Auto-zero operational amplifiers for portable signal conditioning

Abstract

Chopper amplifiers have been used for decades, dating back to the 1960s. Chopper amplifiers were invented to meet the need for ultra-low offset and low drift operational amplifiers that were superior to the bipolar operational amplifiers of the time. In the original chopper amplifiers, the input and output of the amplifier were switched (or discontinuous), the input signal was modulated to compensate for offset errors, and the output was unmodulated. While this technique solved the low offset voltage and low drift problems, it also had other constraints. Since the input to the amplifier was sampled, the frequency of the input signal had to be less than half the chopping frequency to prevent aliasing. In addition to the bandwidth constraint, chopping also causes many large interferences, so these ripples need to be smoothed and filtered at the output.

Later, the chopper amplifier was improved to form a chopper-stabilized operational amplifier through self-calibration. Two amplifiers are used in this architecture, namely a main amplifier and a null amplifier, as shown in Figure 1. The null amplifier corrects its own offset error by shorting the input to ground and applying a calibration coefficient to its nulling terminal, and then monitors and calibrates the offset of the main amplifier. This structure has a great advantage over the old chopper amplifier because the main amplifier can always be connected to the input and output of the IC. The bandwidth of the main amplifier then determines the bandwidth of the input signal. Therefore, the input bandwidth is no longer dependent on the chopping frequency. However, charge injection from the switching action is still a problem, which will cause transients and couple with the input signal, thus causing intermodulation distortion.

Figure 1: Simplified chopper-stabilized functional architecture diagram.

The auto-zero architecture is similar in concept to a chopper-stabilized amplifier with a nulling amplifier and a main amplifier. However, later improvements have been made to reduce noise, charge injection, and other performance features compared to the chopper-stabilized amplifier. Manufacturers use different terms to define this architecture, such as 'auto-zero', 'self-calibrated zeroing', and 'zero drift'. Regardless of the terminology, the basic concept behind it is the same.

Advantages of Auto-Zero Architecture

As mentioned above, the auto-zero architecture continuously self-calibrates the amplifier's offset voltage error. This creates several significant advantages over traditional amplifiers.

Low offset voltage: Because the nulling amplifier continuously removes its own offset voltage and then applies a correction factor to the main amplifier. The frequency of correction depends on the actual design, but it is usually several thousand times per second. For example, Microchip's MCP6V01 auto-zero amplifier calibrates the main amplifier every 100 μs, or 10,000 times per second. Because of the continuous calibration, the offset voltage is much lower than that of traditional op amps. In addition, the process of calibrating the offset voltage also calibrates other DC specifications, such as power supply rejection and common-mode rejection. As a result, auto-zero amplifiers can also achieve better rejection performance than traditional amplifiers.

Low temperature and time drift: All amplifiers, regardless of process technology and structure, will have offset voltages that vary with temperature and time. Most op amps specify temperature drift in V/°C. This drift can vary greatly from one amplifier to another, but for traditional amplifiers, it is usually a few microvolts to tens of microvolts per degree. This temperature drift is a serious problem for high-precision applications and, unlike initial drift errors, cannot be calibrated out using one-time system calibration techniques. In

addition to temperature drift, the amplifier's offset voltage also varies with time. For traditional op amps, this time drift (sometimes called aging) is usually not specified in the data sheet, but it can also produce significant errors as the component ages. Because the drift voltage is continuously

self-calibrated, the auto-zero structure essentially minimizes temperature drift and time drift. As a result, auto-zero amplifiers can achieve much higher drift performance than traditional op amps. For example, the MCP6V01 op amp has a maximum temperature drift of only 50 nV/°C.

No 1/f noise: 1/f noise, or flicker noise, is a low-frequency phenomenon caused by irregularities in the conduction channel and the noise of the bias current in the transistor. At high frequencies, 1/f noise can be ignored because white noise from other noise sources will begin to dominate. If the input signal is close to DC, such as the output signals from strain gauges, pressure sensors, and thermocouples, this low-frequency noise is a big problem.

In auto-zero amplifiers, 1/f noise is eliminated as part of the drift calibration. Because this noise source appears at the input and acts relatively slowly, it appears as part of the amplifier's drift and is compensated for.

Low offset current: Offset current is the current that flows into the amplifier input and offsets the input transistor. The magnitude of this current varies from microamperes to picoamperes and is closely related to the amplifier's input circuitry. This parameter is very important when a high-impedance sensor is connected to the amplifier input. Because this offset current flows into the high impedance, a voltage drop is generated across the impedance, resulting in a voltage error. For these applications, low offset current is a concern.

Virtually all zeroing amplifiers on the market today use CMOS in their input stages, resulting in very low offset currents. However, current injection from the internal switches can result in slightly higher offset currents than traditional CMOS input op amps.

Quiescent Current: For battery-powered applications, quiescent current is a critical parameter. Because the zeroing amplifier and other circuitry need to support the self-calibrating auto-zero structure, auto-zero amplifiers typically consume more quiescent current than traditional amplifiers for a given bandwidth and slew rate. However, significant improvements have been made in increasing the efficiency of this structure. Some op amps, such as Microchip's MCP6V03, provide a chip select or disconnect pin to reduce the quiescent current when the component is in standby mode.

Application Example: Portable Pocket Balance

Above we pointed out several parameters that auto-zero amplifiers can help improve amplifier performance. Here we will discuss an application example using a strain gauge to highlight some of the advantages of an auto-zero amplifier.

A portable balance is a special device used to weigh small items such as precious metals, jewelry, and pharmaceuticals. These devices are battery powered and often require precision up to a tenth of a gram. Therefore, this application requires high-precision, low-power signal conditioning for weighing strain gauges.

In a strain gauge, resistance is used to quantify the amount of stress caused by an external force. There are several different types of strain gauges, but the most common is the metal strain gauge. This type of strain gauge consists of a piece of metal wire or a piece of metal foil. When a force is applied, the strain in the strain gauge changes (either positive or negative), which causes the resistance of the strain gauge to change. By measuring the change in resistance, the amount of applied stress is determined. Typically, strain gauges are constructed as a Wheatstone bridge because this circuit configuration provides extremely high sensitivity. Since the change in resistance is small, the total output voltage of this Wheatstone bridge circuit is also small. For example, we can assume that the full-scale output is 10mV.

Figure 2 is a simplified circuit for application analysis. Note that this circuit is not intended for full characterization, but is simplified to show the advantages of the auto-zero structure. For example, the output of the Wheatstone bridge circuit should be buffered to provide a high input impedance, but this is not shown in the simplified circuit. In this circuit, the amplifier has a differential gain of 500, so that the full-scale output from the Wheatstone bridge will cause the amplifier output to reach 5V.

Figure 2: Simplified circuit diagram of an application utilizing an auto-zero amplifier.

Because high gain is required in this application, the offset voltage drop of the amplifier becomes critical. Any voltage offset will be amplified by the high gain of the amplifier. For example, the MCP606 is a CMOS operational amplifier used to implement nonvolatile memory for input offset voltage correction. Its maximum offset at room temperature is 250 μV. If used in this application, its maximum offset error will cause the amplifier output to have an error of up to 125mV, or 2.5% of full scale. However, if the MCP6V01 auto-zero amplifier is used, its maximum offset at room temperature is only 2 μV. The maximum error at the amplifier output is only 1mV, which is only 0.02% of full scale.

As mentioned above, another advantage of the auto-zero structure is its low time drift and temperature drift. For example, assume that the operating temperature range of this portable weighing scale is 0℃ to 50℃. The temperature drift of the MCP606 is 1.8 μV/℃. The error caused by temperature changes can be as high as 90 μV, which will be amplified by the amplifier gain and produce another 45mV error in the amplifier output. The MCP6V01, on the other hand, has a maximum temperature drift of only 50 nV/°C. Therefore, the offset error caused in the amplifier output is only 1.25 mV, which is 30 times better than the MCP606 amplifier.

As mentioned above, 1/f is a limiting factor in low-frequency applications, such as the weigh scale discussed here. The 1/f noise spectrum of the MCP606 op amp typically has a corner frequency of 200Hz. At this point, 1/f noise begins to dominate, and below 1Hz, the resulting voltage-noise density is as high as 200 nV/Hz. For the MCP6V01 op amp, due to its auto-zero correction structure, there is no 1/f noise, so it remains constant at low frequencies. For weigh scale applications, 1/f is a very critical factor at this time because the load cell output is a very slow-changing signal.

Summary of this article

Although today's auto-zero structure can be traced back to the early chopper amplifier in concept, it has been greatly improved compared to its early products. The old chopper amplifier had many shortcomings that can cause major system-level design problems. The new auto-zero structure is much better and provides much better performance. As in the above application example, the auto-zero structure can provide much better performance than the traditional operational amplifier in such high-precision applications.