Analog Front End (AFE) Principles and Selection Guide

The object of analog front-end processing is the analog TV and analog sound signals given by the signal source. Its main functions include the following aspects:

 

Signal amplification: When the received signal is too weak and cannot meet the system's carrier-to-noise ratio requirements, a low-noise amplifier should be used at the front end to amplify it in order to improve the carrier-to-noise ratio.

 

Frequency conversion: In order to achieve a certain configuration of the transmission channel, and sometimes to avoid certain interference, the front end needs to convert certain channels. For example, early cable TV systems basically transmit signals in the VHF band. For individual programs broadcast in the UHF band, a channel converter can be used to convert them from the UHF band to the VHF band. In addition, for areas close to TV transmission towers, since the TV signal is very strong, the user's TV will directly sense the strong signal. This signal and the same TV signal received by the cable TV front end will enter the TV, but there is a time difference between the two, which will form a ghost image on the TV image that is difficult to eliminate. Therefore, it is also necessary to convert the channel signal into another channel signal.

 

Modulation and demodulation: When receiving satellite and microwave signals, they must first be demodulated to restore the video and audio signals, and then modulated into the radio frequency signal of the selected channel; self-produced programs also need to be modulated before entering the mixer; in addition, some open-circuit signals are also processed using a demodulation-modulation conversion method.

 

Adjacent frequency processing: Cable TV systems use adjacent frequency transmission to make full use of spectrum resources and transmit as many programs as possible within a limited frequency band, but it will also cause adjacent frequency interference problems. Therefore, it is necessary to use various technical measures at the front end to perform adjacent frequency processing to eliminate adjacent frequency interference to the greatest extent possible. Adjacent frequency processing mainly includes surface acoustic wave filtering, phase-locked loop (PLL) frequency synthesis technology, image and audio channel processing, A/V ratio adjustable technology, etc., which are used to complete modulation, demodulation, frequency conversion, mixing and other functions.

 

Level adjustment and control: used to adjust and control the level of each channel so that the level fluctuation within and between channels does not exceed the required range.

 

Mixing: The purpose of mixing is to combine all processed signals so that they can be transmitted using one line.

 

AFE Selection

 

Now many novel analog front ends (AFE-Analog Front End) are aimed at various applications: wired and wireless communications, industrial electronics, consumer products, and special-purpose and general-purpose AFEs have emerged to meet various market needs.

　　AFEs are generally general purpose units. AFEs can be digital-minor, analog-majority devices that use a simple state machine to control multiplexers to route signals to one or more data converters. AFEs can also be digital-majority, analog-minor devices that include one or more data converters with other microcontroller peripherals. The common functional characteristics of all AFEs are their data converters (DACs and ADCs). There is no major difference in the DAC architecture, but the ADC architecture can be delta-sigma, continuous approximation, and pipelined. Each architecture has limitations in terms of throughput, resolution, latency, filtering requirements, power consumption, and silicon footprint. Different converter architectures have different capabilities in different target applications.

　　General Purpose AFE

　　Some companies, including ADI, Linear, Maxim, and Silicon Labs, have added instrumentation ADC programming capabilities using silicon CMOS. The difference lies in the type of programming capability—state machine or microcontroller. For example, Linear's ADC has a pin-shorting programming multiplexer at the input, a feature that is popular with process control system or sensor design engineers.

　　Different vendors have different designs for multi-channel ADC AFEs. These include pin shorting or register control (via a parallel interface) to program single or multiple data converters.

　　Linear's LTC1850/51 AFE has an 8-channel multiplexer on-chip that transmits signals to a 10- or 12-bit continuous approximation ADC (see Figure 1). The AFE has a scan mode that repeatedly cycles through the 8 multiplexed channels and is programmed sequentially with a continuously scanned 16-bit address and configuration. It can also read back the timing memory. All of this is controlled by shorting the AFE pins.

　　Each 8-bit and 10-bit product in the LTC1850 family contains a pair of single successive approximation ADCs and an internal 8-channel multiplexer. The absolute sampling rate can reach 1.25Msamples/s. However, the actual sampling rate depends on the number of inputs sampled. That is, if the design requires only two input channels, then each channel should be connected to 4 inputs of the multiplexer, and the sampling rate of each channel should be 625Ksamples/s. When each multiplexer channel has a different input, the throughput is 156Ksamples/s.

　　ADI's AD7266 uses a similar design concept. The device integrates two independent 12-bit successive approximation ADCs, allowing two channels to be sampled and converted simultaneously with a throughput of 2Msamples/s. Each ADC is preceded by a 3-channel multiplexer and a noise-rejecting, wideband tracking and holding amplifier (capable of handling input frequencies in excess of 10MHz). Each ADC has two analog inputs, programmable for 3 fully differential pairs or 6 single-ended channels. The conversion results of each channel can be read simultaneously on separate data lines or continuously on the same data line.

　　Maxim's 1402 multiplexes fewer signals but with higher resolution, achieving 16-bit accuracy using a delta-sigma modulator and digital decimation filter (Figure 2). The digital filter has a user-selectable sampling factor, allowing for reduced conversion resolution in exchange for higher output data rates. True 16-bit performance is achieved at an output data rate of 480sps. The MAX1402's input multiplexer can be set to manage three fully differential signals or six pseudo-differential signals. Following the multiplexer are two chopper amplifiers, a programmable gain amplifier (PGA, gain 1 to 128), a coarse DAC for system drift elimination, and a second-order delta-sigma converter. The 1-bit data stream is filtered by an integrated digital filter that can be configured as SINC1 or SINC3. The conversion results are available through an SPI/QSPI-compatible 3-wire serial interface.

　　Silicon Labs' C8051 F350 adopts a "more digital, less analog" design. This is the company's latest 8051-compatible MCU with on-chip ADC. It integrates an 8-channel 24-bit 100ksamples/s converter and a 50MHz CPU. Peripherals include dual 8-bit DACs and temperature sensors as well as serial communication peripherals (UART, SPI, SMBus serial ports).

　　Likewise, Microchip's PIC16F684, PIC16F688, and PIC12F683 PIC-based MCUs include a 10-bit successive approximation ADC and an 8-input multiplexer.

　　Even more complex is the Cypress pSoC family, which features a complex set of digital and analog cells that are completely reprogrammable in-circuit. A pSoC IC analog cell can include up to 4 ADCs (6-14-bit resolution, with a choice of pipeline, delta-sigma, and successive asymptotic architectures); 2-, 4-, and 6-pole bandpass, lowpass, and notch filters; 6- to 9-bit DACs; and PGAs. The pSoC design tool includes 2 preconfigured delta-sigma converter models. One has 8-bit resolution, 64X oversampling, and is suitable for 32ksamples/s. The other has 11-bit resolution, 256X oversampling, and is suitable for 7.8ksamples/s.

　　It is relatively simple to use interpolation design in AFE, so it is more desirable to interpolate this function into AFE and simplify its transmission to the digital host chip. This also means that the interface between chips can operate at a lower rate, eliminating possible sources of electromagnetic interference.

　　The AD9862 from Analog Devices is a dual 12/14-bit, 128Msamples/s ADC with sampling filters and digital Hilbert filters. When the filter is enabled, it performs a Hilbert transform to decompose the single-channel input data into I and Q components for use as part of the image rejection structure. The complex data is then further processed using an on-chip digital complex modulator. Some AFEs may also include direct digital synthesis and digital mixers, so in the receive path, the DS conversion includes digital filtering. The DS conversion is particularly useful in narrowband wireless applications because it has high selectivity and very high instantaneous dynamic range.

　　Engineers who design based on the converter structure may be surprised to find that the DS conversion frequency reaches several megahertz. Initially, DS applications were targeted at high-resolution, slow-response applications (such as weighing); later they were applied to the audio field. Process advances have enabled the DS structure to increase the sampling rate to 20Msamples/s, increasing the effective bandwidth to 2.5MHz while providing 16-bit effective resolution.

　　On the other hand, although DS conversion is attractive in narrowband wireless applications (voice communication through discrete PF channels), its architecture is not suitable for wideband applications. Instead, continuous-approximation converters are usually used for high-bandwidth applications in industrial control and measurement. Today, 16-bit 300Msamples/s continuous-approximation converters are more common.

　　Pipeline converters are low cost and can be used in applications that only require 8-bit or 12-bit resolution and 10Msamples/s conversion rate. The pipeline architecture causes waiting time, but has high chip processing efficiency. A 12-bit pipeline converter requires 4095 comparators and a large chip, resulting in high chip power consumption.

　　By contrast, by converting in stages, pipeline converters require significantly fewer comparators, but at the expense of six or seven latency cycles. Latency is only a potential problem in feedback control systems. It is not a problem in communications systems because the latency of the converter is negligible compared to the delay of the entire signal chain.

　　Previously, the reasons for inserting filters before the DAC have been discussed. Although the Maxim MAX1402 includes a decimation filter after the DS converter, it does not have a decimation filter at the output of the pipeline ADC. From an economical point of view, it is possible to add a surface acoustic wave filter and other filters in the analog domain when using an inexpensive ADC.

　　AFE for Wired Communications

　　DSL and other wired communication modes are one of the largest markets for AFEs. Companies such as ADI, TI, and STMicro have many products targeting this application area.

　　Now, with the standard agreement, wired networks represent a new challenging area. ADI's AD9865 supports wired networks, VDSL, and HPNA (Home Phoneline Network Alliance) broadband modems. STMicro has introduced a new AFE as part of a 2-chipset for USB base rate adaptive ADSL modems. The MTC20154 consists of a 12-bit DAC and a 13-bit ADC (both running at 8.8Msamles/s). In the chipset, almost all digital processing is performed by the MTC20455 chip.

　　AFE for wireless communication

　　Bluetooth and IEEE802.11 standards continue to drive the wireless AFE market. STMicro's STLC2150 is a fully integrated Bluetooth single-chip radio transceiver that works with a variety of standard Blue RF interface baseband processors, including the STLC2410 (see Figure 3).

　　TI's AFE8201 (Figure 4) is a more general-purpose AFE used in IF receive channels in software-defined radios. It samples narrowband (2.5 MHz or less) IF signals and digitally mixes, filters, and decimates the signals to baseband.

　　AFE for Industrial Electronics

　　ADI's ADS7869 is aimed at the industrial market. It is a 12-channel 3ADC motor control front end. This AFE provides 3 fully differential inputs, each of which is connected to a window comparator and a sign comparator. The digital interface of the chip's parallel port can be configured to different standards. In addition, there is a serial peripheral interface (SPI) for control.

　　AFE for information appliances

　　Philips' TDA875A is a 3-channel 8-bit video data converter for liquid crystal display (LCD) monitors, projectors, and TVs. The IC receives analog RGB or YUA signals and converts them to digital outputs for high-speed flat panel displays or high-definition direct-view TVs with resolutions up to QXGA (2048×1536 at 85Hz). At the input end of the imaging chain, ADI provides custom AFE products for CCD and CMOS digital imaging applications.

　　Conclusion

　　Mixed-signal AFE chips are extremely suitable for a variety of applications (special or general), and using AFE can speed up system design and product time.