1. Pete Semig Applications Engineer-Precision Linear
Understanding Current Sensing Applications & How to Choose the Right Device
Pete Semig
Applications Engineer-Precision Linear

2. Outline & Presentation Goals
Fundamentals
Devices
Accuracy
PCB Layout & Troubleshooting
Goals of this presentation
Select a current sensing topology (direct vs. indirect, high vs. low-side, etc.)
Select the appropriate device (op amp, difference amp, inst. amp, or CSM?)
Calculate the accuracy of their solution
Troubleshoot a design (is the device out of specification?)
This presentation covers the fundamentals of current-sensing, current-sensing accuracy, PCB layout, and troubleshooting techniques. Fundamental topics include direct vs. indirect sensing, high vs. low-side sensing, and common-mode voltage. The devices that will be discussed include the operational amplifier, difference amplifier, instrumentation amplifier, and the current-shunt monitor. Current-sensing accuracy discusses specifications such as initial input offset voltage, shift, drift, shunt resistor tolerance, gain error, and gain error drift. Finally, PCB layout considerations and troubleshooting techniques are also discussed.
After listening to this presentation you should be able to select an appropriate current sensing topology & device, calculate the accuracy of the solution, and troubleshoot a design.

3. Fundamentals

4. Indirect Sensing Based on Ampere’s and Faraday’s Laws Noninvasive
Inherently isolated
No power lost by the system
Recommend
Very high currents (e.g. >100A)
Very high voltage
Very dynamic load currents
Isolation required
There are two types of current sensing: Indirect and direct. While the presentation focuses on direct sensing, a brief treatment of indirect sensing is in order.
Indirect current sensing is based on Ampere’s and Faraday’s laws. By placing a coil (for instance Rogowski coil) around a current-carrying conductor, a voltage is induced across the coil that is proportional to the current. This allows for a non-invasive measurement where the sensing circuitry is not electrically connected to the monitored system. Since there is no direct connection between the sensing circuitry and the system, the system is inherently isolated. Indirect current sensing typically is used for load currents in the 100A-1000A range. It is also used in applications that have a large common-mode voltage. This type of sensing, however, requires relatively expensive sensors and is not conducive to sensing currents on a PCB.
Rogowski coil-Ampere’s Law.
Air and ferromagnetic cores-Faraday’s Law.
Good paper by Tumanski (Google it).

5. Indirect Sensing-Solutions
Hall Effect Sensor
ADS1208, ADS1205, ADS7861, ADS8361, ADS8365
Application note SLAA286
One common sensor used for indirect current sensing is the hall-effect sensor. Such a sensor can be interfaced directly with an ADC as shown. This figure is from application note SLAA286 on

6. Indirect Sensing-Solutions
VAC Sensor
DRV401 is a custom solution for VAC Sensors
Combine with SAR ADC (e.g. ADS8509)
Less sensitive to temperature drift and initial offset than Hall sensor solutions
TI has developed the DRV401 to specifically interface with VAC sensors.
For more info, visit

7. Direct Sensing Based on Ohm’s Law Shunt resistor in series with load
Invasive
Power dissipated by Rshunt
Sensing circuit not isolated from system load
Recommend
Smaller currents (<100A)
System can tolerate power loss
Low voltage
Load current not very dynamic
Isolation not required (though it can be accomplished)
Direct current sensing is based on Ohm’s law. By placing a shunt resistor in series with the system load, a voltage is generated across the shunt resistor that is proportional to the system load current. The voltage across the shunt can be measured by differential amplifiers such as current shunt monitors (CSMs), operational amplifiers (op amps), difference amplifiers (DAs), or instrumentation amplifiers (IAs). This method is an invasive measurement of the current since the shunt resistor and sensing circuitry are electrically connected to the monitored system. Therefore, direct sensing typically is used when galvanic isolation is not required. The shunt resistor also dissipates power, which may not be desirable. Direct current sensing typically is implemented for load currents <100A.
Differential Amplifier

8. Differential Input Amplifier
Differential input amplifiers are devices/circuits that can input and amplify differential signals while suppressing common-mode signals
This includes operational amplifiers, instrumentation amplifiers, difference amplifiers, and current-shunt monitors
Instrumentation Amplifiers
Operational
Amplifier
A differential input amplifier is a device or circuit that amplifies a differential input and outputs a single-ended signal. Examples include the traditional op-amp, instrumentation amplifier, difference amplifier (notice the term ‘difference’ as opposed to ‘differential’), and current-shunt monitor. Op amps require negative feedback to be usable, otherwise the differential input signal is gained by the extremely large open-loop gain. As a building block, though, the op amp itself meets our definition of a differential amplifier.
Difference
Amplifier
CSM

9. Common-Mode Voltage
Input common-mode voltage is the most important specification when selecting a direct current sensing solution. It is defined as the average voltage present at the input terminals of the amplifier.
This specification is important because it limits our choice of differential amplifiers. For example, op amps and IAs require an input common-mode voltage within their power supplies. Difference amplifiers and CSMs, however, typically can accommodate input common-mode voltages in excess of their power supplies. This is useful in applications where the amplifier senses the shunt voltage in the presence of a large common-mode voltage and must interface with a low-voltage analog-to-digital converter (ADC). In such a scenario the amplifier and ADC can be powered with the same supply voltage regardless of the system’s common-mode voltage.
The first figure is the traditional definition of Vcm. The second figure separates the input signal into two parts: the common-mode voltage and the differential voltage. An ideal differential amplifier only amplifies the differential signal and rejects the common-mode signal.

10. Low-Side Sensing
Shunt resistor placed between the system load and ground
Pros
Vcm≈0V
Straightforward
Inexpensive (may use an op amp)
Cons
Can’t detect load shorts to ground
Single-ended measurement (stay tuned)
System GND is now Iload*Rshunt
When monitoring load current the designer can choose to place the sense resistor either between the supply voltage (Vbus) and load, or between the load and ground. The former is called high-side sensing whereas the latter is called low-side sensing.
Low-side sensing is desirable because the common-mode voltage is near ground, which allows for the use of single-supply, rail-to-rail input/output op amps. The drawbacks to low-side sensing are disturbances to the system load’s ground potential and the inability to detect load shorts to ground.

11. High-Side Sensing
Shunt resistor placed between supply (Vbus) and system load
Pros
Can detect load shorts to ground
Monitors current directly from source
Cons
Vcm≈Vbus
High-side sensing is desirable in that it directly monitors the current delivered by the supply, which allows for the detection of load shorts to ground. The challenge is that the amplifier’s input common-mode voltage range must include the load’s supply voltage, or Vbus. This requirement frequently necessitates the use of DAs or dedicated CSMs, which allow for common-mode voltages outside their voltage supply range.

12. Directionality
Application may require sensing current in both directions
Depending on the application, the supply current may flow either in one direction (unidirectional) or both directions (bidirectional). Unidirectional designs are straightforward in that the output voltage of the amplifier does not need to distinguish direction. If a bidirectional design is required, however, how can the designer distinguish direction of current flow? Bidirectional solutions have an input for a reference, or pedestal voltage. The output of the device is then referenced to this known voltage. Output voltages above the known reference voltage are in one direction while output voltages below the known reference voltage are in the opposite direction.

13. Output
Op amps, difference amplifiers, and instrumentation amplifiers have voltage outputs
Current shunt monitors can also have current-output and digital- output
The output of a current sensing solution can be in the form of a current, voltage, or digital word. Dedicated CSMs can have any of the aforementioned outputs while op amps, DAs, and IAs have a voltage output.
Current output CSMs typically require an external gain-setting resistor (RL) as shown by the INA138/168 diagram. This allows for flexibility in choosing the gain of the device. Also, if the design requires driving a large capacitive load, a properly compensated external op amp can buffer the output of the CSM. However, since the gain of the device depends on the external resistor, the internal resistors must be trimmed to an absolute value which can increase cost.
Voltage output CSMs integrate the gain-setting resistor and output buffer as shown by the INA193-8 diagram. This only requires ratio-metric trimming which is less costly than absolute trimming. The drawback is that the devices have predetermined gains (for instance 20V/V, 50V/V, 100V/V) and the designer does not have access to the feedback path of the buffer amplifier to compensate for large capacitive loads.
Digital output CSMs have integrated programmable-gain amplifiers (PGAs) and ADCs and report the shunt voltage as a digital word via a serial communication protocol such as I2C or SPI, as shown by the INA219 block diagram. While this can be highly desirable for ease-of-use, the full-scale differential input must be restricted to a range of values. Therefore, care must be taken when selecting the value of the shunt resistor.

14. Devices

15. Op Amp-Low Side Only Very large open-loop gain
Must have feedback, so the input will be single-ended
Therefore only useful on low-side where Vcm≈GND
Beware of PCB layout parasitics
Ideal
Actual
The use of an op amp for current sensing is limited by the input common-mode voltage. Due to the design of the input stage, the input common-mode voltage of such a device is limited by the supply voltage (Voa). Additionally, the large open-loop gain of a traditional op amp requires the device to have feedback, which limits its use to single-ended input signals. Such a configuration necessitates its use to only low-side current sensing.
Since the input common-mode voltage of the solution shown above is near ground, the op amp input common-mode range should include ground. It may also be desirable to select an op amp whose output is considered rail-to-rail. This yields the greatest range of load currents that can be accurately sensed. One drawback is any parasitic resistance between the shunt resistor and the ground trace adds to the shunt resistor value.
The voltage developed by a parasitic resistance in this location (for example, PCB trace, solder joint) will ‘pedestal’ the shunt voltage, thereby introducing error. This parasitic resistance may vary greatly in production. For greater accuracy and consistency, a differential measurement across the shunt resistor is required.
Parasitic impedance to ground introduces error

16. Difference Amplifier
Can be used for either high or low- side current sensing
Can tolerate very large common- mode voltages (e.g. INA149 Vcm=±275V!)
This is due to large resistive divider on input pins
However this resistive network loads the system
Ensure system impedance is significantly smaller than DA input impedances
INA149
A traditional DA is simply an op amp with a precision trimmed resistor network as shown. The resistors are typically trimmed during manufacturing so that R2/R1 = R4/R3. The differential gain (Adm) of the device is therefore R2/R1. The reference voltage (Vref) is added to the output voltage (Vo).
Notice that a difference amplifier is seen as a load by the system per the differential and common-mode input impedances.

17. Difference Amplifier Solves PCB parasitic issue
With respect to current sensing, a DA can have input common-mode voltages outside of the supply voltages due to the resistive divider at the inputs. This allows one to use a DA for high-side current sensing. However, a DA places a load on the system bus voltage due to its finite common-mode and differential-mode input impedances. This load draws current from the system bus voltage, which introduces uncertainty in the measurement. In order to reduce the measurement error due to these input impedances, they should be significantly larger than the system load impedance.
Since the common-mode voltage of a low-side current sensing solution is near 0V, a DA can be used for such measurements. This minimizes the effect of the common-mode input impedance, but the differential-mode input impedance is still a factor. The use of a DA for low-side measurements negates the issue caused by parasitic resistance to ground in series with the shunt resistance that was discussed in the op amp section. Finally, DAs have fixed differential gains because the resistor network must be trimmed to maintain good common-mode rejection ratio. Some DAs have an on-chip non-inverting amplifier whose gain can be adjusted. If other gains are required it is advisable to either select such a device or add gain to the output of the DA with an external op amp circuit.
Solves PCB parasitic issue

18. Instrumentation Amplifier
A three op amp IA is made of a DA with buffered inputs
Pros
Large input impedance
Change gain with external resistor
Cons
Common-mode voltage must remain within supply voltage
Usually used for low-side sensing, but can be used for high-side depending on common-mode voltage
Instrumentation amplifiers are typically composed of a DA output stage with buffered inputs. The first advantage over a DA is the ability to easily change the differential gain (Adm) of the device using an external resistor (Rg). Secondly, the inputs are connected to the non-inverting inputs of a buffer amplifier. The non-inverting inputs are high-impedance, which translates to almost no load on the system bus voltage allowing for the sensing of small system currents.

19. Instrumentation Amplifier
Vbus must be within Vcm range given IA supply voltages
Differential measurement solves PCB parasitic issue
One disadvantage, however, is that the input common-mode voltage of IAs is limited by their supply voltage. Therefore, IAs typically are utilized for low-side measurements.
It is possible to utilize an IA in a high-side measurement. The designer must ensure that the system bus voltage is within the input common-mode range of the IA as dictated by its supply voltage.

20. Current Shunt Monitors
Unique input stage topologies (e.g. common-base)
This allows for Vcm values outside of supply voltages AND very large input impedances
But, maximum Vcm value is currently 80V
Typical Op Amp Input Stage
Current shunt monitors are devices that place little load on a system and allow for sensing current under high common-mode voltage conditions. This ability is gained through the design of unique input stages.
Example CSM Input Stage

21. Input and Output Voltage Range
Always ensure that desired input and output voltage ranges are within datasheet specifications
Instrumentation amplifiers can be tricky
For example, can we effectively use the INA826 for a low-side current measurement (Vcm≈0V)?
Care must be taken to ensure than the input and output voltage ranges of all devices mentioned in this article are within the datasheet specifications. The ranges are listed in the “recommended operation conditions” table of their datasheets. Violating the input range of a device, at least, ensures nonlinear operation. Exposing a device’s inputs to voltages outside of datasheet recommendations can also damage the device depending on duration and severity. The output range of a device is directly dependent on the supply voltages. One must ensure that the product of the input voltage and the gain of the device/circuit is within the output range specification of the device.
This is especially important when selecting the supply voltage for IA solutions. Traditional three op amp Ias typically have a plot in the datasheet that shows the output voltage range versus input common-mode range for different supply conditions.
If the INA826 is used for a low-side measurement (Vcm ≈ 0V), the output voltage range of the device extends only from ~0.1V to ~1.2V when Vs = 5V (single-supply), Vref = 0V, and G = 100. This range may be too restrictive to utilize the full input range of a typical ADC.
Powering the device with dual supplies, however, allows for a larger output voltage range. Vs = ±5V allows the output to swing from –4.9V to +4.85V when Vcm=0V for gains of 1 or 100. However, depending on the system and application, this may require the addition of a –5V power supply.

22. Unique Topologies
Some instrumentation amplifiers have unique topologies that allow for RRIO on a single supply!
INA326
Input voltage range (V-)-0.02 to (V+)+0.1
Output swings to within 75mV over temperature! (5mV typical)
Possible drawback: 1kHz BW
Nontraditional IAs such as the INA326 utilize a unique topology that enables true rail-to-rail input and output voltage ranges. With a single-supply voltage of 5V, the INA326’s input voltage range is –20 mV to 5.1V. The output voltage range can swing to within 75 mV of either supply rail. This makes such devices attractive for low-side current measurements.

23. Summary Vcm≈0V 0V<Vcm<Vdevice Vcm>Vdevice Small Load Current
Op Amp IA
CSM
Large Load Current DA
Observations to make:
-Op amps are really best suited for low-side measurements. They are not listed in the 2nd column for they need negative feedback and therefore amplify a single-ended signal.
-Difference amps (due to their input impedance) are well suited for large load currents
-CSMs can be used anywhere

24. Accuracy

25. Accuracy Definition Error Sources Worst case
Probable (Root-sum-square)
The system accuracy of a current shunt measurement is impacted by a plethora of error sources including those shown in the table. Worst-case system accuracy is defined as shown in the first equation where ‘en’ is the error contribution (%) of each error source. A more probable approach to calculating system accuracy, however, is to combine uncorrelated errors in a root-sum-square (RSS) fashion as shown in the 2nd equation.
Since most of the errors listed in the table are referred-to-input (RTI), it is advantageous to discuss accuracy with respect to the input. Errors that are referred to the input of the device can be multiplied by the device gain to determine their contribution at the output.

26. Input Offset Voltage (Vos)
Definition
“The DC voltage that must be applied between the input terminals to force the quiescent DC output voltage to zero or some other level, if specified.”
Typically the largest source of error
Error is calculated with respect to the ideal voltage across the shunt resistor (Vshunt)
Example (INA170)
Vos(max)=1mV
Iload=5A (nominal)
Rshunt=1mΩ
How do we decrease this error?
Decrease Vos(max)
Increase Vshunt
Input offset voltage is typically the largest factor affecting a solution’s accuracy. It is defined as “the DC voltage that must be applied between the input terminals to force the quiescent DC output voltage to zero or some other level, if specified.” [1] An amplifier’s ideal Vos is 0V. Process variations and device design constraints, however, cause non-zero values of Vos.
All input-referred errors are calculated with respect to the ideal shunt voltage. The ideal shunt voltage is the product of the load current and ideal shunt resistor value. Let’s calculate the error contribution of a device’s Vos specification in a system whose nominal load current is 5A and ideal shunt resistor value is 1mΩ, shown in the equation. Assume that we have decided to use the INA170, whose maximum Vos specification is 1mV.
In order to decrease this error, we have two options: increase the Rshunt resistance or decrease Vos(max). Increasing the Rshunt resistance may or may not be feasible due to cost, board space, or power dissipation. Alternately, we could try to find a substitute device with lower Vos.
Finally, note the inverse relationship between load current and error. In our example, we calculated 20 percent with a nominal load current of 5A. If the system load current decreases, the contributed error due to the Vos specification increases. Therefore, a designer should calculate worst-case error at minimum load current.

27. Common-mode Rejection Ratio (CMRR)
Measure of a device’s ability to reject common-mode signals
Otherwise the common-mode signals can manifest themselves into differential signals which add to the error
Specified one of two ways
Linear: μV/V
Worst case is the maximum value
Logarithmic: dB
Worst case is the minimum value
Need the following to calculate error
Worst case specification
Common-mode test condition from datasheet (Vcm-pds)
System common-mode voltage (Vcm-sys)
CMRR can affect a current sensing solution’s accuracy. It is a measure of a device’s ability to reject common-mode signals. This is important because common-mode signals can manifest themselves inside a device as differential signals, thereby decreasing the accuracy of a solution.
CMRR is typically specified in product datasheets on either a linear scale (μV/V) or a logarithmic scale (dB). If reported in dB, the worst-case value is the minimum value. If reported in μV/V, the worst-case value is the maximum value.
In order to calculate the error due to a device’s CMRR specification, we need the following: worst-case CMRR specification from the datasheet, common-mode voltage test condition from the datasheet specification table (Vcm-pds), and the system’s common-mode voltage (Vcm-sys).

28. How to we minimize this error?
CMRR Example
Vcm-sys=50V
Device: INA170
Rshunt=1mΩ
Iload=5A (nominal)
First, convert 100dB to linear scale
Now calculate error contribution
How to we minimize this error?
For example, assume we have a system whose common-mode voltage is 50V (Vcm-sys) and shunt voltage is nominally 5mV. Let’s calculate the error using the INA170, whose worst-case CMRR specification is 100dB (min) and Vcm-pds=12V.
In order to decrease the error contribution due to CMRR we have two options: increase the shunt voltage or select a device with better CMRR performance. Changing Vcm-sys is not usually a realistic option for it is dictated by the application.

29. Power Supply Rejection Ratio (PSRR)
Measure of the change in Vos created by a change in power supply voltage
Calculation very similar to CMRR
Need the following to calculate error
Worst case specification
Supply voltage test condition from datasheet (Vs-pds)
Device’s supply voltage (Vs-sys)
PSRR is a measure of the change in Vos created by a change in power supply voltage. Error due to PSRR can be calculated in a manner similar to CMRR.
In order to calculate the error due to a device’s PSRR specification, we need the following: worst-case PSRR specification from the datasheet, supply voltage test condition from the datasheet specification table (Vs-pds), and the supply voltage that will power the device in the system (Vs-sys).

30. PSRR Example Vs-sys=30V Device: INA170 Rshunt=1mΩ Iload=5A (nominal)
PSRR already specified in linear scale
How do we reduce this error term?
For example, the INA170 has a worst-case PSRR specification of 10μV/V(max) with Vs-pds=5V. If the device is actually supplied with 30V (Vs-sys), the error due to PSRR is calculated as shown. As with the previous examples we assume a shunt voltage of 5mV.
In order to decrease the error contribution due to PSRR we have two options: increase the shunt voltage or select a device with better PSRR performance. Changing Vs-sys is not usually a realistic option for it is typically dictated by the application.

31. Other Errors Gain error Shunt resistor tolerance Vos drift Vos shift
Usually listed in datasheet as a percentage
Shunt resistor tolerance
Usually given as a percentage. Beware of temperature drift, though.
Vos drift
Usually listed in datasheet as μV/˚C
Multiply specification by maximum deviation from 25C
Vos shift
Change in Vos due to time
Usually not specified. Most likely never guaranteed.
Rule of thumb: In 10 years, Vos may shift no more than the Vos(max) specification.
This is in addition to the device’s initial Vos specification.
Some specifications, such as gain error and shunt resistor tolerance, are typically given in percentage form. This makes their incorporation into an accuracy calculation straightforward.
The table from a few slides ago also lists Vos drift and shift. Input offset voltage drift measures the change in Vos due to a change in temperature. This specification is typically reported as uV/C . Multiplying this specification by the largest expected temperature deviation from room temperature (25˚C) will yield the error term.
Input offset voltage shift, however, is not as straightforward. Input offset voltage shift measures the change in Vos due to a change in time. This specification is not usually found in a datasheet and can only be estimated. One way to approximate shift is to understand that a device’s Vos can shift by an amount no greater than the device’s maximum Vos specification over a 10-year period. This shift is in addition to the device’s initial Vos specification.

32. Putting It All Together
Let’s calculate the error of our example due to Vos, CMRR, and PSRR
20% error due to Vos (initial)
7.6% error due to CMRR
5% error due to PSRR
Simply adding each term yields the worst-case error while combining them in a RSS fashion yields a more probable scenario.
For the Vos, CMRR, and PSRR examples given in this article the worst-case accuracy would be 67.4% whereas combining the errors in a RSS fashion yields a more probable accuracy of 78.03%.
Note: This is for the nominal load current. What if 2A<Iload<6A?

33. PCB Layout and Troubleshooting

34. Accuracy Revisited System-Level Errors revisited
Notice these are all specifications
Are they under the control of the designer?
New
Designer Control
Input Bias Current (Ib)
Vos-pcb
Ib is defined as the average of the currents into the two input terminals with the output at the specified level.
Before introducing the effects of PCB layout on accuracy, the input bias current specification (Ib) must be understood. Input bias current is defined as “The average of the currents into the two input terminals with the output at the specified level.”[1] Ideally this specification is 0A, but process variations and design constraints cause this to be a non-zero value. With respect to current sensing, this specification has two implications. First, Ib becomes important when trying to accurately sense ‘small’ currents. For example, attempting to measure a system load current of 100uA with a device whose Ib=35uA would yield great uncertainty in the measurement. Secondly, poor matching between the two input signal currents (or input offset current, Ios) could induce and/or exacerbate a differential voltage at the inputs of the device.
All of the sources of error are device specification (even shunt resistor tolerance). In other words they can be improved simply by choosing a different device. Input bias current along with imbalanced PCB input traces can create a differential signal at the input of the device. The PCB layout, however, is directly under the control of the designer. Poor PCB layout can dramatically affect the accuracy of a solution.

35. Example Design Requirements Can’t use an op amp due to Vcm
Common-mode voltage (Vcm): 70V
Available supply for differential amplifier (Vs): 5V
Load current range: 5A<Iload<10A
Unidirectional
High accuracy
Can’t use an op amp due to Vcm
Can’t use an IA due to Vcm
High accuracy->look at CSMs
No mention of bandwidth->assume DC measurement
INA282
-14V<Vcm<80V
Vos=70μV
2.7V<Vs<18V
Bi-directional
Since the common-mode voltage is considerable and greater than the supply voltage available for the differential amplifier, a current shunt monitor (CSM) should be the device of choice.

36. Sizing the Shunt Resistor
We need the following to size the shunt resistor
Load current range
Available supply voltage
For this example
5A<Iload<10A
Vs=5V
Strategy
Obtain output voltage range from datasheet
Refer it back to the input by dividing by the device’s gain
Use load range minimum and maximum to find range of acceptable shunt resistor values

37. Sizing the Shunt Resistor
Input and Output Range-INA282
Output Range: GND+40mV<Vout<Vs-0.4V
Since Vs=5V, 40mV<Vout<4.6V
Refer to input by dividing by gain (INA282 Gain=50V/V)
Input Range: 800μV<Vin<92mV
Calculate maximum and minimum values for the shunt resistance
Note: Rshunt value is directly related to both power and accuracy
So, use the largest Rshunt that your system can tolerate
In order to properly size the shunt resistor we must examine the input and output specifications of the device. With a supply voltage of 5V, the output of the INA282 can swing from 40mV to 4.6V. Dividing the output range by the gain of the device (50V/V) yields the input range. This equates to an input voltage range from 0.8mV to 92mV.
Given our load current of 5A (Iload-min) to 10A (Iload-max), the shunt resistor must be smaller than 9.2mΩ (92mV/Iload-max) and larger than 0.16mΩ (0.8mV/Iload-min). If the power dissipation of the shunt resistor is tolerable, it is recommended to use a larger valued shunt resistor for accuracy purposes. Therefore, we will select a shunt resistor of 8mΩ.

38. Example Simulation

39. PCB Parasitics
There are 3 parasitic resistances that can affect the accuracy
In series with the shunt (load current)
In series with the device inputs (input bias current)

40. PCB Parasitics Ideally Vshunt=Vsense
Vsense is the voltage at the pins of the device
However, the parasitics cause Vshunt≠Vsense, so we will rename the voltage at the pins to V’sense
Be sure to stress the difference between Vshunt and Vsense. Vsense includes errors induced by the layout, and it’s what the device will actually amplify. That is why it’s important to always measure the input voltage at the pins of the device (not across the shunt).
Ideal Shunt Voltage

41. PCB Parasitics If we solve for V’sense we get
Input bias currents may or may not induce an error
If we solve for V’sense we get
ANY parasitic resistance in series with the shunt resistor will induce an error
The reason why “Ib may or may not induce an error” is because the term is a subtraction. Theoretically they could equal each other and cancel. But, any resistance in series with the shunt will add to the error.

42. Balanced Rpn and Rpp Error
One-ounce copper input traces of 350mils in length and 10mils wide have an impedance of approximately 17mΩ.
This slide and the next 2 may be skipped if you’re running out of time. The errors due to Rpp and Rpn are found to be minimal in comparison to Rps.

43. Unbalanced Rpn and Rpp Error
Assume Rpp=350mils and Rpn=275mils (13mΩ)

44. Rpn and Rpp Error Observations
These errors are generally not significant
Placing the device and shunt on opposite sides of the board can exacerbate the error due to vias and temperature drift
Attempting to size the input traces such that Ibp*Rpp=Ibn*Rpn is not recommended
Ib can vary part-to-part, wafer-to-wafer, and lot-to-lot.
It can also vary based on system parameters such as common-mode voltage and power supply
PCB Layout Recommendations
Make the input traces as short as possible
Make the input traces as balanced as possible
Place the current sensing device and shunt on the same side of the PCB

45. Rps Error ANY resistance in series with the shunt will induce an error
One-ounce copper trace of 10mils in length and 25mils wide has an impedance of approximately 193μΩ.
Recall that Rps is in the high-current path
2.41% error due to 193μΩ!
How did we get resistance in series with Rshunt? Rshunt not Kelvin-connected!

46. Rps Error Example

47. Good! Kelvin-connection
PCB Layout Summary
Bad
Good! Kelvin-connection

48. Rps Error
Many engineers verify this circuit by comparing the output (4.084V) to the product of the shunt voltage (80mV) and the device’s gain (50V/V)
This shows 84mV of error, which is unacceptable for the INA282.
Conclusion: The device’s gain is incorrect…let’s do a failure analysis!
The conclusion is meant to be sarcastic. This goes back to the difference between Vshunt and Vsense.

49. Not the device…no FA required.
Rps Error
Not so fast!
To determine error introduced by the device you must use V’sense
The whole point is that in order to determine the error introduced by the device itself, you must take measurements at the PINS of the device.
Not the device…no FA required.

50. PCB Layout & Troubleshooting Summary
Always ensure that the shunt resistor is Kelvin-connected
Make the input traces as short as possible
Make the input traces as balanced as possible
Place the current sensing device and shunt on the same side of the PCB
To determine device error, look at voltage across device pins (Vsense)
Check to see if device is within specification for given application
Quickly calculate error referred to output due to Vos, CMRR, and PSRR

51. Questions?

52. Additional Reading Current Sensing Fundamentals
Current Sensing Devices
Current Sensing Accuracy
Current Sensing Layout & Troubleshooting

53. Devices
For samples of the devices mentioned in this presentation, please follow the appropriate link
INA170, CSM, voltage output, 2.7V<Vcm<60V, Vos=1mV(max)
INA282, CSM, voltage output, -14V<Vcm<80V, Vos=70uV(max)
INA326, IA, RRI/O, Vos=100uV(max)
INA826, IA, RRO, Iq=200uA(typ), 2.7V<Vsupply<36V
INA149, DA, Vcm=+/-275V, unity gain
INA219, CSM, digital, 0V<Vcm<26V, Vos=50uV(max)
INA138/168, CSM, current output, 2.7V<Vcm<36V/60V, Vos=1mV(max)
INA193, CSM, voltage output, -16V<Vcm<80V, Vos=2mV(max)

54. Backup

55. Load Short