Because there are many formats of signals used in current application systems, such as common VGA, Video, DVI, SDI, and HDMI, etc., when the signals are used in combination, the format unification and conversion problems will inevitably occur, including the following: Several conversion methods.

　　I. Conversion between analog and digital signals

　　The analog signals commonly used in the past include Video, VGA, Y / C, RGB, YUV, YPbPr, YCbCr, etc. Among them, VGA is used in the IT industry and has a corresponding relationship with DVI, while others are basically signals from the Video industry It can correspond to signals such as SDI, HDSDI or HDMI. We do not consider the situation of many protocols in digital signals, but only understand it from the conversion of the signal itself. The conversion from analog to digital is an A / D process (Anylog To Digital). It includes two main concepts: 1. Sampling rate 2. Number of quantized bits. The sampling rate refers to how many times the signal is sampled in one second. For example, if the sample is taken once a second, the sampling rate is 1HZ. If it is taken 1000 times, the sampling rate is 1KHZ. , 96KHZ, 192KHZ, etc., corresponding to the sampling rate of Video, generally Y is 13.5MHZ, U / V is 6.25MHZ, if it is R, G, B color gamut, all 13.5MHZ will be used, etc. Some introductions. The selection of the sampling rate should be based on the desired signal bandwidth. According to the Nyquist sampling law, the maximum bandwidth of the analog signal can reach 1/2 of the sampling rate, which can be understood by corresponding to the introduction in the first part. There must be a corresponding relationship with the line / field, and a corresponding relationship with the color load wave. Therefore, values ​​such as 13.5MHZ, 74, 25MHZ, etc. have appeared. The number of quantized bits refers to a normalized value, such as how many parts are quantized. If 1 is divided into two 1 / 2s, the number of sampled bits is 1, which is 21 = 2. It can describe whether the amplitude of a signal is larger or smaller than 1/2, but if the actual signal is larger than 1/2 and less than 1, you cannot know how big it is, only that it is between 1 and 1/2. The description is rough. If the number of sampling bits is 2, ie 22 = 4, the signal is divided into 4 parts. We can describe that a signal is between 3/4 and 1/2, which is more accurate. If the number of quantization bits is 8, then when we choose a signal, its accuracy can reach an error of 28 = 256, 1/256. If it is a 10-bit number, its error can reach 1/1024, and so on. At present, the audio sample quantization can reach 12bit, 16bit, and the video sample quantization is generally 8bit.

　　D / A (Digital ToAnylog), this is the reverse process of A / D.

　　2. Loss in A / D conversion

　　I learned from the process of A / D and D / A that digital signals can only describe analog signals with as little loss as possible, but cannot be completely restored. Here are a few questions:

　　1. Loss of bandwidth. According to the Nyquist sampling rate, the maximum bandwidth that can be retained is 1/2 of the sampling rate. This data has also been confirmed in experiments. The attached figure is our optical fiber transmission equipment for testing VGA signals. Measurement data, the yellow line is the spectrum composition of the VGA signal, and the blue line is the spectrum obtained through optical fiber transmission (which includes a set of A / D and D / A processes), because digital signals will not be transmitted during transmission. There are other losses, so the losses caused are borne by A / D, D / A. When looking at the 3dB point of the original signal, it is exactly a loss of 6dB. A compound A / D D / A process in theory, a description of 6dB of bandwidth loss.

　　2. The loss of the signal, because the number of sampling bits is limited, the description of the signal cannot be complete, and a lot of information is always lost. According to theoretical calculations, the loss caused by 8-bit sampling on the signal can reach 52-54dB. The loss of 10dB can reach about 60dB, (some signal loss is also a kind of noise for the original signal), so now the number of quantization bits is increasing, from 8bit to 10bit to 12bit, such as DEEPCOLOUR in HDMI and DP This brings an increase in bandwidth.

　　III. Conversion between digital signals

　　Digital signals need to be converted when they are multiplexed due to different coding forms. In principle, because they are all digital signals, there is no A / D and D / A process, and there will be no serious loss, but the actual Because various digital signals are based on different signals in the original generation, such as SDI and HDSDI are based on YUV color gamut space, while VGA, DVI, HDMI, DP, etc. are RGB color gamut space. We know that using the YUV color gamut can use UV bandwidth compression to reduce the bandwidth. Therefore, the signal RGB color gamut based on the YUV color gamut has a small signal bandwidth. The YUV color gamut is converted to the RGB color gamut. letter