A 16.28 ppm/°C Temperature Coefficient, 0.5V Low-Voltage CMOS Voltage Reference with Curvature Compensation
Harshith Reddy, Pankaj Arora · 2025
Read A 16.28 ppm/°C Temperature Coefficient, 0.5V Low-Voltage CMOS Voltage Reference with Curvature Compensation by Harshith Reddy, Pankaj Arora (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-15729.txt.
Introduction
Harshith Reddy and Pankaj Arora designed a low-voltage reference circuit in a 2025 paper. Voltage references give chips a steady yardstick voltage, but modern low-power designs need them to run on half a volt and shrug off heat. Their circuit, built in a 90-nanometer process, works in the subthreshold region where transistors barely turn on, guided by a model that covers all operating regions.
Three currents, two falling with heat, one rising, plus a curvature correction, cancel each other's drift. The output holds 205 millivolts from minus forty to 130 degrees, drifting only about sixteen parts per million per degree. Line sensitivity and supply noise rejection are strong for the class, and the whole circuit draws just 0.67 microwatts.
Catalog record
| CREATOR | Harshith Reddy, Pankaj Arora |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 16 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-15729
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A 16.28 ppm/°C Temperature Coefficient, 0.5V Low-Voltage CMOS Voltage Reference with Curvature Compensation Harshith Reddy, Pankaj Arora Abstract This paper presents a fully-integrated CMOS voltage reference designed in a 90 nm process node using low voltage threshold (LVT) transistor models. The voltage reference leverages subthreshold operation and near-weak inversion characteristics, backed by an all-region MOSFET model. The proposed design achieves a very low operating supply voltage of 0.5 V and a remarkably low temperature coefficient of 16.28 ppm/°C through the mutual compensation of CTAT, PTAT, and curvature-correction currents, over a wide range from -40 °C to 130 °C. A stable reference voltage of 205 mV is generated with a line sensitivity of 1.65 %/V and a power supply rejection ratio (PSRR) of -50 dB at 10 kHz. The circuit achieves all these parameters while maintaining a good power efficiency, consuming only 0.67 μ \mu W. I Introduction In most system-on-a-chip (SoCs) systems, a voltage reference insensitive to temperature, supply voltage, process, and mismatch variations is crucial [ 1 ] . They are essential components in multiple analog circuit applications such as Analog-to-Digital Converters (ADCs), linear regulators, power management systems, sensors, Internet of Things (IoT), and wearable devices [ 2 ] . Most voltage references work by generating two types of currents, namely PTAT (proportional to absolute temperature) and CTAT (complementary to absolute temperature), and summing them to develop a temperature invariant voltage at the output [ 3 ] . The most used bandgap voltage reference circuits use lateral BJTs or parasitic BJTs in CMOS process, and op-amps. In a conventional bandgap voltage reference, the base-emitter voltage ( V_{BE} ) of a bipolar transistor decreases linearly with temperature, exhibiting a CTAT characteristic. A PTAT voltage is generated by the difference in V_{BE} between two bipolar transistors operating at different current densities. A bandgap reference circuit combines these CTAT and PTAT voltages to produce a stable reference voltage ( V_{REF} ) close to the silicon bandgap voltage of 1.2 V [ 4 ] . As CMOS technology advances, ADCs and other circuits require lower operating voltages and power-efficient performance. The start-up voltage of low-power ADCs is constrained by the minimum voltage needed for the voltage reference operation. Lowering its supply voltage extends ADC runtime and allows operation with harvested voltages as low as 0.5 V [ 3 ] . This work presents a purely CMOS-based voltage reference architecture that achieves an excellent temperature coefficient, with a good line sensitivity and a distinctively low operating supply voltage while maintaining power efficiency. Being purely CMOS based makes this design better compatible with a larger set of process technologies and helps achieve operating voltage by the use of low threshold transistor models and subthreshold operation. This paper is divided into four sections. Section II analyzes in detail the working of the proposed architecture. This section is subdivided into four parts to describe the different segments of the circuit – PTAT, CTAT, curvature compensation, and output reference generator. Section III discusses the simulation results achieved in this paper. Concluding remarks are made in Section IV. II Proposed Design Fig. 1 : Schematic diagram of the proposed voltage reference circuit II-A PTAT Circuit The core part of the PTAT circuit as shown in Fig. 1 that generates the PTAT current are two MOSFETs, M_{1} and M_{2} , operating in subthreshold region, having a similar I-V characteristics to that of a BJT, and also because of which power efficiency is significantly improved. The aspect ratios of M_{1} and M_{2} are such that, M_{2} is N_{p} times larger than M_{1} as marked in the schematic Fig. 1 , and the leakage currents, that can be defined by equation ( 1 ), of transistors M_{1} and M_{2} are set equal by the cascode current mirrors M_{3} , M_{4} , M_{5} , and M_{6} . Cascode current mirror helps in improving the line sensitivity of the voltage reference and reduces the voltage drop across MOSFETs M_{1} and M_{2} , for them to persist in the subthreshold region across large temperature and power supply variations. The resistor R_{1} is necessary for maintaining a voltage drop between the gate voltages of M_{1} and M_{2} as a function of temperature. The subthreshold leakage current I_{ds} can be given as [ 5 ] I_{ds}=I_{ds0}e^{\frac{V_{gs}-V_{th}}{nv_{T}}}\left(1-e^{-\frac{V_{ds}}{v_{T}}}\right) (1) where n is a process dependent factor caused by the depletion region characteristics and is typically in the range of 1.3 – 1.7 [ 5 ] . The gate to source voltage is indicated as V_{gs} , V_{ds} is the drain to source voltage, and V_{th} is the threshold voltage. The last term, i.e. (1-e^{-\frac{V_{ds}}{v_{T}}}) , can be approximated to 1 as the V_{ds} of M_{1} and M_{2} will be biased at a voltage of at least 150 mV which is a few multiples greater than v_{T} even at higher temperatures (34.6 mV at 130°C). The \eta V_{ds} term in the equation 2.42 from [ 5 ] caused due to drain-induced barrier lowering has been neglected to make calculations simpler, which can later be curvature compensated including other non-linearities. The coefficient I_{ds0} is given as [ 5 ] I_{ds0}=u_{n}C_{ox}\frac{W}{L}v_{T}^{2}e^{1.8} (2) \mu_{n} , C_{ox} , v_{T} and \frac{W}{L} are as usual the electron mobility, per unit area gate oxide capacitance, thermal voltage and the aspect ratio of the MOSFET respectively. Now, as both the drain currents of M_{1} and M_{2} are set equal, we can say \frac{I_{ds0,2}}{I_{ds0,1}}=e^{\frac{V_{gs1}-V_{gs2}-(V_{th1}-V_{th2})}{nv_{T}}} (3) If we can assume that the two threshold voltages are approximately equal (although practically they would be slightly off, which would just result in an additional constant close to 1 multiplied to our final expression), the numerator in the exponent will just result in V_{gs1}-V_{gs2}=I_{ds1}R_{1} . Take natural log on both sides in equation ( 3 ) we get ln\left[\frac{(\frac{W}{L})_{2}}{(\frac{W}{L})_{1}}\right]=\frac{I_{ptat}R_{1}}{nv_{T}} (4) if we can call I_{ds1}=I_{ds2}=I_{ptat} . As discussed earlier, if (\frac{W}{L})_{2} : (\frac{W}{L})_{1} :: N_{p} : 1 , the final expression for I_{ptat} can be derived as I_{ptat}=\frac{nv_{T}}{R_{1}}lnN_{p} (5) Hence we can say that the leakage currents of M_{1} and M_{2} have a PTAT nature, whose strength and the slope can be adjusted by altering N_{p} or the ratio of the sizes of M_{1} and M_{2} , and R_{1} as can be seen in Fig. 2 . II-B CTAT Circuit The CTAT architecture, as shown in Fig 1 , is a modified version of the voltage reference design proposed by F. Olivera and A. Petraglia [ 1 ] , which exploits the multi-threshold characteristics of a MOS transistor. The unified current control model (UICM) accurately describes transistor behavior across all inversion levels that the conventional square law based models are unable to predict [ 1 ] . The core part of the CTAT circuit is the nature of the gate voltage of the MOS transistor M_{0} , which is copied across the resistor R_{2} (labeled V_{ctat} ) with N_{c} times higher current flowing in that branch. The transistors M_{8}:M_{9} and M_{12},M_{14}:M_{13},M_{15} are sized in the ratio N_{c}:1 and, M_{10}:M_{11} and M_{16},M_{18}:M_{17},M_{19} are equally sized pairs to realize a CTAT current generator from a CTAT voltage. The M_{0} MOSFET, being diode connected, will operate mostly in forward saturation, and its drain current I_{D} can be given as [ 7 ] I_{D}=I_{S}i_{f}\Rightarrow i_{f}=\frac{I_{D}}{I_{S}} (6)