An engine ECU combines a controller, memory, sensor interfaces, power supplies and output drivers. The software interprets inputs and commands actuators, but the external electrical circuits and mechanical components must work as well. The arrangement below is conceptual: actual ECUs may integrate or distribute these functions differently. NXP: example port-injection ECU architecture
Sensors → input conditioning → controller and calibration → output drivers → actuators
↑ ↕
sensor supply vehicle communications
The diagram is an original conceptual illustration; it is not a wiring diagram.
Inputs and signal conversion
Some sensors provide a varying voltage, which is conditioned and sampled by an analogue-to-digital converter (ADC). Others provide switching or pulse signals that can be measured using digital inputs and timing hardware. The ECU applies a sensor-specific transfer function to interpret the signal. A believable number on a scan tool does not guarantee that the sensor or its scaling is correct.
Power, signal ground, reference voltages, shielding and input filtering are part of the measurement path. A wiring or supply fault can affect multiple readings. Use the relevant wiring diagram and test procedure rather than assuming that every input uses the same voltage or pin assignment. Texas Instruments: engine loads and ECU architecture
Separating the signal from the physical condition
A scan-tool value is the ECU's interpretation of an input, and may be filtered, calculated or substituted during a fault. Compare the reported channel, the electrical signal and an independent measurement of the physical quantity when the vehicle's test procedure calls for it. For example, a MAP reading may be wrong because of a sensor, supply, ground, blocked passage or real pressure problem; the MAP diagnostic guide separates these possibilities.
Some analogue sensors produce an output proportional to their supply voltage. HELLA describes this ratiometric behaviour for its angular-sensor example. A supply or sensor-ground problem can therefore affect the signal interpretation; check the actual circuit and reference specified for the vehicle. A clean-looking waveform alone does not establish calibration accuracy. HELLA makes that distinction for its ultrasonic oil-level sensor.
Identify the sensor and test method before connecting equipment. Two-wire active and passive speed sensors can look alike but require different tests; Pico warns that resistance testing can damage the active sensors it describes. See Pico: Hall-effect wheel-speed testing. Do not infer a universal pinout, voltage or resistance from the connector's appearance.
Processing and memory
The microcontroller executes the operating software and uses calibration values to interpret signals and calculate commands. Timers and other peripherals can help capture events and schedule outputs. The program, calibration and temporary working values have different roles, even when stored on the same chip.
- RAM is working memory; ordinary RAM loses its contents without its required supply.
- Non-volatile memory can retain program, calibration or other data through a power cycle. Examples include EPROM, EEPROM and flash, with different erase and programming methods.
- Learned values and fault records may be retained in non-volatile storage, but their location and reset behaviour are specific to the controller.
See automotive EPROMs and how an ECU is remapped for memory and calibration context.
Time, engine position and scheduling
The ECU must coordinate events with a moving engine, not merely calculate a fuel or ignition value. A crank sensor supplies position-related pulses; the controller validates the pattern, estimates speed and establishes synchronisation. A cam signal can identify the engine cycle where the chosen injection or ignition strategy requires it. See port injection modes and injector timing.
In one NXP small-engine reference design, software validates crank teeth and schedules fuel and spark events after synchronisation using timer hardware. A different NXP ECU example uses a dedicated timing unit that tracks crank position and executes commanded injector and ignition events. These illustrate possible architectures, not a claim that every ECU uses the same hardware. A loss of synchronisation, an incorrect trigger configuration or an output-driver fault can prevent the expected event even when a calibration table contains plausible values.
Outputs and feedback
The processor does not normally power an injector coil directly. Driver circuitry switches or regulates the current required by the load. Some drivers monitor their output and report electrical faults, but an electrical check does not by itself establish that fuel was delivered or that a mechanical valve moved correctly. Driver type, current limits and protective features depend on the hardware. Texas Instruments: automotive injector control
Control may be open-loop for one function and closed-loop for another. Feedback, such as measured exhaust oxygen, can help correct fuelling under the conditions where the relevant strategy is enabled. Timing, sensor delays, actuator limits and the quality of the calibration still affect the result. Consult mixture tuning and lambda for measurement limits.
When diagnosing a fault, establish the actual supply and ground, the input seen by the ECU, the commanded output and the circuit response before concluding that the controller itself has failed. Capture evidence under the conditions that reproduce the symptom.