By the time you read this, more than 20 million digital converters will have been purchased and installed in U.S. viewers' homes, and a large number will have been distributed throughout the world as well. Despite their small size and cost, the units pack a considerable amount of technology into a simple-to-use package. While the functions of these units are very similar to those in an integrated TV, the specific application to support legacy analog 4:3 televisions means there are some important differences. This month, we'll take a look at the functional elements of these units, as well as how digital-to-analog (D-A) conversion works at the circuit level.
The functional structure of a typical DTV converter is shown in Figure 1. While some converters may combine several of these functions into integrated silicon, or even implement the functions in software on a digital signal processor (DSP), the basic elements remain the same.
While the tuner in a DTV converter is similar to that in an integrated DTV (or HDTV) receiver, the technical requirements for converters sold under the U.S. converter box program are more stringent. As set forth in the U.S. Dept. of Commerce's NTIA Rules for coupon-eligible converter boxes, these units must meet a technical specification that includes minimum requirements for RF sensitivity, phase noise immunity, interference rejection for co-channel, first-adjacent and taboo channels, burst noise immunity, and multipath rejection.
The transport demultiplexer in these units parse the data stream into its constituent video, audio and data elements, and also conveys the important synchronization data that is used to properly reconstruct the video and audio programs. The video decoder takes the MPEG-compressed video data and converts this to viewable video.
Video format converter
A format converter is needed to display all source material properly on a 4:3 television. This conversion includes several different functions, such as scaling, resampling, interlacing and cadence conversion. Scaling is used, for example, to resize a 16:9 image to fit on a 4:3 display, and can be done in various different ways: linear, squeezed or zoomed.
Because analog television broadcast uses a fixed sampling grid, and digital broadcast allows the use of many other sampling grids, resampling is required when downconverting from the digital source formats. And, resampling requires some degree of digital filtering to avoid objectionable aliasing. However, in order to produce a low-cost product, a very simple filter may be employed, resulting in some amount of visible artifacts. Also, since digital broadcasts can use progressive scan, an interlacer is needed to display progressive video on an analog TV, and this may additionally introduce vertical-temporal aliasing. Finally, a 3:2 cadence converter can be used for film pulldown to 30Hz (for NTSC) because film-rate encoding is an allowed transmission format.
Metadata decoder, graphics processor, modulator
The digital transport stream also contains metadata that conveys program and system information that can be used to provide the user with tuned channel and program information, and other program-related information such as closed captioning. A graphics processor then uses this information to generate a graphical interface for the viewer. Ultimately, a component such as a Digital ENCoder (DENC) converts the digital video to the appropriate analog output interface format, with proper sync and color burst, for presentation to either an RF modulator, or baseband output to a video display. A modulator is needed to interface with a tuner-equipped analog TV; alternatively, baseband video outputs can be used to interface with the composite inputs on a TV.
In order to provide legacy support for analog broadcasters continuing their analog transmissions after the analog cutoff date, an analog passthrough is also provided on some units. This allows the user to bypass the converter and continue to receive analog broadcasts on the analog TV.
D-A video signal conversion
Buried within the video output function (within the graphics processor in our example) will be a basic circuit element — the digital-to-analog video converter itself, called a video D-A converter or DAC. In order to convert the digital representation of a signal to an analog one, a circuit such as the R-2R resistor ladder shown in Figure 2 is often integrated within a D-A chip.
The output signal appearing at Vo will be a stepwise approximation of the video signal, also known as a zero-order hold version of the sampled signal. This signal then requires an analog lowpass filter to remove the sampling harmonics; the filter can also be compensated to counteract the inherent sin(x)/x response of the zero-order hold circuit. In practice, the linearity of this type of D-A converter will depend on the accuracy of the resistor ladder network, something that is not difficult to do using modern silicon fabrication techniques.
Another type of D-A converter is the Binary Weighted DAC. In this circuit, each bit of the parallel digital input switches on or off a weighted current source; these currents are then added together and converted to a voltage, forming the analog output. As before, filtering is needed for proper signal construction. One critical characteristic of D-A conversion is the linearity of the converter. If the voltage associated with each bit is not precisely generated and summed, the resulting signal will have distortion associated with it.
For video signals, nonlinearity can be tolerated to some extent, but not so with audio. With audio signals, an oversampling converter can be used to both simplify the D-A conversion and to lower the quantization noise of the signal. This is because upconverting the signal to a very high sample rate will spread the quantization noise over the entire sampling spectrum; thus, the final lowpass filter will remove a great deal of the quantization noise. In addition, the D-A converter can be constructed using a pulse-width modulation scheme that will result in a very high linearity of the conversion. This is not practical for video circuits, however, due to the extremely high sampling rates that would be required.
In practice, the video D-A converter within a DTV converter or receiver is integrated within the DENC chip mentioned earlier. This device takes 8-bit multiplexed 4:2:2 YCbCr video at a 13.5MHz rate and converts this to Y/C video (for S-Video interface), or CVBS (composite video blanking signal), for composite or RF interface. To support low-end TV sets that are subject to cross-color artifacts, the luminance signal can be notch-filtered at 3.58MHz (NTSC) or 4.43MHz (PAL). A similar notch filter is also available to prevent problems with the sound carrier. Both of these filters, as well as the chroma interpolation filters (needed to convert 4:2:2 sampling to 4:4:4), are implemented digitally within the DENC. A D-A converter on the same chip (usually 10 bits to account for addition of the Y and C signals) then provides the final analog output.
D-A converter boxes are an important part of the digital transition strategy for broadcasters, enabling all consumers to afford keeping up with this technology easily. Within these units, D-A converter technology provides support for analog displays. However, as integrated TVs and digital interfaces such as HDMI become more commonplace, we may eventually arrive at the situation where analog video signals are no longer present in any of these devices!
Aldo Cugnini is a consultant in the digital television industry.
Send questions and comments to: