Car Battery Monitor

Yesterday the car would not start as the battery was flat, so I used a LiPo Jump Start pack to start it, this did not damage the logging device which continues to work flawlessly.

A short update to say this project has now been continuously live behind the front bumper of my car for 8 months now and has survived everything the weather and the cars electrical system has thrown at it.

I have been using the IoT ThingSpeak Monitor Widget to view the current state of the battery on the home screen of my Android phone. The widget obtains the latest battery voltage from my ThingSpeak channel. The widget also contains configurable monitoring functionality so that if the voltage falls below a threshold then an alarm occurs.

'Averaging of N-X ADC samples' technique does not seem to have made an observable difference. I have now move the voltage measurement to before connecting to the Particle Cloud to avoid ADC noise due to network activity.

Using averaging in firmware has made the readings less noisy (see graph after 19th Sept) , so I am going to switch to the 'Averaging of N-X ADC samples' technique described in the application note, to see if it results in a further improvement.

Found this App Note from ST Microelectronics about software techniques for reducing the impact of noise in ADC measurements, given how noisy the measurements appear, I am going to update my code to try the 'Averaging of N ADC samples' method.

How to improve ADC accuracy when using STM32F2xx and STM32F4xx microcontrollers