Some components, such as an error amplifier in an LDO (Low Dropout Regulator), need stable reference voltages for comparative measurements.
The most compact and stable choice of voltage reference is a bandgap voltage reference, which is often made from silicon or integrated onto a silicon die.
These components are highly stable against power supply fluctuations, temperature, and other noise sources in your system.
Gallium arsenide (GaAs) is one material that can be used to build a bandgap voltage reference circuit.
Have you ever looked at a component datasheet and seen a block diagram that shows an input for a reference voltage? Where does this value come from and how is it generated? A battery is unsuitable for use as a reference voltage source as the voltage can droop over time, producing an inaccurate voltage. A typical power supply can have a noisy output and/or some residual ripple. Thus, it also can’t act as a stable reference voltage.
So how can you design a reference voltage source to provide a stable voltage for your components? As it turns out, you can take advantage of the bandgap in a semiconductor to produce a fixed, highly stable voltage for use as a reference. These bandgap voltage reference circuits can have high stability against temperature changes. These circuits are deceptively simple to build, and they reflect important phenomena about the nature of electronic charges in semiconductors.
What Is a Bandgap Reference Voltage?
A bandgap reference voltage is a voltage source that outputs a voltage proportional to the bandgap of a semiconductor. SI bandgap voltage (Silicon) references are most common, which output at ~1.2 V. GaAs can also be used to build a reference source with larger voltage output, thanks to its wider bandgap of 1.42 eV, as demonstrated in a recent research paper. In principle, any semiconductor can be used to create a bandgap voltage reference as long as it can be deposited on standard wafer materials. For this reason, Si bandgap references are normally used as they can be included in an IC with CMOS processes.
Any bandgap voltage reference needs to provide a stable output voltage at a specific value. In order to provide a stable output voltage, an ideal bandgap voltage reference will have the following qualities:
Near-zero temperature coefficient. The output voltage from a bandgap voltage reference will depend on temperature, because the physical properties of the semiconductor also depend on temperature. The mobility of charge carriers in Si has weak temperature dependence, which is another reason this material is typically used for bandgap voltage reference circuits.
Immunity to supply voltage fluctuations. A bandgap voltage reference circuit is generally a 3-terminal device (power, ground, and output). The potential difference between power and ground may vary during operation, and the reference voltage circuit needs to dampen these fluctuations. For this reason, bandgap voltage reference circuits are nonlinear circuits; a large supply voltage change produces a small change in reference voltage.
Designing a Bandgap Reference Voltage
Because the carrier concentration in a semiconductor is a function of temperature, the goal in designing a bandgap reference voltage source is to use at least two elements with opposite temperature coefficients. Bandgap voltage reference circuits involve some combination of diodes, bipolar transistors, and/or op-amps. Note that MOSFETs are not normally used in bandgap reference voltage circuits as their electrical behavior is heavily process-dependent. The example below shows a bandgap reference voltage circuit built from a current mirror.
Bandgap voltage reference circuit built from a stable current source and a current mirror.
The top two PNP transistors regulate the current into the current mirror, and the bottom two transistors provide the output reference voltage. In particular, the total voltage drop across R2 and the base-emitter junction in Q2 is equal to the reference voltage. The base-emitter current acts like a forward-biased diode when the transistor runs in the linear regime, giving a total voltage drop that is proportional to the bandgap.
The base-emitter voltage in Q2 has a negative temperature coefficient, while the voltage drop across R2 has a positive temperature coefficient. If the two coefficients can be chosen to be very similar, the total output reference voltage will have nearly zero temperature coefficient.
Bandgap Voltage Reference Circuits With Op-Amps
An alternative bandgap voltage reference circuit uses an op-amp as a summing amplifier with a forward-biased diode and resistor. In this arrangement, the diode and resistor have opposite temperature coefficients. The summing amplifier adds the voltages, which ensures the output voltage is always directly proportional to the bandgap.
Bandgap voltage reference circuit built from a summing amplifier.
The voltage drop measured across the resistor is normally amplified to achieve the desired output voltage. The current source can be provided from a saturated transistor, just as in the previous circuit. In addition, gain provided by the summing amplifier can be used to boost the output voltage if desired. Finally, the base-emitter voltage depends on the semiconductor bandgap, just as in the previous circuit.
The arrangements shown above ensure the outputs are relatively independent of temperature, and these circuits are preferable to using something like back-to-back diodes. In a back-to-back diode arrangement, the two diodes are unaffected by supply voltage fluctuations because one of the diodes is always running in saturation. The problem with this type of arrangement is that the voltage across the diodes has a negative temperature coefficient, leading to ~600 mV output at room temperature. This shows the advantage of the two types of circuits shown above. The temperature coefficients for the circuits need to be equal and opposite.