Analog video time base correction and processing for nonstandard TV signals - Embedded.com

Analog video time base correction and processing for nonstandard TV signals

Despite the rapid advancement of digital TV, analog TV will remain dominant for both transmission and display for several years to come. Analog video formats, like digital formats, have precise specifications as to how to format video properly. These specifications include how the luma, chroma, and synchronization information is packaged to provide the line, field, and frames of video information that recreate the images observed on TV screens. To display the video information accurately, each of these components must be extracted correctly from the video signal. Similarly, the timing and phase information must be maintained or recreated as it was when the video signal was first encoded at the source.

This article outlines the challenges encountered when decoding and restoring a correct time base to nonstandard input video sources. Examples of where restoring a correct time base is critical to maintaining good image quality include analog tuners (NTSC, PAL, and SECAM) used on most TVs today and the venerable VCR. Since most people today continue to rely on analog tuners and VCRs, it is important that advanced digital video decoders utilize the latest clock reconstruction techniques to produce outstanding image quality in DTVs.

Noise-induced time-base inaccuracies
NTSC, PAL, and SECAM video consist of lines of video information packaged in fields and frames. The lines, fields, and frames are identifiable by embedded synchronization. The correct extraction of this information enables the receiving device–TV, VCR, or projector–to reconstruct the images and provide a visual display.

Video signals consist of various components, each of which can be altered or corrupted within the transmission path, resulting in distortion of some video package aspects. For RF transmitted signals, the synchronization information is normally present on the recovered signal, but its detection and extraction can be difficult or impossible because of excessive noise. It is important to note that even when recovering the synchronization is possible, its detection can be offset due to noise, which in turn introduces jitter on the recovered synchronization information.

Figure 1: Line Synchronization Stream without Noise at Source

Figure 1 shows a representation of a typical stream of line synchronization information. All line lengths are the same and conform to nominal specification requirements. Receiving and extracting the correct synchronization results in a proper and stable display. Figure 2 shows that slicing the synchronization information from this stream results in a stable display.

Figure 2: Stable Synchronization Extraction with No Vertical Jitter on Displayed Image

NEXT: Noisy sync problems
Noisy sync
Noise is commonly introduced in the RF transmission path. The induced noise results in the active video region that will be displayed becoming 'snowy', soft, and corrupted by noise artifacts. However, it is not just the active video data that becomes corrupted. The noise can also corrupt the embedded synchronization necessary for the receiving device to reconstruct the image.

Figure 3 is an example of how noise can distort the synchronization information, resulting in a misinterpretation of the synchronization. This shows the resulting stripped synchronization that can be caused by the noise. Note that the time base is corrupted and jitter introduced. The effect of the jitter on the displayed video is serrations at the start and end of each line (see Figure 4 ).

Figure 3: Noisy Synchronization Information Resulting in Extracted Synchronization Jitter

Figure 4: HSYNC Jitter Resulting in Line to Line Shifting of Displayed Image

A typical noisy input from a tuner source is shown in Figure 5 . It demonstrates the difficulty in determining the synchronization information.


Figure 5: Actual Video Output from Tuner Source Measured on Input to Decoder

The very high noise level within the signal makes it difficult, firstly, to determine where the synchronization information is and, secondly, to determine its exact start and stop positions. A critical requirement for digital video decoders is to meet this challenge. High-performance video decoders can maintain lock to RF video signals with powers of less than 20 dBµV, thus enabling the display device (TV) to show these images without horizontal jitter or vertical rolling. Although noise is an objectionable artifact on any display, its presence becomes much worse when the display loses its time base, resulting in vertical roll and horizontal jitter.

Figure 6 illustrates the decoder output re-encoded into an analog format.


Figure 6: Decoder Output with Correct Noise-Free Synchronization Restored

NEXT: Video amplifiers go awry
Amplifier distortions
The introduction of noise is a common issue for video that has come through a transmission path. It is also common for components of the video signal to be attenuated or amplified, resulting in nonlinear characteristics in the video package.

Many display devices use the synchronization depth as a reference and apply gain until it reaches its nominal value. In a situation where the video package has become nonlinear, this results in incorrect gain being applied to the video image.

An example of a video signal seen at the input to a decoder is shown in Figure 7 . The synchronization level is reduced close to the blanking level, while the other components of the video signal remain at the proper amplitudes.

Figure 7: Input Video Signal with Attenuated Synchronization

Advanced digital video processing will restore synchronization information while maintaining the other signal components at their nominal level, as shown in Figure 8 .

Figure 8: Output Video Signal with Correct Synchronization Level Restored

NEXT: VCR head switching time-base errors
VCR-induced time-base errors
Unlike common transmission path induced errors where synchronization is normally present but can be distorted, video from VCR sources can have missing or incorrect synchronization information.

VCRs are essentially mechanical devices. Variance in the mechanism, including motor speed, belt wear, and head switching, can introduce time-base corruption. In the case of head switching and VCR trick modes, the time base is not only corrupted, but can also be lost (see Figure 9 ).

When head switching occurs, all video and synchronization is lost. The result is a flat dc output for this duration. When such errors occur, the downstream display devices do not receive adequate synchronization information to reconstruct the image correctly.


View full size
Figure 9: VCR Missing Signal During Head Switch

Depending on the amount of missing information and the ability of the decoder to reconstruct the TV images with this information, the effect can vary from a minor top curl artifact to total loss of horizontal and vertical synchronization. Figure 10 shows a typical top curl artifact caused by poor time-base signals from a VCR source.

Figure 10: Output Image with Top Curl

The latest advanced digital line-length tracking (ADLLT) technology used in devices such as Analog Devices' ADV7180, ADV7184, and ADV7188 decoders ensures the correct regeneration of the missing or inaccurate synchronization information ensuring that the top curl like artifacts are eliminated (see Figure 11 ).

Figure 11: Output Image with Top Curl Eliminated

NEXT: Processing nonstandard input video with advanced digital video decoders
Processing nonstandard input video with advanced digital video decoders
Advanced digital video decoders filter the window in which they look to detect the synchronization. In addition, the decoders use HSYNC and VSYNC processor blocks to ensure that the synchronization information is correctly extracted. The filters ensure that the decoder gates the time period in which it looks for synchronization information. Previously, excessive noise outside of this region would have dipped below the slice level and been seen as synchronization. The synchronization PLL and processor blocks ensure that synchronization detected within the gated period is correctly aligned. Because line locked decoders use the HSYNC as a timing reference for color burst detection and subsequent video decoding, its proper detection is essential.

A critical requirement for any decoder is to correctly separate luma and chroma information. This is dependent on the ability of the decoder to extract the color subcarrier and correctly generate the proper number of samples in each horizontal synchronization period. Advanced video decoders use fixed frequency 4x oversampling to digitize the input video. A resampler block within the decoder ensures that a fixed number of samples per line are consistently output. The resampler PLL varies in frequency to obtain this fixed number of output samples. This resampling method that delivers a fixed number of samples in each horizontal synchronization period is referred to as line-locked time-base correction.

With this type of architecture, a simulated line-lock clock (LLC) is generated. Note that although the line-locked time-base correction results in a fixed number of samples per line; samples are not at a fixed 27 MHz rate, but vary with the resampler PLL, i.e., 27 MHz ± 5%.

Pixel information is fed to the output FIFO before it is output from the decoder. Special FIFO control techniques are used to allow a smooth flow of data from the FIFO to the output pixel drivers. Although the long-term clock jitter is still present on the output clock, the short-term jitter variations are smoothed out. Figure 12 provides a representation of output jitter across two fields of video. The peak jitter measurement is the same on both plots, but the short-term jitter in the left plot has been removed. It is the ability to remove this short-term jitter that allows the decoder to now operate in a direct back-to-back configuration with the digital video encoder.

Figure 12: LLC Jitter Performance

Digital video decoders, such as those from Analog Devices, incorporate the functionality of synchronization detection and extraction blocks with a resampler and advanced back end FIFO management.

Summary
Analog TV is still more widely used than digital TV for transmission and display purposes. In order for video information to be displayed accurately on an analog TV, it is important that luma, chroma, and synchronization information is extracted correctly from the video signal, and that timing and phase information is maintained or recreated as it was when the video signal was encoded at the source.

This article described several of the processing problems that can be encountered when decoding and restoring a correct time base to nonstandard input video sources. The ability to restore such a time base has direct consequences for the quality and stabilization of the displayed image. Failing to correctly process the video causes vertical and horizontal synchronization issues that can, in turn, cause rolling and jitter on the displayed image.

The latest techniques for the correct regeneration of the missing or inaccurate synchronization information, such as advanced digital line-length tracking (ADLLT) technology, used in the ADV7188, ADV7184, and ADV7180 decoders, ensures optimized processing of a wide range of nonstandard signal types. Even for the weakest signal strengths or the poorest quality VCR sources, it is now possible to display an image on a user's TV that is both stable and viewable.

About the author
Frank Kearney joined Analog Devices in 1988 after graduation from Limerick School of Engineering. In 2001, he transferred from his role as equipment engineer in the manufacturing group, joining the advanced television (ATV) applications group as a staff engineer. He can be reached at

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.