Let's begin with some broad observations. Precision amplifiers, originally designed for test and measurement equipment, recently migrated to sensor monitoring in automotive and industrial applications. The latest performance-monitoring subsystems in cars and factories need the precision amps' low-input offset voltage and offset current with low temperature coefficients and noise characteristics.
High-speed amps, traditionally boasting at least 50 MHz of bandwidth and at least 100-V/µs slew rates, can be found in high-speed signal chains and analog-to-digital converter (ADC) drive circuits. And with the advent of high-definition television (HDTV), video amps have moved into high-speed territory.
In audio power amps, the venerable Class AB push-pull architecture is being superseded by its bridge-tied load (BTL) cousin. At the same time, super-efficient Class Ds also are challenging push-pull in applications where electromagnetic interference (EMI) from pulse-switching can be tolerated or mitigated.
PRECISION OP AMPS
Automotive OEMs want performance at lower prices than what precision amps used to go for. That means chip makers have had to figure out ways to achieve the same precision they got with ±15 V with only ±5 V or even ±3 V (Fig. 1). This is driving lots of innovation in architectures, trimming techniques, and to some extent, integration in terms of additional circuitry on the die to handle filtering or calibration, auto-zeroing, and digital trimming.
Thanks to shrinking CMOS device geometries, chip makers sometimes can add extra power supplies right on the chip. In certain designs, this permits input voltage swings greater than rail-to-rail. In other cases, negative supplies make it possible to drive resistive loads like headphones without incurring quiescent dc losses by driving ±1.5-V levels.
Also, channel counts in medical systems, from CT and MRI to basic obstetric ultrasound machines, are quadrupling. In terms of volume, amplifiers must keep pace with ADCs. In terms of processes, the 0.25-µm generation seems to be the sweet spot. No one is diving deeply into very fine-geometry CMOS.
HIGH-SPEED OP AMPS
In the high-speed arena, demand grows for low-power video amps in portable applications. HDTV pushes amplifier performance, but with simultaneous downward price pressures because TV is a consumer-based business. That downward pressure on prices is something new for high-speed amps, where customers once were inured to paying premium prices for premium performance.
There's little CMOS in high-speed amplifiers. National Semiconductor has announced its VIP50 process, and Texas Instruments stands by its SiGe BiCOM-III. Analog Devices finds no problem in meeting market-mandated price points with one or another of its many bipolar processes.
Then there's dis-integration. As system digital content expands, it often makes sense to move an amplifier that was once integrated into an application-specific standard product (ASSP) off the ASSP and onto the board. It's a case of intelligent partitioning. For example, cell phones may have an MP3 decoder on a baseband chip set in some ultra-fine geometry, while a headphone amplifier makes more sense in a separate 0.25-µm chip.
AUDIO POWER AMPS
Class D audio amps are beginning to reflect a move from pulse-width modulation (PWM) to pulse-density modulation (PDM) with a delta-sigma modulator on the front end.
Where PWM modulates the width of the pulses that drive the power FETs in the amplifier's output bridge, PDM controls the density of fixed-width pulses. Properly controlled, PDM helps spread the noise spectrum of the Class D switcher, eliminating harmonic spikes in the frequency domain (Fig. 2). The task is to lower energy in frequency bands used for AM and FM radio and cell phones.
Like other spectrum-spreading techniques, PDM doesn't reduce the energy in spurious emissions. Parseval's Law still applies. However, the reduction in big spurs may make it easier to deal with what's left. It takes more than just PDM, though. ADI uses the amp's sigma-delta modulator to shape and direct the out-of-ban noise energy to frequency bands where it may not matter as much in EMI-sensitive applications.
Of course, how the audio subsystem is built is as critical as the digital modulation method. Shaping the out-of-band noise energy is good, but subsystem implementation techniques with power-supply filtering and speaker-wire shielding are also necessary. The guiding philosophy is that conducted EMI eventually will become radiated EMI if nothing is done to shield the conductors.
Other trends include the holistic integration of a Class D amp, codec, and some DSP onto one chip. With so much audio now being stored and distributed in digital form, you will see more examples of the "digital-input" audio amplifier in which the only analog-signal will be the post-filter drive to the speakers or headphones.
Among the papers written for the International Solid State Circuits Conference (ISSCC) set for Feb. 4-8 in San Francisco, there is ample evidence that there's still no final resolution in the audio-amplifier wars between straight-analog Class AB output stages and PWM Class D amplifiers.
On the digital side, engineers from Texas Instruments and the Technical University of Denmark emphasize output power in "A 240W Monolithic Class D Audio Amplifier Output Stage." This chip contains two half bridges. When it's used in BTL configuration, the unclipped output power is 244 W into 4 ½. To the best of their knowledge, the authors say, this power level is unprecedented for monolithic output stages.
Also, TI offers "A Digital Input Controller for Audio Class-D Amplifier with 100W 0.004% THD+N and 113dB Dynamic Range." This controller architecture was designed for an audio Class D amplifier using a PWM switching scheme. Its digital input is provided by an integrated digital-to-analog converter (DAC) that has a special configuration to realize direct connection between the DAC and the feedback loop. It boasts 0.0018% total harmonic distortion plus noise (THD+N) at 10-W output power, 0.004% THD+N with 100-W output power, and 113-dB dynamic range.
On the linear side, Philips and STMicroelectronics are developing bipolar/CMOS/ DMOS (BCD) automotive audio power amplifiers, where a DMOS quad bridge forms the output stage.
Philips authors will present "Frequency Compensation of an SOI BCD Car Audio Power Amplifier." For frequency compensation, Philips researchers modified nested Miller compensation (NMC) using the parasitic gate-drain capacitance of the power transistors for most of the miller capacitance. Philips then built a 46-W amplifier using the scheme. With 10-W output, it achieved THD+N values of 0.005% at 1 kHz and 108-dB signal-to-noise ratio.