@OP
Let us get clarified the meaning of 'Resolution' in the context of DS18B20 Temperature Sensor.
1. There is a programmable ADC (analog-to-digital converter) inside the DS18B20 sesnor. We can command the sensor to take a sample of 'input (how warm it is) signal' and convert it into 9-bit or 10-bit of 11-bit or 12-bit which is later on presented in 0C unit.
2. The following diagram (taken from the data sheets of the sensor) tells about the bit pattern of the ADC value.
Figure-1:
**(1)** The Fig-1 says that the temperature value has two parts -- integer part and the fractional part. This is to say that the temperature is a floating point number (simply float). In Fig-1, the decimal point (the . symbol) is not shown/included; but, it is understood and it is located just after this: 2<sup>0</sup> and before this: 2<sup>-1</sup>. There is also a sign (S) bit which (when assumes Logic-High state) represents -ve temperature. (The float value is encoded in 2's complement form and not in binary32 format.)
**(2)** In Fig-1, we observe that there are upto 4 digits that we can include after the decimal point and these are: 2<sup>-1</sup>, 2<sup>-2</sup>, 2<sup>-3</sup>, and 2<sup>-4</sup> with respective positional values of: 0.5<sup>0</sup>C, 0.25<sup>0</sup>C, 0.125<sup>0</sup>C, and 0.0625<sup>0</sup>C.
**(3)** Now, the meanings of setting up 'Resolution' to 9-bit, 10-bit, 11-bit, and 12-bit' stand as:
**(a)** 9-bit:
1-bit for sign, 7-bit for integer part, and 1-bit for fractional part (1-digit after decimal point)
Temperature range: xxx.0<sup>0</sup>C to xxx.5<sup>0</sup>C in 0.5<sup>0</sup>C discrete step.
The command to present this float temperature (dsTemp) of 1-digit precision (number of digit after the decimal point) when the ADC Resolution is 9-bit.
Serial.print(temp, 1); //temp must be in float; 2nd argument 1 refers 1-digit precision
(Let us observe that there is subtle difference between precision and resolution.)
**(b)** 10-bit:
1-bit for sign, 7-bit for integer part, and 2-bit for fractional part (2-digit after decimal point)
Temperature range: xxx.00<sup>0</sup>C to xxx.75<sup>0</sup>C in 0.25<sup>0</sup>C discrete step.
The command to present this float temp:
Serial.print(dsTemp, 2);
**(c)** 11-bit:
1-bit for sign, 7-bit for integer part, and 3-bit for fractional part (3-digit after decimal point)
Temperature range: xxx.000<sup>0</sup>C to xxx.875<sup>0</sup>C in 0.125<sup>0</sup>C discrete step.
The command to present this float temp:
Serial.print(dsTemp, 3);
**(d)** 12-bit:
1-bit for sign, 7-bit for integer part, and 4-bit for fractional part (4-digit after decimal point)
Temperature range: xxx.0000<sup>0</sup>C to xxx.9375<sup>0</sup>C in 0.0625<sup>0</sup>C discrete step.
The command to present this float temp:
Serial.print(dsTemp, 4);
Hope that the issue is clear!
**The Test Codes:** (tested in UNO)
#include <OneWire.h>
#include <DallasTemperature.h>
#define ONE_WIRE_BUS 2 //DPin-2 of UNO at signal terminal
OneWire oneWire(ONE_WIRE_BUS);
DallasTemperature sensors(&oneWire);
float dsTemp;
void setup(void)
{
Serial.begin(9600);
sensors.begin(); //initialize One-Wire Sensor -- DS18B20
}
void loop(void)
{
sensors.setResolution(9); //before each measurement, set resolution
//sensors.setResolution(10);
//sensors.setResolution(11);
//sensors.setResolution(12);
sensors.requestTemperatures(); // Temp conversion command; waiting here until comversion is done
dsTemp = sensors.getTempCByIndex(0); //read temp data from Sensor #0 and convert to celsius float
Serial.println(dsTemp, 1); //1-digit precision
//Serial.println(dsTemp, 2); //2-digit precision
//Serial.println(dsTemp, 3); //3-digit precision
//Serial.println(dsTemp, 4); //4-digit precision
delay(1000); //sample temperature at 1-sec interval
}