Radio-Frequency Interference

The bands within the tuning range of the VLA which are allocated exclusively to radio astronomy are 1400 - 1427 MHz, 1660 - 1670 MHz, 4990 - 5000 MHz, 15.35 - 15.4 GHz, 22.2 - 22.5 GHz, and 23.6 - 24.0 GHz. No external interference should occur within these bands. Experience shows that RFI is a serious problem only within the 20, 90, and 400 cm bands. At 20 cm, interference is most serious to the D configuration, as the fringe rates in other configurations are often sufficient to reduce interference to tolerable levels.

RFI at the VLA is an increasing problem to astronomical observations. To monitor this increase, and to provide a rough guide to the severity of this interference, the RFI spectrum at all bands from P through K is measured on an occasional basis, using the VLA correlator system. Plots of typical spectra in the L and P bands, taken with 6 kHz resolution, are shown in Figures 6 and 7.

Downloadable plots of all RFI observations from 1993 onwards are available on the Web. For general information about the RFI environment, contact the head of the IPG (Interference Protection Group) by sending e-mail to NRAO-RFI@nrao.edu.

**Figure 6:** **Typical L-band interference Spectrum.**
$\begin{figure}\epsfig{file=Lbandrfi.eps, height=4in }\end{figure}$

**Figure 7:** **Typical P-band Interference Spectrum.**
$\begin{figure}\epsfig{file=PRFI.eps, height=4in }\end{figure}$

Table 10 is a convenient summary of eight such observations taken during 2001. This table lists the `line' frequency, the average equivalent flux density (in mJy) in 50 MHz, and the peak flux density, also reduced to 50 MHz equivalent bandwidth. A significant difference between these columns indicates that the RFI is intermittent. These equivalent flux densities are approximate, and should be used only to give a rough approximation to the severity and likelihood of a problem. A more detailed version of Table 10 is available on the web at http://www.vla.nrao.edu/astro/rfi/rfifreqs.txt.

**Table 10:** **VLA RFI Between 1260 and 1740 MHz**
Frequency	Avg. Flux	Pk. Flux	Source	Comments
1277 MHz	12 mJy	20 mJy	Aerostat Radar	Sometimes absent
1286	2	5	Farmington Radar	Other weak lines nearby
1300	2	5	Internal RFI
1310	100	100	ABQ Radar
1316	???	???	ABQ Radar	Weaker, often absent
1330	45	80	ABQ Radar	Sometimes absent
1381	3	100	GPS L3 IONDS	On 3% of the time
1400	60	60	Internal RFI
1429-1435	15	130	Military	Four separate lines
1444,1453	5	100	High Altitude Balloons	NASA/NSBF
1500	2	5	Internal RFI
1515	15	100	High Altitude Balloons
1525	6	100	High Altitude Balloons
1530-1544	130	200	INMARSAT	Many `lines'
1557-1567	10	20	GPS Sidelobe?	Wide spectrum
1570-1580	500	500	GPS-L2	Wide spectrum
1600	120	120	Internal RFI
1602-1616	500	500	GLONASS	Many separate `lines'
1610-1645	500	500	IRIDIUM!	See text
1620	80	300	?
1650	13	25	Internal RFI
1678-1698	50	100	Radiosondes, satellite	6 variable `lines'.
1710	10	10	?
1714	500	500	Forest Service
1725	10	10	Forest Service
1730	25	25	Forest Service
1735	100	100	Forest Service

Between 1220 and 1255 MHz, very strong and very broad RFI is always present, primarily due to the GPS-L2 signal at 1227.6 $\pm$ 10 MHz, and a border radar at 1252 $\pm 1.25$ MHz near Deming, New Mexico. It may be possible to observe in selected, narrow bandwidths in this region. Numerous other strong radars belonging to the FAA's Air Route Surveillance system exist between 1250 and 1350 MHz in the general vicinity of the VLA. Between 1435 and 1530 MHz, aeronautical telemetry from White Sands Missile Range (WSMR) occasionally will interfere with observing. These transmissions are very occasional, and unpredictable. They are worst in the spring, during the WSMR wargames exercises.

Note that the listed RFI signal strengths are appropriate for the D configuration. These signal strengths are considerably reduced in the larger configurations - an average attenuation of perhaps a factor of 100 will be obtained in the A configuration due to fringe phase winding.

In late 1998, the Iridium constellation of Low Earth Orbit (LEO) satellites was activated. As is well known, this venture was not a commercial success, but it was purchased by a consortium of investors called Iridium LLC, and is being operated by Boeing. Thus these satellites are still broadcasting in the 1621-1627 MHz band, but their signals are strong enough that VLA observations between 1610 and $\sim$ 1645 MHz generally are not possible. Filters have been installed to allow 1612 MHz OH observations (see http://www.vla.nrao.edu/astro/guides/ifilt/).

In general, it may be possible to observe in spectral regions containing strong RFI provided: (1) That the RFI is not so strong as to cause serious gain compression in the amplifiers, and (2) That the RFI is kept out of the correlator through use of a narrow back-end filter. This latter requirement is particularly important for spectral line correlator modes, although use of Hanning smoothing is very effective in reducing the Gibbs ringing (described in the VLA Spectral Line Users Guide, at http://www.vla/nrao/edu/astro/guides/sline/current/). Gain compression in the antenna electronics often can be prevented by using a narrower (12.5 or 25 MHz) ``front-end'' filter (actually present in the IF stage), rather than the default 50 MHz filter. Note that use of these FE filters greatly restricts the range of tunable frequencies. The scheduling program JObserve is aware of these restrictions, and should be used when contemplating use of the narrow front-end filters.

Observers can use Table 10 to assist in deciding which center frequencies and bandwidths are most likely to result in good data at L band. There are very few good combinations for 50 and 25 MHz bandwidths. These are summarized in Table 11, which shows the `statistically' best frequencies to use at L-band with the listed bandwidths. Note that the VLA LO system restricts the selection of frequencies at both 50 MHz and 25 MHz bandwidths. The restrictions are particularly severe at a bandwidth of 50 MHz - for a given IF, the only center frequencies (in MHz) which can be observed with that bandwidth are those ending in 15, 35, 65, or 85. (For example, between 1400 and 1500 MHz, the only permitted frequencies are 1415, 1435, 1465 and 1485 MHz.) With the 25 MHz front-end bandwidth filters, regions 5 MHz wide centered on 1250, 1300, 1350, ..., cannot be tuned. And when using the 12 MHz front-end bandwidth filters, 8 MHz wide regions centered at 1225, 1275, 1325, 1375, ..., and 18 MHz wide regions centered at 1250, 1300, 1350, 1400, ..., cannot be reached.

**Table 11:** **Recommended Center Frequency/Bandwidth Combinations for L Band**
	Class A	Class B	Class C
BW	No RFI Expected	Weak/Occas'l RFI	`Tolerable' RFI
50 MHz	none	1365,1435,1465,1485	1335,1385,1415
25 MHz	1343 - 1347	1290 - 1297
	1353 - 1387	1453 - 1470	Use Table 10
	1413 - 1417	1503 - 1517
	1663 - 1665

Footnotes:

The White Sands Missile Range uses 1357 $\pm$ 3 MHz for airborne telemetry once or twice per week.

The ubiquity of L-band interference makes it difficult for users wishing high sensitivity to find significant bandwidths free of RFI. As a result of monitoring the RFI, we have established a number of standard bands which are more likely to be free of significant RFI. These are shown below in Table 12.

**Table 12:`JObserve` Names of L-band `Standard Frequencies'**
`JObserve`	AC		BD
Name	Center Frequency	Bandwidth	Center Frequency	Bandwidth
LL	1464.9	50	1385.1	50
L1	1364.9	50	1435.1	50
L2	1515.9	25	1365.1	25
L3	1515.9	25	1435.1	25

Footnotes:

The L1 frequency combination was used for the NVSS and FIRST surveys.

Another important form of RFI consists of signals which are generated by each antenna. These signals are picked up by nearby antennas, or by the generating antenna's feed, and produce correlated signals in the visibility data. This form of RFI is especially important in the P and 4 bands when in the C and D configurations. They appear at all multiples of 5 and 12.5 MHz - frequencies divisible by these numbers must be avoided. (It is this spectrum of RFI which limits our P-band bandwidth to 3.125 MHz.) Another family of RFI occurs at multiples of 100 kHz - these are much weaker and can be ignored for continuum work at 90 cm, but are important on many baselines at the 400 cm band.

At P band, the RFI situation can be particularly difficult. Interference is relatively infrequent in the evenings and on weekends, but during the day, very strong interference - sufficient to saturate the receivers - is common. The best advice is to arrange observing to fall outside of regular working hours. Very sensitive spectral line observations at P band require special measures at the VLA site to turn off known sources of RFI. To implement these measures, contact the interference protection group (see Table 16). The situation at 4 band generally is better, as nearly all of the RFI is weak and is generated by the antennas themselves (see above); the self-generated RFI can be removed in processing provided the spectral line correlator mode is employed.

At P and 4 bands, use of the spectral line corelator modes is strongly recommended for all observations to allow diagnosis and removal of internal and external interference. For observations which do not require linear polarization information, use of the correlator modes `1x', `2xy', or `4' is recommended to maximize the spectral resolution. The labels `x' and 'y' are the IF designators, A, B, C, or D. Refer to Section 2 for a description. The current default for 327 MHz is now `4'.

For L, C, K, and Q bands, observers should avoid using an L6 (second LO) setting of 3760 or 3790 MHz, due to an internal birdie produced by those LO settings. For X and U bands, avoid using 3840 and 3810 MHz. The JObserve program is aware of these restrictions, so users should not inadvertently fall victim to this problem.