Advertisement

Blog

Optimizing your overcurrent detection to meet system requirements

One of the most common reasons to measure current in today’s automotive systems is to detect fault conditions and respond to them quickly. I touched on this topic previously in Signal Chain Basics #100: Signal Chain Basics (Part 100): Rethinking System-Level Management with Subsystem Over-Current Detection and Monitoring, when I talked about the benefits of distributed fault identification. In this article, I will discuss how to translate system accuracy requirements into implementation parameters.

A discrete implementation

The first specification you’ll need to determine is the threshold at which the fault must be detected. The maximum current allowable by either a single component or the total current consumable by the system determines this threshold. Regardless of how you define the limit, the accuracy of the overall measurement system will determine its actual trip point. Figure 1 shows a typical overcurrent circuit.

Figure 1

Overcurrent detection implementation featuring discrete operational amplifier and comparator

Overcurrent detection implementation featuring discrete operational amplifier and comparator

The primary overcurrent detection error sources include:

  • Shunt resistor (RS ) tolerance and drift.
  • Amplifier circuit gain error (RI and RF ).
  • Voltage divider (R1 and R2 ) error between the amplifier and comparator.
  • Comparator reference (R3 and R4 ) input error.

For the circuit shown in Figure 1 , let’s assume that all resistors are 2%, 100 ppm/o C except the shunt, which will be 1%, 50 ppm/o C. For this example, Table 1 lists the worst-case error for each of the four sources above at room temperature (25o C) and high temperature (125o C), respectively.

Table 1

Worst Case Errors for discrete implementation shown in Figure 1

Worst Case Errors for discrete implementation shown in Figure 1

The errors due to the shunt, the gain, and the divider between the amplifier and the comparator all contribute to a worst-case current measurement error of ±9% at room temperature and ±13.5% over temperature. This current measurement error is one of the inputs to the comparator; the other input is from the voltage divider. Assuming an ideal supply, the error on the reference input is 4%/6%, respectively. As these two errors are uncorrelated, this could lead to a total error of 13%/19.5% in the actual threshold trip point. Such a level of error means that the trip point needs to be set at a lower level than the application requires in order to ensure that it trips before system damage occurs. Improving any of these error sources can minimize the headroom required and improve the efficiency of an overcurrent detection implementation.

Improving gain error with a dedicated current-sense amplifier

Let’s discuss amplifier gain error improvement first. Using higher-precision lower-drift resistors with the circuit shown in Figure 1 is one option; however, as the accuracy and drift of the external components increases, so does cost. An alternative is to use a current-sense amplifier such as the Texas Instruments (TI) INA185. Current-sense amplifiers integrate a precision matched resistor gain network that cost-effectively reduces the gain error as well as the drift. In the case of the INA185, the gain error is ±0.25% with 8 ppm/o C drift, or ±0.33% over temperature.

It is possible to improve reference error by using improved resistors. Using a comparator with an integrated precision reference such as the TI TLV4021 can also offer significant improvement, with ±0.04% error over temperature. Figure 2 shows a circuit that utilizes both the INA185 and the TLV4021 for an overcurrent detection circuit.

Figure 2

Overcurrent detection implementation featuring precision current-sense amplifier and precision comparator

Overcurrent detection implementation featuring precision current-sense amplifier and precision comparator

Again, assuming that all discrete resistors are 2% and 100 ppm/o C, Table 2 lists the individual error for each of the four items. The total error of the overcurrent detection would be 11.83% over temperature.

Table 2

Worst Case Errors for the INA185 and TLV4021 implementation

Worst Case Errors for the INA185 and TLV4021 implementation

Fully integrated overcurrent detection

This leaves divider error as the primary error source. You can eliminate this error source by using a current-sense amplifier such as the TI INA301, which integrates the comparator and a precision reference, as shown in Figure 3 .

Figure 3

INA301 Functional Block Diagram

INA301 Functional Block Diagram

The INA301 has a precision current source on chip that requires only a single external resistor to set the threshold. The overtemperature error for such an implementation would be 5.8% worst case over temperature based on the individual errors listed in Table 3 .

Table 3

Worst Case Error s for INA301 implementation

Worst Case Error s for INA301 implementation

In either of the first two implementations, if the transfer function of the shunt and amplifier gain is such that the divider is not necessary between the amplifier output and comparator input, the total over-temperature errors will drop to 13.5% and 5.83%, respectively, the latter being essentially the same performance level as the INA301 alone.

Summary

Increasing the accuracy of your overcurrent detection implementation can improve system power efficiency by minimizing the allocated headroom. As with any electronic system, increasing precision typically requires an increase in system cost. Therefore, designing your overcurrent implementation requires trading off the low cost of typical discrete implementations with the increased precision offered by current-sense amplifiers and comparators with integrated references.

References

  1. Download the INA185 data sheet.
  2. Download the TLV4021 data sheet.
  3. Download the INA301 data sheet.

0 comments on “Optimizing your overcurrent detection to meet system requirements

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.