NRAO Home  >  Green Bank  |  Wiki Topic:    GB > Software > PlanOfRecordC22006 > TestSpectrometer

Changes | Index | Contents | Search | Statistics | Go

Test Spectrometer

The goal of this task is to answer some questions about the Test Spectrometer. Basically, think of it this way: you are in the software group, and you know nothing about the test spectrometer, but want to know more about it. Formulate any questions that you would want to ask, and then answer them. Ideas:

• What is the Lab Spectrometer?The Lab Spectrometer is a rump version of the GBT Spectrometer. It uses the same correlator chips, and is controlled in much the same way, via a VME Bit-3 interface to a linux host. The original purpose of the Lab Spectrometer was for ALMA antenna tests. The main difference between the two are as follows:
  • GBT Spectrometer
    • 8 high-speed samplers
    • 32 low-speed samplers
    • 4 quadrants
    • A large number of modes
  • Lab Spectrometer
    • 4 high-speed samplers
    • 1 quadrant
    • only 3 modes
    • 2x bandwidth per quadrant (relative to GBT Spectrometer)

• Who uses it? The Lab Spectrometer has several users. First, the digital group uses it to troubleshoot Spectrometer issues. In addition, receiver engineers and some scientists have expressed a desire to use it to test receiver baselines while the receiver is in the lab, rather than on the telescope.

• Where is it located? The Lab Spectrometer is on a movable rack in room 128 of the Electronics Lab.

• Is it hooked up to a machine?The Lab Spectrometer is connected to ryac, which is the analog to the GBT Spectrometer's host computer, earth.

• How does it currently operate with respect to software? The Lab Spectrometer is controlled by an M&C Manager, which is a version of the GBT Spectrometer's Manager cut-down and adapted for the purpose by David Villa, Rich Lacasse's 2005 summer student. (One key difference is in the way the data is assembled and processed before being written to FITS files.) This means that the Lab Spectrometer can form part of the receiver lab's software installation, and can be controlled with the standard Spectrometer Cleo screen.

• Does it need any M&C software support? The Lab Spectromenter could potentially need support in two software areas: data reduction and visualization (see next question) and control software. The Lab Spectrometer manager has been running for a few months now; it should not require any more or less attention than the GBT Spectrometer's manager does. One issue that can arise is that since the code was copied lock, stock and barrel from the GBT Spectrometer and then modified somewhat, there is still a good proportion of duplicate code that has the potential to get out of sync if changes are made to the GBT Spectrometer manager.

• Is the new engineering spectral line GFM plugin sufficient for data analysis purposes? No. GFM can fill some needs for now, but scientists and engineers want some modifications to GBT IDL to allow them to do more sophisticated work. Paul Marganian believes that this will not require much effort. Below are details on data analysis for the Test Spectrometer:

• • The GFM Spectrometer Engineering Plugin is sufficient for doing Quick Looks at raw lags and FFT data, and for doing basic analysis, specifically simple differences of spectra, and averaging of spectra. It is currently being used in the electronics lab, and it seems to be performing to expectations. There is one issue: the current manager for the Test Spectrometer writes its FITS files in a directory names "LabSpectrometer". This confuses GFM and SDFITS, making it think that this is a whole new breed of backend. This problem is currently being worked around by creating a symbolic link in the project directory named "Spectrometer" that points to the original directory.
• • However, more sophisticated analysis is not easily done in GFM. It would be ideal if the engineers used the GBT's official data analysis tool, GBTIDL, for their analysis needs. However, for this to be possible, the Test Spectrometer data must be first filled with the sdfits filler. Currently this is not possible because the only FITS files present are a ScanLog.fits and Spectrometer FITS files. Small changes are necessary to sdfits to make it possible to fill this incomplete data set. This should only require 1 to 2 days of effort, but should have a scientific sponsor.
• • The one limitation to using GBTIDL is that raw lags are not available. There is no SDD supported analysis package for spectrometer lag data, but this was discussed in a meeting led by RonMaddalena a few weeks ago, and the concensus seemed to be that there are several tools out there for doing this, which astronomer's can show to engineers, including glish scripts, IDL libraries (outside of GBTIDL), and more.

The rationale for this exercise is that in management meetings lately, I've heard often "we need to support the test spectrometer". However, it also seems like no one knows exactly what's needed to provide this "support". In fact, it's possible that no support is needed. But, we need to figure out the real story, and that's why this task has been assigned.

Put your results on this page.

-- NicoleRadziwill - 14 Feb 2006