairs if/lo multiplexes the alfa beams or the single pixel bands to pdev.
Let beams count from 0, sbc mixers) count from 1 then the ordering is:

beam0=sbc1
beam2=sbc2
beam4=sbc3
beam6=sbc4
beam1=sbc3
beam3=sbc5
beam5=sbc7
sbc8 is not connected to pdev.

if we use the same boxes.conf file then b0s0g0 will be sbc1 and b1s0g0 will be sbc3
We probably need to use separate boxes.conf files for alfa and single pixel so the bns0g0 always maps into the correct order.

History: (top)

30nov09:

ALFA outer beam position errors in cimafits and psrfits files.

27nov09

top of table, added section saying that times/positions are for the start of the integration (not the center).

11nov08:

when closing files, zero fill till file length is a multiple of 2880 (fits blk size).
Files previous to this date are not a multiple of 2880 which is not according to the fits standard.

02jun08:

Implemented backendmask, num_ifs for single pixel.
In binary table header added pcalctl,wcalon,wcaloff,wcalphas.

30may08:

added keywords:glon,glat,map_offaz,map_offza, map_offra,map_offdec,

29may08:

added:req_posxx,req_offxx,req_ratexx,req_equinox,rate_dur,req_pnttime

cdelt1,uppersb have the frequency direction set correctly.

28may08:

set rfnum to 17 if using alfa..
added cur_err.
added beam_offraj,beam_offdecj

27may08:

enc_azimuth, elevatio, altel are now interpolated to crval5. I had been using the uninterpolated values because they were closer to the actual values read. The problem is that we get behind by 1 or two seconds waiting for the scramblocks to appear for the normal interpolation and this caused the values loaded in enc_az.. to be off. We also were getting duplicate values then jumping by two seconds since we were no longer synced with the loading of scramnet.

<- page up

home_~phil