Printer Friendly

Maxim Introduces First Core Power Solution with Integrated Dual Drivers for Next-Generation Intel and AMD CPUs.

SUNNYVALE, Calif. -- Maxim Integrated Products (NASDAQ:MXIM) introduces the MAX8809A/MAX8810A, the first single-chip 2-/3-/4-phase controllers with integrated dual driver, designed specifically for next-generation Intel(R) and AMD(R) core power requirements. The MAX8809A/MAX8810A are the latest addition to Maxim's growing portfolio of products for desktop motherboard and server applications.

One need only look back five years to see dramatic growth in the performance capabilities of Intel and AMD CPUs. (For a historical review of the Intel and AMD processors, please see the section below, Background White Paper: CPUs--Increasing Performance Requirements and Tightening Cost Constraints.) Because voltage regulators power those CPUs, power-management ICs must keep pace. The MAX8809A/MAX8810A controllers meet every specification for power, performance, and accuracy in today's demanding CPUs. But just as important, the design of these controllers has been simplified and optimized for flexibility.

"The MAX8809A/MAX8810A provide design engineers with a new approach to addressing high-current Intel and AMD core processor requirements," explains Jonathan Horner, Business Manager for the Company's Non-Portable products. "The combination of technology, simplicity, and flexibility will definitely reduce design time, risk, and total solution costs."

Multiphase Operation and Exceptional Design Flexibility

The MAX8809A operates as a single-chip, 2-phase solution with integrated drivers, making it an ideal stand-alone solution for power requirements up to 65W. It also easily scales to higher power requirements by including a 3rd-phase PWM output on chip; simply add the MAX8552 high-performance driver to support designs up to 90A. The MAX8810A also integrates a dual driver but adds two additional PWM outputs. Combine the MAX8810A with the MAX8523 high-performance dual driver to achieve a low-cost, compact 2-chip solution for 4-phase designs up to 150A.

A selection pin (SEL) allows the designer to program the controller to support three different processors, including Intel VRD10.1 and next-generation CPUs from both Intel and AMD. Included is support for VID codes, VID signal thresholds, proper startup sequencing, and overvoltage protection (OVP) thresholds. This makes it easy for designers to transport a design to new applications, and reduces the different number of controller ICs that must be stocked.

High-Performance Current Balancing

High-power microprocessors require precise voltage regulation to ensure stable performance. The MAX8809A/MAX8810A use an extremely low input bias current (0.1 microamp, typ) differential amplifier to support remote output-voltage sensing at the processor die. This eliminates the effects of trace impedance in the output and return paths. Initial output voltage accuracy is +/-0.4% due to a high-accuracy DAC combined with precision current-sense amplifiers. The precision error amplifier ensures high voltage-positioning accuracy over the full load range.

The MAX8809A/MAX8810A utilizes peak current-mode control to provide fast transient response and inherent current sharing. Applications can use resistor current sensing, or take advantage of the inductor DC resistance (DCR) to save cost and improve efficiency. Well-matched current-sense amplifiers ensure that current-sharing mismatch is less than 5% between phases. Current-sharing accuracy is further enhanced through use of a proprietary Rapid Active Averaging(1) (RA2) scheme, which eliminates errors due to current-sense component tolerances. Current-sense information is fully temperature compensated, which results in accurate voltage positioning and current-limit operation over the full -40 degrees Celsius to +85 degrees Celsius operating temperature range.

Exceptionally Tight Voltage Regulation

Overcurrent protection is extremely robust; current limiting is cycle-by-cycle and based on temperature-corrected average current. The MAX8809A/MAX8810A fold back current to 50% of the maximum value during an overcurrent event, thus reducing stress on input and output components. Other built-in protection mechanisms include overtemperature protection and overvoltage protection (OVP).

Other features include a wide input-supply range (up to 26V), an enable (EN) control pin, and power-good (VRREADY) output signal. Soft-start time is externally programmable with a resistor. Turn-off uses a unique 'soft-stop' feature, whereby the output voltage is regulated down to 0V. This feature prevents negative voltage spikes on the output at turn-off, eliminating the need for an external Schottky clamp diode.

Note: More detailed information on the technology behind the MAX8809A/MAX8810A and the benefits this technology brings to IC designers can be found in the Device Enablers section below.

The MAX8809A is available in a lead-free, 5mm x 5mm, 40-pin Thin QFN (TQFN) package, while the MAX8810A is available in a lead-free, 6mm x 6mm, 48-pin TQFN. Pricing for the MAX8809A and MAX8810A starts at $2.40 and $2.50, respectively (1000-up FOB USA).

Background White Paper:

CPUs--Increasing Performance Requirements and Tightening Cost Constraints

The performance capabilities of desktop CPUs have increased dramatically over the past 5 years, as shown in the table linked below.

The increase in processor performance has driven a corresponding increase in the sophistication, performance, and complexity of the voltage regulators which power those CPU. There are several well-understood performance characteristics required of CPU controllers.

1. Power: One dimension that defines a voltage regulator is the number of 'phases,' or channels that it can accommodate. Each phase can practically deliver 25W to 40W of power, depending on factors such as available space and cooling. While a single-phase voltage regulator was sufficient for the Pentium 3, contemporary CPUs require 3- or 4-phase regulators.

2. Current Balancing: One challenge in designing a multiphase power supply is ensuring that current (power) is properly shared among phases. A significantly disproportionate amount of current in one phase will stress components and degrade their lifetime. Virtually all multiphase voltage regulators must, therefore, incorporate circuitry to actively balance the current among phases.

3. Accuracy: CPUs require that their voltage be regulated to extremely tight tolerances in order to operate at high clock frequencies. These tight tolerances must, moreover, be maintained under both static and dynamic conditions. Static accuracy is achieved by implementing a precise on-chip reference voltage and by minimizing offset voltages and bias currents. Dynamic accuracy is affected by the closed-loop bandwidth of the voltage regulator and the amount of bulk capacitance used on the regulator output. Since no regulator can respond instantaneously to a sudden change in CPU demand, every design will require bulk capacitance. The higher the regulator closed-loop bandwidth, the sooner it can 'catch up' to the CPU demand and 'take over' from the bulk capacitors.

The increasing demands placed on the CPU voltage regulator pose a considerable design challenge. Both the die area and pin count scale with the number of phases accommodated by the regulator. High-accuracy voltage references require sophisticated design and calibration techniques. Amplifiers used for sensing voltage and current, for basic voltage regulation, and for active current sharing must be fast and have low-offset errors and bias currents. They must be stable over process and temperature. Active current-sharing circuitry must also be accurate and not interfere with the basic voltage-regulation operation.

Perhaps the most significant challenge facing high-power CPU regulator designs is cost: price-per-phase for a CPU core voltage regulator has declined 4x or more over the past five years.

MAX8809A/MAX8810A Device Enablers

The MAX8809A/MAX8810A meet all the high-performance requirements of next-generation CPUs, while providing technological and performance improvements over existing regulation (control) schemes.

Accurate Line Regulation

Traditional voltage-mode control has inherently poor line regulation since its error voltage is compared with an internally generated sawtooth. Changes in the input voltage are not reflected in the sawtooth, unless the regulator adds additional circuitry (and complexity) to make this adjustment dynamically. The MAX8809A/MAX8810A use peak current-mode control, which replaces the internal sawtooth with the inductor current ramp from each phase. The duty cycle is controlled by the inductor current ramp, which is a function of both input and output voltage. The line regulation, therefore, is inherently maintained without additional circuitry.

Simple Voltage Positioning

Most high-current CPU core regulator designs use voltage positioning to reduce bulk capacitance requirements. The MAX8809A/MAX8810A use finite gain to set the output load line (see Figure 1).

In regulation the equations for the error voltage are:


where N is the number of phases. Rearranging, we solve for:


The term (VDAC - VOUT)/IOUT is simply the load-line impedance. The current-sense gain (GCA) and the error amplifier transconductance (gMV) are IC parameters;

RSENSE and N are determined by the application. Therefore the load-line impedance is set simply by selecting the proper value of RCOMP, which programs the voltage-error amplifier gain. Competitive solutions generally require multiple resistors to set the load-line impedance.

Fast Transient Response

The voltage-mode approach requires a second control loop for current balancing--the bandwidth of this loop is generally 1/5 to 1/10 of the voltage-loop bandwidth in order to prevent interference. Low bandwidth is sufficient for current balancing, as slow adjustments are usually all that are required. For voltage positioning, however, load transient response is a direct function of current-loop bandwidth. For voltage mode the bandwidth will be low (e.g., 5kHz); for peak-current mode the current and voltage-loop bandwidths are the same (e.g., 50kHz to 75kHz). The difference in transient performance is evident in the scope shots in Figures 2 and 3 (95A load step followed by a 95A load release). Voltage-mode control is also more complex to compensate due to the poles and zeroes formed by the control loop and the output filter. Voltage mode usually requires Type III compensation, compared to peak-current mode, which uses single-pole compensation, and therefore, fewer components.

Highly Accurate Current Balancing

Peak current-mode control has one limitation: any mismatch in the inductance value between two phases (e.g., due to tolerances) will create a DC-current mismatch. The MAX8809A/MAX8810A addresses this problem with a proprietary technology called Rapid Active Averaging (RA2), which averages out the inductor ripple current in each phase. The RA2 circuitry (see Figure 4) 'learns' the peak-to-peak ripple current of each phase over several switching cycles, and then biases the peak current signal down by 1/2 of the ripple current. By moving the 'peak' control point from the inductor current peak to the DC-current point, the benefits of peak current-mode control are retained but with very accurate DC-current matching. Since the RA2 circuitry is not a direct part of the current-sense path used for regulation, it does not slow down transient response.

Accurate Temperature Compensation

The MAX8809A/MAX8810A feature accurate temperature compensation using only a single thermal sensor. A typical method for extracting current information is to use the inductor DC resistance (DCR) as the current-sense element. There is a problem with this approach, however. The DCR will change over temperature according to the positive temperature coefficient of copper. To compensate designs use a resistor with an equal, but opposite (negative) temperature coefficient--an NTC. This NTC is usually part of the resistor network that programs the load-line impedance. The resistor network will also include two additional resistors to linearize it over the temperature region of interest. The drawback to this technique is that current-limit information is not temperature-compensated; the current-limit threshold at room temperature must be scaled upward to account for the increased current signal at higher temperatures. The inductor and MOSFETs must be oversized to handle the maximum current at current limit--at room temperature--which, in turn, leads to a higher solution cost.

The MAX8809A/MAX8810A also use an NTC, but linearization is on-chip, saving two resistors. Sensed current is then temperature-corrected and used internally for voltage positioning and current-limit functions. There is no need to oversize components for temperature effects; the current limit threshold at room temperature is the same at 85 degrees Celsius. Competing products require a second NTC for the VRHOT function, while the MAX8809A/MAX8810A use the same linerarized temperature information for VRHOT, further reducing total solution cost.

To summarize, the features and technology provided by the MAX8809A/MAX8810A will help simplify the design process and reduce total solution costs. Please visit for complete information on other voltage-regulator solutions for desktop-PC and server applications from Maxim.

Maxim Integrated Products is a leading international supplier of quality analog and mixed-signal products for applications that require real world signal processing. For more information, contact Maxim at 120 San Gabriel Dr., Sunnyvale, CA 94086. Telephone: 408-737-7600 or our URL:

For more information on the MAX8809A/10A, please visit:






DATA SHEET: A preliminary Data Sheet for this product is available on the web.

GRAPHIC: The graphic for this product is available on the web in three formats:

PREVIEW image (smaller image to view in Navigator or Explorer):

JPEG (300 dpi, high quality, RGB color)

TIFF (300 dpi uncompressed CMYK TIF, in a .ZIP archive(2))

(1) patent pending

((2) Use Winzip, StuffIt Expander or similar de-archive utility to access the .ZIP file)

Intel is a registered trademark of Intel Corporation.

AMD is a registered trademark of Advanced Micro Devices, Inc.
COPYRIGHT 2005 Business Wire
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2005, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Publication:Business Wire
Geographic Code:1USA
Date:Nov 29, 2005
Previous Article:Fitch Upgrades 21 & Affirms 4 Classes from 4 MLMI Securitizations.
Next Article:Lowe Destination Development Opens Premier Stone Eagle Golf Club in Palm Desert.

Related Articles
AMD announces new AMD Opteron processors, adding more choice and flexibility for leading 32- and 64-bit computing.
ATI delivers industry's first integrated DirectX 9 graphics chipset for AMD Athlon'TM' 64 and Sempron'TM' platforms.
AMD "shatters the hourglass" with the arrival of the AMD Athlon 64 X2 dual-core processor.
AMD unveils world's ultimate PC processor for 3D games.
BEA supports Intel multi-core processor roadmap.
Workstation envy: look at the latest crop of workstations. They're fast, inexpensive, quiet, and built for expansion.

Terms of use | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters