Designing tight performance parameters for today's high-performance
broadband tuners requires careful attention to key design details.
More than 300 million broadband tuners are produced globally each year for operation in the TV band. These tuners are integrated into a variety of consumer electronics, ranging from familiar household standards such as televisions and VCRs to newer, more complex devices that include cable-set top boxes, cable modems, cable telephony systems, WebTVs, PC/TVs and the various implementations of digital television.
Functioning as the RF broadband “gateway,” the basic function of a tuner in these devices is to receive all available channels in the input bandwidth. It is also required to select the desired channel and reject all others, and to translate the desired channel to a standard intermediate frequency (IF). These traditional tuners operate over a frequency range of 54 to 862 MHz, taking into consideration those frequencies used by broadcast television and cable operators.
Tuners that are enabling products to offer PC, television, and Internet functionality have different performance requirements than the traditional TV tuner. As applications become more sophisticated, tuners with higher performance are required. New concerns brought on by the latest tuner applications include smaller form factors, high reliability, compliance with standards (data over cable system interface specification [DOCSIS], OpenCable, PacketCable), and ease of manufacture.
For tuners operating in the TV band, performance can be summarized by five parameters: dynamic range, phase noise, noise figure, spurious responses and image rejection. The relative importance of each parameter and its specification range vary by application. For the designer of the latest communications solutions, understanding these tuner performance parameters and the available configuration options is essential to successful, high-performance, low-cost design.
Dynamic range defines a tuner's ability to handle signals of varying strengths. For example, a TV tuner must be able to receive a weak signal from a distant transmitter while simultaneously handling a strong signal if the transmitter is close. Thus, a tuner must incorporate amplification that varies from strong amplification to strong attenuation. This degree of gain variation is the dynamic range of the receiver.
Different applications require different degrees of gain variation. For example, a tuner for digital terrestrial applications may be required to receive a signal as weak as -83 dBm, and simultaneously be able to handle a signal as strong as -15 dBm. The difference between these two signal levels is 68 dB, or 6.8 orders of magnitude, and that is the dynamic range of the receiver.
In cable applications, a much higher degree of uniformity exists among the channels because they all originate from a common head end and are transmitted at similar power levels. There is, however, tilt in the spectrum, which derives from frequency-dependent losses in the cabling and the methods used to compensate for it. Also, depending on where a home connects to a cable, with respect to the most recent signal amplification, the signal may be either relatively strong or relatively weak. In general, the cable network is regulated so that the minimum analog video carrier signal strength seen is 0 dB/mV and the strongest signal seen is +20 dB/mV. For digitally modulated signals, DOCSIS provides for signal levels that are -15 to +15 dB/mV. Typically, the dynamic range for a cable modem would be 30 dB for an analog cable receiver, and 20 dB and 35 dB for a tuner designed to receive both analog and digital signals.
In the above discussion, two sets of units are used for signal levels: dBm, popular in the terrestrial community; and dB/mV, popular in the cable community. These units are related by 48.75, such that power in dB/mV = power in dBm + 49 (approximately). Also, it is typical in a receiver to divide the gain or amplification over two stages: intermediate frequency and RF. The RF amplifier is in the front end of the tuner, while the IF amplifier is in the IF stage immediately following the tuner. Common practice is to operate the RF amplifier at full gain until the desired input signal level is 0 dB/mV, followed by reducing the gain on a decibel-for-decibel basis as the signal level increases. In conjunction, the IF amplifier operates at full gain for the weakest signals and is decreased until the input signal level reaches 0 dB/mV, where it is set to the minimum gain. Consequently, the dynamic range adjustments for the tuner, excluding the IF amplifier, would be 20 dB for an analog cable application, 15 dB for a digital cable application, 20 dB for a combined analog/digital application, and 34 dB for a digital terrestrial application.
Quality of channel reception
In a communications system, wired or wireless, the tuner is responsible for receiving all available channels in the input bandwidth. It is also responsible for selecting a desired channel and rejecting all others, and translating the desired channel to a standard IF. Implicit in this process is retaining as much of the original signal's fidelity as possible, i.e., not adding appreciable noise and distortion. If the remaining channels are not effectively rejected, the system will experience interference. In an analog TV system this could result in unacceptable video (and audio) quality. In a digital TV system the picture could completely disappear. In a data system, interference can result in signal dropouts and reduced data transmission rates.
The quality of channel reception is greatly affected by the phase noise, noise figure, distortion, and image rejection experienced by the system. And, the relative importance of these qualities varies by application. For instance, analog TV tuners need to perform adequately for low distortion and low noise figure, while their digital counterparts must offer superior image rejection and low phase noise (see Figure 1). Analog cable requires good image rejection and low distortion, but it can tolerate phase noise and a substantial noise figure. Digital cable, on the other hand, requires a reasonable noise figure, but significantly better phase-noise performance than an analog system.
Because the channels are broadcast at high frequencies, it is necessary to downconvert these channels to an IF that is usable by decoding circuitry downstream from the tuner.
All tuners use mixers that are controlled by local oscillators (LO). Figure 2, for example, shows a double-conversion tuner that uses two LOs.
Although it is intended that the LOs oscillate at one pure frequency, real-world oscillators have a spectrum that, while dominated by the desired frequency, also contains undesired frequencies. Phase noise is the relative power of the undesired frequencies with respect to the one desired frequency, and is thus measured in decibels with respect to the carrier (dBc). Carrier in this case represents the desired frequency.
Phase noise has become important recently because of its impact on the reception of digitally modulated signals. One popular digital modulation technique is quadrature amplitude modulation (QAM), where the signal is divided into symbols. A symbol may be thought of as a sampling point of the signal for its amplitude and phase. For purposes of spectral efficiency, the resolution of the amplitude and phase limits the number of bits per symbol. Phase noise is the primary limiter of phase resolution in the signal and, consequently, achieves a large value of bits per Hertz, or spectral density.
The effects of phase noise on signal reception vary by application. Analog television, for instance, experiences little effect from phase noise unless it is excessive. Digital applications, such as cable modems, will not function unless phase noise is below a certain threshold. Tuners have improved recently as designers pay more attention to oscillator design and specifically target low phase noise as a requirement.
All tuners add some noise to the received signal. The noise figure is the measure of the noise added to a signal by passing it through a tuner. More specifically, the input signal has a signal-to-noise ratio (SNR), as does the signal output by the tuner. The difference between input SNR and output SNR is the noise figure. Noise figure is important in terrestrial applications because it defines the minimum detectable signal, or stated differently, the weakest signal receivable. It is important in receiving digitally modulated signals so that the SNR presented to downstream devices is maximized. For terrestrial applications, a 7 dB noise figure is required, while for cable applications, 10 dB usually suffices.
The nature of noise figure is that the front-end RF amplifier of the tuner dominates it. This amplifier is generally referred to as a low-noise amplifier (LNA). If it is low-noise and high-gain, it overcomes noisy downstream components such as mixers and filters. As process technologies advance, tuner noise figures should improve over time.
A spurious product is an undesired signal present in the output of the tuner. The source of this undesired signal may be distortion in the tuner's active circuits or an unwanted coupling from one circuit to another. If the tuner is designed well (the coupling issues are suitably suppressed), then the primary source of the spurious products is distortion.
In tuners, the distortion products that are most frequently discussed are intermodulation distortion (IMD) and cross modulation. Distortion products are generally introduced when handling large signals. For example, in terrestrial broadcast television, a typical problem is the receiving of a weak signal in the presence of a strong one. In cable systems, the problem relates to receiving one out of many relatively strong signals, where the other signals add up to create a large RF envelope.
Distortion is a bigger problem with analog in terms of its source and the visible effect. Analog TV signals are a problem because the bulk of the signal energy is concentrated near the carrier, resulting in strong peaks. By contrast, the energy in a digitally modulated signal is spread smoothly across the channel. Furthermore, the human visual perception system is acutely tuned to detecting patterns in video, which is the manifestation of distortion in analog video. In digital systems, the distortion is not perceptible until the signal becomes so affected that the data stream is compromised and video is lost. For analog systems, it is desirable to have all spurious products at least 57 dB below the carrier, or -57 dBc. For digital systems, there is less agreement on a specific level, but -50 dBc is frequently referred to as a specification.
Tuning involves translating signals in frequency. Suppose that a desired channel is translated to an IF. Then a channel two times that of the IF below the desired channel will be translated to -IF. Negative frequencies differ from positive ones only in the phase of their components, and therefore, interfere with the desired channel at the positive IF. This interfering channel is known as the image channel and must be rejected to a large degree for proper reception. (In broadband systems, the desired channel is translated to -IF, such that the image channel is twice that of the IF frequency above the desired channel.)
Image rejection can be addressed with filters and/or with image-reject mixers. In a single-conversion tuner, a notch filter is used to reject the image channel prior to frequency translation. The performance of such a filter is limited to 50 to 60 dB in the UHF component of the TV band (470 MHz and up). Better performance is possible with dual-conversion tuners, where the first IF filter can suppress the image channel by an arbitrary amount, depending on the cost (and thereby performance) of the filter. For cost-effective, dual-conversion tuning systems, the preferred approach is to use a reasonably priced surface-acoustic wave (SAW) filter at first IF to achieve around 40 to 50 dB by itself, then to complement that with a specialized mixer called an image-reject mixer. Such a mixer can achieve an additional 35 to 40 dB of suppression. The combination allows for consistent image rejection in the range of better than 70 dB.
Image rejection requirements differ by application. In traditional broadcast television in the United States, the Federal Communications Commission (FCC) regulates the spectrum so that a broadcaster in a given locale is guaranteed that no other broadcaster will be transmitting at the image channel. New spectrum allocations for adding digital channels are no longer eliminating this hazard, however, so it will be possible that a digital broadcaster will have a strong analog broadcaster at his image channel. In that case, a tuner must have around 80 dB of image rejection for proper reception. In cable, the image channel nearly always exists and is at a power level similar to that of the desired channel. Consequently, 60 dB of image rejection is good for analog signals, and 50 dB is good for digitally modulated signals.
Designers of tuners suited for particular applications must balance any trade-offs among the key specifications of dynamic range, phase noise, noise figure, spurious products, and image rejection. Designers can choose from a single-conversion or double-conversion tuner design. Currently, tuners are available in subsystems, as modules or as integrated single-chip solutions.
Single-conversion tuners (see Figure 3) select the signal of interest by using a set of variable-frequency filters, called tracking filters because the center frequency of their bandpass characteristic tracks the center of the desired channel. Multiple filters are typically required because of the broad bandwidth of most systems. Because of this filtering approach, it is difficult to design a single-conversion tuner with high selectivity (good rejection of other channels.) Furthermore, these tracking filters require hand tuning during their manufacture to achieve the required filter characteristics.
Until recently, dual-conversion tuners were always used for cable applications because of their superior selectivity, necessary for handling packed (vs. sparse) cable spectrums. Furthermore, dual conversion tuners presented a better broadband impedance match to the cable, so that standing waves are reduced. Also, the LOs in a dual-conversion tuner operate outside of the input spectrum, thus minimizing the potential for interference via leakage onto the cable network. Recently, some cable designs are using enhanced single-conversion tuners because of their low cost. This is likely to be a short-lived phenomenon because nearly all single-chip tuners use dual-conversion techniques.
Regardless of the application or configuration, as communications designs become smaller, faster and more complex, they are requiring next-generation tuners to offer small form factors, high reliability and compliance with emerging standards, while supporting streamlined manufacturing. As tuners have become more specialized, they are extending far beyond their original functions as tuners for analog radio or television to ones that support a wealth of applications, including multimedia PC/TVs, cable-set top boxes, cable modems and digital televisions.
About the author
John Norsworthy is founder and CTO of Microtune Location. He previously served as vice president and general manager of Cirrus Logic's consumer video products division. In 1991, he co-founded Pixel Semiconductor, which developed products and technology that enabled the integration of video into the PC graphics subsystem, and served as vice president of engineering. From 1986 to 1991, Norsworthy served as vice president of advanced technologies for Visual Information Technologies (VITec). He holds 12 U.S. patents and he has a B.S.E.E. from the University of Illinois at Urbana-Champaign. He can be contacted at: www.microtune.com