A microcontroller is a very powerful tool for any embedded engineer, but it's also a very sensitive component. A signal as small as a 4.0V signal is already reaching the maximum threshold of a common 3.3V based microcontroller, and anything greater risks permanent damage to the device.
Galvanic Isolation
The first step to creating a reliable circuit is to identify any potential signals that may introduce excessive voltage spikes and/or signaling levels greater than 3.3V. As you may have guessed by the title, the ignition signal needs to be stepped down and transients (voltage spikes) need to be isolated from the microcontroller.
From the acquired measurements we see a voltage of around 16V, way higher than the 3.3V. This voltage will immediately destroy the microcontroller.
The best way to step down the voltage is to use a transistor. Configured correctly a transistor can act as a robust level translator and provide the necessary operating voltage for a microcontroller. I chose to start with an optocoupler, which is essentially a NPN BJT transistor that is activated by an LED. This device provides true galvanic isolation, guaranteeing that no energy spikes from the RPM signal can propagate to the microcontroller.
Switching Speeds
I chose the Vishay VOMA618A-2X001T, an AEQ-Q101 rated optocoupler. The first parameter to check when picking an optocoupler is to verify that it can handle the switching speeds we expect to acquire. From the oscilloscope capture, it is observed that the positive pulses do not change based on Engine RPM. Therefore the positive pulse will always be around 1.6ms as seen below.
Next, we need to check the shortest duration of the negative pulse. This pulse will vary based on the engine RPM. Therefore we can use our engine RPM formula determined in the previous step to determine the shortest time between pulses:
Engine RPM (max) = 30/Δt
8500 RPM = 30/Δt
Δt = 3.53ms
Now we know that the shortest pulse is 1.6ms and 3.5ms for the positive and negative pulses respectively. This means the optocoupler must be able to handle at least 1.6ms pulses. By referencing the optocoupler's datasheet we can see that the optocoupler is rated at 4.9us turn-on time and 3.3us turn off time, so with a worst-case scenario, the turn-on time is 0.31% of the pulse (1.6ms). This is more than enough margin to allow for a well-defined square pulse to reach the microcontroller through the isolated signal path.
Voltage/Current Protection
The Vishay VOMA618A-2X001T optocoupler is rated at a max forward current of 50mA, therefore the circuit must guarantee that this limit is never reached.
We know that the amplitude of the input signal is around 13.0V and could easily reach 17V with overshoot and/or based on the vehicle's alternator output. To create a known absolute max voltage, a diode is added to clamp the voltage when it gets above the determined limits. The chosen diode for this circuit is the SMF18A-HM3-08. This diode has a breakdown voltage of 20V, meaning the diode will sink the voltage if the IG- signal exceeds expected values. This will protect the diode in the optocoupler from being over-driven, which could cause device failure or reduce it's overall life.
Now we need to determine the normal operating current, we know that the average input voltage is 13.0V, the optocoupler has a typical forward voltage of 1.28V and there is a total of 2kΩ impedance.
Iavg = (Vin - Vf) / R
Iavg = (13.0V - 1.28V) / 2000Ω
Iavg = 5.86mA
Next, the overvoltage scenario must be calculated to make sure the 50mA limit does not get reached. We substitute the average input voltage with the absolute max voltage, which is 20V based on the diode's breakdown voltage:
Imax = (Vin - Vf) / R
Imax = (20.0V - 1.28V) / 2000Ω
Imax = 9.36mA
Lastly, the series resistors will prevent damage to the IG- signal if wired incorrectly (reverse polarity). The diode will sink all the current, but it will not exceed the current determined above, protecting both the vehicle and board from being damaged.
** UPDATE **
The Miata stock tachometer requires a certain amount of current to effectively drive the needle on the instrument cluster. This means that no other devices can draw an excess amount of current. After further testing with the Tach-to-CAN prototype it was found that driving the optocoupler LED was too much current draw causing the stock needle to stop moving. In order to address this an NFET with a sufficient gate-source voltage had to be used to buffer the input signal.