Read Microsoft Word - HET608-Project2.doc text version

An Urban Radio Telescope for Jovian and Solar Emissions at 20.1 MHz

Robert S. French, HET608, Swinburne Astronomy Online, November 2009


Jupiter emits wideband long-wavelength radiation due to the interaction of electrons with Jupiter's strong magnetic field, and the Sun emits similar radiation from surface and coronal activity. These signals can be detected on Earth with a simple and inexpensive radio telescope. Such a telescope, based on the design from NASA's Radio JOVE project, was built for this project in an urban area and attempts were made to detect Jovian and solar emissions. The radio noise generated by the urban environment (> 300,000 K at 20.1 MHz) overwhelmed Jupiter's signals. Attempts were made to isolate the source of the noise at local, neighborhood, and metropolitan scales. The exact source of the noise remains unidentified, but was clearly associated with more distant sources rather than the local house or neighborhood. Individual Jovian emissions are impossible to distinguish from background noise sources using the equipment used in this project. Instead, a statistical approach was taken. The rate of highpower noise spikes during nighttime was 1.99 per hour when Jupiter was predicted to be emitting vs. 1.20 per hour when Jupiter was predicted to be quiet. The rate was 0.34 per hour when Jupiter was below the horizon. This provides compelling evidence that some of the noise spikes were due to predicted Jupiter emissions. The Sun was quiescent during the duration of this project and no solar emissions were detected.


Jupiter emits cyclotron radiation from electrons trapped in its extremely strong magnetic field (Dulk et al. 1999a, 1999b; Rogers 1995). These emissions range from below 1 MHz to ~39.5 MHz with a peak at ~10 MHz. However, the frequencies below ~15 MHz are unable to penetrate the Earth's ionosphere and the intensity of the emissions drops off rapidly above 20 MHz. This makes ~20 MHz an optimal frequency for receiving Jovian emissions on the Earth's surface. Likewise, the Sun emits radiation from surface and coronal activity such as flares and coronal mass ejections that can be detected in the same frequency range. Both Jovian and solar emissions can be detected by a simple, inexpensive radio telescope. Such a project is well within the reach of an amateur astronomer, and this paper describes one such installation. Manmade noise sources pervade the spectrum and, as a result, it is best to locate radio telescopes far from civilization. However, such a location was impractical for the current project and a radio telescope was built instead in the heart of the San Francisco Bay Area, a densely populated region. Manmade interference comes from a wide variety of sources, including electronics (battery chargers, computers, and computer peripherals), arcing high-voltage power lines, and radio transmitters. These sources of interference are much more prevalent in an urban area and significantly affect the ability to detect extraterrestrial sources in much the same way that urban light pollution makes it difficult to see the stars. The effect of interference on a radio telescope was evaluated and is a major result of this project. Even when manmade sources are eliminated, the galactic background still provides a major source of noise. Electrons accelerated by the galactic magnetic field emit continuum radiation with an intensity of

Page 1

~40,000 K (Flagg 2000) and this galactic background should be detectable in a radio-quiet location.1 Jupiter emissions are highly variable but can be in the hundreds of thousands or millions of K, and solar emissions can be even stronger at several million K. This makes them easily detectable above the galactic background. Solar radiation ionizes portions of Earth's atmosphere, as will be described later, making the sunlit half of the upper atmosphere mostly opaque to frequencies in the 20 MHz range. Such ionization can last from sunrise until several hours after sunset. As a result, observation of Jovian emissions must generally be conducted well after sunset and before sunrise. Solar observations must be conducted during daylight, but the intensity of solar emissions is such that they may be detectable even through the mostly opaque ionosphere. In this paper, I will first discuss the telescope's location and environment. I will then describe the receiver and antenna setup, including the effects of local environmental constraints on system design. Next, I will discuss calibration of the system and an extensive investigation into the sources of noise. Finally, I will describe the results of the observations and discuss the practicality of using a radio telescope in this type of environment.


The radio telescope was built during September, 2009 and installed in Sunnyvale, California, USA (37° 20' 13" N, 122° 3' 11" W). Sunnyvale is a small city (population ~132,000) that is part of the large metropolitan San Francisco Bay Area (population ~7,000,000). The location is near the center of "Silicon Valley", which is home to a wide variety of light and heavy industry. The large city of San Jose is nearby to the east, and the south-to-west is mostly uninhabited and looks towards a coastal mountain range and the Pacific Ocean. The relationship of the telescope to its surroundings is shown in Figure 1. Figure 2 shows the visibility (time above the horizon) for Jupiter and the Sun during 2009. Since Jupiter's emissions can generally only be detected when the Sun is below the horizon, August provided the greatest opportunity to detect Jupiter in 2009. After August, the amount of time Jupiter was available during darkness gradually decreased. The current project was conducted from late September to early November, when Jupiter was visible during darkness for approximately six hours each day. Since ionospheric effects take an hour or more to disappear after the Sun sets, the actual useable viewing time was around four to five hours each day. Figure 3 shows the altitude for Jupiter and the Sun on October 1, 2009 at the telescope location. Jupiter is relatively low in the sky this year with a maximum altitude of only ~37°. During the prime viewing time, approximately 02:30-07:00 UTC2 (19:30-00:00 local), Jupiter's average altitude is ~30°. As a result, any directional antenna should be focused around this altitude with maximum gain from approximately 30° left to 30° right of true south.

The power of the signal received by a radio telescope is often given by the antenna temperature in Kelvin (K). This "temperature" has nothing to do with the actual temperature of the antenna, but is instead the amount of power that would be emitted due to thermal noise by a resistor heated to the given temperature. 2 All dates and times given in this paper will be in UTC, unless otherwise indicated. California was on Pacific Daylight Time, GMT-7, during the project.


Page 2

Figure 1: The Greater San Francisco Bay Area surrounding the radio telescope installation (marked by the starburst). True north is indicated. Map from Mapquest (MapQuestweb).

The Radio JOVE Receiver and Data Recorder

The Radio JOVE receiver, designed by Richard Flagg of RF Associates in Honolulu, Hawaii, USA and sold by the NASA-funded Radio JOVE Project as an educational tool, was chosen for this project due to its low cost and special design for receiving Jovian radio emissions. The receiver is supplied in a kit consisting of over 100 different parts that must be assembled and calibrated (Figure 4). It is a directconversion receiver consisting of an initial wide-bandpass filter and preamplifier, a tunable (20.1±0.15 MHz) local oscillator, a narrow 5-pole bandpass filter with a bandwidth of ~3.5 kHz, and a set of audio amplifiers. The output is provided at two 1/8-inch jacks at a level sufficient to drive headphones, an amplified speaker, or the line input of a computer sound card (Flagg 2006). Because the receiver is designed for educational purposes and not as a scientific-quality instrument, no published technical specifications are available on frequency drift, thermal noise sensitivity, or out-of-band noise rejection. However, frequency drift at initial turn-on can be a few kHz in several minutes until thermal equilibrium is reached, and thermal gain drift is sufficiently small that no significant drift is detected at the power levels being used in this project (Flagg 2009).

Page 3

Figure 3: The altitude and azimuth of Jupiter (white arc) and the Sun (yellow arc) during a 24-hour period on October 1, 2009. The labels on the arcs are hours in UTC. The locations of Jupiter and the Sun at 00:30 are shown. Graph from Radio Jupiter Pro 3 (RJProweb).

Figure 2: Visibility (altitude > 0°) of Jupiter (blue-green) and the Sun (red) during 2009. Since Jupiter must be observed during darkness, the best observation time was during August and the worst was during February. The measurements in this paper were taken during October, when Jupiter had moderate observability. Graph from Radio Jupiter Pro 3 (RJProweb). Figure 4: The Radio JOVE receiver near final assembly.

Initial operational calibration was performed using an internal 20.0 MHz crystal oscillator. This oscillator provides the calibrated signal directly into the antenna inputs. First, the tuning knob was calibrated by tuning the local oscillator to match the test signal when the tuning knob was placed at the 10 o'clock position. Then the initial wide-bandpass filter and the preamp were tuned to maximize the delivered signal. At this point, the tuning knob tunes approximately 20.1 MHz when placed at the 12 o'clock position. A more detailed measurement of the exact frequency being tuned is not available. However, due to the wideband nature of Jovian and solar emissions, this is rarely an issue. A final operational test of the receiver was performed by attaching a simple long-wire antenna and detecting a strong carrier at 20.0 MHz, the frequency of WWV.3 However, the overlying voice modulation could not be heard, and this prevented the positive identification of the signal as coming from WWV. For data recording, the receiver was attached to a Lenovo ThinkPad X200s laptop computer. Because this computer has only a microphone (not line-in) input, an external PPA 1455 USB 6-channel sound adapter was used to provide line-in functionality with a linear response. Data were recorded using the Radio-SkyPipe II software available from Radio-Sky Publishing (RadioSkypipeweb). This software allows for calibration, recording, and display of multiple channels of audio and permits the recording of calibrated power simultaneous with the recording of high-fidelity audio for future playback and analysis.

WWV is a station operated by the U.S. National Institute of Standards and Technology. It broadcasts information about the current time, as given by atomic clocks, at 2,500 W from a transmitter in Fort Collins, Colorado (WWVweb).


Page 4

Power was computed by squaring the raw voltage values from the sound card with calibration factors included as described below.

The Antenna

The antenna used for this project was based on the recommended design from the Radio JOVE project (Radio JOVE Project Team 2004), which consists of two dipoles in a phased array. Because Jupiter is in the southern sky, and the greatest gain of a dipole antenna is perpendicular to the wire, the dipoles are mounted with the wires running east-west. The local streets, and the walls of the houses, in the area near the telescope are built on a true north-south-east-west grid, making alignment straightforward. The orientation of the antenna was verified by sighting off Polaris to the precision permitted by "eyeball measurement". Due to the poor directionality of the antenna, as discussed below, small errors in orientation would have a negligible effect. A single dipole in free space is an isotropic radiator in altitude with nulls off the long axis of the wire. However, when a real ground is included, the ground reflections produce an altitude asymmetry (Figure 5a) with the altitude of maximum gain and exact gain pattern dependent on the distance between the antenna and the ground. A particular antenna height can thus be selected to produce a desired altitude of maximum gain (Straw et al. 2007). A dipole can be made more directional by adding a second, parallel dipole in a phased array (Straw et al. 2007). This is the approach taken by the Radio JOVE design. The two dipoles, each of which consists of a pair of #14 gauge bare copper wires connected to insulators at each end, have a total end-to-end length of 7.09 m (making the half-wave dipole resonant at 20.1 MHz4). According to the original Radio JOVE design, the dipoles should be mounted on 6.1-m masts and placed 6.1 m apart perpendicular to the wires. The signals from the two dipoles are transmitted through 75- RG-59/U coaxial cable to a TRUSPEC DSU-2P power combiner (TRU-SPECweb). A W2DU-style balun (Straw et al. 2007) consisting of three ferrite toroid cores were used at each antenna feed point to eliminate signals that may be carried by the coax shield. The southern dipole has an additional 3/8- (3.69 m with a coax velocity factor of 0.66) of coax before the combiner, resulting in a 135° phase offset. This design results in a strong south-facing gain lobe (9.49 dBi) with an elevation of 35.0° and a south-to-north gain ratio of 1.79 dB (Figure 5b). As discussed earlier, the 35.0° elevation is ideal for receiving Jovian signals in the current epoch at the location used for this project. It is not as ideal for solar signals (since the Sun has an altitude of 40-50° during this period), but since they are correspondingly stronger they should still be detectable. The area in which the phased dipole array for this project was installed was unable to accommodate a full 6.1-m distance between the two dipoles and a distance of 5.7 m was used instead. This changed the gain pattern slightly (Figure 5c), resulting in a 9.62 dBi maximum gain at an elevation of 34.0° with a south-to-north gain ratio of 2.01 dB. On the surface, this would appear to be an improvement. However, the lower impedance (29.98 vs. 32.06 ) and thus greater mismatch with the 50- receiver, and higher resulting SWR5 (1.671 vs. 1.571), will likely negate any benefit.

In an ideal world, using infinitely thin wires, the antenna would be /2=7.463 m long. However, the finite diameter of the wires and capacitive end effects require the antenna to be approximately 5% shorter to be resonant (Straw et al. 2007). 5 SWR is the Standing Wave Ratio, a measure of the ratio of the maximum electric field strength in the cable to the minimum electric field strength, or, alternatively, the forward-transmitted power to the reflected power. High SWRs (>> 1) are due to reflections from mismatched transmission lines and result in radiative losses in cable transmission. These losses are greatest


Page 5


South max gain 7.18 dBi @ 37.0°, S/N = 0 dB, Z = 83.79-42.58j , SWR = 2.263 b)

South max gain 9.49 dBi @ 35.0°, S/N = 1.79 dB, Z = 32.06+3.29j , SWR = 1.571 c)

South max gain 9.62 dBi @ 34.0°, S/N = 2.01 dB, Z = 29.98+1.726j , SWR = 1.671 d)

South gain 8.71 dBi @ 37.0°, S/N = 0.48 dB, Z = 31.69+4.15j , SWR = 1.596

at high frequencies (e.g. UHF or VHF) and in high-loss cables. At the frequencies and cable lengths being used in this project, these modest SWRs should not result in significant cable loss (Ford 1994).

Page 6


South max gain 9.42 dBi @ 40.0°, S/N = 3.54 dB, Z = 112.5-25.38j , SWR = 2.390 f)

South max gain 4.97 dBi @ 30.0°, S/N = -5.58 dB, Z = 104.6­16.59j , SWR = 2.160

Figure 5: Antenna gain patterns for various designs. Elevation gain through the 0°-180° azimuth plane (left) and 3-D gain pattern (right) with south on the right-hand side in both cases. Also included are maximum south-facing gain and associated elevation, south/north maximum gain ratio, impedance, and SWR with a 50- load. a) A single dipole (6.1-m high); b) The original JOVE design (6.1-m masts, 6.1 m apart); c) The JOVE design as installed (6.1-m masts, 5.7 m apart); d) The JOVE design as installed (6.1-m masts, 5.7 m apart) with the power line 5.7 m south of the south dipole; e) Only the south dipole connected; f) Only the north dipole connected. All antennas were modeled using EZNEC Demo v5.0 (EZNECweb).

The actual installation area consisted of various additional obstructions (Figure 6) that must be modeled to produce an accurate gain pattern. One obstruction in particular is both important and relatively easy to model: a power line that is exactly parallel to the dipoles and located 5.7 m south of the southern dipole. The power line is attached at the east end, approximately 20 m away from the east end of the southern dipole, to the local power grid and at the west end, approximately 15 m away from the western end of the southern dipole, to a streetlight. In addition, the power line hangs in a catenary curve from about 7.5 m above ground at the streetlight to about 4.5 m above the ground in the middle. Because the power line is approximately the same distance from the southern dipole as the two dipoles are from each other, and the height is approximately the same as well, the power line acts as the reflector element of a Yagi. In a normal Yagi design, the reflector (which is a longer element than the driven element) is placed in the direction where the least gain is desired. In this installation, the maximum gain is desired to the south, which is also the location of the power line/reflector element. The result is that the directionality of the antenna is reduced as the array phasing and reflector effects cancel out, resulting in a gain of only 8.71 dBi vs. 9.62 dBi for the same antenna without the power line present (Figure 5d).

Page 7

Figure 6: An aerial view of the dipole antenna installation. The red lines illustrate the locations of north and south dipole. The long cyan line indicates the location of a twisted 120 VAC power line that powers a street light on the left side of the image. True north is indicated. Imagery data from Google Maps (GoogleMapsweb).

Figure 7: The altitude and azimuth of Jupiter (orange arc) and the Sun (yellow arc) during a 24-hour period on October 1, 2009 in relation to the antenna gain pattern from Figure 5d as seen from the radio telescope location. The labels on the arcs are hours in UTC. The locations of Jupiter and the Sun at 00:30 are shown. Graph from Radio Jupiter Pro 3 (RJProweb).

However, the directional effects of the power line do offer an opportunity that will be exploited later. If only the north dipole or only the south dipole is connected to the receiver and the other is allowed to act as a director in the Yagi design, then a significant change can be made in the directionality of the antenna (Figure 5e and f). This permits measurement of noise sources to be made in both the north and south directions with relatively high sidelobe rejection. In addition, it is possible to use the antenna with only the south dipole connected (Figure 5e) to receive Jupiter signals. The gain is higher than the normal phased array (9.42 dBi vs. 8.71 dBi) and the north-facing sidelobe rejection is higher, but the impedance is more poorly matched and the SWR is increased. In addition, the elevation of maximum gain is increased (from 35.0° to 40.0°), moving it further away from Jupiter's location. Because of these disadvantages, the use of this configuration was not explored for this project. Unfortunately, many of the other obstructions in the area are impractical to model with the information and software available for this project. These include the effects of the primary house structure, the surrounding trees, the imperfect ground plane, and the neighbors' houses. The effects of these obstructions will remain unmodeled for the remainder of this paper, and we must simply accept an additional, and unquantifiable, amount of error in future discussions. Photographs of the building and installation of the receiver and antenna are available on the author's web site (Photoweb).

Page 8


Calibration of the receiver and recording software to absolute antenna temperature requires two primary steps: the subtraction of the intrinsic noise level from the sound card and the calibration of the resulting sound card values to power antenna temperature based on a calibrated noise source. The sound card intrinsic noise level was detected by turning off the Radio JOVE receiver, with all cables still attached, and averaging the sound card values for 10 seconds. This average value was then subtracted from all future readings before computation of antenna temperature. The absolute antenna temperature was calibrated using a RF 2080F calibrated noise source produced by RF Associates (Flagg 2009). This device is a 25,000 K noise source that is connected directly to the input terminals of the Radio JOVE receiver. Calibration was performed in software by measuring the average sound card value for 10 seconds with the calibration device attached, the receiver tuned to ~20.1 MHz, and the receiver volume turned to the ½ scale position. The Radio-SkyPipe II software then computed an initial "power conversion factor" value, which assumes a linear response of the sound card to varying signal input. This initial conversion factor, however, only indicates the antenna temperature at the receiver antenna terminal, not at the antenna itself. In order to properly calibrate for the antenna temperature at the antenna feed point, we must further take into account the loss in the coax between the antenna and the receiver. The experimental setup used for this project included 14.78 m of RG-59/U coax between the antenna and the receiver. At 20.1 MHz, this coax has a loss of 1.5 dB per 100 feet (30.49 m) and thus the total cable loss is ~0.7 dB (Radio JOVE Project Team 2004). In addition, the power combiner used to combine the two antenna feeds has a loss of 0.6 dB at 20.1 MHz (TRU-SPECweb), resulting in a total loss of ~1.3 dB. This loss is taken into account in software, resulting in a final calibration that records a 25,000 K antenna temperature corresponding to the same temperature detected at the dipole feeds. It is important to note that an explicit assumption is being made that the sound card used has a linear response across a wide range of sound levels. In the ideal case, the calibration would be performed at a wide range of antenna temperatures, resulting in a calibration curve-fitting function instead of a linear approximation. Unfortunately, no such calibration equipment was available. However, the calibration noise source used is close to the known galactic background noise (~40,000 K) and thus measurements near this noise floor should be reasonably accurate. For the purposes of this project, the absolute calibration of measurements at higher antenna temperatures is not critical. It is most important, instead, to measure large spikes in the noise background, such as those emitted by Jupiter or the Sun, which can have an equivalent temperature several orders of magnitude higher than the galactic background (Flagg 2000).

Noise Sources

The presence of high levels of background noise near 20.1 MHz was obvious immediately after starting measurements. While highly variable with time, the average background at night was ~300,000 K, and the average background during the day was several million K. This section will discuss the attempts made to isolate the sources of noise.

Page 9

Figure 8: Mean noise from 03:00-12:45 (night, left hashed bars) and from 15:45-00:00 (day, right hollow bars) with 1- error bars.

Figure 9: Noise measurements around sunrise on October 4, 2009. Approximate sunrise (14:06) is marked in red.

Figure 10: Noise measurements around sunset on October 5, 2009. Approximate sunset (01:46) is marked in red.

Diurnal Patterns

The most obvious background variation was diurnal. Figure 8 shows the dramatic difference in mean temperature between day and night measurements. To avoid the day-night transition time, each measurement excludes the time approximately 1.5 hours on either side of sunrise or sunset. Clearly, daytime correlates with a greatly higher noise floor. But what causes this increase? One possibility is that the interference is caused by local human activity, since people are generally more active during the day and are probably driving more cars, using more heavy equipment, and using more computers. While this may contribute to the problem, it is clearly not the main source. This can be seen by looking at the noise floor during the period several minutes either side of sunset or sunrise (Figure 9 and Figure 10). The interference changes abruptly around this time, and it would be difficult to argue that human activity follows exactly the same pattern, especially in a large urban area where people work a wide variety of schedules.

Page 10

A second possibility is that the interference is coming directly from the Sun. The Sun emits bursts of radio energy across a wide range of frequencies (Burke & Graham-Smith 2002). However, most of these solar bursts are narrow-band emissions of short duration, which does not explain the uniformly increased noise floor. A more reasonable explanation is that this is an ionospheric effect. The ionosphere is that region of the Earth's atmosphere above ~50 km where solar radiation (in the form of ultraviolet light and X-rays) is sufficient to ionize atoms and molecules. The structure of the ionosphere is complex and varies greatly over time, but is generally divided into three distinct layers. The lower two layers, D and E, are unlikely to have a significant effect on this project because they only affect much lower frequencies. However, the top layer, F, will have a significant effect. The F layer, which exists from 160 km to more than 500 km above the Earth, is tenuous due to the low air density and is highly ionized. The main effect of the F layer is to reflect terrestrial signals back to Earth while preventing the transmission of extraterrestrial signals. This works over a wide range of HF frequencies and is very effective at ~20 MHz. The exact ability to reflect, and the geometry of the reflection, is highly variable based on the angle and amount of solar incidence and the molecular content of the F region. This effect is most likely responsible for the bulk of daytime interference in this project as manmade signals from around the world are reflected onto the radio telescope in a random and timevarying fashion. During periods of high solar activity, the F layer can remain ionized for several hours after sunset. However, solar activity was at a minimum during this project, and thus the F layer quickly lost its ionization immediately after sunset. The F layer will also reflect extraterrestrial signals and prevent them from reaching the Earth's surface. As a result, Jovian emissions are not generally detectable during daylight hours, regardless of the noise background. Jovian studies are conducted mainly at night, and, since Jovian emissions are less intense than solar emissions and more likely to be covered by noise, the remainder of this section will deal solely with nighttime noise sources.

Self Interference

Self interference is generally defined as interference that is caused by the facility that is actually operating the radio telescope. In this case, I will define it as interference originating from the house where the telescope is located. Such interference is often caused by low-power devices, such as microwave ovens, computers, printers, disk drives, CRT monitors, and lab equipment, that nonetheless emit signals strong enough to be detected by a nearby antenna (Millenaar & Stiepel 2004). A thorough search for self interference was conducted the night of October 9-10, 2009. All circuit breakers in the house were turned off, along with all uninterruptible power supplies. The Radio JOVE receiver was powered by a 12-V deep-cycle marine battery, and the recording laptop was powered by its own internal battery. The circuit breakers were then turned on one at a time and the noise level was measured immediately before and immediately after power was applied. For circuits where multiple pieces of potentially interfering equipment were present, the individual pieces of equipment were powered down and then turned on one at a time. The ambient background with the entire house turned

Page 11

off was 552,610±147,202 K6 for the two minutes prior to the experiment, making it a particularly noisy night. In the end, it was found that the majority of household equipment, including heavy appliances (e.g. hot tub and swimming pool pumps, refrigerator, washer, and dryer) and computer equipment (desktop computers, laptops, printers, network switches, and wireless routers) did not contribute noticeably to the noise level and no increase in noise was detectable above the existing background variations. In addition, the contribution of the adjacent low-pressure sodium streetlight was measured immediately before and immediately after it turned on with no difference in noise level. However, a major source of noise was discovered in an unexpected place: the charging unit for the iRobot Scooba® 380 floorwashing robot (Scoobaweb). The 10-second average with the Scooba® charging unit unplugged was 804,338±69,068 K, and the 10-second average with the charging unit plugged in was 1,732,361±132,082 K, an increase of over 900,000 K. The charging unit was left unplugged for the remainder of the project. The internal noise contributed by the receiver and sound card was easy to quantify. The sound card noise was measured during the calibration process, with the receiver turned off, and this average background was subtracted from all future measurements. The receiver noise was measured by disconnecting the antenna and attaching a dummy load to the antenna port. The average noise was ~2,400 K, several orders of magnitude below the general measured noise level.

Neighborhood Noise

The night of October 7-8, 2009, the radio telescope location and surrounding neighborhood experienced a power outage lasting several hours. Power distribution in the vicinity does not follow a regular layout and the area without power had an irregular shape spanning several blocks. In addition, any batterypowered items, including laptop computers and mobile phones, would still be emitting signals. Nevertheless, the power outage provided a unique opportunity to determine the importance of local interference. During the power outage, radio measurements continued through the use of an uninterruptible power supply. Power to the area was restored at 03:29:50, and at 03:32:45 the main circuit breaker to the house was pulled, thus allowing measurement of neighborhood interference without contamination from the house. The 5-minute noise immediately before power was restored was 326,164±36,561 K, and the 5-minute average immediately after the house power was disconnected was 362,518±64,753 K. Thus, the neighborhood would appear to contribute ~36,000 K to the antenna temperature, but the error bars are sufficiently large that the measurements are also consistent with no change at all. Either way, the neighborhood is clearly not the primary contributor to the background noise.

Regional Noise

With solar effects, self interference, and neighborhood interference eliminated as primary noise sources, the only remaining sources of noise are local battery-powered devices and manmade interference outside of the immediate vicinity of the telescope. The probability of the primary interference coming from battery-powered devices was reduced by moving these devices to different locations in the house,


Measured temperatures will be presented as the mean over a specified time period followed by the 1- standard deviation over the same period.

Page 12

including very close to the antenna, without noticeable change in the detected noise level. Interference from other manmade sources outside of the local neighborhood seems highly likely, given the population and level of industry in the area, but is by no means a foregone conclusion. As shown in Figure 1, the majority of potential interference sources are to the north and east of the telescope. Thus, one way to implicate manmade inference is to find a strong directional bias to the interference. An attempt was made to determine the location of the interference by using the directional biases of the dipole array as shown in Figure 5e and f. Measurements were taken October 17, 2009 from 02:21-02:37. Using the configuration in Figure 5e, with only the south dipole connected, a 90-second measurement produced a temperature of 434,112±164,170 K with a south-facing gain of 9.42 dBi, while using the configuration in Figure 5f, with only the north dipole connected, produced a temperature of 898,569±247,962 K with a north-facing gain of 10.55 dBi. Considering the gain difference, the northfacing configuration produced a temperature ~2.0 dB higher. While this difference is suggestive of a north bias to the interference, the complexities of accurately modeling the environment around the dipole antennas are beyond the scope of this project, as described earlier, and additional evidence is needed. To provide this evidence, two local amateur radio sites were enlisted: the Stanford Amateur Radio Club (W6YX) located in Palo Alto, CA (about 13 km to the northwest of the dipole array telescope) and the home of Dewey Churchill (KG6AM) located in Sunnyvale, CA (about 50 m north of the dipole array). W6YX has a tri-band (10/15/20 m) Force-12 C-31XR Yagi antenna (Force12web) mounted on a 10-m mast. The antenna, which has four elements on 15 m (~20 MHz), a forward gain of 14.3 dBi, and a front-to-back gain ratio of 22 dB, is located on the top of a hill ~150 m above sea level with an unobstructed view of the entire Bay Area. Measurements were taken on October 21, 2009 04:16:10-05:24:32. KG6AM has a tri-band (10/15/20 m) Mosley MP-33 Yagi antenna (Mosleyweb) mounted on a 2.75-m mast on the roof of a single-story house for a total height of 10 m. The antenna, which has three elements, a forward gain of 9.0 dBi, and a front-to-back gain ratio of 20 dB, is located at approximately sea level with a poor view of the Bay Area. Measurements were taken on October 22, 2009 03:33:40-04:38:18. Each antenna can be rotated a full 360° in azimuth. The differences in altitude gain pattern were not modeled. In both cases, the Yagi was attached to the Radio JOVE receiver and rotated in 15° increments while taking two-minute signal averages at each position (Figure 11). The W6YX measurements were taken at 20.1 MHz. However, the KG6AM site showed an intense (~2,000,000 K) noise source, which had the distinctive audio buzz of an arcing power transformer, in all directions at 20.1 MHz. Because of this, the KG6AM measurements were made at ~20.2 MHz, which seemed mostly free of local electrical interference. Due to unknown differences in cable loss, the resulting measurements needed to be calibrated to absolute antenna temperature. To do this, a conversion factor was computed by taking the average of the Yagi measurements for each antenna (excluding the 75°-105° and 255°-285° bearings to approximate the east-west nulls of the dipole) and dividing by the antenna temperature measured at approximately the same time by the dipole array. This conversion factor is only approximate, due to the natural variation in the background noise during the time required to transport the Radio JOVE receiver from the Yagi site to the dipole site and the approximate modeling of the dipole directional gain, but should be at most 10-20% off. Absolute errors are not important here, however, since only the relative signal strength is used to determine directionality.

Page 13

Figure 11: Noise measurements from W6YX vs. true bearing from the antenna taken October 21, 2009 04:16:10-05:24:32.

Figure 12: Noise measurements from KG6AM vs. true bearing from the antenna taken October 22, 2009 03:33:40-04:38:18.

Background variation likely occurred during the time required to make a complete 360° sweep (~50 minutes). To quantify this variation, at W6YX eight equally spaced bearings were pre-sampled ~45 minutes before the primary measurement. The maximum difference between these pre-samples and the corresponding later measurement was 9.4% and thus we can place an approximate bound of 10% on the amount of variation during the sweep. A similar process was followed at the KG6AM site, with several directions visited multiple times during the one-hour observation time to verify that the readings did not vary by more than 10%. The W6YX measurements (Figure 11) show a clear peak in the range 60-75° with the noise level dropping by 3 dB at ~30° and ~105°. A second, diffuse, peak is evident at ~240°. Because the second peak is 180° out of phase with the main peak and this direction is pointed towards the Pacific Ocean (which is unlikely to be a major source of radio noise), it is most likely caused by the rear-facing gain lobe of the Yagi antenna. The KG6AM measurements (Figure 12) are much noisier and show a broad peak in the range 45°-90°. The peak is not sufficiently high that a 3 dB drop is a practical measure. Instead, I arbitrarily chose a 1 dB drop to define the extended peak, and this runs from 15°-105°. No second peak is obvious. Potential errors due to background variation and imprecise calibration for absolute temperature are not included in the error bars in Figure 11 and Figure 12. The 13 km distance between the two antenna sites allows a loose form of triangulation. Figure 13 overlays the peak noise regions on a map of the Bay Area. The intersection of the core peaks is in uninhabited park land. However, Mission Peak, located in the upper-right corner of Figure 13, contains a large number of transmitting and relay towers. While none of these transmitters is likely to be targeting 20 MHz specifically, it is possible that unfiltered noise from a transmitter is causing the main interference. It is also possible that the noise is coming from northern San Jose, which is in the overlap region of the two Yagis, although well outside of the core of W6YX's peak detection. Finally, it is possible that either W6YX or KG6AM were near local noise sources that distorted their measurements. In order to isolate the noise source further, a third triangulation would need to be performed, ideally from Fremont (top center of Figure 13). However, in the absence of such additional information, the

Page 14

Figure 13: Triangulation of noise from the W6YX (green) and KG6AM (red) sites. The core peak area and 3dB (W6YX) or 1dB (KG6AM) down areas are marked. Map from Mapquest (MapQuestweb).

most likely conclusion is that the interference is not coming from the local house or neighborhood, but instead is coming from a source farther away and not under the control of the author.

Detection of Jovian Emissions

Jupiter emits two distinct types of signals from its magnetic storms that are distinguished most easily in an amateur setup by sound: S-bursts, which sound like pebbles hitting a tin roof, and L-bursts, which sound like waves crashing against a shoreline. Neither type is emitted continuously. Although the exact cause of the different signal types and their times of emission is unknown, tens of thousands of observations have allowed an empirical model to be developed to predict when these storms are detectable on Earth. These models are based on two primary inputs: Jupiter's System III longitude (which is based on a 9 hour, 55 minute, 29 second rotation period corresponding to Jupiter's magnetic field) that is pointed at Earth, and the orbital position of Jupiter's moon Io relative to a line between Jupiter and Earth. Taking Jupiter's longitude into consideration, there are three regions during which Jupiter's emissions are likely to be detectable on Earth (called A, B, and C), with each pointing towards the Earth every ~10 hours. Each region has both an Io-dependent and an Io-independent portion. However, even when Jupiter (and Io, if necessary) is properly aligned, the probability of detection can still be fairly low (< 30%) and thus observations may need to be made over time to actually detect a magnetic storm (Rogers 1995; Flagg 2000).

Page 15

Figure 14 shows an example of a potential S-burst that was selected by ear because it sounded similar to S-bursts recorded at other observatories (AstroSurfweb). Unfortunately, it is nearly impossible to differentiate between a Jovian emission and other types of interference, even by ear, without either using a spectrograph or correlating the observation with other stations that have similar visibility but different noise sources. Neither approach was practical for this project. Instead, a statistical approach was taken.

Data were collected over 16 nights. The nights were divided into two time periods: 0300-0630, when Jupiter was near the meridian and highest in the southern sky, and 0945-1245, when Jupiter was below the horizon. For each time period for each night, an automated procedure was used to detect events. An event is defined as a signal of greater than 3,000,000 K lasting for at least ½ second. If multiple events occurred near each other, only a single event was recorded if the detections were less than five minutes apart. Storm predictions were performed using Radio Jupiter Pro 3 (RJProweb). Assuming the predictions of Jovian storms are accurate, there should be a noticeable increase in detections during times when storms are predicted versus times when they aren't predicted. In addition, there should be a significant increase in detections during times when Jupiter is above the horizon versus below the horizon. The results are presented in Table 1. A total of 83 events were detected during 41.8 hours of predicted storms, yielding a rate of 1.99 events per hour. A total of 17 events were detected during 14.2 hours of predicted quiet, yielding a rate of 1.20 events per hour. Finally, a total of 19 events were detected during the 56.0 hours Jupiter was below the horizon, yielding a rate of 0.34 events per hour.

Figure 14: A series of potential S-bursts, each ~1-5 ms in duration, on October 6, 2009 at 05:31. The plot covers ~60 ms.

As hypothesized, there is a noticeable bias (1.66×) of detected events during times when Jupiter is above the horizon and Jovian storms are predicted versus when Jupiter is above the horizon but storms are not predicted. While it is not possible to determine which of those detected events actually originated from Jupiter, it is nonetheless compelling evidence that some of them were from Jupiter. There is an even greater bias (5.85×) of detected events during times when Jupiter is above the horizon and Jovian storms are predicted vs. when Jupiter is below the horizon. There are two possible explanations for this much larger bias. The first is that Jupiter is emitting radio noise even when storms are not predicted. The second is that the general level of terrestrial interference is simply lower much later at night. This could be due to reduced human activity or further deionization of the ionosphere longer after sunset.

Detection of Solar Emissions

As discussed earlier, daytime interference at 20.1 MHZ is extreme. Figure 15 (left side) illustrates the noise during 3.5 hours on a randomly selected day. Dramatic noise events occur on a regular basis. The details of one such event are shown in Figure 15 (right side). These bell-shaped noise events are common and their origin is unknown. As these types of noise do not occur at night, the most likely possibility is variability in the ionosphere reflecting strong, distant noise sources. Unfortunately, the presence of these events makes the detection of solar emissions nearly impossible. In addition, the Sun was quiescent during most of this project, and the WAVES experiment on the STEREO spacecraft (STEREOweb) did not record any significant high frequency (> 10 MHz) solar radio emissions during

Page 16

Jupiter Above Horizon Date 10/2 10/3 10/4 10/6 10/7 10/15 10/18 Detected Events 05:01:30 05:30:32 Predicted Emissions 0300-0347 0614-0630 0300-0630 0300-0525 0300-0630 0521-0630 0300-0630 0300-0630 Det. Predicted # Hrs 1 1.0 0 3.5 0 2.5 4 3.5 2 1.1 6 10 3.5 3.5 Not Predicted # Hrs 2 2.5 0 0.0 0 1.0 0 0.0 4 2.4 0 0 0.0 0.0

Jupiter Below Horizon Detected Events # 0 0 0 0 0 3 0 Hrs 3.5 3.5 3.5 3.5 3.5 3.5 3.5


10/19 10/20

04:17:00 03:05:09 05:33:38 03:19:51 05:58:49 03:34:31 04:59:49 06:15:31 03:25:03 04:48:05 03:06:26 04:01:24 04:48:00 05:33:17 06:09:28 03:27:19 04:09:32 04:51:18 04:36:18 03:03:11 03:51:13 04:29:12 05:23:40 03:01:13 03:59:11 04:46:15 05:39:13 03:00:09 03:44:08 04:30:04

05:03:34 03:24:57 05:44:05 03:53:48 06:17:44 03:58:11 05:37:54 03:45:30 03:12:21 04:07:19 05:01:22 05:45:54 06:25:20 04:21:00 04:58:41

05:16:58 04:01:32 04:07:07 04:13:21 05:43:46 04:05:50 03:33:25 04:26:37 05:07:18 05:55:20

05:31:02 04:59:11 04:41:57 04:21:18 05:56:34 04:22:53 03:39:20 04:36:19 05:27:21 06:01:16

10:55:37 11:10:29


0511-0630 0300-0630

0 18

1.3 3.5

5 0

2.2 0.0 09:33:24 10:09:23 10:45:22 11:21:21 11:57:20 12:33:19 09:50:15 08:33:56 09:39:20 10:15:19 10:51:17 11:27:17 12:03:16 12:39:15

0 12

3.5 3.5

10/23 10/24 10/25 10/26 10/28


0300-0600 0418-0630 0300-0630 0300-0329

3 1 1 1 12

3.0 2.2 3.5 1.0 2.1

0 1 0 1 2

0.5 1.3 0.0 2.5 1.4

0 1 1 0 0

3.5 3.5 3.5 3.5 3.5

05:58:24 03:16:17 04:02:17 04:44:16 05:40:36 03:23:16 04:20:14 04:52:10 05:58:01 03:07:05 03:56:08 04:46:16 03:23:17 04:08:39 04:51:12 03:30:12 04:26:10 05:00:12 03:27:09 04:02:04 05:21:14 03:35:53 04:21:16 05:01:13 03:52:15 04:39:18 05:17:39 03:34:05 04:24:08 06:20:33





















Total Rate (events per hour)

83 1.99


17 1.20


19 0.34


Table 1: Result of 16 nights of measurements. Jupiter above horizon events cover 03:00-06:30, and Jupiter below horizon events cover 09:15-12:45. Detected Events presents the list of discovered events during each period. Predicted Emissions shows the times during which any Jovian emissions are predicted (A, B, or C, with or without Io). Det. Predicted shows the number of detected events that occurred during predicted storms and the number of hours of storms that were predicted. Not Predicted shows the number of detected events that occurred outside predicted storms and the number of hours that storms were not predicted. Rate shows the number of detected events per hour for the categories detected and predicted, detected but not predicted, and detected but Jupiter below the horizon.

Figure 15: A sample of daytime noise on October 3, 2009. The left figure covers approximately 3.5 hours and the right figure is the details of the large spike from 20:51 to 20:59.

Page 17

the month of October. A rare potential solar burst was detected at 16:08:50 on October 31, 2009 by the LGM Radio Observatory in Alachua, Florida (Greenman 2009) and reached a temperature of ~570,000 K but was not detected by this project's radio telescope. As a result of these issues, it is reasonable to say that no solar emissions were detected during the duration of this project.


The amount of noise present in an urban environment, coupled with an antenna with poor directionality, makes the detection of Jovian emissions nearly impossible. However, even in a radio-quiet environment, there is no accepted means of distinguishing Jovian emissions from local interference. As a result, a rigorous investigation of Jovian emissions would require three main attributes: a radio-quiet environment, a highly directional antenna, and the ability to correlate with other observatories. A radio-quiet environment is a necessary prerequisite to any viable observations. Even the use of a directional antenna cannot fully eliminate the detection of noise through side lobes or local environmental reflections. Jovian emissions are often only slightly more powerful than the galactic background. This means that the observatory must have a local noise floor that is at or below the galactic background level. A highly directional antenna is necessary to provide evidence that detected signals are actually from Jupiter and not from other, off-axis sources. Such an antenna combined with a radio-quiet environment should provide sufficient sensitivity and selectivity to detect Jovian emissions with high confidence. As a final check, detected signals can be correlated with other observatories. Such observatories must also be in radio-quiet locations, ideally have directional antennas, and be in a similar time zone and at a similar latitude to provide an equivalent time span during which Jupiter is visible. Correlated signal detection provides strong evidence that the source of the signal is not local. However, absence of such correlation does not guarantee that the signal is local. The non-uniform nature of the ionosphere can prevent the detection of Jovian emissions at different locations at the same time. Despite the difficulties in isolating the origin of individual signals, it nevertheless appears to be possible to show, in a statistical sense, that some of the detected signals are of Jovian origin. This can be done, as was done in this paper, by comparing the number of detected events during periods when emissions were or were not predicted, or when Jupiter was simply below the horizon. A more thorough investigation could be conducted if more nights of data were available by performing the same basic comparison against individual predictions for Jupiter A, B, and C emissions (and their Io counterparts). However, such predictions are still probabilistic in nature. To make the statistics more rigorous, observations from other sites could be used to determine, in a general sense, when emissions are actually taking place. In addition, it is not possible with the current data set to provide error bars for the statistics. In order to compute error bars, it would be necessary to be able to predict the number of background events in a given amount of time. These events come from currently unknown terrestrial sources and are undoubtedly affected by time of day, time since sunset, and weather. An extended observation campaign across a wide range of seasons would need to be performed in order to build up a sufficient database with which to predict background event counts. Detection of solar emissions has similar issues. Because the Sun's maximum altitude is often different from Jupiter's maximum altitude, any directional antenna should be adjusted to account for the difference. In addition, it is obviously more reasonable to attempt the detection of solar emissions during a time of high solar activity. At the end of 2009, the Sun is just beginning to emerge from an unusually long dormant period, and the activity maximum should next peak around 2012. Page 18


A simple radio telescope consisting of a Radio JOVE receiver and a phased-array dipole was constructed to detect Jovian and solar emissions at 20.1 MHz. The telescope was placed in an urban environment. Noise from undetermined sources in this environment made detection and isolation of Jovian and solar emissions difficult. Significant effort was expended to isolate the origin of this noise, and it was determined that the majority of the noise was coming from non-local sources outside of the author's control. Rather than detecting individual Jovian emissions, a statistical approach was taken with success. The number of detected signal spikes was significantly higher during times when Jupiter was above the horizon and predicted to be emitting signals. This showed that, while the background noise made the identification of individual signal bursts difficult, Jovian emissions were likely being received. The same could not be said, however, of solar emissions. The STEREO spacecraft did not detect solar emissions in the 20 MHz range during the duration of the project, and thus it is clear that any detected signals were simply interference. A variety of improvements to the experimental setup could be made to improve future results. These include relocation to a radio-quiet area, use of a highly directional antenna, and correlation with other observatories. Without these improvements it will be impossible to tell for certain whether or not a particular signal detection is of extraterrestrial origin or is simply local interference.


Any observation project is a cooperative effort, and this project could not have been completed without the help of the following people. CJ Smith and Patti Andrews provided many hours of labor installing the dipole antennas. Sawson Taheri (KG6NUB) and the Stanford Amateur Radio Club (W6YX) provided access to their Yagi for noise measurements. Dewey Churchill (KG6AM) provided access to his personal Yagi for noise measurements. Dave Typinski (AJ4CO) and Wesley Greenman provided valuable data from their Radio JOVE installations in Florida for comparison. Chuck Higgins shipped the Radio JOVE kit quickly so that the project could start on time. Richard Flagg of RF Associates, designer of the Radio JOVE receiver, answered many questions about the receiver's specifications and noise calibrations. Jim Sky of Radio-Sky Publishing answered questions about the Radio-SkyPipe software. CJ Smith and Dave Typinski provided valuable comments on early drafts of this paper. Finally, my Swinburne project advisor, Eduardo Manuel Alvarez, provided continuous and valuable guidance throughout the project.


AstroSurfweb: Audio files of Jupiter and Saturn emissions, (accessed 2 November 2009) Burke, B. F., & Graham-Smith, F. 2002, An Introduction to Radio Astronomy (2nd ed.; Cambridge: Cambridge University Press) Dulk, G. A., Leblanc, Y., Sault, R. J., & Bolton, S. J. 1999a, A&A, 347, 1039 Dulk, G. A., Leblanc, Y., Sault, R. J., Bolton, S. J., Waite, J. H., & Connerney, J. E. P. 1999b, A&A, 347, 1029

Page 19

EZNECweb: EZNEC Antenna Software by W7EL, (accessed 18 October 2009) Flagg, R. 2006, JOVE RJ1.1 Receiver Kit: Assembly Manual (Murfreesboro, TN: NASA Jove Project) Flagg, R. 2009, personal communication Flagg, R. S. 2000, Listening to Jupiter: A Guide for the Amateur Radio Astronomer (Louisville, KY: Radio-Sky Publishing) Force12web: Force 12, Inc. Amateur and Commercial Antennas and Towers, (accessed 25 October 2009) Ford, S. 1994, QST, April, 70 GoogleMapsweb: Google Maps, (accessed 27 October 2009) Greenman, W. 2009, personal communication MapQuestweb: MapQuest, http:/// (accessed 27 October 2009) Millenaar, R. P., & Stiepel, H. J. 2004, On Self-Generated RFI at Radio Astronomy Sites (Westerbork, The Netherlands: WSRT) Mosleyweb: Mosley MP-33 Antenna, (accessed 29 October 2009) Photoweb: Photographs of the Radio JOVE project construction and installation, (accessed 26 October 2009) Radio JOVE Project Team 2004, RJ1.2 Antenna Kit: Assembly Manual (Murfreesboro, TN: NASA Jove Project) RadioSkypipeweb: Radio-SkyPipe Strip Chart Program, (accessed 20 October 2009) RJProweb: Radio Jupiter Pro 3, (accessed 18 October 2009) Rogers, J. H. 1995, The Giant Planet Jupiter (Cambridge: Cambridge University Press) Scoobaweb: iRobot: iRobot Scooba® 380, (accessed 20 October 2009) STEREOweb: STEREO / WAVES Experiment, (accessed 2 November 2009) Straw, R. D., Cebik, L. B., Halliday, D., Jansson, D., Lewallen, R., Severns, R., & Witt, F. 2007, The ARRL Antenna Book: the Ultimate Reference for Amateur Radio Antennas, Transmission Lines Page 20

and Propagation (Newington, CT: American Radio Relay League) TRU-SPECweb: TRU-SPEC DSU-2P Power Passing Splitter, (accessed 20 October 2009) WWVweb: NIST Radio Station WWV, (accessed 3 November 2009)

Page 21


Microsoft Word - HET608-Project2.doc

21 pages

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate


You might also be interested in

URSI General Assembly 2008
Microsoft Word - HET608-Project2.doc