In power design for communications equipment and data centers, achieving fast transient response, miniaturization, and high voltage accuracy simultaneously is emerging as an important challenge.

To address these needs, the high-precision serial active voltage positioning (AVP) technique applied to Analog Devices' (ADI) μModule® regulators is attracting attention.

This approach enables the implementation of all-ceramic capacitor-based power solutions with fast load transient response characteristics and minimal board area.

Compared to conventional shunt type AVP structures, this series AVP has excellent load line accuracy, resulting in significantly improved output voltage accuracy.

To help users, we also present the results of actual load transient response measurements using the technique.

■ Preface

Active voltage positioning (AVP) or active droop control is a technology that variably adjusts the power output voltage depending on the load current size.

This method maintains the output voltage high under light load conditions and lowers the output voltage under medium or high load conditions.

The key advantage of AVP control is that it improves load transient response performance while reducing the required output capacitance.The point is that it can reduce the amount of stress.

This is because it provides the power supply with voltage margin to cope with sudden load changes.

μModule regulators are pre-validated, integrated power module solutions widely used in telecommunications equipment and data center applications.

In these systems, fast transient response, minimal board area, and all-ceramic capacitor solutions are preferred.

However, it is difficult to meet all these conditions with existing non-AVP control methods.

This article describes how to implement a high-precision series AVP by adding two resistors to the feedback control loop.

A major advantage of this approach is that the load line characteristics are virtually unaffected by the gain variation of the GM amplifier.

On the other hand, other AVP implementations, such as shunt AVP1, may suffer from significant load line accuracy degradation when the gain variation of the gm amplifier is large.

By applying the series AVP technique, the output capacitance can be reduced by up to 50%, and the peak-to-peak transient response characteristics of the output voltage are also slightly improved.

As the output capacitance requirement decreases, a configuration using only ceramic capacitors becomes possible.

Ceramic capacitors are much more reliable and cheaper than aluminum electrolytic capacitors, which is advantageous in terms of improving system reliability and reducing costs.

Additionally, AVP control provides the effect of reducing the power consumption of the load by lowering the output voltage under high load conditions.

Experimental results for the LTM4650-2 circuit show a power saving effect of 1.4W for a single output and 5.6% for the entire circuit, which can also extend the operating time in battery-powered systems. Contribute.

■ Serial AVP implementation principle

AVP is a control method designed to intentionally change the output voltage according to the load current.

In the conventional non-AVP structure, the output voltage is maintained at the nominal value (nominal VOUT) under all load conditions, as shown in Figure 1.

However, when the AVP method is applied, the output voltage gradually decreases as the load current increases.

Under light load, the output voltage is set slightly higher than the nominal value, and under medium load, it is adjusted slightly lower than the nominal value.


When the load current suddenly increases, the output voltage starts at a level higher than the nominal value, which may result in a larger droop, and remains within the set voltage range.

Conversely, if the load current decreases abruptly, the output voltage starts at a level lower than the nominal value and experiences a larger overshoot, but this is also limited within the specified voltage range (between VMAX and VMIN).

Figure 2 shows an AVP circuit using series compensation.


In this circuit, the internal reference voltage (VREF) and the VOUT feedback signal are connected to the positive (+) and negative (-) input terminals of the error amplifier, respectively.

VHI (or INTVCC) applied to RHI provides an appropriate DC voltage to the amplifier output (ITH or COMP) to prevent the output from saturating.

The feedback resistor (RLO) is connected from the output (ITH) to the negative (-) input terminal (FB) of the error amplifier, and this RLO has the greatest influence on the gain of the gm amplifier.

The resistance values of RHI and RLO must be sufficiently larger than R1 and R2.

The load line formula 1 is as follows:


Here, Ki is the current sensing gain, and RSENSE represents the current sensing resistor value or the inductor DCR value in the DCR sensing method.

Compared to the AVP shunt compensation circuit1, the biggest advantage of the series compensation circuit is that the load line is affected by the R1/RLO ratio and is hardly affected by the transconductance (gm) tolerance of the error amplifier.

Due to the nature of the semiconductor process, the gm value of some ICs can vary by up to ±30% depending on the product, and the shunt compensation circuit AVP is vulnerable to this variation because it is directly proportional to the 1/gm gain.

■ LTM4650-2 Application Case

The LTM4650-2 is a current-mode synchronous buck μModule As a regulator, it can supply a load of up to 25A at 1V output and has a load transient response window of approximately ±8% (160mVp-p).

In conventional non-AVP configurations, external RC filtering circuitry is required to achieve fast Type II control loop compensation.

Five 100μF ceramic capacitors and two 470μF POSCAP capacitors were used in the output stage.

The measured output voltage transient response under conditions of 19 A load step (75% of full load) and 19 A/μs slew rate was 136 mVp-p (Figure 3).


▶ Improved transient response performance
When implementing AVP, the AVP compensation circuit is applied to COMP, as shown in Figure 4, but RC compensation is not required. At half load (12.5 A), the output voltage was set to the nominal value (1 V) by fine-tuning R2. At this time, the load transient response was measured to be 95 mVp-p (Figure 5), confirming improved transient response performance compared to conventional methods.

▶ Power saving effect
When the output voltage is set to 1V at 25A (full load), the load power is 25W, but when AVP is applied to lower it to 0.945V, the load power decreases to 23.6W. This results in a power saving of 1.4W for single output and a total of 2.8W for dual output.



▶ Implementation of all-ceramic capacitors
By implementing AVP, the two existing POSCAPs can be replaced with two ceramic capacitors, resulting in a total of seven 100μF ceramic capacitors used for COUT1. Using ceramic capacitors offers the advantages of lower equivalent series resistance (ESR), lower equivalent series inductance (ESL), lower cost, smaller size, and higher reliability. The transient response performance was improved, and the measured result was that the VOUT transient response was 104 mVp-p, as shown in Figure 6.


Table 1 compares the load transient response Vp-p measurements of Non-AVP (reference), AVP, and AVP with only output ceramic capacitors for comparison.


■ Conclusion
Implementing the AVP series compensation circuit in the LTM4650-2 μModule regulator improved the load transient response characteristics and reduced load power consumption under high load conditions.

Output capacitance has been reduced by more than half, and cost savings and board area minimization have been achieved by replacing POSCAP with ceramic capacitors.

This AVP technique can also be applied to other μModule regulators that have an external compensation pin with an external RC compensation network (e.g., LTM4630-1, LTM4626, LTM4636, LTM8055-1, etc.)

※ About the author
Sin Keng Lee joined Analog Devices (ADI) in May 2022 and is currently a power module design engineer. He holds a bachelor's degree and a doctorate in antenna, RF, and microwave communications engineering from Northumbria University, UK.

Zhijun (George) Qian is a Senior Manager in Power Module Design at Analog Devices, responsible for all LTM80xx products and select LTM46x/LTM47x products. He holds a bachelor's and master's degree in power electronics from Zhejiang University and a doctorate in power electronics from the University of Central Florida. He joined ADI in early 2010.