AN607

Rev. 1.7

33

The Si7013 has the capability to apply a lookup-table-based non-linear correction to voltage measurements. This

correction is invoked by writing a “1” to bit 5 of user register 1. Note that humidity measurements should not be

performed when this bit is set. In the discussion below, “input” refers to the A/D voltage measurement result, which

is a 16-bit signed integer, and “output” refers the output after the non-linear correction, which is assumed to be a

16-bit unsigned integer.

The non-linear correction is based on a 10-point table lookup linearization. Each point consists of the ideal output

for a given expected A/D measurement result. Table 6 is stored in the Si7013 memory, which must also have the

slope at points 1–9. Slope is multiplied by a scaler of 256.

Only nine of the input/output pairs need to be in the table because the 10th pair is determined by the slope

equation. Overall, the Si7013 has 27 16-bit numbers in its table (54 bytes). This table is stored in non-volatile

memory of the Si7013 and must be programmed based on the desired look-up table.

The actual output is determined by extrapolation:

If in >in2, out = out1+slope1 x (in-in1)/256

Else if in >in3, out = out2+slope2 x (in-in2)/256

Else if in >in4, out = out3+slope3 x (in-in3)/256

Else if in >in5, out = out4+slope4 x (in-in4)/256

Else if in >in6, out = out5+slope5 x (in-in5)/256

Else if in >in7, out = out6+slope6 x (in-in6)/256

Else if in >in8, out = out7+slope7 x (in-in7)/256

Else if in >in9, out = out8+slope8 x (in-in8)/256

Else out = out9+slope9 x (in-in9)/256

slopeN

256

outputN

1

outputN

–

+



inputN

1

inputN

–

+



------------------------------------------------------------------



=