Diode forward voltage drop is just as useful as diode rectification, it can vary greatly with temperature, resulting in increased losses and tolerances in the power supply.
While it is impossible to eliminate losses, diodes can be used to reduce tolerance errors in some applications. This article will demonstrate how to achieve this through three examples.
You can build a simple low current regulator using a resistor and a zener diode. This type of regulator is usually suitable for non-critical applications, such as internal bias voltage. Generally, the Circuit will control the tolerance of the output voltage to about ±10%, but it is possible to improve regulation by connecting a diode in series.
Figure 1 shows a zener diode circuit in series with a diode, and the curve plots the temperature coefficient of the zener diode for different voltages. When the voltage of the Zener diode is greater than 4.7V, the temperature coefficient gradually becomes a positive number, so when the operating temperature increases, the voltage of the Zener diode increases accordingly. If paired with a diode with a negative temperature coefficient, by reducing the diode forward voltage, the increased voltage from the Zener is canceled out, thereby eliminating the temperature error.
When the Zener diode voltage is less than 4.7V, the corresponding temperature coefficient is negative, and a diode in series will actually increase the regulation error.
Figure 1: Placing a positive temperature coefficient Zener diode in series with a negative temperature coefficient diode reduces temperature error.
For example, a 7.5V Zener diode has a temperature coefficient of +5mV/°C, while a conventional diode (BAT16) has a temperature coefficient of about -1.6mV/°C at 10mA. With very small diode currents, the temperature coefficient gets progressively smaller (-3mV/°C), so be sure to check when the zener diode is passing current. Ideally, the positive and negative temperature coefficients completely cancel each other out, but this is impractical and unnecessary, and a simple improvement would suffice. In the case of diodes with high voltages and higher positive temperature coefficients, two (or more) diodes can be used to improve cancellation.
Figure 2 shows the voltage regulation deviation calculated in Figure 1 with different Zener diodes in the case of no diode in series, one diode in series, and two diodes in series over an operating temperature range of 25°C to 100°C Comparison of output voltages. The vertical line in Figure 2 shows that the temperature-dependent error can be reduced by 3~5% at 7.5V output voltage with the addition of series diodes.
Figure 2: Placing one or more diodes in series with Zener diodes with voltage values over 4.7V can reduce voltage regulation errors.
The second example uses a converter that requires a level shifter to send output voltage information to the control circuit.
Figure 3 is an inverting buck-boost circuit with negative input to positive output. The control circuit is referenced to the -Vin rail and the output voltage is referenced to ground. In order for the control circuit to precisely adjust the output voltage, the level shifter recreates the differential “Vout to GND” voltage between “FB and -Vin”. In this implementation, a current source approximately equal to (Vout – Vbe Q1)/R flows from Vout to Vin. Current flows in the lower resistance, recreating the output voltage referenced to -Vin. Adding Q2, configured as a diode, can recover the Vbe drop loss generated by Q1. At this point, except for a small error related to beta, the level-shifted voltage at the FB pin nearly replicates the voltage between Vout and GND.
One benefit of adding a “diode” Q2 is that the forward voltage of Q2 and the voltage of Q1 are very close, since the currents flowing through both are almost exactly the same. To get the best voltage matching Q2, use the same resistors as Q1. Another benefit is that both resistors have the same temperature coefficient, allowing the two to more accurately track each other’s forward voltages. The temperature errors associated with Vbe changes are significantly reduced because they cancel each other out (VFB ~ Vout – Vbe Q1 + Vbe Q2). It is important to place Q1 and Q2 next to each other so that both are at the same temperature, use a two-transistor package if possible.
Figure 3: A level shifter uses Q2 to cancel out Q1-related changes.
The third example of Figure 4 shows a boost converter with a set of charge pump stages, each stage “n” adding approximately “V1” to the total output, resulting in “Vn + 1”.
Figure 4: Charge pump diode voltage drops can cancel each other out.
An approximation of the total output voltage is:
In formula (1), it can be seen that Vn+1 is largely determined by the multiple of n, but is affected by the “error term” related to the forward voltage drop of the diode and the ripple voltage of the charge pump conversion capacitor, which will has decreased. Assuming all diodes are of the same type, their forward voltage is equal to:
VD1 = VDa = VDb, resulting in formula (2):
In equation (2), the “error term” on the right makes the output voltage n+1 times lower than ideal. To improve this, VDa and VDb use Schottky diodes, while VD1 uses a conventional diode, and the forward voltage drop is equal to:
VDa = VDb = VD1/2, resulting in formula (3):
As can be seen from equation (3), it is possible to further increase the output voltage by reducing the error term associated with the diode drop. But equation (3) is still only an approximation, the concept of output voltage increase is valid.
Diode forward voltage and temperature variations can often, but not always, degrade circuit performance. These design examples demonstrate methods that have the potential to offset or minimize diode temperature-dependent errors.