Not all precision op amps are created equal:
Kevin Tretter, Senior Product Marketing Engineer, Analog & Interface Products Division Microchip Technology Inc. assesses the benefits of different precision op amp architectures.
Whilst there is little doubt that higher-precision operational amplifiers (op amps) can eliminate the need for system calibration, either during manufacturing or in the field, designers still face a choice of which low-offset architecture is best for each application. Choosing the right op amp means assessing the advantages and disadvantages of each type of architecture. This article outlines the pros and cons of op amps using EPROM trimming, laser trimming, auto zeroing and on-chip calibration.
A precision op amp is typically one in which some form of input-offset voltage correction is implemented. The input offset voltage is the voltage difference between the amplifier’s inverting and non-inverting inputs, which can vary from microvolts up to millivolts. The amount of offset is largely dependent upon how well the amplifier’s input transistors are matched.
In addition to the initial input offset voltage, other conditions can also affect the behavior of this error voltage, including changes in common-mode voltage, operating voltage, output voltage, temperature and even time. Depending upon the application, these external conditions can determine the best amplifier architecture for each design.
Some precision op amps use non-volatile EPROM fuses to correct the input offset voltage. In many cases, this is carried out in-package during the final testing stage, and is a very cost-effective way to achieve an amplifier with low initial offset voltage. Since the amplifier is trimmed post-assembly, any assembly-related offsets can be corrected. The other advantage of this architecture is that the amplifier is trimmed by the manufacturer and does not require any further trimming by the customer. The down-side, however, is that the EPROM fuses take up space on the chip, therefore EPROM-trimmed devices are unlikely to be available in ultra-small packaging.
Also, just like a general-purpose amplifier, this architecture will be sensitive to environmental conditions such as temperature, as well as changes in common-mode or operating voltage.
Another method often used to increase the accuracy of an operational amplifier is laser trimming. This process involves using a laser to adjust the resistance value of thin-film resistors within the silicon wafer. The accuracy of this approach can be relatively high, since the trimming process is continuous, rather than the series of discrete steps used in EPROM trimming. Another advantage is that thin-film resistors are inherently very stable over temperature, adding to the overall accuracy of the amplifier across a wide temperature range.
Laser trimming, however, must be implemented at wafer level and cannot be carried out within a packaged device.
The processes of sawing the wafer into individual die, placing the die in a package, and bonding the die to the packaged pins can all cause mechanical stress on the wafer that will negatively affect the overall accuracy of the device.
Such assembly-related changes cannot be accounted for with laser-trimmed amplifiers and, therefore, add to the error of the amplifier. Like non-volatile EPROM fuses, laser trimming is done only once during device manufacturing, with no option to re-trim the device. Changes in external operating conditions, such as temperature and operating voltage, will therefore adversely affect the accuracy of the amplifier and could directly impact the performance of the overall design.
Auto-zero op amps
The auto-zero architecture is a continuously self-correcting architecture that uses a null amplifier to correct the offset voltage of the main amplifier. This architecture enables ultra-low offset error which can be 100 times better than an EPROM-trimmed amplifier. It also achieves low offset drift and eliminates 1/f noise while providing superior power-supply and common-mode rejection. Since this architecture continuous self-corrects the input offset voltage, it is inherently insensitive to the environment. Changes over temperature and ageing, as well as changes in operating or common-mode voltage, will have very little effect on the accuracy of an auto-zero amplifier. As the self-correcting circuitry is integrated on-chip, no customer input is required. From a system-level perspective, an auto-zero op amp, such as the MCP6V0 shown in Figure 1, looks and functions just like a standard op amp, but with the added advantage of exceptional performance.
Despite all of these advantages, the self-correcting auto-zero architecture does have some limitations. The continuous switching of the internal correction circuitry does produce switching noise and also results in a higher quiescent current for the given bandwidth. Finally, due to the ultra-high precision of this type of device, testing time can be relatively long, resulting in a more expensive device to manufacture.
Another alternative is to use a high-precision operational amplifier with an on-chip calibration circuit. mCal calibration technology, from Microchip, enables op amps to achieve the same very low initial offset voltage as other architectures but, unlike EPROM- or laser-trimmed amplifiers, calibration is active upon power-up, or via an external calibration pin. This enables the user to recalibrate the amplifier as often as required.
Frequent recalibration can make the accuracy of the amplifier insensitive to the environment. For example, if a customer is very concerned about drift over temperature, drift error can be minimised by recalibrating the device every time the temperature changes by five degrees. Whilst this can significantly reduce the drift of the amplifier over temperature, it requires the user to actively initiate a calibration routine by toggling the calibration pin on the amplifier.
Most applications can benefit from using a higher-precision operational amplifier but, in order to choose the right amplifier, designers must understand strengths and weaknesses of each of the architectures used to achieve low offset. Whilst all of the architectures described above achieve low initial offset voltage, environmental conditions can also impact on amplifier accuracy. The use of an amplifier with continuously self-correcting architecture, such as an auto-zero amplifier, or one with the ability to recalibrate using mCal technology, allows the adverse effects of external conditions to be minimised.
Table 1 outlines the trade-offs which must be considered when evaluating which amplifier architecture is best for a particular application.