Cascaded power management unit characterization for TEG-based IoT devices in 65 nm CMOS

Dima Kilani, Mohammad Alhawari, Baker Mohammad, Hani Saleh, Mihai Sanduleanu, Mohammed Ismail

Research output: Contribution to journalArticlepeer-review

7 Scopus citations


This paper presents a detailed characterization of several power management units (PMUs) targeting wearable healthcare IoT device powered by a thermoelectric generator (TEG). The characterization is based on silicon measurements where the TEG-PMUs are fabricated in 65 nm CMOS. Power efficiency, area and voltage ripple have been measured for different PMU topologies. In this paper, three various TEG-PMUs are characterized and compared: design 1: switched inductor (SI) boost converter followed by a switched capacitor (SC) buck converter, design 2: SI followed by LDO regulator and design 3: SI followed by two voltage regulators in series: SC and LDO. The measured results show that the end-to-end power efficiency is constant for the three PMU design options at a certain load power giving that the input power is fixed due to MPPT. The maximum end-to-end power efficiency is 65%. However, the voltage regulation range of the LDO is better than the SC due to the design of the LDO that is capable of generating a load voltage equal to or smaller than the input voltage. Design 2 and design 3 which have the LDO regulator as the last stage has lower voltage ripple of 12 mV compared to 35 mV in design 1 where SC drives the load. The SI boost converter consumes an area of 0.036 mm2 and it requires off-chip inductor. The SC regulator occupies the largest area of 0.495 mm2 whereas the LDO regulator occupies the smallest area of 0.0357 mm2.

Original languageBritish English
Pages (from-to)285-296
Number of pages12
JournalMicroelectronics Journal
StatePublished - Aug 2019


  • LDO
  • Power management unit
  • Switched capacitor buck regulator
  • Switched inductor boost converter


Dive into the research topics of 'Cascaded power management unit characterization for TEG-based IoT devices in 65 nm CMOS'. Together they form a unique fingerprint.

Cite this