ATMEGA328P Internal Temperature Sensor Bias
While testing on it’s own, without any peripherals, the ATmega328P does a good job in temperature measurements. The question that comes to mind is, “What happens when this chip starts to source current?”.
For my little project, I am looking at 15mA pulse for the radio Tx and an occasional Tx LED blink of 13.6mA.
Since there are three LEDs on the board already, 40mA should be enough to test with.
Using the same sketch as the Hidden Response Accuracy in the ATMEGA328P, I commented out the LM35 measurements and added a Blink without Delay for the (3) LEDs.
Tests included were:
- Measure every second and toggle the LEDs every 10 seconds
- Measure every second and toggle the LEDs every 2 minutes
- Measure every second with the LEDs always ON
While there is an obvious bias with the “Always On” chip heating, the reduced duty cycles can help dissipate some of the heat generated internally. I would imagine 100mA total sourced current will make this a problem for a project requiring <3°C accuracy.