Engine Knock Sensors, Part 2

We continue our investigation of engine knock sensors with a look at one and two-wire sensors.

In the previous issue of Counter Point, we discussed the various causes of engine knock. Prior to the age of electronic engine control, an engine designer had a limited number of effective tools to guard against engine knock. Principle among these were combustion chamber design, the octane of the fuel used and mapping of the spark advance curve. The designer couldn’t risk the possibility of engine damage caused by knock, so it was always necessary to keep the engine well short of the point where it might begin. This assured engine longevity, but hurt performance.

Modern engines are now fully controlled by electronics. Engineers no longer have to settle for a conservative preset spark advance curve. Spark advance can now be controlled dynamically. This dynamic control allows the engine control module (ECM) to take into account changing engine operating conditions as well as the available octane of the fuel, then use that information to extract the maximum amount of engine performance without running the risk of damaging spark knock.

The key sensor used to maintain dynamic control of spark advance and extract maximum performance is the engine knock sensor. For any given operating situation, the ECM attempts to deliver the maximum available performance by advancing the ignition. If this advance were unchecked, it would inevitably lead to engine knock.

The control unit needs a sentinel to report knock as soon as it begins. Ignition timing is then retarded by the ECM, and the knock stops. The ECM repeats the process by steadily advancing the timing until knock is detected, then retarding the timing until the knock stops. This closed loop process allows the engine to deliver maximum performance under all conditions, without the risk of damage or lost performance caused by knock.

Two major knock sensor designs are used today: broadband single-wire and flat response two-wire knock sensors. Both sensor designs use piezoelectric crystals to produce and send voltage signals to the ECM. The amplitude and frequency of this signal varies, depending upon the vibration levels within the engine. Broadband and flat response knock sensor signals are processed differently by the PCM. Broadband sensors use a single-wire circuit. This sensor type can respond to knock frequencies up to 1000 Hz from the design frequency value. This allows the sensor to accommodate shifts in engine knock frequency with changing engine operating conditions. The sensor’s high voltage output allows the use of a single, non-shielded output wire and a low impedance measuring circuit, while providing reduced susceptibility to electromagnetic interference (EMI).

Some PCMs output a bias voltage on the knock sensor signal wire. The bias voltage creates a voltage drop that the PCM monitors and uses to diagnose knock sensor faults. The knock sensor noise signal rides along this bias voltage. Due to the constantly fluctuating frequency and amplitude of the signal, it will always be outside the bias voltage parameters. The PCM in many applications will learn the knock sensor’s normal noise output. The PCM uses the noise channel and the knock sensor signal that rides along the noise channel similar to bias voltage systems. Both systems constantly monitor the sensor output, watching for a missing signal or one that falls within the noise frequency channel.

Flat response knock sensors use a two-wire circuit. This is a self-generating piezoelectric design that requires no power to the sensor. The sensor has a flat frequency response over the range of 5 to 18 kHz. This allows the sensor to be used on different engines by adjusting the filter frequency of the signal processing electronics to match the knock frequency of the engine. The sensor responds to knock frequencies that are higher than the primary knock frequency, allowing the higher knock frequencies to be used by a control system, either individually or combined with the primary knock frequency.

The signal rides within a noise channel which is learned by the PCM. The noise channel is based upon the normal noise input from the knock sensor and is known as background noise. As engine speed and load change, the noise channel upper and lower parameters will change to accommodate the knock sensor signal, keeping the signal within the channel. When there is knock, the signal will move outside the noise channel frequency and the PCM will reduce spark advance until the signal moves back inside the noise channel frequency.

Both the number and position of the knock sensors must be carefully selected so that knock from any cylinder or cylinders can be recognized under all conditions, with special emphasis on high loads and engine speeds. The knock sensor mounting position is generally on the side of the engine block or under the intake manifold. Four cylinder engines are normally equipped with one sensor, five and six cylinder engines with two and eight, ten and twelve cylinder engines with two or more knock sensors.

The sensor signals are evaluated by the PCM. A reference level is formed for each cylinder, which is continuously and automatically adapted to operating conditions. A comparison with the useful signal obtained from the sensor signal for every combustion process in every cylinder allows the PCM to determine whether knocking is occurring. If so, the ignition point is retarded by a fixed amount, 3° of crankshaft rotation for example, for the cylinder involved. This process is repeated for every cylinder for every combustion process that has been recognized as knocking. Once the knock subsides, the ignition point is advanced in small steps until it has returned to its spark advance map value.

Since the knock limit varies from cylinder to cylinder within an engine and changes dramatically within the operating range, the result is an individual ignition point for every cylinder. Cylinder-selective knock recognition and control makes possible the best optimization of engine efficiency and fuel consumption. If the vehicle is designed for operation with unleaded premium fuel, it can also be operated with regular unleaded fuel with slightly reduced performance and without the risk of internal engine damage.

In dynamic operation, knock frequency will increase under such conditions. To reduce knock, an individual spark advance map can be stored in the electronic control unit for the two fuel types. After start-up, the engine operates with the “premium” map. The PCM is switched to the “regular” map if the knock frequency exceeds a predetermined limit. The driver is not aware of this switchover; only power and fuel consumption will be slightly reduced.