The TMP110 temperature registers use a 12-bit format. The 12 bits are aligned to the left side, or most significant side, of the 16-bit word. The four unused bits are on the right side, or least significant side. For this reason, a shift is needed to discard the extra bits. 2’s Complement is employed to describe negative temperatures. C code can easily convert the 2’s Complement data when the data is typecast into the correct signed data type. Q notation describes the number of bits which represent a fractional result. 4 bits of fractional data, known as Q4, offers 0.0625°C resolution.

/* 12-bit format will have 4 bits discarded by right shift
 q4 is 0.062500 resolution
 the following bytes represent 24.5C */
uint8_t byte1 = 0x18;
uint8_t byte2 = 0x80;
float f = (((int8_t) byte1 << 8 | byte2) >> 4) * 0.0625f;
int mC = (((int8_t) byte1 << 8 | byte2) >> 4) * 1000 >> 4;
int C = (int8_t) byte1;

Similarly, in extended mode, the temperature register is extended to a 13-bit format with the same resolution (Q4). This changes the range and effective bits but the resolution remains the same. For this reason, the bit shift also changes. Encoding and C code examples for extended mode are shown below.

/* 13-bit format will have 3 bits discarded by right shift
 q4 is 0.062500 resolution
 the following bytes represent 24.5C */
uint8_t byte1 = 0xC;
uint8_t byte2 = 0x40;
float f = (((int8_t) byte1 << 8 | byte2) >> 3) * 0.0625f;
int mC = (((int8_t) byte1 << 8 | byte2) >> 3) * 1000 >> 4;
int C = (((int8_t) byte1 << 8 | byte2) >> 3) >> 4;