Interesting new applications continue to emerge within the world of amplifiers. In some cases, the applications themselves are novel. In others, the combination of applications and their influence on architecture constitutes the story. Sometimes, the application will provide unexpected insight into consumer behavior.
For reasons of battery life, efficient Class D amplifiers are squeezing their way into cellular phones, particularly as mobile phones become repositories for downloaded music and video. What's surprising is that more than one Class D amplifier is needed in this application. The second is used to drive ringtone speakers.
How come? In some subsets of the class "Things I Don't Do Myself With a Mobile Phone," owning the latest and most distinctive ringtone is a mark of status. The possessors of such intellectual property like to flaunt their avant-garde-ness by letting everyone within earshot hear said ringtones loudly and for the full duration of the performance. Of course, they want the longest possible battery life (despite the pencil-thin form factor of their hip jewelry), so Class D efficiency is the answer.
THE RACE CONTINUES
Class D has its attractions in cell phones and personal media players, but there is no clear winner yet in the race between Class D, Class AB, and bridgetied- load (BTL) topologies for audio power amplifiers.
One reason there is still a race is that electromagnetic interference (EMI) from the switcher is a concern for Class D amplifiers. One demonstration of how deeply this concern is felt is that last October, National Semiconductor used the first paragraph of its press announcement to declare that its 2.65-W LM4675 monophonic amp "reduces EMI more than 11 dB below the United States Federal Communications Commission (FCC) limit" (Fig. 1).
Spread-spectrum techniques for reducing (or at least redistributing) switcher energy is becoming an important part of chip companies' intellectual property. One indication of that is that National's datasheet for the new chip limits itself to a statement that "The LM4675 features a filterless spread-spectrum modulation scheme that eliminates the need for output filters, ferrite beads, or chokes. The switching frequency varies by +/-30% about a 300-kHz center frequency, reducing the wideband spectral content, improving EMI emissions radiated by the speaker and associated cables and traces."
Class D's efficiency is appealing in the portable applications National is targeting with the LM4675. In addition, the National design team's achievement of 11 dB below the FCC spec is notable.
In terms of efficiency, National says, "with a 3.6-V supply driving an 8-Ω speaker, the amplifiers' efficiency for a 100-mW power level will be greater than 80%, reaching 89% at 400 mW." The chip operates from 2.4 to 5.5 V and consumes 1.6 mA from a 2.7-V supply. Another trend represented by the LM4675 is the chip designer's effort to deal with the shrinking supply voltages that go with small-geometry logic chips. In response, the LM4675 uses differential signaling to preserve signal- to-noise ratios.
A representative example of other companies taking a non- Class D approach to mobile apps would be ON Semiconductor's 1.3-W NCP2892 BTL audio power amplifier. Internally, the chip uses two identical power amplifiers. One allows gain-setting via external resistors, while the other is internally fixed in an inverting unitygain configuration. The load is driven differentially, which eliminates the need for an output coupling capacitor.
AMPS IN SURPRISING PLACES
A particularly novel application integrates amplifiers into data cables (Fig. 2). W.L. Gore & Associates offers high-datarate, low-power active cable assemblies for InfiniBand SDR, DDR, and CX-4, applications. They triple the reach of standard InfiniBand and 10 Gigabit Ethernet cables, and they're skinnier and lighter, which helps simplify routing.
The cables use a combination of technologies. GORE Eye- Opener technology uses passive equalization built into the conductor. Quellan's Q-ACTIVE devices compensate for signal attenuation. The devices are nested completely within the cable assembly and consume 60 mW.
Quellan's family of signalintegrity- enhancing ICs for the physical layer provides fixed, adjustable, and adaptive equalization, active crosstalk cancellation, and what the company calls "collaborative signal processing" (CSP), with "analog and digital signal processing in tandem to achieve the desired functionality."
Details are sparse, but the CSP chip is based on a five-tap finiteimpulse response (FIR) filter rated for up to 10-dB cancellation of near-field reflections. It applies feed forward equalization, modulation, and symbolization to analog signals on backplanes.
Linear Technology isn't a slave to trends, insists CTO Robert Dobkin. He says the company sticks to pushing performance and offering a wide range of high-performance products that command decent margins. Even so, some of Linear's recent offerings combine performance with some simplification of the circuit designer's task. For example, the new design topology of the LT6411 differential amplifiers lets Linear integrate precision gainsetting resistors on-chip.
The LT6411 is intended for use as a high-speed analog-to-digital converter (ADC) driver, twistedpair line driver, or single-ended to differential signal converter. Inside the chip is a pair of current- feedback amplifiers (CFAs) with matched 370-Ω feedback and gain resistors. Using no external components, simply strapping the pins in different ways allows the amplifier to provide well-balanced differential signalling with gains of 1, -1, or 2 or single-ended to differential conversion.
The performance story is a 650-MHz -3-dB small-signal bandwidth (it's flat, 1 dB to 200 MHz), -77-dBc harmonic distortion at 30-MHz, and a slew rate of 3300 V/µs. The LT6411 can be used on split supplies as large as 6.3 V or on a single supply as low as 4.5 V. Current draw for each amplifier is 8 mA when enabled. When disabled, either amp's output pins become high impedance, and its current draw drops to less than 350 µA.
DYNAMIC RANGE AND LOW OPERATING VOLTAGES
There's clearly a challenge in delivering performance in the face of shrinking supply voltages. National provides an example of an operational amplifier designed for rail-to-rail operation at both the input and output with power supplies from 3 V down to 1 V. While 1-V operation is guaranteed, National recommends 3 V, but says the chip will operate even at 0.9 V. Typical specs include 2.7-MHz unity-gain bandwidth when driving a 500- pF capacitive load, 25-nV/?Hz voltage noise from -40°C to 125°C, and 80-dB commonmode and power-supply rejection ratios.
SIGE DELIVERS PERFORMANCE
At the other end of the voltage spectrum, the use of silicon-germanium (SiGe) process technologies is growing. Texas Instruments used its 36-V SiGe BiCom3HV bipolar process for the first time commercially to create the OPA211 and OPA827 operational amps for driving precision ADCs in data-acquisition systems for test and measurement, instrumentation, imaging, medical, audio, and process-control applications.
The rail-to-rail output OPA211 achieves an 80-MHz gain bandwidth product (GBW) with 1.1- nV/?Hz voltage noise with a 3.6-mA supply current. Offset voltage is 100 µV, with 0.2- µV/°C offset voltage drift and less than1-µs settling time.
The OPA827 is a JFET-input op amp for applications with a high source impedance. Its dc characteristics include 4.5-nV/?Hz voltage noise, 250-µV offset voltage, 1-µV/°C offset voltage drift, and 400-nV p-p frequency noise. Its ac specifications include 18- MHz GBW, 22-V/µs slew rate, and 0.0004% total harmonic distortion (THD) at 1 kHz.