ATSC compliance and tuner design implications By Nick Cowley Chief RF Systems Architect DHG Group Intel Corp. E-mail: nick.cowley@zarlink. com Robert Hanrahan National Semiconductor Corp. Applications Engineer Technical Staff Member E-mail: robert.hanrahan@nsc. com As the U.S. government pushes ahead with plans to switch off the analog broadcast system in favor of an all-digital transmission, manufacturers must navigate through a growing number of government standards for television performance and design new sets that better support over-the-air DTV and HD viewing. The Advanced Television Systems Committee (ATSC) has developed a receiver specification to ensure that DTV receiver equipment receives over-the-air signals as reliably as analog channels. Today, ATSC A/74 is only a recommended practice. However, as U.S. consumer demand for over-the-air DTV and HD viewing increases, coupled with an aggressive government drive to reclaim analog spectrum, many believe it will become the de facto standard for quality ATSC receiver equipment. Digital modulation format ATSC uses a digital modulation format based on a multilevel coding, uniformly distributed around a small DC offset. This composite spectrum is then amplitude-modulated onto a carrier and further processed to produce a vestigial modulated carrier, where the level of vestigial sideband is proportional to the DC offset (typically, 0.3dB). This modulation is referred to as 8-level vestigial sideband (8-VSB). The ATSC standard also applies data randomization, which effectively codes the data so the transmission appears as a noise-like signal with uniform power spectral density, forward error correction (FEC) in the form of Reed Solomon coding and Trellis coding both of which enable the demodulator to detect and correct received data errors and data interleaver, which spreads packet data over time to increase packet immunity against burst noisegenerated errors. Tuner performance As described above, ATSC uses 8-VSB digital modulation with various coding means to enhance noise immunity. From this, a theoretical carrier-tonoise (C/N) can be calculated, which the demodulated data bit error rate (BER) after error correction will be deemed unobjectionable to the viewer. The FCC Advisory Committee on Advanced Television Service (ACATS) has determined the threshold of visibility (TOV) or quasi-error-free (QEF) level to be a BER of 3E-6. At this level, the error protection will correct most data errors to deliver a quality picture without macroblocking blocks of corrupt data within the video picture. At BER levels above this limit, there is no perceived improvement in QoS. Conversely, as the BER increases, the error protection can no longer correct data errors, leading to macro blocking. To achieve TOV with 8-VSB modulation, the theoretical value for C/N should be 15dB. However, perfect performance cannot be achieved in any practical implementation, so a level of 15.5dB is typically applied as the TOV value. A margin of 0.5dB is allowed over the theoretical limit to account for receiver impairments. Per A/74, it is predicted that the presence of typical multipath will require a further 3dB improvement in C/N ratio to 18dB over the theoretical limit to achieve TOV. To provide adequate reception of ATSC transmissions in a typical multipath environment, an 18dB C/N ratio must be achieved where the noise can either be received noise, tuner additive noise, or a combination of both. This C/N requirement has to be carefully considered when implementing a front-end tuner function. A/74 specifications The A/74 specification describes the expected performance from a complete tuner and demodulator receiver front-end to deliver acceptable QoS. The ATSC standard allows digital channels to be transmitted in taboo channel locations relative to existing analog services. These taboo channels should not be occupied by other analog services, as intermodulation effects and other impairments will lead to channel degradation. This is generally managed through system planning. New digital services in these taboo channels are transmitted at a lower power level to avoid interference with existing analog services. This is possible because a digital service requires a much lower C/N than analog for the same level of picture quality. A/74 describes a minimum desired-to-undesired channel ratio (D/U ratio) for each taboo channel, under which the Figure 1: A negative D/U ratio implies that the desired channel is weaker than the undesired adjacent by the specified decibel, ratio.
DTV into DTV A74/RP D/U requirement vs. desired channel power -90-80 -70-60 -50-40 -30-20 -10 0 Figure 2: An example curve demonstrating linear interpolation. receiver should exceed TOV. Three scenarios are described to cover weak, moderate and strong desired channel powers. Linear interpolation between specified values is generally assumed for other channel powers. The three scenarios for DTV and NTSC (analog)-adjacent interfering channels are shown in Figure 1, where a negative D/U ratio implies that the desired channel is weaker than the undesired adjacent by the specified decibel ratio. An example curve demonstrating linear interpolation is shown in Figure 2. Finally, A/74 does not currently define a specification for a multicarrier overload condition rather, it draws the designer s attention to the fact that the front-end must work in real environments with a mix of high- and low-amplitude carriers. As ATSC gains popularity, more transmitters will be located in close spectral proximity. As a consequence, ATSC receivers will be subject to more multicarrier adjacent energy, placing much higher demands on the receiver. Designers today need to consider situations that will place higher demands on a receiver that will inevitably exist as the FCC repacks spectrum (Figure 2). Per ATSC A/74, the minimum suggested operating range in an echo-free environment with no applied incident noise from the receiver is - 83dBm to -8dBm. Some industry members have argued that this range is limited on the top end and needs to be extended to emulate real-world situations. Lower sensitivity limit The receiver s lower operating limit is defined by the tuner noise figure (NF). The tuner NF is a measure of the noise added to the received signal relative to the source impedance noise. This is sometimes also referred to as the tuner additive white Gaussian noise (AWGN). The required NF can be determined by calculating the Power dbm 0-5 -10-15 -20-25 -30-35 -40-45 -50 total noise power described by the NF and relating this to the minimum operating power by the required C/N (Listing 1). In field use, this NF should be substantially achieved with Power level in congested areas at receiver (Hollywood, Florida) Channel Figure 3: An example curve demonstrating linear interpolation. -5-10 -15-20 -25-30 -35 the A/74 taboo channel D/U ratios applied as described in Figure 1. The relevant taboo channels for consideration will depend on the tuner architecture. In a single conversion tuner, there will be protection for channels at greater than typically two channel offsets due to the frontend tracking filters. In double conversion solutions that do not use tracking filters, there will be no protection. It is far more challenging to meet the NF requirement in the presence of taboo undesired channels with a double conversion architecture compared to a single conversion. For example, the relative levels of N+2 and N+6 taboo channels is 13dB for DTV into DTV. For a double conversion architecture to meet A/74 at the sensitivity point, this demands a 13dB greater signal handling hence, a 13dB greater dynamic range Analog broadcast Digital broadcast (Listing 1)
Minimum operating requirement P (-83dBm) Required CN for TOV (15.5dB) Allowed noise for TOV defined by P and CN (-96.5dBm) NF R (-93.5+105.5 = 8dB) (Listing 4) Integrated noise floor (-105.5dBm) Thermal noise floor, N T (-174dBm) is required. The A/74 D/U ratio should be substantially achieved at the sensitivity point. This in turn defines the additional allowed impairment to the C/N from the tuner due to N+/-1 spectral splatter. The additional impairment from spectral splatter can be calculated from the root mean square (rms) addition of tuner thermal noise and spectral splatter noise, which can be calculated from the tuner IPIP3 and desired and undesired signal powers. The allowed spectral splatter can be determined by calculating the additional noise arising from the spectral splatter, which when added in an rms manner to the tuner thermal noise, results in the above defined 1dB increase in total composite noise and in a 15.5dB C/N, and finally in TOV. This can be calculated from Listing 2: Integration over channel bandwidth (10log :5.38MHz) - <87.5dB) Figure 4 : The required NF can be determined by calculating the total noise power described by th e NF and relating this to the minimum operating power by the required C/N. First, calculate the integrated thermal noise power from the NF. Now derive the allowed spectral splatter (Ps) by performing an rms addition of the thermal noise power (P n ) and P s, which leads to composite noise (P c ) (Listing 3). (Listing 3) Considering the above example of 1dB greater than the lower operating limit, which defines an allowed increase of composite noise of 1dB, maintaining a 15.5dB C/N, For this test condition, the desired is at -82dBm (1dB above the A/74 minimum operating limit) and the adjacent undesired is -49dBm. Thus, the spectral splatter that results from the undesired at -49dBm (per A/74 adjacent channel D/U ratio) must be a maximum of -104.5dBm or around -56dBc (-104.5 + 49) relative to the undesired channel. This implies that an IPIP3 of circa -21dBm (-49 + 56/2) minimum is required at the defined operating limit. The implication of this requirement on the tuner automatic gain control (AGC) attack must also be considered where the AGC attack is the level at which the tuner frontend AGC begins to reduce protecting latter stages from strong signal conditions. It is standard practice to determine this value on the composite received signal (the desired plus interfering) as this will ensure that the optimum C/N plus intermodulation (C/N+IM) is obtained under all received signal conditions, for example, Defined operating requirement Required C/N for TOV Composite noise Tuner additive noise power NF Integrated noise floor with and without taboo interferer present. In theory, the AGC attack could be set at the sensitivity point. Assuming the NF increases at 1dB per db of gain back off, as might be the case with some AGC architectures, the C/N achieved at the sensitivity point will be maintained for all signal conditions, thus maintaining TOV. However, this is not a desirable implementation, since there will be no margin to accommodate any other tuner impairments. Instead, it is normal practice to define the AGC attack above the sensitivity limit. A typical value for AGC attack may be 5dB above sensitivity limit, meaning the above derived IPIP3 requirement of -21dBm would be 5dB higher, or -16dBm. Similar analyses can be considered for other taboo channels, and since they are of greater amplitude, the signal-handling requirement Allowed degradation in noise through RMS addition of tuner additive noise and spectral splatter Spectral splatter noise level Thermal noise floor (Listing 2) Figure 5: The spectral splatter that results from the undesired at -49dBm must be a maximum of -104.5dBm or circa -56dBc relative to the undesired channel.
Undesired channel power Desired channel power Allowed level of image after conversion Thermal noise floor A/74 desired D/U ratio to image channel Integration over channel bandwidth Desired image reject ratio ure for the composite signal power of eight channels at -8dBm. This effectively determines a maximum mean input amplitude of +1dBm, which the input stage should be able to support. This requirement is easier to meet in a single conversion tuner implementation than in a double conversion. This is due to the protection afforded by the tracking filters, which will suppress some of the eight carriers. The designer s task is also eased by deploying an FET stage and MOPLL with a high IPIP3 and P1dB. Figure 6: Ideally, the tuner should achieve a maximum image cancellation across its operation range of 72.5dB and a minimum of 68dB. may be greater. When operating at maximum gain, which corresponds to the sensitivity limit, the tuner has counter demands of low NF and high IP3. In a typical single-conversion tuner, which uses a dual-gate field effect transistor (FET) LNA/AGC stage and a silicon mixer oscillator PLL (MOPLL) IC, the NF will be dominated by the dual-gate FET and the intermodulation by the MOPLL. This is because the gain of the FET will reduce the MOPLL thermal noise contribution when referred to the tuner input, while the gain will increase the signal amplitudes seen at the MOPLL and thus render this stage more susceptible to intermodulation generation. Consider the above -16dBm IPIP3 requirement in the case where the FET has 20dB of gain. Assuming the IPIP3 is limited by the MOPLL, this must achieve a minimum of +4dBm IPIP3. As can be seen when considering tuner implementation, demands on the designer are eased by deploying an FET stage and MOPLL with a high IPIP3 and P1dB. The receiver higher operating limit is defined by the tuner P1dB, which relates to the signal-handling performance, and the IPIP3, which defines the level of spectral splatter from both the desired and adjacent, and the FET AGC range. When considering the above described application, where the AGC attack is set at 5dB above sensitivity point with an N+1 adjacent present, then the input power which defines AGC attack is -83 + 5 + 33dBm = -45dBm. Given the A/74 maximum operating limit of -8dBm, the RFAGC range requirement is circa 40dB. Maximum signal handling When considering a desired only or desired at strong condition with maximum amplitude undesired per A/74, the maximum input level is -8dBm. This level can be higher when considering a multicarrier environment (Figure 3). Distribution experts have proposed a representative fig- Adjacent channel N Third-order nonlinearity, IPIP3 Third-order nonlinearity will lead to spectral splatter from N+1 DTV onto the desired DTV channel. The worst-case condition for this will be a strong desired signal with N+1 taboo per A/74. By extrapolation from the value for IPIP3 determined at the sensitivity operating point, a value of +13.5dBm can be deduced when operating at A/74-specified strong condi- Figure 7: The graph shows performance achieved from a tuner incorporating an image-rejection MOPLL with conventional tracking filter technology.
tion. This figure is derived from an increase in desired power from -82dBm to -28dBm and an increase in D/U ratio from -33 to -20dB per A/74. This places further demands on the tuner. The designer s task is again eased by implementing a single conversion design, using an FET stage and MOPLL with a high IPIP3 and P1dB. Image cancellation The image cancellation requirement is perhaps one of the least understood specifications in that the A/74 specifies a requirement for -50dBm. However, there is a caveat within A/74: This value should not be interpreted as just a tuner image-rejection value in that it applies to the entire receiver. In simple terms, the D/U requirement for the image channels does not specify the minimum image cancellation required from the tuner. Instead, it defines the applied D/U ratio under which the front-end shall deliver TOV performance. In practice, the image channel will fold on top of the desired channel and appear as a noise-like degradation to the C/N. Considering no other noise sources, then the tuner image cancellation requirement can be calculated from the following formula: With no other impairments, a minimum image cancellation of 65.5dB must be achieved. However, in real-world situations, other impairments will exist. Multipath, for example, will increase the image-cancellation requirement by a further 2.5-68dB. As discussed earlier, it is also a stated objective to achieve the A/74 requirement substantially at the operating sensitivity. Using the same definition as substantially at sensitivity being 1dB above minimum operating limit, the allowed level of image signal will be -104.5dBm. This was the signal level calculated for a second noise source, which raises the tuner additive noise power by 1dB. The required image cancellation in this situation is the difference between the A/74- defined image channel level and this derived allowed level of image channel converted to the IF output leading to a 1dB rise in composite noise. This can be calculated from the formula in Listing 5. The tuner should achieve a maximum image cancellation of 72.5dB across its operation range and a minimum of 68dB. Meeting the requirement will either compromise the desired channel filter flatness or require additional tracking filter stages to achieve additional image suppression. Neither of these is desirable. The former may render the tuner performance unacceptable for other distribution means such as analog terrestrial or cable, while the latter will involve additional design complexity and increased manufacturing cost. These issues are compounded when aging and temperature variation effects are considered. A solution to these disadvantages is to apply active image cancellation within the tuner by using image reject mixer technology in the MOPLL section. With this approach, it is feasible to achieve 30dB of image suppression, which will ease the tuner tracking filter requirement to 42.5dB. This helps lower manufacturing costs and is also far less susceptible to environmental aging. PnP performance Plug and Play (PnP) applications are typically addressed by deploying a single conversion tuner for ATSC and a double conversion tuner for digital cable-ready (DCR). This approach increases component and design costs as well as board space. Neither of these solutions is optimum for both applications. For example, the single conversion tuner with traditional MOPLL technology will have neither sufficient signal handling nor channel flatness to meet the requirements of DCR. The channel flatness issue is exacerbated when the tracking filters are modified to try and meet the full image rejection requirements of A/74 due to the higher Q factor required. Conversely, a tuner optimized for cable will have neither sufficient narrow band signal handling nor NF to meet the A/74 s requirements. As previously described, a single conversion tuner based on an image rejection MOPLL will require 30dB less image cancellation. The net consequence of this is that a significantly flatter passband compatible with DCR requirements can be achieved. In addition, the higher signal handling, which also enhances A/74 D/U performance, is of benefit in meeting the DCR intermodulation requirements. It is now technically feasible to develop a single hybrid PnP tuner that is fully compatible with both A/74 and DCR requirements, while also reducing overall size and system cost. Designing a tuner that achieves A/74 performance is challenging. Using MOPLL technology that incorporates both image rejection and high signal-handling capability will (Listing 5) ease design and help speed time-to-market for better-performing products targeting a high-volume consumer market. A designer will find that traditional single conversion receiver architectures can be used with an MOPLL that uses image rejection with enhanced signal-handling capability. Figure 7 shows performance achieved from a tuner incorporating such a product with conventional tracking filter technology. This design exceeds the A/74 minimum recommended D/U performance under weak desired signal conditions by between 6dB to 10dB (depending on channel offset) to enable quality reception for the consumer. This new MOPLL technology replaces the two tuners currently required in PnP receivers with a single conversion architecture that meets the demands of both ATSC and DCR.