This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

My Topic: Digital Fuel Quantity Gauge - need advice (not yet red)

hello friends and keil men

I need your advice for starting a medium complexity project: a Digital Fuel Quantity Gauge.
As you know, most older cars have a very rudimentary analog gauge, showing full (F), empty (E) and a warning light for low on fuel. They measure the voltage send by a sender unit, which is a variable resistor attached to a float inside fuel tank.
NOTE: there are cars with senders having low resistance when tank is empty, and also other cars have the opposite.

My idea is this: design the simplest CORTEX circuit that is basically a non-linear digital voltmeter (due to irregular shape of fuel tank) and has the following specifications:
- 4 buttons operated: UP, DOWN, SET, MODE.
- calibration mode: specify up to 10 points of calibration
- measuring mode: continuously interpolate between calibration points
- gauge made using the smallest 7-segment LED displays (bi-color, 3 pieces showing 88.8 in liters)
- temperature will be changed (blue if cold, yellow if warm, red if cold)

In calibration mode the user should start with an empty fuel tank. He then pushes the MODE button to enter in calibration mode (indicated by flashing the comma point), selects 00.0 using UP/DOWN buttons, then pushes the SET button to memorize 1st calibration point.
In a similar manner, the user enters up to 10 points after pouring the respective fuel quantity in his fuel tank. At the end, he pushes the MODE button to switch to measuring mode.

As a finishing touch, the display digits should turn RED at a preselected low fuel quantity.

Please guide my approach, either by pointing ready made similar projects published in electronics magazines, or by giving me some tips from your experience.

Looking forward to hear from my friends.

  • Not sure about any need for similar projects.

    Every nontrivial project can be broken down into smaller pieces. And then suddenly most projects will be common to other existing projects.

    So one tiny block is the sensor.
    One tiny block is driving 7-segment digits.
    One tiny block is scanning buttons (with debounce).
    One tiny block is reading/writing for nonvolatile memory for handling calibration data.
    ...

    Suddenly you do not need to look for existing projects of similar type because it is actually irrelevant. Most things can be done directly from datasheets and manufacturer application notes.

  • hello Per Westermark

    You make it look very simple for me. I will read and come back with progress and many more questions.

  • You are assuming that the dial on the dash is the limiting factor on the accuracy. I somehow doubt that.

    "In calibration mode the user should start with an empty fuel tank"

    How will they ensure that the tank is truly empty?

    "He then pushes the MODE button to enter in calibration mode (indicated by flashing the comma point), selects 00.0 using UP/DOWN buttons, then pushes the SET button to memorize 1st calibration point.
    In a similar manner, the user enters up to 10 points after pouring the respective fuel quantity in his fuel tank. At the end, he pushes the MODE button to switch to measuring mode."

    This sounds totally impractical!

    "As a finishing touch, the display digits should turn RED at a preselected low fuel quantity"

    But you've already used the display colour to indicate temperature!

  • I assume he intents a RGB-LED for the temperature since the 7-segment digits was specified as bi-color red/green.

    A nicer way to calibrate is to have a flow sensor and have the system fill the tank full. The system then measures the tank sensor while integrating the amount of fuel passing the flow sensor. This would tell the system the non-linear behaviour of the tank sensor. When the user then fills the tank, you could optionally use this to calibrate the flow sensor. Rinse and repeat multiple times and the system would get better and better at it but remember that people do not always run the car dry before filling it up.

    Even if you never have the user tell your meter when and how much they fill, you would still be able to auto-learn relations between amount of fuel consumed through your flow sensor and expected measurements of the tank sensor totally without inclusion of any buttons. But your measurements but always be limited by the lowest tank state you have ever seen, since you can't guess the tank sensor reading for an empty tank until you finally measure with empty tank.

  • Most people make a very serious effort not to allow the car to run dry before filling up!

    It is a very bad idea to run a fuel-injection system (petrol or diesel) dry...