Output Stage Surge Protection

previous Input Overvoltage Considerations
Most low speed, high precision op amps generally have output stages which are protected against short circuits to ground or to either supply. Their output current is limited to a little more than 10mA. This has the additional advantage that it minimizes self-heating of the chip (and thus minimizes DC errors due to chip temperature differentials).
If an op amp is required to deliver both high precision and a large output current, it is advisable to use a separate output stage (within the loop) to minimize self-heating of the precision op amp. A simple buffer amplifier such as the BUF04, or a section of a non-precision op amp can be used.
Note that high speed op amps cannot have output currents limited to low values, as it would affect their slew rate and load drive ability. Thus most high speed op amps will source/sink between 50-100mA. Although many high speed op amps have internal protection for momentary shorts, their junction temperatures can be exceeded with sustained shorts. The user needs to be wary, and consult the specific device ratings.
Offset Voltage Trim Processes
The AD860x CMOS op amp family exploits the advantages of digital technology, so as to minimize the offset voltage normally associated with CMOS amplifiers. Offset voltage trimming is done after the devices are packaged. A digital code is entered into the device to adjust the offset voltage to less than 1mV, depending upon the grade. Wafer testing is not required, and the patented ADI technique called DigiTrim™ requires no extra pins to accomplish the function. These devices have rail-to-rail inputs and outputs (similar to Fig. 1-33), and the NMOS and PMOS parallel input stages are trimmed separately using DigiTrim to minimize the offset voltage in both pairs. A functional diagram of the AD8602 DigiTrim op amp is shown in Figure 1-34 below.
AD8602 (1/2) CMOS op amp showing DigiTrim™
Figure 1-34: AD8602 (1/2) CMOS op amp showing DigiTrim™
DigiTrim adjusts the offset voltage by programming digitally weighted current sources. The trim information is entered through existing pins using a special digital sequence. The adjustment values can be temporarily programmed, evaluated, and readjusted for optimum accuracy before permanent adjustment is performed. After the trim is completed, the trim circuit is locked out to prevent the possibility of any accidental re-trimming by the end user.
The physical trimming, achieved by blowing polysilicon fuses, is very reliable. No extra pads or pins are required, and no special test equipment is needed to perform the trimming. The trims can be done after packaging so that assembly-related shifts can be eliminated. No testing is required at the wafer level because of high die yields.
The first devices to use this new technique are the Analog Devices' AD8601/02/04 (single, dual, quad) rail-to-rail CMOS amplifiers. The offset is trimmed for both high and low common-mode conditions so that the offset voltage is under 500μV over the full common-mode input voltage range. The bandwidth of the op amps is 8MHz, slew rate is 5V/μs, and supply current is only 640μA per amplifier.
At this point it is useful to review the other popular trim methods. Analog Devices pioneered the use of thin film resistors and laser wafer trimming for precision amplifiers, references, data converters, and other linear ICs (Reference 5: Richard Wagner, "Laser-Trimming on the Wafer," Analog Dialogue, Vol. 9, No. 3, pp. 3-5). Up to 16-bit accuracy can be achieved with trimming, and the thin film resistors themselves are very stable with temperature and can add to the thermal stability and accuracy of a device, even without trimming. Thin film deposition and patterning are processes that must be tightly controlled. The laser trimming systems are also quite expensive. In-package trimming is not possible, so assembly-related shifts cannot be easily compensated. Nevertheless, thin film trimming at the wafer level provides continuous fine trim resolution in precision integrated circuits where high accuracy and stability are required.
Zener zapping uses a voltage to create a metallic short circuit across the base-emitter junction of a transistor to remove a circuit element (References 4 and 6: George Erdi, "A Precision Trim Technique for Monolithic Analog Circuits," IEEE Journal of Solid-State Circuits, Vol. SC-10, December, 1975 pp. 412-416 and Donn Soderquist, George Erdi, "The OP-07 Ultra-Low Offset Voltage Op Amp," Precision Monolithics AN-13, December, 1975). The base-emitter junction is commonly referred to as a zener, although the mechanism is actually avalanche breakdown of the junction. During the avalanche breakdown across the base-emitter junction, the very high current densities and localized heating generate rapid metal migration between the base and emitter connections, leading to a metallic short across the junction. With proper biasing (current, voltage, and time), this short will have a very low resistance value. If a series of these base-emitter junctions are arranged in parallel with a string of resistors, zapping selected junctions will short out portions of the resistor string, thereby adjusting the total resistance value.
It is possible to perform zener zap trimming in the packaged IC to compensate for assembly-related shifts in the offset voltage. However, trimming in the package requires extra package pins. Alternately, trimming at the wafer level requires additional probe pads. Probe pads do not scale effectively as the process features shrink. So, the die area required for trimming is relatively constant regardless of the process geometries. Some form of bipolar transistor is required for the trim structures, therefore a purely MOS-based process may not have zener zap capability. The nature of the trims is discrete since each zap removes a predefined resistance value. Increasing trim resolution requires additional transistors and pads or pins, which rapidly increase the total die area and/or package cost. This technique is most cost-effective for fairly large-geometry processes where the trim structures and probe pads make up a relatively small percentage of the overall die area.
It was in the process of creating the industry standard OP07 in 1975 that Precision Monolithics Incorporated pioneered zener zap trimming (Reference 6, again). The OP07 and other similar parts must be able to operate from over ±15V supplies. As a result, they utilize relatively large device geometries to support the high voltage requirements, and extra probe pads don't significantly increase die area.
Link trimming is the cutting of metal or poly-silicon links to remove a connection. In link trimming, either a laser or a high current is used to destroy a "shorted" connection across a parallel resistive element. Removing the connection increases the effective resistance of the combined element(s). Laser cutting works similar to laser trimming of thin films. The high local heat from the laser beam causes material changes that lead to a non-conductive area, effectively cutting a metal or conductive polysilicon connector. The high-current link trim method works as an inverse to zener zapping—the conductive connection is destroyed, rather than created by a zener-zap.
Link trim structures tend to be somewhat more compact than laser trimmed resistor structures. No special processes are required in general, although the process may have to be tailored to the laser characteristics if laser cutting is used. With the high-current trimming method, testing at the wafer level may not be required if die yields are good. The laser cutting scheme doesn't require extra contact pads, but the trim structures don't scale with the process feature sizes. Laser cutting of links cannot be performed in the package, and requires additional probe pads on the die. In addition, it can require extra package pins for in-package high-current trims. Like zener zapping, link trimming is discrete. Resolution improvements require additional structures, increasing area and cost.
Summary of ADI trim processes
Figure 1-35: Summary of ADI trim processes
EEPROM trimming utilizes special, non-volatile digital memory to store trim data. The stored data bits control adjustment currents through on-chip D/A converters. Memory cells and D/A converters scale with the process feature size. In-package trimming and even trimming in the customer's system is possible so that assembly-related shifts can be trimmed out. Testing at the wafer level is not required if yields are reasonable. No special hardware is required for the trimming beyond the normal mixed-signal tester system, although test software development may be more complicated. Since the trims can be overwritten, it is possible to periodically reprogram the system to account for long-term drifts or to modify system characteristics for new requirements. The number of reprogram cycles possible depends on the process, and is finite. Most EEPROM processes provide enough rewrite cycles to handle routine re-calibration.
This trim method does require special processing. Stored trim data can be lost under certain conditions, especially at high operating temperatures. At least one extra digital contact pad/package pin is required to input the trim data to the on-chip memory. This technique is only available on MOS-based processes due to the very thin oxide requirements. The biggest drawback is that the on-chip D/A converters are large—often larger than the amplifier circuits they are adjusting. For this reason, EEPROM trimming is mostly used for data converter or system-level products where the trim D/A converters represent a much smaller percentage of the overall die area.
Figure 1-35 above summarizes the key features of each ADI trim method. It can be seen from that all trim methods have their respective places in producing high performance linear integrated circuits.
Share this article :


Post a Comment

Please wait for approval of your comment .......