ca-forum #21 minutes - 13/2/97
Interference
Barbel Koribalski provided some plots of narrow band data (8 MHz bandwidth centred on 1406 MHz) which showed substantial interference. Two forms were present - a birdie at 1402 MHz, and a variable frequency component. The latter covered roughly 2 MHz, centred on 1405 MHz. It is strongest on the shortest baselines. It was present on several days, and persisted for long periods (> hours).
It has also been seen by other observers. Ron Beresford has installed a wide-band yagi on the roof, along with RF amplifier and spectrum analyser; this could be used to help search for the source of this interference.
Lister reports that interference of this nature was seen a number of years ago, but that it then disappeared.
Warwick Wilson has been asked by Marcus Price to develop a correlator to search for interference at Parkes. Narrabri could consider joining forces with parkes to obtain a similar device.
Narrabri already have a guide to interference - a plot of the more serious sources is available in the control room. G.Baines also maintains a RFI file, listing all the observer reports of interference. More detailed data is required: duty cycles and some characterisation of the nature of the interference is needed. RDE signalled that some support in the characterisation efforts may come via the 1KT project.
Many of the reports on interference are after the fact - which makes difficult the tracking down of the source.
Active Measures
There are a number of ways the interference can affect the observations:- a signal within the band; an aliased signal; intermodulation products. A filter could help reduce the problem; a filter in front of the the LNA would be the ideal, but seems unrealistic, as it would compromise the system temperature. Further, it is argued (G.Graves) that the LNAs have sufficient head-room to avoid intermodulation products. Some options:
[G.Graves adds: Observers wanting two simultaneous windows within the one band (eg, H and OH) pose a problem. Filters that are tunable both in width and frequency will be a challenge]
In principle these filters could be concatenated without performance loss.
Passive Measures
Does it matter? Some instances of interference may look bad on the VIS plots, and yet be benign in the final image. It would be useful to be able to provide observers with some indication on this - which interference can be ignored, and which cannot.
Summary
At this stage we need more information on the nature of the interference:
Circular Polarisation Tests (RS)
Bob Sault ran some tests to check whether the sense of circular polarisation is treated in a consistent manner at all AT bands. A number of technical difficulties were experienced; the results are unclear because it was not possible to recover the x-y phases from the syscal tables. The experiment will be repeated.
Warning Flag The experiment did reveal a source of locally generated interference. A strong birdie will be seen if the L4 module of the second conversion chain is set to the frequency of the first conversion chain. For example:
If you were to ask for 6390/128 and 8000/128, then the L4 setting of the second conversion chain is set to 6390 MHz, and a strong birdie will be seen in the first conversion chain : 100 Jy in a 4 MHz channel.
When the LO conversion chains were originally set up the frequency bands were rather narrower, and care was taken to ensure that these clashes did not occur. A number of overlaps are now possible with the extended frequency coverage. The observer still has to be unlucky to be affected, but we have increased the odds.
>> SCHED should warn observers if a collision has occurred. Data Quality Observations (MW)
These are 12 hour runs made after each reconfiguration. See DQO URL for more details. One of their aims was to provide a check that the reconfiguration had been successful - it seems that the reconfiguration check list is adequate for this purpose. The secondary aim was to provide a long term monitor of the array's performance, as a safeguard against slow degradations. It was hoped that some science might follow from the observations - from, for example, the many observations of a standard blank field.
It seems that the latter aims are not well met, and that we might make better use of the time. Some suggestions :
This topic will be re-addressed at the next forum - suggestions are solicited.