A 30V Step-Up Regulator with Shunt-Current-Reuse Controller for >85% Efficiency over 200µA-100mA Loading Range

For applications requiring both low-noise and high-efficiency such as sensors, data converters, and RF [1], we propose a combined inductor-boost / LDO step-up converter scheme employing output-referenced feedback control. The boost stage is operated with high voltage-conversion-ratio (VCR) to maximize efficiency while a quiet rail is provided at the output of the LDO for high supply-noise rejection. Choice of LDO drop-out voltage (V<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DROP</inf>) is a compromise of switch noise rejection vs. power loss. At high VCR, dynamic loss can be reduced by increasing the inductor peak current while reducing the switching frequency but at the expense of larger V<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">BST</inf> ripple noise. Thus, for applications that require a high-voltage and low-noise supply, an optimized V<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DROP</inf> range can be found with respect to efficiency, as shown in the bottom right corner of Floura 1. For applications needing >20V supply, the optimized <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{v}_{\text{DROP}}^{\circ}$</tex> region (≈0.5V) is also sufficient for circuits to operate. Therefore, in this work, the outout of the Boost stage (V<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">BST</inf>) and the output of the LDO stage <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(\mathrm{V}_{OUT}$</tex> serve as a local supply rail and ground rail, respectively, as shown in Figure 1. The control circuits for these cascaded blocks, including the bandgap (BGR), loop-regulation circuitry, and the top-switch-control-logic are all realized between the two rails (i.e “output-referenced”) with local effective supply voltage equal to <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\rightarrow\leftrightarrow\cdot \mathrm{r} VDROP$</tex> As a result, the shunt current consumed between the two rails is re-used by the output load. When the load current <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$l_{LOAD}$</tex> is higher than the current consumed by the circuitry in the V<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">DROP</inf> rail domain, the current efficiency of the LDO stage approaches 100%. Figure 1 includes an efficiency comparison between a conventional boost architecture and the proposed method; for high VCR and light load conditions, the current reuse scheme offers notable improvement (up to ≈85% vs. ≈75%). The proposed Shunt-Current-Reuse structure is therefore an optimum choice for LiDAR-like applications that operate under high supply voltage (> 20V) and with a prolonged light load <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(&lt; 10\text{mA})$</tex> condition [2].