STMicroelectronics Current Sense Amplifiers Instruction Manual

STMicroelectronics Current Sense Amplifiers

overview

This document explains two approaches to estimating measurement error for a shunt-based current sensing system—an extreme conservative worst-case sum of maximum error sources, and a more realistic statistical RSS (root-sum-of-squares) approximation—using an example with the TSC213 current sensing amplifier.

Concrete parameters from the example

How the example sets up and performs error calculation

1. Define the measurement objective: bidirectional current from -4 A to 4 A with a 13.8 V battery using a shunt-based solution powered from 5 V.
2. Select the shunt resistor value to limit shunt power dissipation (10 mΩ used in the example).
3. Choose an amplifier gain (50 in the example) to exploit the full ADC output range.
4. Set the output common-mode level using an external 2.5 V reference so the TSC213 output sits centered in the ADC’s output range (mid-range common-mode).
5. Estimate errors using either:
6. Compute worst-case error by adding offset-type and percentage-type error contributors from current sense amplifier parameters and external components.
7. Compute a more realistic RSS error by combining the same error contributors through root-sum-of-squares instead of direct summation, under the statistical assumptions described.

User guide

This presentation provides a written explanation of the topic covered in the tenth chapter of the video: “Error calculation methodology #10”, which is available on our YouTube channel at the following link:

Recommendations

In this document you’ll find key concepts explained through informative tables, graphs, sequences of graphs, followed by a detailed explanation.

• When you see this icon: the topic is explained in detail on the following page(s)
• Each explanation page shows the page or page range where the topic is explained: (page # to #)
• If you see this icon: the explanation continues on the next page(s)

Typical application sensing amplifier

Typical application using the TSC213 current sensing amplifier

Explanation

• To illustrate this point, start by defining an application.
• The objective is to measure a bidirectional current delivered by a battery with a voltage of 13.8 V, ranging from −4 A to 4 A, using a shunt-based current sensing solution powered by 5 V. To limit power dissipation in the shunt, a value of 10 mΩ is selected.
• A current sensing amplifier with a gain of 50 is required to exploit the full scale of the output voltage. For precise measurement, special care is taken with the shunt, and the TSC213 precision current sensing amplifier is selected. The TSC213 operates with a
• common-mode voltage up to 26 V and offers a maximum input voltage offset (Vio) of 100 µV.
• We use an external 2.5 V voltage reference, such as the LM4040A, to set the output common-mode voltage in the middle of the output range.
• The TSC213 output voltage is then processed by a differential analog-to-digital converter (ADC).
• This example focuses on accuracy in the positive current measurement range, between 100 mA and 4 A. The analysis of accuracy in the negative current range, between −100 mA and −4 A, is identical.
• From a theoretical point of view, the expected output voltage of the TSC213 is between 2.55 V and 4.5 V
• Current sense amplifiers

Worst-case-error estimation

The simplest worst-case method is to add all error sources from the current sense amplifier and its external environment, such as the shunt and the voltage reference. Although very pessimistic, this approach helps estimate the maximum error. Here, the impact of each parameter is evaluated for a reference current of 2 A. As shown in the previous chapters, several error sources affect current measurement. We first consider the errors acting as an offset, which are the dominant contributors at low current. The main contributor is the input offset voltage and its temperature drift. At 2 A and 125∘C, the combination of ??? and ???? adds an error of 0.8% relative to the theoretical current measurement expressed by ??ℎ??? × ??????.

1. The simplest worst-case method is to add all error sources from the current sense amplifier and its external environment, such as the shunt and the voltage reference. Although very pessimistic, this approach helps estimate the maximum error. Here, the impact of each parameter is evaluated for a reference current of 2 A.
   As shown in the previous chapters, several error sources affect current measurement. We first consider the errors acting as an offset, which are the dominant contributors at low current. The main contributor is the input offset voltage and its temperature drift. At 2 A and 125∘C, the combination of ??? and ???? adds an error of 0.8% relative to the theoretical current measurement expressed by ??ℎ??? × ??????.
2. PSRR is evaluated in the same way as CMRR, but with respect to the supply voltage. In this example, ??? = 5 V, which is the same as the datasheet test condition. Its impact is therefore null and can be neglected.
3. The input bias current can also affect measurement accuracy in some cases. For the TSC213, however, the input bias currents are very small and more than three decades below the minimum current to be measured, so their impact is negligible.
4. Another offset-type contributor comes from the external shunt resistor: the thermal EMF. This very small voltage, in the ?V range, is generated by a temperature difference Δ??2? across the resistor. In this example, assumingabout 10∘C temperature difference between the two ends of the shunt, thermal EMF adds about 0.3% error to the measurement of a 2 A current at 125∘C.After offset-related terms, we must also consider the errors acting as a percentage, which generally have a stronger impact at high current.
5. The first and most important of these is the gain error. Since the gain is integrated inside the current sense amplifier, internal resistor matching must be as accurate as possible. Because gain error is already expressed as a percentage, its impact on the current measurement is straightforward to evaluate.
6. Linearity describes how constant the gain remains over the amplifier input range. In general, the linearity error is much smaller than the gain error and can often be neglected. In this example, its contribution is only 0.01%.
7. The shunt resistor tolerance must also be included, even though it is not correlated with the current sense amplifier itself, because it is part of the measurement chain. In this example, the selected shunt adds 1.5% error to the current measurement at 125∘C.
8. One last term can be added for the output common-mode voltage, considered here as an offset-type error. The TSC213 is used in bidirectional mode, with the output common-mode level fixed by an external ????. In this example, the ADC measures both ???? and ???? in differential mode, so the error linked to ???? is cancelled.
9. By summing all error contributions from the current sense amplifier and the external components, the total output error of the TSC213 reaches 3.65% for a measured current of 2 A at 125∘C.At first glance, the full error equation may look long and complex. In practice, it is simply a summary of the main contributors, added step by step.If the same calculation is repeated for each current point, the expected error can be plotted over the full current range. The plot shows that the percentage error is much higher at low current, mainly because of the offset-related terms.To reduce the percentage error at low current, there are two options:
   • increase the shunt resistor value, which raises the input voltage level and reduces the relative error;
   • choose a current sense amplifier with better parameters than those listed here.

Once again, this calculation method is highly pessimistic and should be considered an extreme worst-case estimation. As shown in this chapter, two methods can be used to estimate the error on a current sensing measurement.The first is a very conservative worst-case approach, which consists in summing all maximum parameter values from the current sense amplifier, the shunt, and the reference voltage datasheets. This method is highly pessimistic, but it provides large design margin and ensures that the estimated error should never be reached. In this example, the worst-case estimation is represented by the pink curve.The second method, which is more realistic, is the statistical RSS approach, represented by the dark blue curve. The graph compares both estimation methods at 25∘C for the example discussed in this chapter.You can also use ST’s eDesignSuite tool to estimate the error of a current sensing system using the RSS method. The tool allows you to select a current sense amplifier from the ST portfolio, enter the required measurement conditions and shunt characteristics, and automatically generate the corresponding error curve. eDesignSuite – High side current sensing Current sense amplifiers

© STMicroelectronics – All rights reserved.ST logo is a trademark or a registered trademark of STMicroelectronics International NV or its affiliates in the EU and/or other countries.© STMicroelectronics – All rights reserved. ST logo is a trademark or a registered trademark of STMicroelectronics International NV or its affiliates in the EU and/or other countries. www.st.com/m/traddemaemarks  For additional information about ST trademarks, please refer to www.st.com/trademarks.All other product or service names are the property of their respective owners.For additional information about ST trademarks, please refer to www.st.com/trademarks.All other product or service names are the property of their respective owners.