Fullscreen HTML5Med Res (??MB)HTML4

The effect of the introduction of FWA networks on C-band satellite reception in Australia Death by O by Garry Cratt Last month, we told you about the introduction of Unwired – a wireless broadband system which (along with several other new players in the game) brings you the internet without the need to plug in to ADSL or cable services. H OWEVER, as we mentioned at the end of that story, every silver lining has a cloud . . . in this case, it’s C-band satellite television reception. In October 2000, the Australian Communications Authority (ACA) auctioned off part of the 3.4GHz spectrum to three companies, raising over $112 million. In doing so, they set in stone a progressive rollout of a huge fixed wireless network designed to carry broadband internet services in most populated areas of Australia, one that is already disrupting the reception of over 100,000 C-band satellite viewers across the country. The ACA had, some time earlier, devised three bands within what they termed the 3.4GHz spectrum. These three bands cover 3.425GHz to 3.575GHz, despite the fact that international satellites such as Apstar 2r, Thaicom 3, Insat and Palapa C2 were delivering signals into Australia on the same frequencies. In the roll-out prior to the commencement of service in August/ September of this year, scores of sites had been equipped with terrestrial transmitters to ensure virtual saturation coverage of Sydney. Because these sites rely on micro74  Silicon Chip wave or fibre optic connectivity, they have been installed at many GSM (mobile phone) towers and two-way-radio communication sites, where this connectivity already exists. In addition, high-rise buildings have been used to ensure optimum coverage. Wireless internet users could now experience the benefits of broadband- like Internet speed without connection to a wired service provider. Houston, we have a problem But for C-band satellite viewers in Sydney (and soon in most population centres in Australia), a blank TV screen is the result. Analysing the problem revealed that From last month’s feature, the estimated Unwired coverage area of Sydney. If you’re in the red zone, Unwired should be available. If you’re close to an Unwired tower OR someone close by connects to Unwired, chances are your C-band satellite reception will be affected, if not obliterated. siliconchip.com.au Overload the most prolific operator is “Unwired Australia” whose wholly owned subsidiary, AKAL, had paid $95 million for licenses in all areas where the spectrum had been made available. The company claims to own close to 100% of the 100MHz of spectrum that was made available in Sydney and Melbourne, along with the majority of spectrum in Australia’s other capital cities. They utilise the combination of wireless broadband product from US supplier, Navini Networks, backhaul equipment from Airspan Networks, network operations services from Ericsson Australia Ltd, router and IP switching equipment from Cisco Systems and fibre connectivity from Uecomm. The network uses the 802.16e protocol, a proprietary type of WiMAX technology, yet to be ratified by the IEEE. The map of Sydney printed last month and reproduced opposite shows the wide coverage of the system. Unwired have licenses allowing them to operate at all sites until December 13, 2015. So the problem is here to stay. The ACA have a website, www.aca. gov.au, which contains a comprehensive list of transmitter sites, power siliconchip.com.au levels, frequencies and radiation patterns, so it is possible to research the probability of interference by location. The site indicates that there are at least 196 separate transmitters, operating from 68 sites licensed to operate in Sydney. The company is targeting the home and soho market: a potential of 3.5 million people in 1.2 million households and 240,000 small businesses in Sydney. Unwired’s own website states that users are able “to connect within a range of up to 10km from a tower”. Two sources of interference Not only is interference generated by Unwired towers, the wireless mo- An Unwired modem: itself a source of interference to C-band satellite reception, as are the Unwired towers. dem supplied to consumers by the company also operates at C-band. This means that wireless broadband users located in very close proximity to a C-band satellite system can also generate interference anywhere within the allocated bands. The downlink remedy Spectral display of signal from satellite Thaicom 3 showing LNB overload as a result of Unwired. The obvious result of operating a terrestrial transmitter on exactly the same frequency as a satellite signal means satellite reception is simply not possible. Even operating at a power level of a few watts, the terrestrial signal obliterates a satellite signal of a few picowatts. Satellite systems are designed to receive very weak signals and the amplifiers (LNBs) used cannot tolerate having a huge signal forced down the December 2004  75 At left is a single polarity feedhorn; centre shows a waveguide filter which is inserted between the feedhorn and the LNB (low noise block converter) shown at right. Weatherproofing gaskets are used at both ends of the waveguide filter. input port… it’s death by overload!! In a case of severe interference, the presence of a huge signal on a frequency within the pass band of the LNB means is that the LNB is severely overloaded, sometimes driven into compression. This results in distortion in the amplifier section of the LNB. The amplifier produces a range of output signals other than the desired one from the satellite, which are mixed with the local oscillator (5.15GHz) – the result being a huge, broad output signal up to 500MHz wide. In some cases of mild interference, moving the dish to a location shielded from the source of the interference may help. In all cases, changing the LNB from one covering 3.4-4.2GHz (extended C-band) to one covering 3.7-4.2GHz (normal C-band) and ensuring that quad shielded RG-6/U cable has been used will help. Of course this means giving up the “extended” part of C-band but that is a small price to pay. In more severe cases, changing the LNB to the desired frequency range may not cure the problem and it may be necessary to change the LNB to a single polarity type and use a single polarity feed horn. This means that one polarity will not be available from the satellite. As the interference is vertically polarised, satellite signals of the same polarity suffer more. Satellite channels having a frequency closer to the interfering frequency will be more adversely affected. In severe cases of interference (the majority of satellite users), it will be necessary to use a special LNB called a phase locked loop (PLL) type, as well as a waveguide filter to eliminate the interfering signal before it gets to the LNB input. For satellite systems where both polarities must be received, an othomode coupler must be used with two PLL LNBs. The graph below (from Microwave Filter Company USA) shows the steep attenuation of signals falling outside the filter bandpass limits, in this case 3.7-4.2GHz. Typically, these filters offer 70dB attenuation at the interfering frequency. While the best remedy is to remove the interference before it enters the LNB input, some relief from the effects of mild interference can be obtained by “conditioning” the signal after the LNB and before it arrives at the receiver. A satellite receiver is designed to accept a 500MHz wide band of input signals and, as such, it does not have any great selectivity. So feeding a high-level signal into the receiver can cause the tuner to be overdriven. In many cases an improved situation can be obtained through the use of an in-line 10dB attenuator. As well as reducing the satellite signal, the attenuator “deafens” the tuner to some degree, reducing the total amount of input signal and allowing the tuner to operate normally. Where the interfering signal is converted by the LNB (along with the satellite signal), it will appear as an IF signal, normally just outside the nominal 950-1450MHz IF band generated by 3.7-4.2GHz LNBs. An IF filter centred on the interfering frequency can be used to advantage without adversely affecting the satellite signal. Such filters can be inserted in the coaxial line to the receiver and can be configured to pass the DC voltage necessary to power the LNB. A dual-band combiner also has filtering properties that can be used to C-band feedhorn fitted with two wave guide filters and LNBs, The combination of waveguide filter and LNB for each polarity ensures reception for all signals. 10dB attenuator - it must be rated to 2GHz and must have DC power passing capability to ensure the LNB is powered. Actual frequency response graph (taken from “MFC” website) shows sharp cutoff below 3.7GHz and above 4.2GHz. 76  Silicon Chip Step 2 – the IF remedy siliconchip.com.au Ozitronics www.ozitronics.com Tel: (03) 9434 3806 Fax: (03) 9011 6220 Email: sales2004<at>ozitronics.com 40 Second Message Recorder Record as many messages as will fit in 40 seconds. Messages stored in non-volatile memory. Features message looping option. Dual-band combiner has good IF filtering at 1600MHz. Examples of single and twin simple open-circuit stub filters. advantage. Designed to accept signals from two dishes, these combiners have a high level of filtering to keep the two output blocks of IF frequencies separated. This means that the 9501450MHz section is very well filtered from the adjacent 1500-2000MHz block of output signals. So an interfering signal at 3.55GHz (IF 1600MHz) is greatly reduced on the “bypass” (950-1450MHz) port. A simple IF filter can be constructed to reduce the effects of an interfering signal. By connecting a quarter wavelength open circuit “stub” (tuned to frequency) to the coaxial cable feeding the receiver, 25dB of rejection can be obtained. Construct two of these filters and place them a half wavelength apart and 70dB rejection can be achieved. Unfortunately, while the mechanical construction of the filter is simple, measuring the frequency to which it is tuned requires a signal generator and spectrum analyser capable of operating from 1400-2000MHz. One drawback of the stub filter is the bandwidth that can be achieved. This type of filter can only achieve 10dB of rejection for every 10MHz of bandwidth, so a filter producing 70dB of rejection will be something like 70MHz wide. If the interfering frequency is less than 70MHz away from the desired satellite signal, the filter will attenuate the satellite signal as well. The other problem is that the stub length is critical. A change in length of 1mm results in a shift in the centre frequency of 100MHz! It is possible to construct elaborate cavity filters offering narrow bandwidth and sharp frequency response, but the effort and cost put into this would be better spent on a filter before the LNB. The picture below shows such a multi-cavity filter and the frequency response achieved. Use of F-type compression plugs ensures consistent, low-impedance earth connection –vital for minimising interference. Highly effective laboratory-built IF notch filter prototype has narrow bandwidth and deep-notch capability, as shown by its response graph at right. siliconchip.com.au Step 3 – the receiver remedy Finally we come to the receiver itself. There is a huge difference in tuner performance between receivers. Generally it, is true to say that the cheaper satellite receivers will probably give poor results compared to the more expensive models. Cabling and connectors are another area where attention to detail can help eliminate interference problems. In cases of high-level interference, K146 - $30.80 120 sec version K64 --$46.75 Prices include GST – shipping extra. Full documentation available from website. Over 130 kits available – check website. where RF energy not only drives the LNB into compression but also enters the coaxial cable, it may be necessary to run the vertical sections of coaxial cable in metal tubing for maximum protection. This is particularly noticeable when a waveguide filter has already been used. Even wrapping connectors in aluminium at the point of ingress may assist. Having eliminated the primary route of the interfering signal, other more minor problems that still contribute to the overall problem are more readily noticed. Crimp connectors can give problems because they fail to provide continuity of shielding and a low impedance bond to the coaxial shield. Instead of a continuous bond between the coaxial shield and the body of the connector, crimp connectors provide only six points of earth bonding. The use of compression connectors resolves this issue. Part Two of this series, to be published next month, examines available remedies to the problem. SC December 2004  77