The linearity is not taken into account in the error calculus as it represents 0.03% of error only and it is negligible.

Nevertheless, as the gain error has been calculated thanks to the best fit line approach, it gives the information

that the gain error can be relatively constant throughout the linear input range of the TSC2011H.

Equation 10 has been described for a temperature of 25 °C. For sure with a temperature variation, Dvio/DT error

term must be added. And if the power supply is susceptible to change, the SVR parameter must also be taken

into account.

•

Example

Let us consider that the maximum total error can happen on the output of the TSC2011H.

•

Use case:

–

Vcc = 5 V

–

Vicm = 24 V

–

Vocm = 2.5 V

–

Temperature = 25 °C

–

Iload = 5 A

–

Shunt 5 mΩ with 1% accuracy

Theoretically the expected output voltage should be Vout = Rshunt * Iload *60 + Vocm = 4 V.

From the equations above, all the error terms are detailed by using the maximum value of the electrical

characteristics (when available), in order to express as much as possible, the worst case condition. The % error

on output of the following table is expressed in reference to Vout – Vref, so in this typical example: 1.5 V.

Table 6. Gain error

Error source

Calculus

Output voltage error

% error on

output

Gain error

60*5.10−3*5*0.3%

4.5 mV

0.3%

Vio error

60*500µV

30 mV

2%

CMRR error

60*24V−12V

109020

22.7 mV

1.5%

Vocm error

2.5*0.1%

2.5 mV

0.2%

Noise

60*29nVHz 10kHz*ln10k −ln0.1 +750kHz*π2−0.1Hz 1.98mVRMS 0.4%(1)

Total

60 mV

+1.98 mVRMS

4.4%

1. The percentage is based on voltage peak value, which is 3 times RMS value.

So the maximum output voltage in the worst case condition at ambient temperature is 4.060 V + 1.98 mVRMS

instead of 4 V expected. This represents an error on the current reading of about 4.4%. 1% more must be added

due to the shunt accuracy.

This calculus comes from all the maximum values and all the error terms which have been added to each other,

meaning that the chance of getting 4.4% precision in the use case above is extremely low and on the whole

population, the error is largely smaller.

TSC2010H, TSC2011H, TSC2012H

Application information

DS13791 - Rev 2

page 31/49