NRAO Home  >  Green Bank  |  Wiki Topic:    GB > Knowledge > PulsarRFI

Readme | Changes | Index | Search | Go

What are pulsars and why (and how) do we observe them?

Radio pulsars are neutron stars with strong magnetic fields that rotate rapidly -- from once every few seconds to many hundreds of times per second (the so-called millisecond pulsars). They emit broadband radio waves continuously in beams from their magnetic poles. Because the stars are spinning, though, if we happen to be in the path of the rotating beam, we see the radio emission pulse each time the star rotates, much as we see a lighthouse "pulse" here on earth.

Those radio pulses arrive at earth with atomic clock-like precision, allowing use to use pulsars as "tools" to study a wide variety science that would be otherwise impossible. For example, we can measure the curved space-time around ultra-dense objects like black holes, and thereby test Einstein's theory of General Relativity, by watching how pulsars orbit those objects. We can also use pulsars to give us information on how matter behaves at densities much higher than we can achieve in any laboratory on earth, in effect doing nuclear physics with "experiments" located throughout the Galaxy.

Pulsars emit most of their observable radiation in the radio spectrum between a hundred MHz and a couple GHz. However, that emission is so faint that if you were somehow able to gather all of the radio energy ever collected from the nearly 2000 known pulsars by all of the radio telescopes in the world, it would only be the equivalent of the kinetic energy in a few falling snowflakes. The only reason we can see these extremely weak radio sources at all is because we use the largest radio telescopes in the world with extremely sensitive radio receivers. For the faintest pulsars we also need very long observation times (measured in hours) and large amounts of radio bandwidth (or fractions of the radio band, typically hundreds of MHz).

Why is pulsar data so susceptible to interference?

A big problem with these observations is that this portion of the radio spectrum, between 100 MHz and 3 GHz, contains most man-made radio signals as well as almost all the pulsar emission. Unfortunately, the man-made signals can often be very powerful and/or very nearby, resulting in interference in our data that is much stronger than the faint pulsar signals. Because our observations monitor large portions of the radio band for up to hours at a time, that means that we often see many different kinds of interference during an observation -- at different radio frequencies and at various times.

Another big problem that affects pulsar observations but not other kinds of radio astronomy observations is that pulsar signals are periodic in time, just like most man-made radio signals. Almost all modern communications devices, from radios to cell phones to wireless networks, encode the voice or data that they are transmitting in a periodic way. In our astronomy data, these man-made periodic signals can contaminate the pulsar signals from known pulsars or they can masquerade as pulsars when we are searching for new ones.

Typically, radio frequency interference (RFI) that affects pulsar observations comes in one of three types:

1. Narrow Band Continuous These signals can be very strong and are typically periodic at specific radio frequencies. They also tend to be present either continuously or for long amounts of time. Examples are radio stations, satellite transmissions, and wireless networking signals. If the signals are not too numerous or too strong (so that they affect our instrumentation and not just the data) we can typically "clip" those small portions of the radio band from our data and still do our science.
2. Short Duration Broadband These signals tend to be short duration (as short as a millisecond but possibly lasting seconds or minutes) but covering large portions of the radio band. Lightning is a good example of a very short duration and broadband (but natural) form of interference. Other examples include capacitor discharges associated with taking a digital picture or the arcing that occurs during welding. If these signals are not too numerous or frequent, pulsar astronomers can typically clip out the small portions of time in their data that are affected, and once again, still do their science.
3. Broadband Continuous These signals are perhaps the most problematic for pulsar astronomers. They are present throughout the radio band and throughout the duration of most observations. A good example of a signal of this type is the 60 Hz signal present in your wall outlets and in power lines. Since most all electronics and many mechanical systems run on this 60 Hz signal, it tends to be present at some level in all pulsar data, despite our best efforts to prevent it. Luckily, since it is usually fairly weak and we know where to look for it (at 60 Hz!) it typically does not seriously affect our science (although note that there are no known 60-Hz pulsars!). However, broadband signals which are unknown to us are very difficult to remove or ignore.

One of the worst things that can happen to pulsar data is if a very strong narrow-band signal gets effectively transformed into a very-strong broadband signal. Such a thing can happen if the signal overloads our extremely sensitive instrumentation at the radio telescope. Once that happens, there is almost nothing that can be done to "fix" the data, and it is ruined. As the number of wireless transmitters (e.g. cell phones, laptops with wireless networking, satellites) continues to increase, such situations will become much more common. The RFI example that I show in the next section is one of this type.

Nearby Wi-Fi and a Ruined Pulsar Observation

Wireless networks operate at 2.4 GHz over short distances, typically a few hundred feet. However, given how sensitive the GBT is, if a wireless network is close enough and/or located in the right spot, it can destroy an observation.

Since 2004, we have been making a series of highly successful pulsar observations using radio frequencies between 1650-2250 MHz. These observations look at dense clusters of stars (called globular clusters) where numerous millisecond pulsars are thought to occur. The observations last for 7-8 hours and are used to search for new millisecond pulsars to be used as tools to do "exotic" science. The searches use what is called a Fourier Transform to separate all of the periodic components out of the incoming data.

The following image shows a piece of a Fourier Transform from some fairly interference-clean GBT data from this project (taken in Feb 2008). You can easily see several bits of RFI sticking out above the noise as well as 3 of the 32 known pulsars in this cluster (the other pulsars are either at higher pulsation frequencies or completely buried in the noise such that advanced signal processing techniques are required to see them). Note the "power" scale on the left of the plot.

The observing band between 1650-2250 MHz was chosen such that it would not be affected by satellite radio stations and wireless networking, both of which are at slightly higher radio frequencies. However, the receiver that is used for these observations is still sensitive at those higher frequencies. So if a particularly strong signal outside the 1650-2250 MHz band adversely affects the receiver itself, all of the data from that receiver, including that in the lower band, will be contaminated.

The following image shows a similar piece of a Fourier Transform from an observation of the same cluster using the same exact observing system as in the previous example. In this case, though, a wireless network was in use within approximately 1 mile from the telescope. Its signal was so strong that it modulated the observing system and completely contaminated the data. Once again, note the power scale on the left of the plot. The same pulsars in the example above are in these data, however, their power levels haven't changed. What has changed is the level of the surrounding noise. This 7-hour GBT observation was effectively completely ruined by this interference.

The last image shows a zoomed in region around the first pulsar visible in the top image. You can see that there is a forest of interference lines that completely dominates over any pulsar signal in the data.

-- ScottRansom - 14 Feb 2008