For amplifier applications which require overvoltage protection and also call for low distortion, low noise, and high bandwidth, engineers must pay careful attention to the design of the overvoltage protection. An overvoltage may be induced by human error such as shorting the amplifier input to a higher supply voltage, or may be inherent to the application, such as a transducer that routinely produces voltages higher than the amplifier supply rails.
Most amplifier overvoltage protection methods use diodes to shunt overvoltage fault current to ground or to the supply rails. These diodes have capacitance and leakage current that contribute to distortion and limit bandwidth. This article explains the basics of reverse-biased diodes, discusses several protection strategies, and provides a few solutions for reducing parasitic leakage and capacitance. Operational amplifiers are used to illustrate protection methods, although many of the methods are useful for discrete amplifiers as well.
Reverse-biased diode basics
After inspecting of the diode equation
you might assume that a reverse-biased diode draws a reverse current IR equal to IS. In practice, however, the current is much higher than IS and is not constant across temperature and reverse-bias voltage. IR is proportional to the volume of the space-charge layer in the PN junction, and because the space-charge layer volume depends on applied reverse voltage, IR is typically modeled by the equation:
where n can vary roughly from 2 to 4, depending on the manufacturer.
Normally, IR vs. VR curves are found in the diode datasheet. A widely accepted rule-of-thumb states that the reverse current of a PN junction doubles for every 10° C rise in temperature.
From this rule, and a reference point, we can establish an equation that relates reverse current to temperature:
I0 is the reverse current as specified at a temperature T0. Usually IR vs. T curves are found in the diode datasheet.
The capacitance of a diode below its built-in potential (approximately 0.7 V for silicon) is modeled by the following equation:
Cj0 is the PN junction capacitance at 0 V, Φ0 is the built in potential, and M is the grading coefficient, which is unitless and quantifies the abruptness of the P material where it meets the N material. Equation 4 assumes VR is negative for a reverse bias voltage and positive for a forward bias voltage.
The equation is a good model for the reverse bias capacitance and a good model for the forward bias capacitance up to about half of the built-in potential. Normally CR vs. VR curves are found in the diode datasheet.
Basic diode protection
Most all ICs have some form of internal Electro-Static Discharge (ESD) protection. The most popular internal protection circuit employs ESD clamping diodes connected to the supply rails, thereby shunting ESD strikes to the power supplies. You can argue that these diodes are sufficiently robust to handle an overvoltage if the current is limited through a series resistor; however, each IC is different and the ESD protection topologies may vary considerably.
It's best to add external clamping diodes to the supply rails to minimize or eliminate the amount of overvoltage current that flows into an IC terminal (see Figure 1).
Figure 1: Basic Diode Protection
(Click to Enlarge Image)
The basic diode-protection method shown in Figure 1 clamps the input voltage of the amplifier to VCC + VFBD and VEE - VFBD, where VFBD is the forward-biased voltage drop of the diode.
The overvoltage current is limited by RLIMIT as per the following equation:
where VSUPPLY is VEE or VCC. This protection method also works for the inverting op-amp configuration, where RLIMIT serves as a gain-setting resistor.
The forward voltage drop of an ordinary silicon diode will be close to that of an internal ESD diode. This means that in an overvoltage condition, both the internal and external diodes will share the overvoltage current. Because the forward-drop matching between the two types of diodes is unknown, it is good practice to assume that all the overvoltage current will flow through the internal ESD diode. An industry rule-of-thumb is to select RLIMIT so that no more than 5 mA will flow through the IC input.
Schottky diodes exhibit a lower forward drop (0.3 V) and are frequently used in this protection topology to shunt most of the fault current. The lowest-leakage Schottky diodes, however, exhibit orders of magnitude more leakage current than the lowest-leakage silicon diodes. In applications where input bias currents approach nanoamperes or lower, Schottky leakage becomes overbearing. Also, a Schottky diode's forward drop can easily increase to 0.7 V with temperature and forward-bias current. The popular 1N5711 Schottky diode has a maximum forward drop of 1 V with 15 mA of bias current at room temperature.
Protection diode reverse-bias leakage becomes important when amplifier input bias currents are small. Ideally, the leakage currents in both protection diodes are equal and no offset is introduced. However, in practice the diodes aren't perfectly matched and the leakage currents will change with input voltage and temperature, creating offset error and non-linearities. A good rule of thumb is to keep the maximum reverse-leakage current ten times smaller than the input bias current of the amplifier.
Protection diode reverse-bias capacitance, CR, is also an important design criterion to consider. This reverse-bias capacitance across each diode, combined with RLIMIT, creates a low-pass filter with a cutoff frequency
Recall that CR is a function of the applied voltage, so if the input voltage swing is large, significant non-linearities may be introduced.
Protection circuit overvoltage recovery time may also be an important criterion. When a diode is forward biased, charge is stored in the PN junction's depletion region. To turn the diode off, the charge must be removed from the depletion region. Although manufacturers of high-speed switching diodes generally specify a reverse-recovery time trr, manufacturers of lower-leakage diodes generally do not. Reverse recovery time can be measured if trr is not specified in the datasheet.
Many IC companies package diode arrays that have fairly good reverse leakage/capacitance specifications. For example, the MAX3202E ESD protection diode array exhibits a 1 nA (max) leakage current, with 5 pF of capacitance per channel. For lower reverse leakage, diode-connected 2N3904s are a good choice. The PAD1 diodes by Vishay exhibit even lower reverse-leakage current and capacitance: 1 pA (max) and 0.8 pF (max), respectively.
Part 2 of this article looks at diode protection to ground, differential diode protection, and more. It also includes additional schematics. You can access Part 2 by clicking here.
References
1. Op-Amp Applications Handbook (by Walt Jung, Analog Devices)
2. Overvoltage Effects on Analog Integrated Circuits (by Adolfo Garcia, Wes Freeman, Analog Devices)
3. Analysis and Design of Analog Integrated Circuits (by Gray, Hurst, Lewis and Meyer)
4. Intuitive IC Op Amps (Thomas M. Frederiksen)
5. Noise Reduction Techniques in Electronic Systems (Henry W. Ott)
About the authors
Erik Anderson and Eric Schlaepfer are with Maxim Integrated Products, Sunnyvale, CA.
