Looking at the development of embedded DVR from the perspective of chip technology

The history of DVR technology development can be simply summarized as a process from single-channel to multi-channel, with increasing integration and decreasing cost per channel. Chip technology plays an extremely important role in this process. This article mainly analyzes the development process and future trends of DVR from the perspective of chip technology and shares it with readers.

With the rapid development of the national economy, embedded hard disk recorders, or DVRs, have made great progress and have been widely used. Among them, the hardware architecture of DVRs has also undergone some changes in the past few years of rapid development. Below we will analyze the history, present and future development trends of DVRs from the perspective of chip technology.

Basically, the history of DVR technology development can be simply summarized as a process from single-channel to multi-channel, with gradually increasing integration and each-channel cost from high to low. What we can see now is that in this technological change, the biggest driving force is cost reduction, and there is no sudden and independent technological evolution. Chip technology plays an extremely important role in this. It can be seen that the market is the only standard for testing DVR technology.

Next, we will introduce the basic specifications of DVR and its internal architecture. The main hardware specifications of DVRs currently on the market are that they must support both video and audio, USB interface, optional VGA output, network output, SATA hard disk connection, and mouse control is now also a standard configuration. Other software specifications include video resolution must reach more than 400 lines, and some even require 430 lines; compressed images generally require various formats and sizes to be convertible; the bright and color separation part must be done very well, without crosstalk, etc.

Mainstream DVR solutions

The main chip in the DVR solution design can be roughly divided into two parts, one is the front-end video decoding part and the picture synthesis part; the other is the back-end video compression and decompression part. As early as before 2003, most of the mainstream hardware solutions for Chinese DVRs were Philips Semiconductor's SAA7113+PNX13 series. (SAA7113 is the most popular video A/D conversion chip in the early days of Philips Semiconductor. It was launched very early, but due to the lack of upgrades for a long time, there is no comb filter, and the A/D quantization is also eight bits, so the performance is relatively poor now. The PNX13 series is also Philips Semiconductor's early Trimedia series DSP, with several models such as 1300/1301/1303, and the total resources are probably able to compress one CIF real-time image.) Let's call it the first generation of Chinese DVR solutions.

This solution has been popular for quite a long time, and even now it is still the mainstream solution in some companies. Its advantages and disadvantages are very obvious. The advantages are simple structure and high flexibility; the disadvantages are that it uses Philips' SAA7113 decoding chip, which has low resolution and no comb filter, affecting the overall video quality. The capability of PNX1300 is also weak, and it can only process one CIF channel. The integration of the whole solution is also low. In any case, the Philips semiconductor chip solution has the greatest influence in this era. In the same era, in addition to the Philips solution, there are also some ASIC solutions.

The advantage of this solution is that it uses Techwell's four-in-one video* for the first time, with high integration and good video quality; the disadvantage is that the ASIC compression method is not flexible enough at this time, and the specifications are still a little far from the best application. In the early stage of the formation of this DVR solution, there are some factors that deserve our attention.

First, from this period onwards, the mainstream compression method of China's DVR solutions gradually transitioned from the earliest Motion-JPEG and MPEG-1 to MPEG-4;

Second, a large number of professional DVR design, production and manufacturing companies have been formed in China.

Third, during this period, quite a number of semiconductor companies entered the Chinese DVR field, including Philips Semiconductor, Techwell, ADI, South Korea's Intime, Pentamicro, etc.

The second generation of DVR solutions in China is the mainstream solution provided by four chip solution providers: Techwell (Tewei Semiconductor), ADI, TI, and NXP. As a front-end video decoding chip provider, Techwell has occupied more than 90% of the front-end decoding market share by relying on its excellent video decoding quality and market-oriented technical support services. This can also be called a historical turning point in China's DVR industry.

In the industry, a single chip was the first to integrate four-channel audio and video decoding, screen segmentation and synthesis, motion detection, OSD, Scaler, etc. From this time on, Techwell emerged as an important player in the DVR market, replacing the previous monopoly of Philips decoding chips and changing the entire DVR market landscape.

In terms of DSP, ADI's DSP has the best price/performance ratio. Currently, most of Zhejiang Dahua adopts this solution, which occupies a pivotal position in the existing DVR solutions. Subsequently, TI emerged in the DVR field, relying on its strong DSP computing power and gaining a relatively large market share in the back-end compression part through Hangzhou Hikvision. NXP, as the successor of Philips DSP, still occupies a considerable market share in the back-end compression part. Most DSP users are still using NXP's Tremedia DSP series because it is simple and convenient to use and has great technical inheritance with the previous generation of popular PNX13 series DSP.

In addition, many DVR manufacturers still use it as the main control CPU because of its ease of use. Below we briefly introduce these solutions.

The reason why this solution is competitive is that it utilizes the high integration of TechwellTW2835 (TW2835 is a highly integrated front-end chip launched by Techwell at the end of 2006. It has four-channel audio and video decoding and integrated picture synthesis functions, etc. It is a DVR front-end dedicated chip) and the low cost of ADIBF561 (BF561 is ADI's Blackfin series DSP, and its total resources can compress 4-channel real-time CIF images), so that the cost of the entire solution is controlled while the quality is greatly guaranteed.

Compared with the previous Philips solution, it uses TW2835 to greatly reduce the number of main chips in the entire eight-way system, making the system architecture simpler, and the image resolution reaches more than 430 lines. The performance of the comb filter also reaches 4H, and the processing capability of complex images is greatly improved.

This solution is very cleverly designed because it uses the unique functions of TW2815, which makes the previous DM642 (DM642 is TI's largest DSP in the DVR industry, and its total resources can compress 8 channels of real-time CIF images) solution smoothly change from one-core four-channel to one-core eight-channel, which greatly reduces the cost of TI's solution per channel. However, due to TI's market strategy, the cost of DM642 is very high, resulting in this solution not being recognized by all manufacturers. Hangzhou Hikvision is the biggest supporter of this solution.

The advantage of this solution is that it greatly improves the integration. It uses Techwell's four-in-one decoding plus a single-chip solution for image splitter, which integrates resources to the maximum extent and improves efficiency. In addition, this solution uses the latest Philips 15 and 17 series DSPs at the back end, and the image processing capability has been increased to 8-channel real-time CIF or 2-channel real-time D1, MPEG-4, and Trimedia is used as the main control chip. This solution brings some new concepts to the Chinese DVR industry, such as image splitting, one chip with eight channels, multi-channel playback, non-real-time applications, etc.

The third generation of DVR solutions in China, which are the solutions currently being designed and the solutions to come, are the TW2835+SOC solutions that are popular and cost-effective. The SOCs in these solutions can be Taiwan's Faraday, the United States' Mobilygen, China's Hisilicon, and South Korea's Pentamicro, etc.

The advantages of this solution are obvious. The entire solution only uses two main chips, TW2835+SOC.

The high integration of TW2835 and the low cost and high cost performance of SOC (In the mid-1990s, inspired by the use of ASIC to implement chipsets, the idea of ​​integrating all different functional blocks of a complete computer directly on a silicon chip at one time emerged. This chip was initially named System on a Chip (SoC), and its literal Chinese name is system-on-chip.) are combined to make the entire solution the most cost-effective.

At present, there are dozens of SOC manufacturers on the market, all of whom originally switched from consumer electronics to the security industry. Their common feature is that they have strong compression capabilities, but the chip specifications are not comprehensive, so they all chose Techwell as their front-end strategic partner to make up for this shortcoming. Techwell's TW2835 plus Zhiyuan's FIC8180 solution may be the one with the largest number of mass-produced customers in the current DVR market. It has a better time to market, the integrated ARM speed is faster, and the market price is more competitive;

Another popular ASIC solution is Techwell's TW2834 plus HiSilicon's Hi3510. This is because Hi3510 is a SOC with a high degree of integration, and HiSilicon has good technical support. As for Korean chip solutions, they were popular in the past few years, but with the increase in the number of security semiconductor companies, their competitiveness has become increasingly low, and they have almost disappeared. This is because they are weaker than American companies in terms of technology, inferior to Taiwanese companies in terms of cost, and have poor localization, resulting in technical support and market understanding being lost to local companies. American chips are almost all very powerful in compression, but due to cost reasons, they may eventually occupy some high-end ASIC markets.

Development Trend of DVR Chip Technology

In the future, there are two directions for DVR that we have to pay attention to. One is the new SOC with more comprehensive and powerful functions; the other is the DSP that continues to work at a relatively faster main frequency. In general, SOC will definitely become the mainstream in the future, mainly for two reasons: first, once the DVR specifications are determined, the compression capability of SOC will increase very rapidly, and it will be easier to mass-produce, which will make it more cost-effective; second, the threshold for SOC development is relatively low, which makes it possible to quickly launch products and shorten the cycle of new finished products. As for DSP, they will definitely continue to occupy the high-end market in the future, and the market needs more flexibility, but the number of shipments will drop on a large scale.

In addition to these traditional DVRs, there are some other technical trends that deserve our attention. First, the focus of future DVR research and development will shift from the compression part to the display part, because the compression part has become increasingly mature, and the display part is the biggest reference factor for end customers to purchase DVRs. At present, 1080P will become an important guiding concept in the future; second, the back-end chip part will gradually shift to centralized control and intelligent analysis.