Initially, the video filter is a passive LC circuit, and now, the combination of amplifier and RC filter can be smaller, more efficient design. In addition, the twentieth century, developed in 60 years the sensitivity analysis and predistortion methods have overcome the weaknesses of poor early performance of the filter. High-performance op amp and the PC, special software to facilitate the design of broadband active filter, but it does not address any specific application problems.Video filters, and signal application-specific standard for each circuit design have brought subtle differences. The following are two main types of video applications:
Anti-aliasing filter: this type of device to be placed before the ADC is used to attenuate the signal above the Nyquist frequency components higher than the ADC sampling rate is half of the signal components. The filter frequency response of the transition zone is usually designed to be as steep as far as possible the purpose of the signal above the cutoff frequency of all filtered out.On the ITU-601 and other applications, this performance is through the analog filters, digital filters, and combined with the use of oversampling ADC to achieve. The image of the class of PC applications, a small amount of filtering is sufficient.
Reconstruction filter: Also known as (sin x) / x or zero-order to maintain correction, these devices are placed after the DAC, to eliminate ghost caused by the sampling, DAC clock is not to eliminate crosstalk. Few reconstruction filter anti-aliasing filter that as the selectivity, because the DAC has to maintain the function of the role of filtering - This effect reduces the selectivity of the request, but to respond to their toll. The existing video formats, including RGB, component video, composite video and RGB PC images.
All applications and Zhishi want the video filter can achieve the "linear phase", which is the group delay (the delay of different frequency) characteristics required. Phase linearity required depending on the application and video formats. For example, anti-aliasing filter and weight than standard composite video reconstruction applications and more stringent requirements. With the standard requirements of various applications by the NTSC, PAL / DVB, ITU, SMPTE, and ESA requirements. This article will provide a comparison of different filters for a given application or standard to determine the best design.
Filter and its characteristics
Whether used for anti-aliasing or reconstruction, the filter must have low-pass characteristics, so that the video frame rate can be adopted. Therefore, designers need to carefully consider the AC coupling. Low-pass filter characteristics according to their magnitude or the polynomial used to describe its classification (Bessel, Butterworth, Chebyshev or Cauer). Figure 1 shows normalized after 1-rad-bandwidth characteristics of these filters. Should normally be selected with the best selectivity and the minimum number of poles (reducing costs) filter, but the additional phase linearity requirements limit the range of choice.
Filter determines the phase linearity or group delay envelope delay (GD) and frequency. Flat group delay indicates that all frequencies are delayed the same time, in the time domain waveform which is conducive to maintain. Therefore, the group delay difference is more important than the absolute group delay. Another independent parameter called a key difference between channels, between channels used to measure the "time according to the degree" should not be confused with the group delay. The video is concerned, how much difference in group delay is acceptable, and why? The answer depends on the application and video formats. For example, ITU-470 composite video signal group delay requirements are very loose. However, ITU-601 will have a lot stricter in order to ensure the stability of the video playback, both for the MPEG-2 compression, but also to control the phase jitter before serialization. So, to ensure phase linearity, should be considered what kind of filter characteristics it?
Figure 1, the group delay curve in the cut-off frequency there is a peak. This is near the cutoff frequency due to the steep phase change. In order to have a general concept to a 3-pole, 6MHz Butterworth filter, for example, in its group-delay variation within the bandwidth is about 20ns-25ns. Increase the selectivity of the filter poles or the difference will increase. Group delay is used to reduce the number of other differences include a more specific filter Bessel, phase approximation, Thompson-Butterworth, and LeGendre and so on. Nevertheless, in the video is still the most commonly used Butterworth filter characteristics.
Anti-aliasing filter design
For anti-aliasing filter, selectivity provided by the ITU-601 to determine the template, as shown in Figure 2. Bandwidth is defined as 5.75MHz (0.1dB, insertion loss at 6.75MHz Department to 12dB, 8MHz Department to 40dB, 0.1dB bandwidth of a group-delay difference (3ns. So that the performance requirements for a simple analog filter is too difficult in terms of , but by four times over-sampling can be adjusted to the requirements of 27MHz at 12dB, 32MHz at 40dB.
Use of software or normalized curve, we find-3dB bandwidth of 8.45MHz 5 pole Butterworth filter will be able to meet the selectivity requirement, but did not consider the group delay. For the latter, requires a delay of class, for this circuit, the need to focus on the op amp parameters as 0.1dB, 2VP-P bandwidth. The value used in type 1 and 2 receive the exact design. Figure 3a and 3b shows a schematic for such applications, and with the curve shows the gain and group delay characteristics, the design is based on four times over-sampling system. As MAX4450/4451 excellent transient response and large signal bandwidth (2VP-P when the 175MHz), so select it.
DAC reconstruction filter after the not well understood is one of the applications. Many designers believe that the introduction of reconstruction filter is to filter out the sampling clock, but this is far from the truth. When the signal is sampled, the sample contains a number of duplicate image signals, respectively, all aimed at sampling clock harmonics. Reconstruction filter to filter out the baseband sampling in addition to all other external signals. If the anti-aliasing filter to achieve its purpose, then the DAC output looks like the image in Figure 4 A, then it should be the right of all samples filtered out. Therefore, the requirements for the reconstruction of similar anti-aliasing, but only stay because each sampling instant, DAC will be for one clock cycle for each sample, so that to generate the approximation that we are familiar with a particular slash step-like waveform. Functional equivalent of a digital filter to maintain its characteristics similar to the Butterworth or Bessel filter response of half of the sampling frequency is reduced by 4dB. The second purpose of reconstruction filter is to compensate for this loss, which requires the circuit in Figure 5a that the magnitude of the equalizer. The equalizer circuit in a time delay based on the composition of Bessel filter with a similar response. It can be in accordance with the DAC sample rate (FS) to design. Figure 5b shows the rate of adoption and not use the DAC when the equalizer frequency response. And the delay level is similar, it can be included in any reconstruction filter.
There is also a response aimed at maintaining the sampling clock of the pole, you can completely eliminate the clock. However, most re-application or attenuation of the clock as its quality factor. Now understand the function of the reconstruction filter, we can proceed to design.
NTSC / PAL video reconstruction of the most common requirement is 13.5MHz at the attenuation> 20dB, at 27MHz at> 40dB, cut-off frequency depends on the video standard adopted. For two reasons, we chose three Sallen-Key pole Butterworth filter structure. First, the gain (+2) for driving back-terminated cables. Second, it can be adjusted between group delay, so that we do not need to delay equalizer group delay can be adjusted for optimal performance.
Active Video practical problems in filter design
Both manual design, or software-aided design, or combination of these two methods are filters, the actual response curve may not be fully in line with expectations. One reason is that the calculated response and the use of standard components is worth to the discrepancies between the actual response. Through the selection criteria (5%) capacity values derived resistance value capacitors can minimize errors. The reason is practical - you can buy 1% or 2% tolerance capacitor, but the capacitance of only 5% accuracy, while the resistance, the 1% and 1% tolerance precision resistors are readily available. This choice of components can provide the best range of consistency and the most accurate response.
Filter form, the possible situation of instability and oscillation. If this happens, the input shorted to ground, to see whether it continues to oscillate. If the oscillation stops, indicating that resistance is too high, reduce the design resistance oscillations can be eliminated. However, if the oscillations continue, please note that the oscillation frequency is close to the filter cutoff frequency or the cutoff frequency below. If so, the oscillations may be caused by components or parasitic cause. If the oscillation frequency is higher than the cutoff frequency, then the op amp may be caused or circuit board layout.
Good layout is like a work of art, but it is only based on some simple principles. Provide a clean supply voltage and a solid grounding is very important, which means that capacitors with low ESR filter, sometimes a regulator. Bypass capacitor connected from the circuit must be as small as possible, otherwise the parasitic inductance and capacitance of the resonator. Good ground plane is critical for good analog design, but as bandwidth increases, it may bring parasitic capacitance, the filter disorders. To avoid such problems, will affect the bottom line of components and away the ground plane can be removed.