
Analog IP for multimedia SoCs: an eye on a world of essential analog features
The increasing consumer demand for smaller electronic devices has been pushing the semiconductor technology roadmap to smaller process nodes over the last decade. Today’s consumers demand high-quality (e.g. Hi-Fi, 1080p, etc.), multi-functional and feature-rich electronic products at a low price. As a result, product differentiation is now, more than ever, achieved by increased functionality, higher performance, improved efficiency and more application-specific features. In this article, multimedia analog IP (intellectual property) refers to audio codecs and video analog front-ends (AFEs).
While in the past video AFE content consisted of 90 percent data conversion and 10 percent auxiliary circuitry, today’s advanced HDTV/xVGA video AFEs contain 25 percent data conversion and 75 percent application specific functions and low power modes. For example, features such as sync processing that are capable of generating the pixel rate clocks and multi-mode clamping circuits, among others, are needed to enable the multiple functions necessary in next-generation devices.
Similarly, audio IP used to be data-converter centric but, to dramatically increase product differentiation and address ever evolving quality and function requirements, today’s high-end audio codecs feature a myriad of signal-conditioning, signal-path selection and power-efficiency functions. These include high efficiency class-D and class-G drivers, PLL-less and cap-less operating modes and digital sound effects.
In the past, designers would turn to simple and compact IP functions (lower quality) or to discrete chips (higher quality) to solve this problem but today, there is a new breadth of analog solutions that allows designers to efficiently bring the highest possible quality into a chip through the integration of analog IP.
This article unveils how the analog functions that are built into an analog IP multimedia subsystem should be carefully selected to address the particular application requirements while guaranteeing the smallest silicon area and the lowest power consumption, a two-fold challenge that is commonly faced today by both system-on-chip (SoC) integrators and IP providers:
- Some of the new analog features and performance levels available in today’s analog IP are paramount to meet the high quality requirements of the most demanding multimedia SoCs, so they cannot be eliminated from the circuits as a way to reduce costs through area savings without jeopardizing performance. To better illustrate this point, examples of video AFE and video digital-to-analog converter (DAC) key features are presented to show the impact of the end result.
- Also, some of the new analog functions should not be substituted by digital functions as it runs the risk of compromising quality and sometimes functionality despite the fact that it would appear to be another way to gain area savings. This concept will be supported by examples of audio features that, when implemented, generate either advantages or issues depending on whether they are implemented in the analog or in the digital domain.
Video analog front-ends: the rise of a new set of functions
Video AFE circuits (also known as video analog-to-digital converters (ADCs) or video decoders) are essential components of digital home and personal entertainment systems, especially displays: Flat Panel TVs (Figure 1), PC monitors and more. A quick look at the back of a typical flat-panel digital TV shows different connectors that correspond to the different ways analog video signals are brought into the TV, such as RGB, YPbPr and S-Video.
Due to their high image quality and low power consumption compared to other popular digital video interfaces out there, there is a thriving market for consumer devices that use analog video as their primary connection. Additionally, computer displays and home entertainment equipment will continue to support the large installed base of set-top boxes, VCRs, DVD players, digital cameras and other equipment that only provide video in an analog format.
Figure 1 : Analog Video Inputs (back of a DTV)
To meet different application requirements, SoC designers must incorporate high-performance video AFEs that can capture and digitize video from both xVGA computer video sources and a wide range of analog display formats, including the increasingly popular 1080p high-definition TV (HDTV or simply HD) format while keeping silicon area and power consumption to a minimum.
As described in Figure 2, the video AFE “receives” the analog video signal that is typically sourced from a video DAC and transmitted over a 75Ω cable.
Figure 2 : Typical implementation of an HD analog video interface
Although this configuration is commonly used, video analog signals may also be originated differently. Another possible arrangement is used in TV broadcast, in which the video AFE can be used after the radio frequency (RF) down conversion and the resulting video signal can be fed directly to the analog front end. Overall, as the signal source and transmission channel is often unknown and not normalized, Alternating current (AC) coupling transmission is needed to create electrical isolation between the two sides of the system.
At its simplest form, a video AFE consists of one or more ADCs and a series of clamping, signal conditioning and filtering circuits. In some applications, such as HDTV and PC graphics, the video AFE also contains synchronism (sync) processing circuitry that is used to extract the timing and frame sync information embedded within the analog signal, as illustrated in Figure 3.
Figure 3 : Simplified video analog front-end block diagram
The sampling frequency and resolution of an ADC varies by application, with standard-definition TV (SDTV) requiring 10/12-bit at 27/54 megasamples per second (MS/s) and 1080p HDTV signals requiring 10-bit at 148.5 MS/s. Computer display applications require 8/10-bit resolution, with sampling frequencies as high as 205 MS/s to capture the full resolution of a WUXGA/60 Hz or UXGA/75 Hz resolution display. The variations in ADC requirements, however, are relatively small compared to the differences in the requirements for the analog portion of the video AFE’s signal chain.
Regardless of the application it is supporting, the video AFE’s analog signal chain is critical to its overall performance and, in most cases, a more demanding portion of the overall design effort than just the ADC. To understand why, let’s look at the things that must be done to a raw video signal before it can be cleanly digitized by the ADC:
- Clamping: A video signal often has a direct-current (DC) component that must be removed or brought to a normalized level that is acceptable for the ADC input. In SDTV applications, this is handled by a charge pump clamp, but more sophisticated techniques (such as bottom or mid-scale clamping, sync-tip and others) are also required for computer monitors and HDTVs.
- Gain and offset control: A programmable gain amplifier (PGA) is used to adjust the incoming signal’s swing to match the ADC’s input range and control each color component’s saturation (in component video systems) or the image brightness/contrast (in composite video systems).
- Clock recovery and sync generation: These circuits extract the sync signals embedded within the analog video signal and provide the digital data stream with the timing required to produce a properly framed image. The sync circuit’s ability to recover timing information from weak or noisy signals plays a major role in the video AFE’s overall performance and its ability to function under real-world conditions.
- Filtering: Programmable filters and de-glitching circuits are used to clean up the sync signal(s) being sent to the video AFE’s sync and clock recovery circuits. In addition, HDTV applications require a very sharp (third order) low-pass anti-aliasing filtering to be applied to the video stream to suppress unwanted sampling artifacts and improve noise performance by strongly attenuating out-of-band noise that could otherwise be folded back to baseband during the analog-to-digital conversion.
A video AFE’s capability to deliver high-quality video images relies not only on the performance of the ADC but also depends greatly on the remaining functions’ quality and capability to operate according to different video format requirements.
In a word, today’s video AFEs are no longer just high speed analog-to-digital data converters – a breadth of equally important analog functions are pivotal to guarantee high quality images and full compliance with multiple HD TV/xVGA formats as well as lower quality legacy video sources, and cannot be simply eliminated from the circuits as a way to reduce costs without jeopardizing performance and usability.
Hi-Fi Audio Codecs: multiple applications drive multiple functions’ line-ups
Today’s popular consumer gadgets need to be portable. This means they need to have a long battery life and be small. Audio processing is used to help achieve these goals.
It is essential to consumer electronic applications, such as mobile phones, MP3 players, set-top boxes and a host of other products in which size and power consumption are the critical design criteria. Additionally, high-end products, such as digital TVs, set top boxes, gaming consoles and smart phones, have a competitive edge with high quality, high-fidelity (Hi-Fi) audio capabilities.
Since different applications have a multitude of different requirements, determining the functions’ lineup that optimally serves a particular target application, while keeping silicon area and power consumption to an absolute minimum, is a challenge commonly faced by both SoC integrators and IP providers.
To understand the different requirements possibilities, let’s take a look at two different examples – a smartphone and a DTV (Figure 4).
On a typical smartphone, it is possible to identify multiple human audio interfaces – built in microphone, earpiece, hands free loudspeaker and a jack for external headsets and microphone. All these correspond to audio inputs and outputs that are different in nature and as such, have different function requirements.
Figure 4 : Analog audio interfaces on a a) smartphone (upper)
and b) DTV (lower)
While the built in earpiece and the external headset require drivers capable of driving down to 16 Ohms, hands free loudspeakers typically present 8Ω or 4Ω to the driver and require higher output acoustic power. These different requirements establish the need for drivers with different characteristics. Taking into consideration the mobile nature of a smartphone, in which battery life is one of the most important characteristics, high efficiency drivers should be used. This commonly results in the headset driver to be a class-G type amplifier, and the loudspeaker and the earpiece, in most cases to be class-D type amplifiers.
Conversely, most DTV sets have built in loudspeakers. Some DTVs also have a stereo line output which can be used to connect the TV to an A/V receiver or external loudspeakers.
The built in loudspeakers on a DTV typically deliver over 10 W (sometimes even more than 50 W) of electrical power. The result of this is that the audio drivers needed to deliver such power levels cannot be integrated in 3.3 V digital SoCs and need to be external. As such, the codec embedded benefits of the non-inclusion of drivers other than simple and compact line drivers, mostly class-A or class-AB high linearity amplifiers, significantly reduce its silicon area.
The right selection of the driver’s lineup for a given audio codec is pivotal to guaranteeing that the silicon area used by the IP is exactly the required area, not more, and is just one of the several decision points when dealing with analog audio.
Another typical example of function tradeoffs, in this case between analog or digital implementation, is volume control. Audio levels can be adjusted using volume controls. These allow one to match the input levels to the ADCs range, adjust the playback volume or to adapt to varying signal levels. Volume controls can be implemented in digital domain, analog domain or both.
At a first glance, it may seem that it would be more advantageous to implement volume controls in the digital domain, thus benefitting from higher gate density available at deep submicron processes and hence, smaller silicon area. However, digital volume controls cannot substitute the analog volume controls in the functions of adapting the input signal levels to the ADC range or adjusting the playback volume. The reason is that the resolution of the data converters would not be scaled in the same ratio as the signal and a loss of dynamic range would result, as represented in Figure 5.
Figure 5 : Differences between analog and digital volume controls for the record (ADC) channel (left) and playback (DAC) channel (right).
The two examples above – selecting the appropriate driver’s lineup and the balancing volume controls between the analog and digital implementations – tell us that audio analog functions cannot always be replaced by digital functions without compromising quality and functionality and that these functions should be carefully selected for each application requirement in order to guarantee the smallest possible silicon area.
More analog multimedia functions
There are other equally important analog multimedia functions worth considering in multimedia SoCs – one example is the video DAC.
A quick look at the back of a typical set-top box or DVD player (Figure 6) will reveal a number of different connectors that correspond to the different ways analog video signals are transmitted.
Figure 6 : Back of a Set Top Box with, among others, an analog video output (Video DAC)
For example, there may be a VGA connector, a triple-component video RCA connector, a composite video connector, an S-video DIN connector and, depending on the region, a SCART connector. In spite of the various connectors normally featured in video devices, a single signal source that is routed to the connector is desirable from a system-cost point of view – a video DAC. This single source must be able to transmit without compromising the signal integrity of any of the video formats, from the narrowband SDTV standards (PAL, NTSC, SECAM) to the broadband HDTV and VGA graphics video formats.
The most flexible and power-efficient solution is built around a current-steering DAC that can drive the output directly to the video cable. Current-steering DACs also have several characteristics that make them the ideal solution for integrated SoC video systems.
Although it may look like a multimedia video, DAC is a single-function analog IP, and to a certain extent the vast majority of available IP solutions are in fact just current-steering DACs. There are a breadth of essential functions and performance levels that actually determine whether the result is a high quality and crisp video signal (image) or a lower end product.
As an example, video DACs are widely used in mobile applications, such a digital video cameras (DVC). As such, power savings are paramount to extend battery life. One of the functions that, although fundamental, are often not present in video DACs is the cable detection capability. Cable detection allows the video DAC to automatically go into a low power mode whenever there’s no video cable connected, which results in dramatic power savings. Saving silicon area by not adding such function will result in a lower grade product.
In sum, analog functions are pivotal to differentiate the next generation products. They cannot be eliminated from the circuits as a way to reduce costs through area savings without jeopardizing performance and quality of the end applications.
Conclusion
In today’s multimedia world, consumers demand high quality and feature-rich products at a low price, and product differentiation is now more than ever.
Analog functions play a two-fold role in this world: they provide SoCs with functions that are essential to address today’s power and quality requirements and they enable functions that are paramount to address higher end and higher value product requirements.
The analog functions built into an analog IP multimedia subsystem should be carefully selected to address particular application requirements in order to guarantee the smallest possible silicon area. The examples cited earlier are ample evidence that there is a path to a product’s success that requires keeping analog functions in the analog domain, not moving them to digital implementations. SoCs that are intended for higher end applications and market positioning also benefit from the addition of analog functions and if removed (e.g., to save silicon area), severely degrade product differentiation and the ability to address fundamental application requirements.
About the author
Navraj Nandra joined Synopsys in February 2005 and is the senior director of marketing for the DesignWare Analog and mixed-signal IP products. He has worked in the semiconductor industry since the mid 80’s as an analog/mixed signal IC designer for Philips Semiconductors, Austria Micro Systems, (San Jose & Austria) and EM-Marin (Switzerland). He has been responsible for the complete design of a number of analog front ends in application areas such as digital audio, RFID and automotive. He joined Synopsys from Barcelona Design where he was Director of Application Engineering. During his four years at Barcelona he was responsible for pre- and post-sales support for Barcelona’s analog synthesis technology.
Navraj holds a masters degree in Microelectronics, majoring in analog IC design, from Brunel University and a post-graduate diploma in Process Technology from Middlesex University. He has presented at numerous technical conferences on mixed-signal design, analog IP and analog synthesis/EDA.
