When selecting a current sensing resistor, engineers typically review four key specifications from the datasheet: rated power, temperature coefficient of resistance (TCR), resistance tolerance, and pulse withstand capability. Many treat these as independent checkboxes; if all four pass individually, the selection is considered complete.
The problem lies in the fact that these four parameters are linked along a single thermal path: Power dissipation determines temperature rise → Temperature rise amplifies TCR drift → Drift combined with initial tolerance constitutes total error. Pulse withstand corresponds to a different, shorter thermal path—pulse energy injects heat into the resistive film, causing local hot spots that lead to permanent resistance shifts or open circuits. If each parameter has margin but the system error exceeds limits, it is likely because they were treated as independent knobs rather than coupled variables.
Misapplication of Parameters: Resistance Value Determines Actual TCR
Consider a 0.5Ω, 1206 package, ±1% tolerance thick-film sensing resistor. At 20mA, the power dissipation is only 0.01W, providing ample power margin. However, regarding TCR: the front page of the datasheet may list ±100ppm/°C, while the detailed table for low-resistance ranges lists ±600ppm/°C. For a 30°C temperature rise, this results in drifts of 0.3% and 1.8%, respectively. The latter already exceeds the initial ±1% tolerance.
In the same datasheet, two values differ by a factor of six. The parameters are not wrong; the error lies in referencing the row that does not apply to the specific resistance range.
Power Verification: Temperature Rise Criteria Under Ambient Correction
Resistance value is first determined by signal strength and allowable self-heating range: large current main loops typically use 0.5–2mΩ, while signal-level sampling uses 10–50mΩ.
The approach to power verification needs correction—it is not about whether "calculated power is less than rated power," but rather "under worst-case ambient temperature + maximum continuous current, does the device's temperature rise reach its limit?" TI engineering references suggest that continuous current should not exceed 2/3 of the rated value, dropping to as low as 1/4 under insufficient heat dissipation. This translates to using only 44% of the rated power.
For a 2512 package resistor with a 2W rating, at 85°C ambient, 1.65W is allowed; at 105°C ambient, 1.18W is allowed. An 0.8W load at 85°C represents a 48.6% load rate—numerically there is margin, but thermally, the design may actually be failing.
Copper area, via count, PCB layer count, and air cooling are not just layout finishing touches; they are integral parts of component selection. Approximately 90% of self-generated heat is conducted away through PCB copper foil. The same resistor can experience a 20°C difference in temperature rise on different boards.
TCR Dependency on Resistance: Boundaries of Front-Page Nominal Values
This is the most frequent parameter misapplication. Taking the Ta-Ii CRH thick-film series as an example:
| Resistance Range | TCR |
| 10Ω–1MΩ | ±100ppm/°C |
| 1Ω–10Ω | ±400ppm/°C |
| 100mΩ–1Ω | ±600ppm/°C |
| 50–100mΩ | ±800ppm/°C |
| 10–50mΩ | ±1800ppm/°C |
Within the same model prefix, TCR varies by a factor of 6–18 across different resistance ranges. The "±100ppm/°C" repeatedly seen on front pages and e-commerce listings applies only to resistances above 10Ω. This is not poor labeling rigor—low-resistance segments require a higher ratio of metal particles in the paste to reduce sheet resistance. The resulting thin film with coarse grains naturally leads to a higher TCR.
Conversion habit: 1% = 10,000ppm. A TCR of 600ppm/°C means 1% drift occurs every 16.7°C; 1800ppm/°C requires only 5.6°C. When seeing milliohm-level resistors with TCRs above 600ppm/°C, first estimate the board's temperature rise magnitude.
Initial Tolerance: Priority Positioning Under Calibratable Compensation
The full definition of ±1% is: resistance tolerance at 25°C, factory state, before soldering. It does not include thermal drift, soldering thermal stress, long-term aging, or lead and joint resistance. After superimposing these factors, field measurements showing 3% error are not uncommon.
There are significant differences between material systems: Alloy resistors have a TCR of ±50ppm/°C and annual aging <0.5%, costing 2–5 times more than thick-film resistors; Thick-film resistors typically have a TCR of ±200–500ppm/°C and annual aging of 1–3%.
Among the four parameters, initial tolerance is the only one that can be directly compensated via production line methods. Single-point calibration eliminates initial tolerance; multi-point calibration partially compensates for thermal drift; Kelvin four-wire layout eliminates lead parasitics. TCR can only be addressed by changing materials or reducing temperature; power margin can only be improved by modifying copper pours—calibration cannot solve these last two.
Sacrificing thermal design margin to achieve ±0.5% initial tolerance puts the cart before the horse.
Pulse Withstand: Energy-Time Curves and Repetitive Pulse Criteria
Surges, short circuits, hot-plugging, and lightning strikes share one mechanism: Energy injection → Film heating → Local hot spots → Resistance shift or open circuit.
Key distinction: A 1ms, 100A pulse with high peak power, if repeated every 10ms, is governed by average power—under continuous pulses, average power must not exceed rated power consumption. For sparse single pulses with high energy, the device may remain unaffected provided the peak temperature does not burn through the film.
When checking pulse capability, refer to the curves and note three prerequisites:
- Curves must specify the allowed resistance change (commonly ±0.25% R); otherwise, "withstanding" has no judgment standard;
- Confirm if the curve covers the entire resistance range or a single value; milliohm and kiloohm ranges do not share curves;
- Thick-film layers have poor uniformity and small effective area, offering the weakest pulse capability; Thin-film is better, and MELF thin-film is optimal.
Pulse capability is determined by material and package volume. During selection, you can only "choose the material"; layout optimization and calibration cannot improve this metric.
Selection Decision Chain: Ranked by Remediability of Failure
The criterion is simple: If a parameter exceeds its boundary, can it be remedied by other means?
| Priority | Parameter | Failure Consequence | Remediable? |
| 1 | Power & Thermal Design | Burnout, Open Circuit | Irreversible; requires material change or redesign |
| 2 | Pulse Withstand | Permanent Shift or Open Circuit | Irreversible; determined by material volume |
| 3 | TCR | Thermal Drift Error | Remediable: Derating, larger copper pours, alloy substitution, multi-point calibration |
| 4 | Initial Tolerance | Factory Tolerance | Easiest to remedy: Production line calibration, four-wire layout, software calibration |
Five operational steps, order cannot be reversed:
- Determine resistance: Set range based on signal strength and allowable self-heating;
- Calculate power: Use max continuous current + worst-case ambient temp; check derating curves, do not use typical values;
- Check TCR: Look up the tiered table for the specific resistance range; calculate drift as a percentage of the error budget (Temp Rise × TCR);
- Verify pulses: Check pulse energy, width, single vs. repetitive, and allowed change specified in curves;
- Compare accuracy & cost: Only then compare initial tolerance grades and procurement costs.
Boundary Conditions: Applicable Scenarios and Exceptions to the Priority Chain
Microamp leakage detection: Dominant errors are noise, amplifier offset, and thermoelectric potential; TCR priority shifts down;
Forced temperature suppression (large copper area + air cooling keeping rise within 10°C): TCR impact scales down proportionally; choosing thick-film for cost savings becomes feasible;
High-volume consumer electronics: Thick-film + production line calibration replaces 2–5x material cost difference with one-time calibration labor; viable at scale;
Pulse-dominant conditions (short-circuit protection, surge absorption): The order flips directly; pulse capability takes precedence.
This priority chain is a default starting point, not a fixed formula. The judgment basis remains constant—can the failure be remedied?
Returning to the initial 0.5Ω resistor: None of the four parameters were labeled incorrectly; the error was treating them as four independent switches to tick off. Viewed within the thermal chain, power, temperature rise, TCR, and accuracy are four links on the same line, while pulses form another shorter line. Prioritize irreversible failures first, then address compensable ones. Selection transforms from parameter comparison to engineering judgment.
