ACSIS Information
August 29, 2008

Current Status

ACSIS is available for use with HARP and other receivers. The system - ACSIS/receivers/OCS - is still very much being worked on e.g. fixing outstanding problems such as total power spikes, measuring and implementing K-mirror pointing offsets, measuring and implementing actual pixel positions, improving calibration, baselines and adding observing modes. In particular multi subsystem modes have been added. Note that frequency switching has not yet been implemented. Time permitting frequency switching will be debugged during 2008 - this assumes that some of the other issues are solved and that SCUBA2 not will consume to much effort.

Available Bandwidth Modes

HARP
ACSIS has in total 32 down converter modules (DCMs) and correlator cards. A DCM "picks out" a 250 or 1000MHz wide subband from an IF input and feeds the signal to a correlator card.

HARP only has 16 IF outputs while there are 32 DCMs and correlator cards. The "excess" of DCMs and correlator cards can be used in two ways.

Chaining two correlator cards together doubling the number of channels in a 250 or 1000MHz subband. Only 16 DCMs are used in this case.
Use two DCMs to extract two frequency subbands from each of the 16 IF inputs and feed each subband to a correlator card. This creates two spectral windows or spectral regions to follow the OT terminology. All 32 DCMS are used in this case.
The OT support a special case of option two - by selecting a single spectral region with 440MHz or 1860MHz bandwidth the OT automatically set up two subbands that overlaps slightly. These bands can be merged later in the data reduction (it will also be done by the pipeline in the future). Note: the merging used to done by the ACSIS DR system - for further details see footnote*. When using 440 or 1860MHz bandwidth you might be better of not centering the line in the wide band - since that places the line in the overlap region. New options are the 1600 and 1800MHz bandwidths, which have larger overlap between the 2 subbands (but the same spectral resolution) than the 1860MHz bandwidth. This is useful when the spectra contain very broad lines. Likewise there are 400 and 420MHz bandwidths available.

An example of the second use is to observe 13CO 3-2 and C18O 3-2 using two spectral regions (subbands). A number of such setups are pre-prepared and can be selected as "spectral configurations" in the OT. The spectral window and bandwidth possibilities for HARP are summarized in the table below.

An example of the third option is to select 1860MHz mode for a wide extra galactic source. It this case it is less error prone to let the OT do the setup than "manually" do the setup using the frequency editor. Further, it has the advantage that both spectral regions use the same rest frequency.

Spectral windows	BW mode	Channel Spacing	Usable Bandwidth
1	250	30.5kHz	~220MHz
	1000	0.488MHz	~930MHz
	1860 (1600,1800)	0.977MHz	~1860(1600,1800)MHz
	440 (400,420)	61.0kHz	~440(400,420)MHz
2	any 250	61.0kHz	~220MHz
2	any 1000	0.977MHz	~930MHz

Table 1: HARP BW modes

E.g. if two spectral windows are selected in the OT one with a bandwidth of 250MHz and the other with a bandwidth of 1000MHz, the result is one spectrum of usable bandwidth ~220MHz and channel separation 61.0kHz and one spectrum with usable bandwidth ~930MHz and channel separation 0.977MHz. Also not that if the 1860 or 440MHz option is selected only one spectral window is allowed - the 32 DCMs and correlator cards has already been used up! As can be seen from channel spacing in the table the 1860 and 440MHz mode are using 2 subbands. Conversely, if you select 2 subbands only the bandwidth options 250 and 1000MHz exists.

RxA & RxW

RxA or RxW are not able to use all the DCMs and correlator boards in ACSIS - see the HARP section. A maximum of 4 DCMs/correlators can be feed from the same IF in a usable way. Thus, you can select 1-4 spectral windows for each IF output from RxA or RxW with the BW mode summarized in the table below.

Spectral windows	BW mode	Channel Spacing	Usable Bandwidth
1-2	any 250MHz band	30.5kHz	~220MHz
	any 1000MHz band	0.488MHz	~930MHz
	any 440MHz band	61.0kHz	~440MHz
	any 1860MHz band	0.977MHz	~1860MHz
3	A spectral window as in one of the four rows above
	any other 250 MHz band	61.0kHz	~220MHz
	any other 1000MHz band	0.977MHz	~930MHz
4	any 250 MHz band	61.0kHz	~220MHz
4	any 1000MHz band	0.977MHz	~930MHz

Table 2: RxA & RxW bandwidth modes

Note that it is possible to configure ACSIS with four 250MHz subbands with the centers separated by 220MHz to generate a ~880MHz wide spectrum. Non of the current frontends has an IF bandwidth large enough to make it fruitful to implement the analog setup using 1000MHz subbands - the OT will not allow you anyway. Another possibility is to combined 3 250MHz subband to a ~660MHz spectral window and a separate 250/1000MHz spectral window. These possibilities are not explicitly supported by the OT or covered in table 2.

For the curious - with 3 spectral window and 4 DCMs/correlator cards the DCMs/correlator cards can only be distributed 1+1+2 over the three spectral windows. The spectral window with 2 DCMs/correlator cards can support higher resolution or higher bandwidth.

* The merging in the ACSIS DR occasionally created artificial features where the spectra were joined. The reason were to small overlaps and that the merge was weighted by the total power drop off. In principle Tsys weighting would be better but the estimated Tsys also get very uncertain at the edge. Further, merging the spectra was an irreversible process. Thus, if there was a problem we could not go back and reprocess the data. Of these reasons we do now store the unmerged raw spectra and the merging is done in the post processing.