Aircraft and Satellite Echoes from TV transmitters at VHF and UHF

©Ian Roberts, ZS6BTE. Randburg, South Africa.  February 2010

 

The investigation discussed below is aimed at detecting aircraft and low altitude satellites using the South African Broadcasting Corporation’s VHF and UHF TV transmitters. These high power transmitters (EIRP around 162 kW) are scattered around the country on VHF and UHF and appear ideal for illuminating aircraft and satellites.

See Table 1 below for reference regarding the transmitters used.

 

TV Transmitter Exact Frequency Control

To reduce mutual interference to the TV screens of viewers, the SABC has the various TXs at offsets of a nominal -26.025 (20m) or +26.025 kHz (20p) from the “0” offset channel centre frequency. This is further refined into a “precision offset” regime where the TXs are again offset under experimentation to reduce this interference to a minimum. For example the Kimberly ch 4 TX on 20m is not at 175250000-26025 Hz (=175,223,975 Hz) but at 175223996 Hz (or +21 Hz) from the nominal offset of 20m.

All the VHF TXs at the 20m offset are set to this precision offset under rubidium oscillator control. From my home I can only confirm the strongest of these TXs, the rest are masked on the exact frequency.

However, by examining aircraft reflections from these unseen “partner” TXs, one should be able to confirm they are on these frequencies. This is because the aircraft reflections should show simultaneous positive and negative Doppler shifts on the TX carrier frequency, and its 50 Hz sidebands if powerful enough.

 

Most of the high power VHF TXs in South Africa are on this precision offset scheme because the separation distances are not great for VHF TV broadcasting. Particularly prone to interfere on viewer’s screens are neighbouring coastal TXs due to tropospheric ducting along advancing cold fronts: East London, Port Elizabeth, Durban, Villiersdorp, van Rhynsdorp. And if the cold front goes inland, interference to those viewers as well.

The rubidium oscillator controlled precision offset frequencies for Band 3 ch 4 (20m, “0”, 20p) are in Table 1 below (these are my measurements off-air on long tropospheric scatter paths using a precision oscillator as reference – they are not official frequencies).

 

Ch

Site, frequency offset, direction and distance  (20M = -26.025 kHz; 20P = +26.025 kHz)

Polarisation

Service, Frequency and Date measured         

4

175.250

Potgietersrus 20p

Welverdiend 0

East London 20m

Port Elizabeth 20p

Queenstown 0

Durban 20p

Kimberley 20m

Beaufort West 20p

Oudtshoorn  0

Van Rhynsdorp 0

Villiersdorp 20p

Elands Heights

30°/245km

260°/84km

181°/760 km

195°/902km

190°/634 km

147°/492km

223°/428km

216°/864km

212°/1003km

233°/1097km

221°/1192km

distant low power

H

H

H

H

H

H

H

H

H

H

H

V

TV2         175.276089.5 01/01/10

TV1         175.250093.3 25/01/10

TV3         co-channel 20m

TV1         co-channel 20p

TV1         co-channel “0”

TV2         co-channel 20p

TV2         175.223996.0 22/01/10

MNET     co-channel 20p

TV3         co-channel “0”

TV1         co-channel “0”

MNET     co-channel 20p

TV1

Table 1: Frequency allocations on VHF channel 4. Directions and distances from Randburg. Reference TXs, dates and frequencies in red

 

Tropospheric Scatter

Radio frequencies propagate further than the visible horizon, typically to k4/3 radius of the earth (i.e. 1.3x), Figure 1. In Southern Africa this value seems to be a bit lower and is around 1.1-1.2, possibly due to the drier climate.

Figure 1: Bending of radio beam due to refraction (a = true Earth’s radius)

 

Figure 2: Tropospheric Scatter Path Geometry

In Figure 2 the zero elevation lines between the TX and RX intersect and form the Scatter Angle. The smaller the better. Each additional 1 degree in the scatter angle increases the already high path loss due to the obstructed path by another 10 dB. In professional practice this is kept below 4 degrees if possible.

The “volume” around the intersect point is referred to as the “scatter volume”. The properties of the scatter volume have an enormous effect on the quality of the tropo scatter path. Typical height of the scatter volume is listed in Table 2.

 

Distance, km

Height of Scatter Volume, m agl

Favourable vs. unfavourable paths

150

300-2000

300

600-3000

600

3000-20000

Table 2: Height of the scatter volume, meters above smooth ground level

 

Detection of Aircraft and Satellites

To better understand how all this fits into the detection of aircraft and satellites refer to Figure 4: Detection Geometry.

 

 

Figure 4: Detection Geometry – side view

 

Here a similar situation exists. X is the scatter angle with associated scatter volume, while TX and Home represent the TX and RX sites separated in the scale drawing by 1000 km. Two arcs need to be discussed: ARC A at 500 km is optimistically taken as the satellite height above ground – there are relatively few satellites this low – fortunately the huge International Space Station complex is at 350 km above ground.

ARC B represents the maximum height of aircraft, around 35000ft (10.7 km).

 

Suitable TV Channel to Use, Receiving System Requirements

VHF is advised because of the RF advantages.

A TV channel occupied locally cannot be used due to in-channel noise and there is likely to be a high-power TV TX on an adjacent channel in the same polarisation.

It follows those scanner-type receivers with wide front ends cannot be used as they will overload, as will an unfiltered preamplifier. The wanted signal level is barely above 0 micro volts into 50 ohms….

The required dynamic range can be as high as 95-100 dB depending on distance to the local TV TX. The dynamic range of scanners varies from 45-60 dB – they need filtering if they are to be used otherwise the observed result is an inter-modulated mess.

The ambient RF noise level is extremely high, as in all city localities with high power, in-band TV transmitters and computer hash. The receiving system’s noise temperature can be several hundred or even several thousand Kelvin; preamps at the antenna might help provided they are filtered and are not driven into gain compression by excessive input levels.

I used an ICOM IC-R8500, on VHF a home-brew log periodic antenna with 11.5 dBi of gain on TV ch 4, and low loss cable, no preamp. This setup allows continual detection of the weak Kimberley ch 4 transmitter distance 428 km, for example, over an amateur style tropospheric scatter path. The IC-R8500 has a dynamic range of about 80 dB – it is not enough under all circumstances and spurious results may be observed.

 

Signal Processing

The sound card based Spectrumlab software was used in waterfall mode and an FFT of 262k at a sound card sampling rate of 44.1 kHz. This reduces the noise floor by nearly 30 dB compared to “straight” analogue detection. This is a colossal improvement in receiving system performance. However, the sound card/sound chip should be able to maintain reasonable performance over a dynamic range of about 70-80 dB when the input volume is set correctly. If not, severe over-modulation of the waterfall trace will result, causing false screen image effects.

 

Questions to Answer, Facts to Confirm, Assorted Information

1.     An important question is: can an observer at Home detect aircraft above TX, the TV transmitter, in Figure 4? The answer is “no”, see point 1 Observed Data below.

2.     Conversely, can an observer at Home detect aircraft above Home using the TX radio beam? The answer is “yes”; see point 2 Observed Data below. In Table 2, the scatter volume heights available to amateurs are the unfavourable values as amateurs are unable to utilize mountain tops for TX and Home and will typically have obstructions to the horizon a few km away, increasing the required scatter angle and height above ground at X.

3.     It follows therefore that aircraft at X, even on a shorter path of 600 km for example, have to be at the maximum height of X above ground to reflect the signal from TX to Home.

4.     Thus, it seems that in Figure 4, aircraft within a couple of hundred km from Home in the direction of TX (Table 2: Height of Scatter volume at ~300 km = 3000m; easily in aircraft cruising height), should produce strong echoes at home.

5.     While over the rest of the path TX - Home, particularly the central bit below X, there will be aircraft echoes if aircraft are in the scatter volume, producing considerable enhancement.

6.     Occasionally tropospheric ducting will occur; this is an enhancement mode which will open the entire path, up to thousands of kilometers, enabling aircraft reflections over that distance, providing the tropo duct is high enough above ground level.. However, it will have no effect on the detection of satellites other than possibly attenuating the TX beam in the far field by bending the beam along the earth at low altitude and making satellites more remote.

7.     Regarding satellite detection, how much power do TV TXs put over the horizon? They have around 13 dBi gain and a vertical beam width of only a couple of degrees – does enough power go over the horizon to illuminate satellites at low elevations?

 

Observed Data

1.     Using the Johannesburg and Pretoria transmitters on local channels at distances of 11 and 44 km respectively, there are no reflections from overhead aircraft. This is because aircraft, even just a few thousand feet above ground, are above the TV TX beams. These beams are omni-directional and aimed below the horizon, i.e. at a small negative angle around 1-1.5 degrees.

2.     However, distant TXs (428 km), well over the horizon and propagating by tropo scatter, produce strong echoes on overhead aircraft where the just detectable TV carrier is enhanced by as much as 20 dB. Confirmed in this case by visual observation of these aircraft – they are flying over my house (explanation: the scatter angle is reduced due to the aircraft’s height above ground).

3.     A TX 84 km away (Welverdiend to the s/w) on the “0” offset produces a display of aircraft reflections showing Doppler shifts, Figure 5. There are no visible responses from the further afield TXs on the “0” offset listed in Table 1 – the distances are just too great. In Figure 5, the stronger trace going to the left is a negative Doppler shift indicating the bistatic range in the Johannesburg area using the near by Welverdiend TX is increasing, while the Queenstown TX produces no visible Doppler shift on these aircraft. More on bistatic range later.

 

Figure 5: Aircraft echoes received using the Welverdiend and Queenstown TV TXs. TV carrier frequency in centre, with 50 Hz symmetrical sidebands each side

 

4.     The Kimberley ch 4 20m offset TX, 428 km s/w, produces a number of aircraft reflections if these aircraft are on the direct line of sight path Rx - Tx, while the only other TX on that offset, East London 760 km s/e, cannot be detected and produces no echoes: Figure 6.

5.     Detection of aircraft using a UHF high gain antenna array and masthead preamplifier produces similar aircraft detection as when using VHF TXs, a low gain antenna and no preamp.

6.     Aircraft flying directly along the line of sight between the TX and RX produce little or no Doppler shift.

Figure 6: Local aircraft echoes, above QTH near Johannesburg, received from the Kimberley TV TX. Note enhancement as aircraft cross the carrier frequency

 

More on Path Geometry

Using these results, it is apparent that aircraft can be detected to a typical range of about 200-400 km if all the possibilities of a tropospheric scatter path are used. Continual traces of 15 minutes, corresponding to a distance of 150-200 km for aircraft traveling around 600-800 km/h, are common on waterfall displays such as Figures 5 and 6.

This implies that aircraft within 150 km of Home (see Figure 4) may be tracked, but possibly not over the central portion of the path in the scatter volume which may be above aircraft height. In this case, on the probable maximum 650 km path there may be no reflection from aircraft beneath the central 350 km or so portion centered under X depending on the altitude of this volume at X.

But all this is over obstructed paths where the path loss is high.

What about satellites?

 

Bistatic Doppler Shift on Satellites

We have already established there is a null pattern above a TV transmitter due to good suppression of any vertical radiation from the array. In Figure 4 there will be no satellite detection along the line Home – X to the satellite arc as that line of sight does not intersect ARC A, the satellite orbital arc above ground, until the far side of the TX at around 1500 km from Home. In any case the elevation angle from TX to the satellite arc discounts all possibility of an echo from satellites in the vicinity of TX.

Therefore the only alternative is above Home along the sight line Home-BC approximately, keeping the possibility within the red zone. This provides also the best suppression of unwanted aircraft echoes which could cause confusion of the observed result. Note that this possible sight-line forms an ellipse around Home.

The detection of satellites (and aircraft) using this bistatic, passive “radar” technique, is highly dependent on the Radar Cross Section of these objects. Too high a frequency (UHF or SHF) used on a satellite complex such as the ISS, with all its modules joined to the main body in all directions, causes a reduction in effective RCS due to the shorter wavelength passing through the gaps without being reflected. A lower VHF frequency is ideal as the longer wavelength will cause the beam to be reflected back, i.e. the RCS is more efficiently used. Satellites, as for aircraft, produce a Doppler shift.

 

Result of Satellite tracking using TV Transmitters

The calculated Doppler shift is high for an object such as the ISS, around 14 kHz peak-to-peak due to the high velocity of 27700 km/h at low range. This is troublesome when using a SSB bandwidth of 2.5 kHz, offset by some 2-3 kHz to obtain a convenient beat frequency around 1 kHz at the minimum Doppler shift point. Is there enough power to detect reflections from a target such as the ISS over a typical straight-line distance of 1100-1500 km? See the Appendix for a signal-to-noise ratio calculation.

After a few attempts at detecting echoes from the ISS at its 350 km altitude under very favourable conditions using various transmitters on ch 4 scattered around South Africa, various possible echoes (“noises”) were received, but many attempts produced no recognizable results. Essential to success was a prior determination of the satellite path across Southern Africa in conjunction with a rough estimation of the expected bistatic Doppler shift and azimuth for antenna pointing, the whole while noting the elevation angle as seen from the TV transmitters to be used. An elevation angle from a transmitter much above 3-5 degrees will result in zero detectable echoes from a satellite due to the restricted vertical beam width of the TX antennas (“acceptance angle 5 deg”, my estimate, Figure 4).

 

The ISS as a reflective body at high VHF might have a lower RCS than a typical jet liner. This may be explained by the numerous flat solar panels, playing such as important part in visual observation of the ISS and making up a large portion of its apparent RCS, having little role in reflecting RF, much as the flat panels on stealth aircraft do not typically reflect radar signals back to the radar receiver. These solar panels might function effectively as anti-reflection screens. On the other hand, should they line up briefly the RCS will be very large. The idea is to capture these brief moments.

 

Figure 7: The ISS seen through an amateur’s telescope – take away those screening solar panels and there’s not much left….

 

Are these echoes from the ISS?

After an orbital prediction, equipment was set up to receive the ISS. Unfortunately nothing was received presumably due to the horizon cut-off pattern of the TV antenna beams – I was informed by a former colleague at the SABC that these beams have a vertical beam width of only a couple of degrees and are aimed at an angle of around -1 degrees to the horizon.

 

General Conclusions

1.     While conducting these experiments various persons in the USA particularly maintained they were receiving satellite echoes from TV transmitters. I am happy to dispute these claims. In all cases images sent to me indicated the echoes were aircraft.

2.     Distant aircraft (~150 km or so) or local aircraft, when crossing the direct line between TX and RX, enhance the received TX signal by 20 dB or more for a period lasting possibly some minutes - the more distant, the longer the enhancement in time.

3.     The bistatic Doppler shift from aircraft wobbling around by only a few meters readily shows up on a recording waterfall – the technique is very sensitive.

4.     At times the signals are enormous or extremely weak and will test the dynamic range of the best receivers.

5.     Aligning TV broadcast antennas at a negative angle of around -1.5 to -2.0 degrees to the horizon is almost certainly international practice – it would not make sense to aim at the horizon as half the EIRP would disappear uselessly into space. This makes satellite detection using TV broadcast TXs as illuminating sources unlikely if not impossible as the ERP at an angle of 3-5 degrees or so is likely to be around 15-20 dB down .

6.     In Figure 5 as the aircraft cross the carrier frequency where the Doppler shift equals zero, considerable enhancement of the path’s signal level occurs and echoes from the 50 Hz sidebands become visible. This corresponds to an enhancement in signal level of at least 25 dB as the first sidebands in a properly modulated TV transmitter are around 25 or more dB down on the carrier level.

 

Appendix – calculations

S/N of a satellite echo

The signal to noise ratio for the predicted total bistatic path to the ISS may be calculated to check the feasibility of the link, note the pessimistic RCS of 50 sq m indicating dimensions of only approximately 7 x 7 m – it might be larger than this. Note also the calculation assumes the antenna pattern is centered on the horizon, we now know it is not and the SNR will have a large negative value:

 

s/n    =       PTGTGRt0λ2σFTFR  /   (4π)3 kTSRT2 RR2

 

Pt           =       Transmitter power, W (10,000)

Gt          =       Transmit antenna gain as a factor (19.9)

Gr          =       Receive antenna gain as a factor (14.1)

t0       =       integration time in signal processing (FFT 262k “window time”), (5.9 s)

λ        =          wavelength, m (1.71m~175.25 MHz)

σ        =       Radar Cross Sectional area, sq m (50sq m)

FT          =       Transmit antenna to medium coupling factor, for yagis 0.75

FR          =       Receive antenna to medium coupling factor, for yagis 0.75

π        =       3.41

k        =       1.38 x 10-23  Boltzman’ s constant

TS          =            System noise temperature (2790 Kelvin), see below

RT     =       Transmitter to target distance (1100000m)

RR         =       Receiver to target distance (1100000m)

 

s/n    =       (10000)(19.9)(14.1)(5.9)(1.71)2(50)(.75)(.75)

 /

(4x3.41)3 (1.38 x 10-23 )(2790)(1100000)2 (1100000)2

         

          =       1361472748/143053056

          =       9.51

          =       9.78 dB, this is ample to detect an echo

 

RT  and RR of some 1500 km (they are not normally equal as used in this calculation) are necessary to keep the elevation angle from the TX below 3-5 degrees if possible; otherwise the ISS will not be illuminated.

 

Bistatic Range

RTX  = Transmitter to target range, meters.

RRX  =  Receiver to target range, meters.

L     =  Distance between TX and RX, meters

 

Bistatic range  = RTX RRX – L

Around the ellipse bistatic range does not change, or when the target moves along L, so bistatic Doppler shift = 0.

Figure 8: Parameters of Bistatic Range

Source: WWW. Wikipedia.com

Bistatic Doppler calculation

Bistatic Doppler shift is proportional to the rate of change of bistatic range in period “t1-t” seconds.

Thus, two bistatic ranges are calculated, firstly at time “t”, then at time “t1” seconds.

During this time period the target has moved and increased or decreased the bistatic range.

The two values are subtracted to provide the change in bistatic range as at time t1.

 

          Change in bistatic range        ΔR = (RTX RRX – L) – (RTX RRX – L)1

 

If the range has increased the Doppler shift will be negative, and positive if the range has decreased. The shift can be negative, even if the target is moving closer to RX.

         

Bistatic Doppler shift, Hz        f = 1/λ. d/dt. (ΔR)

The Bistatic Doppler shift in period t+t1 seconds can be read off a waterfall display as captured in Figure 5 where a RF frequency of 175.25 MHz (λ=1.71m) was used.

The change in bistatic range can be isolated as:

                             ΔR, m         = f/1. λ/1. (t1-t)

Thus in Figure 5, the target causing the negative 50 Hz Doppler shift in time 22:57:35 – 22:55:45 (approximately 110 seconds as read off), has changed the bistatic range by:

(50s)(1.71m)(110s) = 9404m or an average speed of 85.5m/s, 308 km/h

 

Bistatic Doppler shift using multiple TV transmitters locked onto a common frequency (there will be two bistatic Doppler shifts in this case) can be read off directly and direction, velocity and distance calculated. This has not been documented here.

 

Tropospheric Path Calculations

These calculations will provide an indication of the signal to be expected from a distant VHF or UHF transmitter. A tropospheric scatter path will provide a nearly constant signal around the calculated value. The steps to follow are:

1.     Scatter angle

2.     Path loss

3.     System noise temperature

4.     Signal to noise ratio

Tropospheric Scatter Angle

The Kimberley ch 4 TV transmitter will be used in this example.

R                 =       4/3 earth radius (8446 km)

d                  =       great circle path distance 428 km

h1 and h2          =       respective antenna heights above sea level, 1.65 and1.385 km

h11 and h12      =       height of radio horizons above sea level, 2 and 1.235 km

d1 and d2      =       great circle distance between radio horizons and respective

antennas, 5 and 100 km

scatter angle θ                  =       θ0 - θ1 – θ2 radians,

where:

θ0 = d/R

θ1 = h1 - h11/ d1 + d1/2R

θ2 = h2 - h12/ d2 + d2/2R

So scatter angle  = 428/8448 – (1.65-2/5 + 5/16896) – (1.385-1.235/100 + 100/16896)

             = 0.050663 – (-0.069704) – (0.007419)

             = 0.112948 rad

             = 6.5 degrees

Tropospheric Scatter Path Loss

LP                          =      tropospheric scatter path loss, dB

LFS                        =       free space path loss, 92.4 + 20 log d +20 log fGHz

LS                          =       Normalised all year scatter loss at NS, 57+10log(θ-1)+10 log(f/0.4)

NS                          =       surface refractivity index, around 300 in South Africa

LP          =       LFS + LS – 0.2(NS – 300) dB

                    =      (92.4+20log428+20log0.17525)+(57+10(6.5-1)+10log(0.17525/0.4)

          =      (129.9) + (112.0) + (-3.58)

          =      238.32 dB

 

System Noise Temperature

Tsys  = α(Ta) + To (1- α) + T1 + Tm /gm-1

where:

α        =       line transmission coefficient as a factor, 0.8 for a 1 dB loss

Ta          =       temperature of transmission line, k, equal to Ta for uncooled lines

To          =       ambient temperature, k, typically 290

T1          =       temperature of 1st RF amplifier stage, 2500k (NF of RX-no preamp used)

Tm         =       temperature of 2nd RF amplifier stage

gm-1       =       gain of stage T1

Since no masthead preamp was used the noise figure of the Rx is taken and Tm /gm-1 is set to 0.

so      Tsys      =       0.8(290) + 290(1-0.8) +2500

                    =       2790k

Signal to Noise Ratio over tropospheric scatter path

SNR            =       Po+ Gt + Gr – Lp –Pn 

Po                          =       output power of TV TX, 10 kW (40 dBW)

Gt                          =       gain of Tx antenna, dBi (12)

Gr                          =       gain of Rx antenna, dBi (11.5)

Lp                          =       tropospheric scatter path loss, dB (245.7)

Pn                          =       10 log(kTB) noise power ratio of Rx (-160.1), k is Boltzman’s

constant 1.38 x10-23 and B = 2500 Hz in USB, T = Tsys

So SNR       =       40+12+11.5-238.3-(-160.1)

                   =       -14.7 dB

An FFT of 262k provides an improvement in SNR of about 30 dB:

So SNR       =       30-14.7

                   =       15.3 dB

This is in good agreement with the variable 3-20 dB above noise carrier received from the Kimberley Tx on ch 4 when a “waterfall” display is used with a FFT of 262k.

 

My contact information