The consumer electronics market has seen an impressive growth in portable products that require both audio playback and recording. Desirable attributes of these products include long battery life, small and lightweight form factors, and high-quality audio performance. Not only must these requirements be achieved but they must also be met while meeting the consumerís expectation for low-cost consumer electronics.
The most viable path taken to meet these expectations is the integration of the multiple functions required in these products into a minimal number of integrated circuits. The advantages of integration have been proven many times over the years, in many different markets and applications. An obvious advantage is the lower cost of an integrated solution compared to a solution composed of several discrete components. Another obvious advantage is that the integrated solution requires less space on the printed circuit board, thus minimizing size, height and weight. However, there is also another advantage that is not necessarily obvious, but equally important. Integrated solutions typically have greater power-management capabilities and the ability to power-down unused functional blocks and functions, thereby maximizing battery life. As an example, if a portable audio recorder/player is being used to listen to MP3 files, there is no reason to power the analog-to-digital converter and microphone circuits used for voice recording. These circuits can be individually powered-down to conserve power and increase playback time.
The opportunities for integration generally fall into one of two categories. The first is the analog and mixed-signal functions. These include the audio converters (A/D and D/A), analog input multiplexers, volume control, microphone preamps, headphone amplifiers, speaker drivers and other similar functions. The second is the system digital functionality, which includes the audio/video decoder, the user interface, USB input/output, the display interface and similar functions. Unfortunately, the characteristics of state-of-the-art sub-micron digital processes are less than optimal for analog and mixed-signal functions. These limitations force a decision, and related compromise, on the level of integration that can be targeted. As a result, there exist two general paths of integration. The first is to accept the analog limitations of the advanced processes and head down the path to a single-chip solution. There are many product areas where these analog limitations are acceptable. Examples include the very low cost MP3 players currently on the market; cost is the primary concern, and audio quality is secondary. A similar scenario exists in very low-end DVD players and set-top boxes. The second path retains a higher prioritization on audio quality. Advanced sub-micron digital processes are ideally suited for highly integrated digital functions in portable products, and there are very good processes available for analog functions; they just arenít the same ones. As a result, higher-quality portable audio products are evolving towards dual-chip solutions.
System Power Supplies
One of the more interesting and effective techniques available to minimize power consumption and extend battery life is to lower the system power supply voltages. This approach also aligns well with the latest advances in silicon process technologies, which continue down the path to lower and lower supply voltages. It is interesting to study the relationship between the system power supply voltage and power consumption. In analog integrated circuits, power consumption as a function of supply voltage is basically a linear relationship. Essentially, there is roughly a 45% power savings by implementing a 1.8 V analog system compared to a comparable 3.3 V system. The power savings in a digital IC are even greater as a result of the additional decrease in current consumption. With the additional decrease in current, there is roughly a 75% decrease in power consumption by going to a 1.8 V system when compared to a 3.3 V system.
A rather obvious point on power supplies: the fewer the better. A minimal number of system supply voltages minimizes the number of components within the system and reduces the complexity of the printed circuit board designs which minimizes system cost and size.
A 0.25 micron process is a very suitable choice for quality, low power analog and mixed-signal ICs. This process operates very well at typical supply voltages down to 1.8 V. Therefore, based on the design goals to minimize costs and power consumption while maintaining quality audio performance, a single system supply voltage of 1.8 V is very desirable. Unfortunately, there is a catch. The majority of headphones used with these systems have an impedance of 32 ohms, which dictates the need for a power supply greater than 1.8 V to achieve an acceptable listening volume.
A common requirement for headphone amplifiers in battery powered products is that they must operate from a single positive power supply. The fundamental drawback to a single-supply amplifier is that the output of the amplifier is biased at a DC voltage that is approximately Ĺ the supply voltage. Consider for a moment that the DC resistance of the headphone voice coil (or any speaker for that matter) is determined solely by the resistance of the wire used in the windings. This resistance is very low and will allow a significant amount of current to flow if a DC voltage is applied. The least of the problems this can cause is degradation of audio quality, but the most obvious and likely outcome is permanent damage to the headphones. Suffice it to say that applying a DC voltage to a headphone or speaker is generally considered a very poor design practice.
DC-Biased Headphone Amplifiers
The most common technique employed to block this DC current is a decoupling capacitor, as shown in Figure 1.
Inserting a capacitor in the signal path creates a high-pass filter with a corner frequency set by the impedance of the headphone (R) and the capacitance (C) of the decoupling capacitor as defined by 1/ (2 RC). It is generally accepted that the human ear can hear frequencies as low as 20 Hz. With this lower limit and the typical impedance of a headphone, it is an easy task to calculate the required capacitance value, which is 500 F for a 20 Hz corner-frequency. Obviously, this is an unacceptable solution due to cost and the area required on the printed circuit board for these physically large capacitors. A commonly accepted compromise is the use of a 220 F capacitor. This capacitor places the corner frequency at approximately 45 Hz and consumes less board area. While this compromise has traditionally been employed in many designs, it remains a compromise. It is also important to note that this topology does not lend itself to the implementation of a single 1.8 V system, since a 1.8 V amplifier does not have the ability to drive headphones to an acceptable listening level. Despite the limitations of cost, PCB space and frequency response, the capacitively coupled headphone amplifier has been the only viable solution for many years in portable audio products. While this topology has been employed in many designs, it is a compromise that has become a significant limitation as system requirements have become more stringent.
The Ground-Centered Headphone Amplifier
A headphone amplifier with a ground-centered output is the ideal solution. As shown in Figure 2, with the amplifier output at the same DC potential as the headphone return (ground), DC current will not flow through the headphone. The ground-centered output amplifier has the advantages of optimal low-frequency response and significant savings in both PCB space and component costs when compared to the DC-biased amplifier. However, one of the system requirements for a ground-centered amplifier is the presence of both positive and negative power supply voltages.
Cirrus Logic and other manufacturers solved this dilemma by integrating a charge-pump to generate a negative supply voltage for the headphone amplifier. These charge pumps are relatively efficient and require minimal external components. Not only does this topology negate the need for the headphone decoupling capacitors, it is now possible to get the required amount of power to generate a sufficient listening volume from a 1.8 V system. The charge-pump generates the negative 1.8 V supply to effectively produce a 3.6 V-capable amplifier.
Benefits of a Ground-Centered Amplifier
Minimizes Required PCB Area. Figure 3 compares an actual PCB layout of an audio codec requiring DC-blocking capacitors and an audio codec with an integrated ground-centered amplifier.
Removing capacitors C3 and C4 provided more than 50 mm2 of extra board area. Capacitors C5 and C6 in the ground-centered layout are required for the internal charge pump. These capacitors are comparable to typical power supply de-coupling capacitors.
Optimal Low Frequency Performance. Since an HPF with a large series capacitor is not required after the ground-centered amplifier, it achieves optimal response at low frequencies. Figure 4 compares the actual frequency response of a headphone amplifier requiring DC-blocking capacitors and a headphone amplifier with a ground-centered output.
Provides Appropriate Listening Levels with a Single 1.8 V System Supply Voltage. The charge-pump generates a negative 1.8 V supply that, when combined with the system 1.8 V supply, effectively produces a 3.6 V-capable amplifier. This architecture makes it feasible to operate a system with a single 1.8 V supply to minimize power consumption and retain the ability to generate a sufficient listening volume.
Freedom from Power-On and Power-Off Transients. In addition to all the benefits that would normally require a tradeoff in a DC-biased amp configuration, the ground-centered configuration is also immune to the clicks and pops resulting from the power-on and power-off transients associated with charging and discharging the DC-blocking capacitors.
Example Integrated Mixed-Signal Audio CODEC
The Cirrus Logic CS42L51 is an example of a highly integrated mixed-signal audio codec designed for low-power, high-quality audio products. The device is capable of operating from a single 1.8 V system supply and has an integrated charge-pump with a ground-centered headphone amplifier. The advantages of a highly integrated solution are also evident in the photograph shown in Figure 6. The PCB shown includes all the power supply decoupling, charge-pump filter, and other components required for the CS42L51.
The latest product announcements at the Consumer Electronics Show illustrate the incredible growth rate of portable, battery-powered audio products. The trend of integrating additional functions and features, while extending operational time and decreasing product size and weight, will also continue. This trend can be seen in the latest offerings in audio and video media players, smart phones and personal digital assistants, wireless headsets, gaming consoles, handheld satellite radio / GPS products and other products we have yet to envision!
The increased functionality and portability in these devices will continue to drive mixed-signal integration of additional functions, including phase-locked-loops for supporting non-audio clocking requirements, voltage regulators, and improved power-management capabilities. The high-quality audio required in many of these applications will continue to drive demand for separate digital and mixed-signal integrated products. Efforts are also underway to develop the next generation of technologies to achieve the desirable attributes that these products require, including even longer battery life, smaller and lightweight form factors, and high-quality audio performance.
Blog Doing Math in FPGAs Tom Burke 21 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...