Important changes to forums and questions

All forums and questions are now archived. To start a new conversation or read the latest updates go to forums.mbed.com.

Bug in course lesson 'Serial Communications with I2C'?

Topic last updated 11 May 2016, by Tom Gilbert. 0 replies
Tom Gilbert
11 May 2016

Bug in course lesson 'Serial Communications with I2C'?

I've noticed a problem with the sample code for the TMP102 temperature sensor which shows up in my dev environment (mbed-classic, nRF51-DK). The sensor returns temperatures as the first 12 bits of the two byte read, in twos compliment format.

Given:

float temp;
char temp_read[2];
tempsensor.read(addr, temp_read, 2);       //read the two-byte temp data
temp = 0.0625 * (((temp_read[0] << 8) + temp_read[1]) >> 4);  //convert data

The 'convert data' code doesn't properly handle negative temperature values. Looks like the compiler needs a hand with type inference, as this fixes it:

temp = 0.0625 * ((int16_t)((temp_read[0] << 8) + temp_read[1]) >> 4);

For example if I test with:

temp_read[0] = 0xE7;
temp_read[1] = 0x00;
int16_t sum = (((int16_t)(temp_read[0] << 8) + (int16_t)temp_read[1]) >> 4);
int16_t sum1 = (((temp_read[0] << 8) + (int16_t)temp_read[1]) >> 4);
pc.printf("sum: %i sum1: %i\r\n", sum, sum1);

I get: sum: -400 sum1: 3696