Information for observers and their GBT friends.
Quick links:
The Ka band receiver is built according to a pseudo-correlation design intended to minimize the effect of 1/f gain fluctuations for continuum and broadband spectral line observation. A block diagram of the Ka band receiver is shown below.
The design and operating principles of pseudo-correlation receivers in general and this receiver in particular are discussed by
Simply put, this architecture allows a beamswitch to occur before the first stage HEMT amplifiers (in effect) without the excessive price and spectral band limitations that can be incurred by placing traditional dicke switching circulators at this point in the system. One disadvantage is that due to imperfections in the RF hardware, some crosstalk is evident in channels that nominally see opposite feeds when the system is used in total power mode; crosstalk does not affect beamswitched photometric measurements. The IF conversion scheme for the Ka band receiver is also different from the existing GBT high frequency receivers, in order to accomadate delivery of the full receiver band to a future wideband spectrometer.
At 32 GHz the beam fwhm is about 25 arcseconds. The beam separation is 1.35 arcminutes.
The Ka band receiver was commissioned in the Fall and early Winter of 2004. Basic properties of the receiver (linearity, spectral response, and system temperature) were nominal. Beamswitched baselines on blank sky were seen to be almost negligible; while there is a 20 to 30 MHz ripple it appears to average down with the noise over tested integration times up to one hour. Under reasonable observing conditions (PWV < 15mm and no precipitation) system temperatures at zenith in the band center should be under 40 K. Repeatability of the relative calibration of spectral data on continuum sources is better than 20% in all known cases (users should bear in mind that even under benign conditions, the GBT antenna itself isn't guaranteed to deliver better than a 5% RMS in flux density measurements at these frequencies).
Baselines on blank sky, one hour of "clock on the wall" time (not all phases included in the average):
Baselines against a continuum source (calibration transferred from another continuum source) also showing the LO1 x2 harmonics near 29.33 GHz:
Some known issues:
- Only two of the four spectral line channels currently exist. With the receiver in its nominal beamswitching state (SIG, ie, straight-through), these two channels give: Feed 1 Left polarization, and Feed 2 Right polarization. For spring 05 this implies a penalty of root(2) in sensitivity if dual-feed dual-polarization simultaneous measurements were assumed. We aim to have all four channels available in Fall 2005. (The two unimplemented channels correspond to those on the top half of your Ka band CLEO screen)
- The 2nd harmonic of LO1 shows up in the spectrum as a sharp spike at 29.3 GHz and several harmonics spaced by multiples of 10 MHz around it. This will be addressed in the future Ka band LO upgrade that adds the other two spectral channels.
- We have not had good success reconciling laboratory Tcals with commissioning calibration data in detail. Therefore the system calibration should be taken as only indicative, and carefully checking your calibration against a known calibrator-- with the calibration data taken in the same fashion as your astronomy data-- is as ever essential procedure.
- Analysis procedures have not yet been coded into any official user package, however interim IDL scripts are being developed (see below).
- the config_tool should work normally for observations with the Ka band receiver. A set of recommended configuration scripts for pointing & focus, 4x800 MHz beamswitched spectroscopy, and 4x800 MHz total power spectroscopy, are in the table below.
- The recommended method for observing point sources is to do a double beamswitch (beamswitch and NOD). The 4x800 MHz beamswitched spectroscopy scripts below set the system up for this. They also configure the system to use just a single Cal diode (which is what the interim IDL data reduction scripts assume also).
- The total power mode is appropriate for example for: On/Off observations (if you are in a confused part of the Galaxy say or your source is extended) or On-the-Fly maps. The total power mode configures both cals to be used in a manner identical to other GBT receivers.
- In the past the receiver LO power levels have not been as stable as we'd like. It is recommended to verify that the blue "LO power level" box on the Receiver CLEO screen be strictly kept between 4.5 and 5.1 volts, levels above 5.1 cause compression in the LO amplifiers and can increase the level of the 29.3 GHz second harmonic. You might need to disable LO1A "auto level" in the LO CLEO screen and set the power level by hand, but this should only be a last resort.
- For general observing procedures refer to the GBT Observer's Help Page
CONFIGURATION SCRIPTS
NOTE: Turtle scripts and config tool scripts only differ in trivial ways now, as you can see by comparing (absence of "g." etc, which are standard turtle things and nothing special to Ka band).
Frank also has equivalent, but now independently maintained, scripts for configTool only at http://wiki.gb.nrao.edu/bin/view/Data/ConfigurationCases
Procedures to analyze Ka band data have not yet been implemented in the "Official" GBT IDL package or AIPS++, so in the meantime Frank Ghigo is developing a set of IDL scripts that will serve temporarily. These will use SDFITS to gather the data together and are still being worked on.
The AIPS++ MSFILLER appears to fill Ka band data although it's not been checked for correctness.
If you wish to do the analysis yourself, the basic operations required to get your data into IDL are actually quite simple.
- run sdfits to make a single FITS file with all your data. You should run it in "RAW" mode since sdfits will not properly calibrate Ka band data, and do the calibration yourself in IDL. Example syntax to fill data from scans 42 through 50: "sdfits -backends=acs -scans=42:50 /home/gbtdata/MYPROJECT mySdFitsFileName"
- add /users/astro-util/idl/astrolib/pro/ to your IDL_PATH environment variable
- Start IDL and read the sdfits file you just made in via "myTable = MRDFITS('mySdFitsFileName.fits', 1, myHeader)"
- Proceed from there. For example to average all but the first integration of the SIG/CAL OFF phase from bank A, sampler 1 in scan 43 do this:
myTable = MRDFITS('mySdFitsFileName.fits', 1, myHeader)
scan= 43
isSig='T'
isCal='F'
bankSamp='A1 '
q=where(myTable.sampler eq bankSamp and myTable.scan eq scan and myTable.sig eq isSig and myTable.cal eq isCal)
qs=size(q)
qList=q[1:qs[1]-1]
dataArr=myTable[qList].data[*]
avgSpec=avg(dataArr,1)
plot,avgSpec
Info for GB experts: how this receiver differs from other receivers
- IF Scheme: The Millimeter Converter comes between the Receiver and the IF rack, performing an extra stage of frequency conversion. More details about the IF conversion scheme are available at http://wiki.gb.nrao.edu/bin/view/Projects/KaLoIf which describes the modifications to config tool that were implemented to automatically configure Ka+mmconverter via standard scripts. You can also see the block diagram
- Direct RF detectors: Like the prime focus receiver, this receiver has direct RF detectors in the receiver, whose voltage is piped straight into the DCR bypassing the entire GBT IF chain. There will eventually be 16 of these, all used by the Caltech Continuum Backend but presently there are 4 of them. However one can also configure "normal" continuum observations using the detectors in the IF rack (or analog filter rack for that matter), and this is the default implemented in the config tool scripts above.
- Beamswitching: Beam switching is implemented with "phase switches", which must be properly configured to track the SIG/REF signal distributed by the GBT switching signal bus. This is done by the config tool scripts above.
- Cals: because this receiver implements a proper beam switch, it is desirable to be able to inject a power difference to the feeds for calibration. Therefore the two CAL diodes can be configured individually to track or not track the CAL signal in the switching signal bus.
- LO1: the mixer for this receiver is more sensitive than many , and the LO1 power level in the receiver CLEO screen wants to be kept between 4.5 and 5.1 volts without exception, levels above 5.1 cause compression in the LO amplifiers and can increase the level of the 29.3 GHz second harmonic.
- Ka receiver manager writes one FITS file per scan (not just the calibration FITS file with Tcals written once per project) that has various information about the receiver config useful for checking things.
- beamswitching is L1 <-> R2, L2 <-> R1, ie between opposite circular polarizations.
- Continuum tunings: Due to peculiarities of the IF conversion scheme, we recommend that you avoid continuum tunings above 36 GHz. Such tunings are possible but subtle. The pointing and focus script above implements a 32 GHz tuning using the detectors in the IF rack, which is what you should do unless there are really compelling reasons to the contrary.
Information for Experts: FAQ
- Why is there an extra frequency conversion again? (i.e., why do we have these "Millimeter Converters"?) For other GBT high frequency receivers, the first LO is tuned by many GHz in order to deliver the 6 GHz IF that the IF rack wants. For Ka band, in order that the full Receiver band can be delivered at IF to a future wideband spectreometer, the first LO is almost static at 14 and 2/3 GHz. "Almost", because doppler tracking is performed through the first LO. The millimeter converter does the further conversion necessary to translate this full receiver IF band, into an IF band that the rest of the GBT IF system (IF rack etc) needs.
- What do I need to do to write a configuration script that differs from those above? Eg, if a narrowband tuning is desired. The only thing you need to be aware of is that the entire band of interest in a given tuning must be confined to one of the following three ranges: 26-31 GHz; 30 - 37 GHz; 36-40 GHz. This is a constraint imposed by the common millimeter converter, and if you try to violate it the config_tool should default to something reasonably sensible, but warn you. Other than this detail all is as for other GBT receivers-- the configuration keywords are the same as for any other GBT receiver, etc.
- Can ka be used for total power observations? Yes: just set up an ordinary tp phase table and enable both cal diodes to track the CAL signal-- this is done
- Can ka be used for mapping? Yes, but a) we've not tried this; b) the feed-leakage mentioned above will either need to be established as negligible in your maps, or removed by some appropriate algorithm (eg; make a beamswitched map and EKH it). This observing mode is not currently supported and any efforts are completely experimental in nature.
- How are the spectral baselines again? On blank sky, pretty good if you use a full double beamswitch and single cal for calibration. That means there's a 20 MHz ripple at about twice the thermal noise level that seems to go down with the noise, at least over one hour. On strong continuum sources (say a Jy), not as good. That means fine scale structure with an amplitude of maybe +/-20% the continuum level.
- Why does the beamswitch switch between opposite polarizations, of opposite feeds? There was a design concern about crosstalk causing oscillations in the HEMTs which this was thought to mitigate. Assuming the atmosphere and your source are not strongly circularly polarized this will have minimal impact on your data.
THINGS WE ARE STILL CLEANING UP (ought to be done before march)
- GFM TrackBeam? problem (for now: select ONE dcr channel that corresponds to "Beam 1", which is what you should also select in GO) -- fixed
- S/R track buttons turn yellow for no apparent reason (we don't see an effect from this in the data)
- The receiver monitor point labeled TEC300K does not have an actual sensor connected yet.
- bogus "phase switches under computer control" message
- Auto level power levels on LO1 not so good (giving 5.3 V at RX which is a red box) -- it's ok to disable auto level.
- Rcvr FITS files don't appear in the ScanLog?
- Do we need to change the names of the receptors in the database, so the spectral line data can be accessed by MSFILLER?
-- BrianMason - 13 Jan 2005
Revision r1.18 - 11 Dec 2005 - 17:06 GMT - GalenWatts
Parents: WebHome > KaBandReceiver
|
Content copyright © 1999-2007 by the contributing authors.
All material on this collaboration platform is the property of the contributing authors.
|
| |
FullBlockDiag04.jpg 
