Integrating voltage dividers and networks into common packages, rather than selecting discrete resistors, can ultimately produce analog circuits with better performance and reliability. As a matter of practical application, the low temperature coefficients of resistance (TCRs), tight ratio tolerances, and excellent load-life stability necessary for peak performance are better achieved in matched pairs, networks, and resistor arrays than with discrete resistors (Fig. 1).
However, engineers must keep up with resistor technology advances, or they won’t extract the best performance from their amplifiers and other IC circuits. In fact, component obsolescence forces some engineers to perform redesigns. To help avoid these costly redesigns, this article discusses the limitations of existing precision resistor technologies and techniques, new advances in precision resistor networks, and critical applications that demand the highest-precision products (see the table). As used in the table, the term “thermal stabilization” means the time required to reach within 10 ppm of the steady-state value.
Table of Contents
- Present Techniques And Limitations
- Widening The Gap Between Precision Resistor Networks
- Tracking With Temperature And Other Ratio-Stability Factors
- Critical Applications for Resistor Networks
Precision resistors are typically combined to maintain a stable resistance, or more specifically, to maintain a stable resistance ratio. Six different methods can achieve a stable resistance ratio.
The resistor manufacturer matches a group of resistors for resistance values and/or TCR and then bags them as a set. This runs the risk of mixing one set with another at assembly or the loss of one resistor, rendering the entire set unusable. Another risk is potential electrostatic discharge (ESD) damage to one or more resistors during handling or assembly. It could change the tolerance of one resistor in the set and render the entire set unusable.
On the factory floor, the manufacturer identifies the resistor’s value by the appropriate bin of resistors. The risk here is that any one empty bin prevents shipment of the unit, and/or a large inventory of unused resistors remains at the end of the contract. This also limits the circuit’s performance because the bins aren’t matched for tolerance or TCR, and the ratios can be no better than the random match and TCR among the selected resistors.
The designer specifies a tolerance that’s half the ratio allowance, ensuring adequate matching of all resistors without selection or set control. Although there’s no risk in this case, total cost still may be a concern. In addition, the total error budget may need to be increased to performance inconsistencies since there’s no guarantee of similarity among the resistors. This holds true particularly through temperature changes, because individual TCRs aren’t matched.
The resistor manufacturer combines discrete resistors in the specified combination, and then it either packages or molds them together with the specified pinouts. Factory-floor risk is minimal, but installation can be slow, coupled with an increase in size and cost.
The resistor manufacturer selects and combines chips of the appropriate resistance values in hermetic or non-hermetic packages. Interconnections in hermetic packages employ one-mil gold wire. Trimmed and/or matched, individual chip-resistor assemblies reside in a plastic or hermetic enclosure. These devices allow for resistance variation from circuit to circuit without the non-recurring engineering cost (NRE) of specialized networks. When using foil chips, it’s important to keep in mind that all foil resistors have very low absolute TCR. This ensures excellent TCR tracking, even among randomly selected chips.
Thin-film resistor manufacturers deposit the complete device on a substrate, attach leads, laser-trim to value, and encapsulate the ensemble. The resistors share the heat during the unequal self-heating. However, they’re made from the same resistive layer, so they’re expected to have the same TCR. This isn’t always the case, though, as they may not always move in the same direction with time. Nonetheless, they do respond more uniformly than individually produced and later selectively combined resistors.
Foil chips, on the other hand, will always move together in the same direction. On a single substrate, they will behave even more uniformly and with tighter TCR tracking than individual chips—as low as 0.1 ppm/°C. In quantity, this represents the most cost-effective way of producing high-performance resistor networks. Regardless of the technology employed, single-substrate networks require upfront costs (NRE) for masking, fixturing, and testing—all resulting in extended delivery time for the first deliverable product.
For applications that require less precision, thick-film networks also are created by screening pastes of mixed ceramic and metal particles, which are then fired and solidified into “cermet” resistors. Currently, thick-film resistor networks are the most widely used—by a large margin. Still, thin-film and foil components offer the advantage of lower TCR and tighter tolerances.
Unfortunately, “thin film” isn’t unique. Devices falling under this classification may be made using any one several processes and materials : vacuum depositing chromium cobalt on glass; evaporation of Nichrome (Ni-Cr alloy) onto glazed ceramic; sputtering of NiCr onto ceramic; deposition of NiCr on silicone; deposition of tantalum/nickel (Ta/Ni) on silicone; sputtering of combined layers of tantalum nitride (TaN) and NiCr onto ceramic; sputtering of Ta/Ni onto ceramic; silico-chrome (SiCr) on ceramic and Cermet, ion beam deposition; and ion implantation.
Obviously, not all of these methods can lead to exactly the same performance parameters. Moreover, categorizing a specified device in critical applications should go beyond “thin film.” Otherwise, divergent performances will likely be applied to the same drawing. Several factors should be considered in the total error budget of a resistor network (Fig. 2). However, performance inconsistencies between resistors may require an increase in that budget.
The gap between prior thin-film networks and the new generation of foil networks applies primarily to absolute TCR, stability, and value range and/or cost. Precision thin-film resistor networks are limited to 10 ppm/°C absolute TCR and TCR tracking to 2 ppm/°C. In some cases, screening will achieve a 5-ppm/°C absolute TCR. A technology that delivers <2 ppm/°C absolute TCR consistently over a value range from 100 Ω would substantially widen the performance gap.
New state-of-the-art foil networks accomplish that feat. Through a proprietary cold-rolled method of foil production and employing the principles of constant-sheet resistivity and variable real estate, performance is optimized whereby absolute TCR of <2 ppm/°C is possible regardless of the resistance value. Other benefits include very low current-noise levels, very low thermal EMF, fast thermal stabilization, ESD immunity, and a very high stability through thermal and mechanical stresses.
Two resistors in the same network can be at different temperatures due to proximity to external heat sources, or differential self-heating from uneven power dissipation. Take, for instance, circuit elements consisting of an operational amplifier and two resistors. If the amplification ratio is 10:1 at room temperature, what will it be when, after powering up the equipment, one resistor’s temperature increases by 60°C while the second only rises by 6°C? To answer this, we need to know the make and TCR of the two resistors.
Changes in ratio resulting from the model can expand repeatedly depending on the circuitry. How important is a shift of 1 ppm? The practical application and importance of choosing a fundamentally low absolute-TCR resistor pair to maintain ratio stability is clear. The degree of demand may vary, from demanding military applications to the scaling requirements of commercial applications.
Other factors may cause ratio changes. For example, a precision voltage divider might require a voltage divider with ratio R1 = 10 Ω to R2 = 9990 Ω (total resistance R = 10 kΩ). Even if large temperature swings aren’t expected in this application, a difference of about 0.5°C from one lead or termination to another is likely. The 0.5°C difference results in a thermal EMF error superimposed on the output voltage.
To visualize that effect, suppose this divider had a 5-V input with a 5-mV output. If the 0.5°C were in the 10-Ω resistor leads, a typical thermal EMF of 8 µV/°C would change the output by 4 µV, or an error of 0.08% of an expected 5-mV output.
New foil resistors have a thermal EMF of <0.1 µV /°C, which would result in an output change of only 0.05 µV, or an error of only 0.001%—80 times less than the typical resistor. Whatever system is being fed, this divider’s output would start off with 80 times more error using typical resistors instead of foil resistors. Say the typical divider for an op amp requires ratio matches of 0.005% to 0.01%. The thermal EMF error would then amount to 10 times the design ratio using a typical resistor, versus only one-tenth the design ratio using a foil resistor.
Many circuits require stability over time and temperature, under varying power, and through environmental changes. Design engineers can attempt to meet those criteria by choosing resistors and networks with proven histories of reliable performance in similar applications.
Damage caused by electrostatic discharge is a concern in any precision application. There are three types of ESD-related damage: parametric failure, catastrophic damage, and latent damage.
Parametric failure takes place when the ESD event alters one or more device parameters (resistance in this case), causing it to shift from the required tolerance. Catastrophic damage occurs when the ESD event causes the device to immediately stop functioning. Latent damage occurs when the ESD event causes moderate damage to the device, which isn’t noticeable because the device still appears to be functioning correctly. However, the load life of the device reduces dramatically, and further degradation caused by operating stresses may cause the device to fail during service.
Overall, foil resistors withstand high ESD spikes more reliably compared to their thin-film and thick-film counterparts.
Resistors are the passive building blocks of an electrical circuit. They may be used to drop the voltage, buffer the surge when the circuit is turned on, provide feedback in a monitoring loop, sense current flow, etc. When the application requires stability over time and load, initial accuracy, minimal change with temperature for more than 200°C, resistance to moisture, and a number of other characteristics, only the latest-generation foil resistors offer the necessary attributes.
Recently, demand has grown for precise, stable, and reliable resistors that can operate in harsh environments, especially at extreme temperatures to +220°C. Many analog circuits for industrial, military, aerospace, alternative energy, down-hole, oil-well, and automotive applications require passive components. Such passives include resistors that have a minimal drift from their initial values, particularly when operating above 175°C and in humid environments. In these applications, the most important factor is end-of-life tolerance, which is part of the stability, and to a lesser extent, the initial tolerance.
Foil resistors provide stabilities well under the maximum allowable drift necessary to withstand thousands of hours of operation under harsh conditions. Examples include the extreme temperatures and radiation-rich environments of down-hole oil-well logging applications, in the frigid arctic, under the sea, or in deep space. The devices receive stabilization processing, such as repetitive short-term power overloads, to ensure reliable service through the unpredictable stresses of extreme operation
For applications that require high reliability, the resistor production process includes MIL-style testing consisting of electrical and environmental stresses to screen out and identify any parts or lots that exhibit performance variations. The level of testing needed depends on how critical the circuit is to the overall objective of the project. Parts testing may vary depending on the specification’s requirements. However, the tests are usually separated into groups. Some undergo screening tests performed on 100% of a production lot, while other lots undergo destructive tests performed on a sample of units from within a specific lot to guarantee the performance of the rest.
For space-project applications (or even military and avionics), a more specific and relevant testing pattern developed by NASA may be used as a basis. The EEE-INST-002 guidelines titled “Instruction for EEE Parts Selection, Screening, Qualification, and Derating” are used as a basis for the development of several specifications that use foil resistors. Subjecting the resistors to the aforementioned sequence of tests can improve the confidence level of a part to perform in a critical space application. NASA developed the EEE-INST-002 guidelines to be flexible enough to accept any technology and part type. Consequently, there are foil devices with a number of different specifications suitable for these types of critical projects.
Some military applications require a part that’s not qualified to a QPL specification. The Defense Supply Center of Columbus also maintains several up-screening plans oriented around high-demand products that haven’t been QPL-qualified. These specifications, known as DSCC specs, consist of a 100% screening test (Group A testing) and a destructive series of tests (Group B testing). DSCC specifications are easily recognized by engineers, distributors, and QA organizations and come in a wide variety of part types. Hermetically sealed precision foil-resistor networks are available with screen/test flow in compliance with EEE-INST-002 (Tables 2A and 3A, Film/Foil, Level 1) and MIL-PRF-83401.
For applications in isolated locations with minimum access, especially in high-temperature and high-humidity environments, equipment must operate on its own until the specified recalibration. Resistor networks used in these high-reliability applications must provide stable operation for long periods of time without any means of adjustment or recalibration.
The best long-term tracking stability for thermally coupled resistors is guaranteed by mounting the resistors in the same hermetically sealed package. The ceramic package, which has the advantage of electrical isolation on the underside and high-heat dissipation capability (heatsink effect), combined with the package’s hermeticity and chip location, helps preserve uniformity. The electrical specs in a hermetically sealed network hold their tight TCR ratio under the combined influences of temperature, load, and time.
Designers can avoid costly redesigns by implementing new-generation foil resistors into circuits. As demonstrated, significant performance differences exist among the families of products categorized as precision resistors. Resistors built on foil technology offer several advantages over thin and thick film, including absolute TCR, stability, thermal EMF, current noise, ESD immunity, rise time and others. In critical applications that have high-temperature, high-reliability, or stringent military, medical and space requirements, foil resistors have become the device of choice.
Vishay Foil Resistors (various resistor configurations).