- Thu May 07, 2015 2:56 pm
#16835
pedro wrote:As you’ve mentioned that we can replace this sensor with LM75, I guess that won’t be tough since both have the same mode of temperature sensing. However, since you’ve already used both, could you please let us know about the accuracy of these sensors? I was wondering if they can be accurate up to 0.1K. I’ve been using LM185, not satisfied though. The output of this sensor is sometimes noisy tough, specifically at higher temperatures.
LM185 is not an adjustable band-gap voltage reference diode?
If you stay under 100°C, you can have, in terms of accuracy:
- under ±0.35°C - MCP9808
- arround ±2/±2.2°C - LM75
Over 100°C (max temperature 125°C):
- ±1°C - MCP9808
- ±3°C - LM75
They are many factors than can influence badly reading process, from power supply to decoupling and PCB design. Overall MCP9808 is clearly better than LM75, and, off course, more expensive.
If you need even higher temperature measurement, one of the circuits suitable for high precision temperature measurement is a Wheatstone bridge with an NTC thermistor used as one bridge leg. Depending on your range you nedd to do some linearizing of the R/T characteristic and/or use lookup tables.
Another solution is a thermocouple based one. But in this case you need cold junction compensation. In goes up to +1000°C
What is the desired temperature range ?