Ignition systems and spark plugs

An ignition system has to start a stable flame kernel in the compressed mixture, at a repeatable crank angle, with enough spare voltage to do it again when the plug gap, mixture and cylinder pressure are worse than they were when the engine was new. Spark plugs are one part of that system. The rest is the energy source, the coil, the device that switches the coil, and the way the high tension reaches each plug. Choosing the crank angle of the spark is a calibration job of its own. The method is in Ignition tuning.

Spark plugs

A plug is a controlled gap in a threaded shell. The center electrode is insulated from the shell. When the coil voltage is high enough to ionise the gap, current flows and the plasma kernel can light the mixture. The plug also has to live in the chamber: sealing gas pressure, shedding heat into the head, and keeping the insulator nose clean enough that the spark does not track down the deposits. The same basic parts, a thread, a metal shell, a ceramic insulator and two electrodes, are what a modern plug still is.

History

Étienne Lenoir used an electric spark plug in his gas engine in 1860, and he is generally credited with the invention. Earlier ignition on gas engines had been a flame or a hot tube. Those were too vague to time a piston engine properly. Patents around 1898, including ones associated with Nikola Tesla, Frederick Richard Simms and Robert Bosch, sit in the decade when the high-tension system was being made practical. In 1902 Bosch engineer Gottlob Honold produced the first commercially workable high-voltage plug as part of a magneto ignition. That is the point at which the spark-ignition engine stops being a laboratory idea.

The industry that followed is a crowd, not a single inventor. Albert Champion was making plugs in the United States by about 1905. The company that kept his name and the AC company that used his initials split into two histories, which is why Champion and AC both exist. In Britain, the Lodge brothers manufactured their father's igniter, and Kenelm Lee Guinness developed the KLG plug. Spark-plug collectors count thousands of brand names between the early 1900s and the mid-1930s. Almost all of them were still a porcelain or mica insulator, a centre wire and a ground electrode on a threaded shell.

The insulator is the part that had to grow up. Mica, stacked and compressed, was common on early and aircraft plugs. Leaded petrol from the 1930s coated it and short-circuited the path. Sintered alumina replaced it: hard, a good enough conductor of heat for a ceramic, and able to be made in one piece. NGK came out of a ceramics works. Magoemon Ezoe visited a Detroit plug factory in 1920, and NGK Spark Plug was set up as its own company in the 1930s. NGK dates its copper-core wide-range plugs to 1965, its V-groove centre electrode to 1979, platinum plugs to 1982 and iridium to 1996. A separate industry account puts copper-cored centre electrodes in the late 1970s, as the cure for a solid nickel electrode that could not be cold enough for a motorway and hot enough for town. Both claims are about the same problem. Copper moves heat. The precious metal lets the firing tip stay small without burning away.

One historical dead end is worth remembering so it is not revived by a catalogue. Firestone sold polonium plugs from 1940 to 1953, on the idea that a little radiation would ionise the gap. The activity was small, the half-life is short, and deposits soon covered the surface that was supposed to help. They are not a type to look for.

The discharge

The gap is an insulator until the voltage exceeds the dielectric strength of the compressed mixture. Published descriptions put ordinary firing voltages in the range of about 12 to 25 kV, and the demand can reach about 45 kV when the gap is wide, the cylinder pressure is high, or the electrodes are worn and rounded. The arc channel is described as reaching on the order of 60,000 K for the brief life of the spark. That temperature is the spark, not the heat range of the plug. What is left after the arc is a kernel of burning mixture. A small kernel behaves like a late spark. A large one behaves like an early spark. Turbulence and the mixture between the electrodes decide the kernel as much as the coil energy does.

The centre electrode is normally the negative one, the cathode. It runs hotter than the ground strap, and electrons leave a hot, sharp surface more easily. Firing the plug the other way, centre electrode positive, is described in service material as needing up to about 45 percent more voltage. Wasted spark does exactly that on one plug of each pair. A pointed wire would fire at the lowest voltage and would also erode in seconds, so production electrodes are a compromise between a sharp edge and a metal that survives. As the edge rounds off, the required voltage climbs and the misfire arrives under load first.

Parts of a plug

The terminal on top may be a solid SAE stud, a removable nut for a screw-on eyelet, a bare thread as on many motorcycles, or a cup that lets the ceramic be longer in the same space. Coil-on-plug boots often need that cup or a specific stud. The wrong terminal is a loose contact and a misfire that looks like a failed coil.

The insulator on a modern plug is sintered alumina, glazed on the outside. Ribs lengthen the surface from the terminal to the shell so moisture and dirt have a longer path to track across. The nose that enters the chamber is the same ceramic, unglazed. Its length is most of what the heat range is. Older aircraft plugs used stacked mica instead of a one-piece ceramic.

Internal seals on a current plug are a fused glass and metal powder. The external seal is either a crush washer or a taper seat. The shell takes the tightening torque, carries heat from the insulator into the head, and is the earth for the ground electrode. Threads are rolled. A marine shell may be coated because of the environment. The centre electrode connects to the terminal through an internal path that, on a resistor plug, includes a ceramic resistance specifically to cut radio noise. The metal of the tip is nickel alloy, a copper core, or a precious-metal firing pad. Conductivity of the pad barely matters. It is in series with the resistor and the lead. Melting point and wear are the reasons for platinum and iridium, and a fine wire also steals less heat from the kernel.

Thread, reach, seat and gap

Common automotive thread diameters are 10 mm, 12 mm, 14 mm and 18 mm. The hex size does not follow the thread in a way you can guess, and a 14 mm plug can be a standard hex or a small hex for a narrow plug well. Reach, the length of thread, has to match the head. Catalogue reaches run from a few millimetres to well over 20 mm. Too long meets a piston or a valve, or leaves threads in the chamber where they glow and then fill with carbon. Too short buries the spark in the hole and runs the plug out of its designed heat range. A gasket seat and a taper seat seal in different ways and are not interchangeable. Taper-seat plugs are tightened to a lower torque than a crushed washer. An already crushed washer does not seal a second time.

Automotive gaps commonly fall somewhere between about 0.6 and 1.8 mm. That is a description of what is catalogued, not a gap to set. The engine specification wins, and the gap in the box is the plug maker's gap, which is often not the engine gap. Compressed natural gas is often specified with a narrower gap than petrol on the same engine family, because the mixture is harder to ionise. Measure a gap with a wire gauge. A flat feeler bridges the curvature and lies. On a fine-wire or precious-metal plug, do not treat the gap as freely adjustable. Bending against the centre electrode, or against the welded pad, cracks the insulator or pulls the pad off. Many of those plugs are made at the gap they are supposed to keep.

Heat range

Hot and cold describe the heat path. They do not describe a hotter or colder spark. The firing end needs to stay hot enough to burn deposits off the insulator, and cool enough that the tip does not glow and light the mixture before the spark. NGK puts that window at about 500 to 850 °C at the firing end. A hot plug has a longer insulator nose, so heat has a longer path into the shell and the tip runs hotter. A cold plug has a short nose and dumps heat into the head more quickly. A cold plug does not cool the engine. It only cools its own tip. A hot plug can start pre-ignition, and that secondary burn is what then heats the engine.

NGK states that one full heat-range step changes the plug's heat rejection by about 70 to 100 °C, and that the numbering is not shared between brands. On NGK, Denso and Pulstar a higher heat-range number is colder. On Champion, Autolite, Bosch, Brisk, Beru and ACDelco a higher number is hotter. An NGK racing plug is numbered differently again: the heat range sits after the hyphen, and a higher number is still colder. Compare a complete part number, including reach, seat and gap, rather than the heat digit alone.

A plug that is too cold fouls. The insulator stays black, the gap shunts, and the cylinder misfires, often first at idle or on a short trip. A plug that is too hot blisters or erodes the electrodes and can cause pre-ignition. NGK's guidance for a modified engine is to change heat range on its own, not in the same step as fuel, timing or boost, and to err colder: a cold plug fouls, a hot plug can damage the engine. Their rule of thumb is one heat range colder for roughly 75 to 100 hp added. That is a starting point for a plug of the same family, not a substitute for the engine's reach, seat, thread and gap. Before electronic fuel control it was normal to keep a hotter plug for short trips and a colder plug for sustained high speed. Closed-loop mixture control made that unnecessary on road cars. Racing engines still need the heat range chosen for the load they actually run, and some very old race engines carried a hot set for starting and a cold set for the session.

Reading the nose

Reading colour is only useful after a loaded run, with the engine shut down before a long idle washes the evidence. A light tan or grey-brown nose is the usual petrol target. Dry soot points to a cold plug, a rich mixture or a weak spark. A wet plug is fuel or oil. A chalky white or blistered nose, with eroded or melted electrodes, points to a hot plug, a lean mixture, or detonation. A sand-blasted look on the firing end is described as light, persistent detonation that may not have been loud. Heavy detonation can break the insulator before the colour chart looks dramatic. On a long-life fine-wire plug the nose colour is a weaker clue than it was on a copper-core plug that was removed every service. Cleaning a fouled plug with a wire brush leaves metal on the insulator and a new tracking path. Filing a precious-metal tip to sharpen it removes the pad that was the point of the plug.

Electrode designs

The spark temperature is set by the coil energy and the gap, not by the metal stamped on the box. The metal decides how the electrodes wear and how fine a wire can survive.

• Nickel alloy with a copper core. Good heat transfer and a wide choice of heat ranges. The electrodes wear faster, so the service interval is short. Usual on older engines and on builds where the plug is inspected often. The copper may be in the centre electrode, the ground electrode, or both, depending on the generation of the plug.
• Single platinum. A small platinum pad on the centre electrode. Long life on systems that always fire that electrode with the same polarity, which is a distributor or a sequential coil-on-plug.
• Double platinum, or iridium on both electrodes. The usual specification where one plug of a pair fires with reversed polarity. Wasted spark does this. A copper plug in that job wears the ground electrode on the reversed plug.
• Fine-wire iridium. A small centre electrode lowers the voltage needed to jump the gap, takes less heat out of the kernel, and the wear rate stays low. Common on coil-on-plug engines with long plug intervals. NGK dates its iridium range to 1996.
• V-groove. NGK introduced a V-shaped cut in the centre electrode in 1979 so the spark leaves from the edge of the wire rather than the middle of the face. It is an ignitability shape, not a different heat range and not a multi-spark.
• Several ground electrodes. The spark still uses one path, the easiest gap at that moment. When one gap wears, the spark moves to a closer strap, which is why the design lasts. The straps can also shroud the kernel, and the gaps cannot be reset to one common figure. Extra electrodes are a wear surface, not several sparks.
• Projected nose. The insulator and the gap stand further into the chamber, so the spark meets the mixture instead of firing in a pocket. The projected nose also runs hotter. It is the wrong choice if the piston or a valve comes close at the top of the stroke.
• Retracted or racing. A short nose and a cold heat range, for high cylinder pressure and a short service life. The spark is less exposed. Survival is the point.
• Surface discharge. The centre electrode is almost flush and the shell itself is the ground. The gap does not grow as a strap wears, and there is no strap to break off. Wankel engines use this form because an apex seal would hit any plug that stood proud of the chamber, while a plug buried in a hole would not see the mixture. The spark path wanders around the face over the life of the plug.
• Resistor and non-resistor. The resistor is in the plug, in series with the centre electrode. Most ECU cars specify it, because the noise upsets crank sensors, radios and the control unit. A non-resistor plug belongs only where the ignition, often a racing magneto, asks for it. The absence of the resistor is sometimes the missing R in the part number, but the catalogue is the check, not the letter alone.

Indexing

Indexing means tightening the plug so the open side of the gap faces the chamber rather than a wall, and the ground strap does not sit between the spark and the mixture. The thread and the strap are not in a fixed relationship on an ordinary plug, so each plug is marked, trial-fitted, and shimmed with indexing washers until the mark lands. A few plugs are made with a known orientation for very small engines, where the strap is a large part of the chamber. On a car engine the gain is small beside mixture, timing and a plug of the right reach. Washers also change reach and heat transfer. They are not a way to make a short plug into a long one.

Prechamber plugs

A prechamber is a small volume around the plug, open to the main chamber through several orifices. The spark lights the mixture in the prechamber. The pressure drives jets of burning gas into the main charge, so the main chamber is lit in several places at once rather than from a single kernel. A passive prechamber is filled by the compression stroke. An active one has its own fuelling, so the prechamber can be kept near stoichiometric while the main chamber runs lean. Reviews of the layout put the prechamber at a small fraction of the clearance volume, and they describe it as a way to keep a lean or heavily diluted burn stable. It is a combustion-chamber design that still uses an ordinary electric plug. It is not a different electrode metal, and the orifice sizes from one gas-engine study do not transfer to another head.

Laser ignition plugs

A laser plug is a research and prototype device. A short laser pulse is focused in the mixture, and the plasma kernel forms at the focal point rather than between two pieces of metal. The usual reasons given for trying it are that no electrode sits in the kernel to quench it, that the kernel can be placed away from the wall, and that one source can be split into more than one focus so a lean mixture is lit in several places. There is no dwell and no spark gap. The problems move to the window, the lens and the laser.

It has been run on engines. A European project built a prototype laser plug and, in 2018, trialled it on a four-cylinder multipoint-injection petrol engine running lean mixtures, with a reported improvement over an electric plug on those lean tests. The same line of work treats large gas engines and specialist propulsion as the realistic applications, and treats a production car as something that still needs a manufacturer to industrialise the laser, the cooling and the service life.

A 2025 stationary-engine programme, with the University of Bayreuth, Fraunhofer IOF and INFLPR, fitted laser plugs to a gas engine used for combined heat and power. The laser was a very short-pulse device, under a nanosecond, in a resonator on the order of 11 mm. At 200 kW, on a highly lean mixture, it matched or beat an electric prechamber plug: a few percent higher peak pressure, a faster burn, and lower NOx. At 250 kW the two were close. At the engine's 275 kW maximum the electric plug was ahead. A focal length of 8 mm broke lenses. A focal length of 11 mm did not. Sapphire windows collected debris when a prechamber was used, and stayed clean without one. Some debris burned clear in the beam, which is a self-cleaning effect, not a service plan. The published conclusion is that this is a step toward a laser plug for stationary engines, not a part for a road-car service schedule.

Until that hardware is a catalogued, cooled, window-cleaning production system, a laser plug is not a substitute for a hotter coil or a finer electrode. A cracked window or a lens full of deposit is a misfire. It will show the same high hydrocarbons and high oxygen as a dead coil, and it will not be cured by a heat-range change.

Plugs with a pressure sensor

A measuring spark plug is an instrument that still has to fire. A piezoelectric cylinder-pressure sensor is built into the shell, so the existing plug hole becomes the indicating bore. Kistler's type 6118C is an M14 × 1.25 example. It carries a 3 mm sensor, flush with the chamber, with a measuring range of 0 to 200 bar and an overload of 250 bar. The datasheet rates the insulation to 45 kV and uses a platinum-platinum firing end. The centre electrode is offset to make room for the sensor. Heat range, spark position and insulator diameter are chosen to suit the engine, in 9 mm or 10.5 mm ceramic. The sensor cable unscrews. The sensor and the insulator are replaceable, and Kistler does that replacement at a service centre rather than on the bench.

Flush mounting is what makes the trace usable at high rpm and for knock. The 6118C datasheet quotes a sensor natural frequency above 120 kHz and an acoustic natural frequency around 65 kHz. An older 6117B variant was described specifically for setting ignition and injection on a racing engine near the knock limit, and as the one to prefer above about 5,000 rpm because its natural frequency stayed higher. The point of the tool is a pressure trace against crank angle: indicated mean effective pressure, the location of peak pressure, the burn rate, and the spikes of knock. That trace is the measurement behind the phasing targets on the ignition-tuning page. One cylinder is not the brake torque of the engine. It is the cylinder you can no longer guess about.

These plugs are not a hotter or colder road plug, and they are not a closed-loop production sensor. Kistler recommends calibration after each campaign, and at least within a year. Thermal shock moves the trace. A plug left in the head as a permanent fitment will drift, and the spark side still has to match the reach, the seat and the heat range or it will pre-ignite like any other plug that is wrong. Diesel indicating often uses a sensor in a glow-plug adaptor rather than a spark plug. That is the same measurement in a different hole.

The plug as an ion sensor

Saab's Direct Ignition cassette, in production from the 1988 9000 and later part of Trionic, uses each spark plug as a sensor between firings. After the spark, a voltage is held across the gap. Combustion leaves ions, the gap conducts a small current, and the shape of that current follows the pressure and temperature of the burn. Knock adds spikes. A misfire leaves the current flat. Trionic also uses the current to decide which cylinder of a pair has just fired, which is how those systems do without a separate cam sensor, and it uses the same signal in place of a separate knock sensor. The plug itself is a normal plug. The sensing circuit lives in the cassette and the controller. A resistor plug, a rich mixture or a fouled nose changes the current, so an ion-sense system is picky about the plug part number in a way a plain coil-on-plug car is not.

Ion current is not cylinder pressure. It correlates with the burn well enough for knock, misfire and sync. A piezoelectric plug measures the pressure directly. The two are easy to confuse because both turn the plug hole into an instrument.

Two plugs in one cylinder

A second plug shortens the distance each flame front has to travel. Combustion finishes sooner, so the engine can often use less advance, tolerate a leaner mixture, and leave less unburned fuel. It is not free power. The gain depends on where the second plug sits and on whether the two sparks are simultaneous.

Mercedes-Benz M112 and M113

The M112 V6, introduced for the 1998 model year, and the related M113 V8 use two spark plugs in every cylinder. The head is a three-valve layout, two inlet valves and one large exhaust valve, which is what leaves room for the second plug. The launch description of the M112 says the two plugs are fired one after the other, with the offset depending on load and speed, and that the offset can be changed from one cycle to the next. A double-ended coil cannot do that, because both plugs would fire together. The S-Class installation is described as one coil assembly per cylinder, feeding both plugs.

The same launch material says warm-up uses a very lean mixture and ignition retarded by 5 to 10 crank degrees, to raise exhaust temperature and light the catalyst sooner. It also says that if one plug circuit fails, the other can still burn the charge. A failed plug circuit on these engines can therefore be a small loss of efficiency, a rough idle or a partial burn that still feeds the catalyst, rather than a dead cylinder. Those timing and mixture figures are the introduction's description of the new engine. Later M112 and M113 calibrations still have two plugs per cylinder, but the exact warm-up offset belongs to the calibration, not to a fixed mechanical advance.

The early V6 is described as Bosch Motronic ME 2.0, including the phase-shifted ignition. Treat that as the launch controller. Confirm the unit on the car before assuming a later M112 or M113 is the same Motronic generation. ECU timing maps and knock control are a separate subject from the hardware. The ME7 ignition and knock reference is one example of how a Bosch controller splits base timing, temperature, knock and torque intervention. The general method for choosing the angle is Ignition tuning.

Alfa Romeo Twin Spark

Alfa Romeo used two plugs per cylinder on the 1914 Grand Prix car, again on 1960s racing engines, and then on road cars from the mid-1980s. The road engines people mean by Twin Spark are inline fours. The 8-valve engines, in the 75, 164 and 155, are Alfa twin-cam units with two identical plugs, placed symmetrically, because a two-valve chamber has no room for a central plug. The published description says those two plugs fire together, so each flame travels less distance and the engine needs less advance and can run leaner. Later 8-valve cars are also described as wasted spark, pairing 1 with 4 and 2 with 3. Earlier 8-valve cars are not all that layout, so the cap, leads or coil count has to be read off the engine in front of you.

The later 16-valve Twin Spark engines, used in the 145, 146, 155, 156, 147, 166, GTV, Spider and GT, are Fiat Pratola Serra fours with an alloy head. Each cylinder has a 14 mm plug in the centre and a 10 mm plug at the edge of the chamber, between an inlet and an exhaust valve. Both are long-life platinum plugs, with a published interval of 100,000 km. The side plug adds little peak power. The stated benefit is light-load combustion, including a smooth idle on mixtures as lean as 18:1. The JTS fours that replaced them dropped the second plug and used direct injection.

The 16-valve coil wiring changed, and the two generations are easy to mix up:

• Earlier CF1 and CF2. Four coils. Each coil fires the plug under it and, through a short lead, a plug in the cylinder that is 360 degrees out of phase. Cylinder 1's coil also sparks cylinder 4, and the other coil of that pair sparks the second plug. Each cylinder is fed by two coils. One failed coil leaves the other plug in that cylinder working. This is a wasted-spark arrangement, even though there are four coils for four cylinders, because each coil still fires every crank revolution.
• Later CF3, from the 2001 Euro 3 engines. Each coil fires both plugs in its own cylinder. The published description says this need not be wasted spark: the coil can fire every 720 degrees instead of every 360, so it has more time to charge at high rpm, and a Bosch controller can trim that cylinder on its own. Individual trim is impossible on CF1 and CF2, because each cylinder depends on two coils.

Confirm which layout the car has before replacing leads or simplifying the wiring. Cross-connecting a CF3 coil as if it were CF2, or the other way round, does not preserve the two flame fronts the head was built for.

Magnetos

A magneto makes its own ignition energy. A permanent magnet rotates past a coil, or the coil rotates in a fixed magnet, and that movement generates the primary current. Opening the points, or an electronic switch, collapses the field and the secondary produces the high tension. A battery is not required to make the spark. The battery on a vehicle that still has one may only be there for the starter, lights and instruments.

Magnetos are the normal ignition on many small industrial engines, older tractors, some classic cars, and certified piston aircraft. Aircraft practice is two independent magnetos and two plugs per cylinder, so either system can keep the engine running. Aircraft plugs are often shielded, with a metal body over the insulator and a harness lead, because an open high-tension lead is also a radio transmitter. The thread and the reach still have to match that head. Many aircraft magnetos use an impulse coupling so the spark is retarded and mechanically snapped over while the engine is being started, then left behind once the magneto is up to speed. Advance on a classic-car or tractor magneto is mechanical, weights or a fixed timing, rather than a load and speed map in a controller.

The kill switch works by grounding the primary, which is the opposite instinct from disconnecting a battery ignition. A magneto that has been timed by guessing the impulse position, or that has weak magnets, gives a spark that falls away with rpm. There is no alternator voltage and no ECU dwell table to compensate.

Distributors and the distributor coil

A distributor system has one coil for the engine. The coil secondary feeds the centre of the cap. A rotor passes the spark to one plug lead per firing event, in the order of the cap. The rotor gap is a second spark gap, so some of the coil voltage is used before the plug ever fires. That is one reason a worn cap, a carbon-tracked cap or a long corroded lead shows up as a misfire under load.

Triggering is either contact points and a condenser, or an electronic pickup in the distributor: a Hall vane, an inductive reluctor, or an optical chopper. The condenser on a points system gives the primary current somewhere to go as the points open, so the field collapses quickly and the points do not arc themselves away. A shorted condenser is a weak spark and burned points. An open condenser is burned points and radio noise.

Advance is mechanical. Centrifugal weights add advance with rpm. A vacuum canister adds or, on some engines, subtracts advance with manifold vacuum, which is a load signal. Stuck weights, a seized plate or a split vacuum pipe leave the timing wrong across the range even though the idle timing mark looks plausible. On a points distributor, dwell and timing are set in that order. Opening the point gap reduces the closed angle and also moves the moment the points open, so the timing mark moves with the feeler gauge. Why that advance has to keep changing, and how to choose it when the weights are replaced by a map, is the ignition-tuning page.

Older coils are often ballast-resisted. The coil is built for a lower running voltage, and the resistor is bypassed during cranking so the spark survives a low battery. Bypassing that resistor all the time overheats the coil. Replacing a ballast coil with a 12 V coil and leaving the resistor in series gives a weak spark. Match the coil to the resistor that is actually in the circuit.

Wasted spark

Wasted spark removes the distributor. One double-ended coil feeds two plugs whose cylinders reach top dead centre together. One cylinder is on compression and the spark lights the mixture. The other is on exhaust and the spark does nothing useful, which is why it is wasted. On a four-cylinder with firing order 1-3-4-2 the pairs are 1 with 4 and 2 with 3. A six-cylinder or eight-cylinder pairs the cylinders that are 360 degrees apart in the cycle. The secondary of the coil is in series with both plugs, and it is not tied to chassis earth the way a distributor-coil secondary is. One plug fires with the centre electrode negative. The other fires reversed, centre electrode positive. Reversed polarity needs more voltage, described as up to about 45 percent, and it wears the ground electrode when that plug is the one on compression.

The arrangement needs a partner cylinder at TDC. A three-cylinder or five-cylinder engine does not have one, so it cannot use a double-ended pair. Those engines use a distributor or a coil per plug. Firing a single plug again on the exhaust stroke, from its own coil, is a different and less common meaning of the same words.

Wasted spark was the usual distributorless system through the 1980s and 1990s, including GM DIS and Ford EDIS. It needs a crank reference and does not need a cam sensor to decide which plug of the pair is on compression, because both fire. It cannot time the two cylinders independently. Both run at the same advance. At high rpm each coil has half the charge time of a sequential coil, because it fires every crank revolution rather than every second revolution. The gain against a distributor is the opposite comparison: several coils share the work that one coil and a rotor gap used to do, and there is no cap to track moisture across.

Coil-on-plug

Coil-on-plug puts one coil on one plug and fires it once per cycle, on the compression stroke. Coil-near-plug is the same idea with a short lead or a boot between the coil and the plug. The controller charges and fires each coil, so dwell and timing can be trimmed per cylinder. A failed coil, a cracked boot or a plug well full of oil takes out one cylinder. The misfire is then a single-cylinder hydrocarbon and oxygen rise, and on a modern controller a misfire counter for that cylinder, rather than a pair of cylinders going weak together as they do with a shared wasted-spark coil.

There is no rotor gap and almost no high-tension lead, so more of the coil voltage reaches the plug. The coil also has the longest possible time to charge. The parts that actually fail are the boot, the spring between coil and plug, and oil or water in the plug well, which fire the spark down the outside of the insulator. A coil swapped onto another cylinder is the clean test: the misfire moves with the coil, or it stays with the plug, the injector and the compression.

Dwell

Dwell is how long the coil primary is turned on before the spark. The spark happens at the moment the primary is switched off. Energy stored in the coil rises with primary current, and current rises during the on time, faster when the supply voltage is higher. Enough dwell reaches the current the coil was designed for. More dwell after that point mostly heats the coil and the driver. Too little dwell is a small spark, a misfire under load or at high rpm, and the five-gas pattern of unburned fuel plus unused oxygen. Dwell is not the ignition angle. Lengthening the dwell does not advance the spark. The controller, or the points, has to start the dwell early enough that the current is up at the angle you actually want.

On points, dwell is specified as the closed angle in distributor degrees. The distributor turns at half crank speed. Each cylinder therefore owns 90 distributor degrees on a four-cylinder, 60 on a six and 45 on an eight. A larger point gap closes for fewer of those degrees. Set the specified angle, then set the timing.

On an ECU or an ignition module, dwell is a time, and it is coil-specific. It is lengthened at low battery voltage, including cranking, and shortened as voltage rises. The time available also shrinks with rpm, because the coil must finish the spark before the next charge starts. The interval follows from the speed and how often that coil has to fire:

The interval in milliseconds is 60,000 divided by rpm, divided again by the number of sparks that coil produces per crank revolution. The 8-cylinder row matches the worked example in the MegaSquirt sequencer coil notes: 2.5 ms with one coil, 10 ms wasted spark, 20 ms sequential. Those numbers are the window the spark and the charge must share. They are not a dwell setting. A commanded dwell longer than the window cannot happen; the coil is still on, or has only just fired, when the next cylinder needs it. Single-coil engines run out of window first, which is why high-rpm misfire on a distributor engine is often a dwell limit rather than a mysterious lean mixture. Wasted spark sits in between. Sequential coil-on-plug has the most time, which is also why copying a coil-on-plug dwell map onto a wasted-spark pair overheats the coils: each coil is being asked to charge twice as often.

Use the dwell specified for that coil at that voltage. A dwell table from a different coil, or a longer dwell added for a stronger spark after the coil is already at its design current, heats the primary without a useful increase in spark energy.

Ignition angle

The angle is how many crank degrees before top dead centre the spark is commanded. The mixture needs a fairly fixed amount of time to burn. At higher rpm the crank covers more degrees during that time, so the spark has to move earlier if the pressure peak is to land in the same place, a little after top dead centre, where it can still push the crank. At higher load the denser charge burns faster, so the spark usually comes back. Knock then removes the advance you would otherwise keep. A lean or diluted mixture burns more slowly and wants more advance until it misfires. A cold start often wants the opposite, a late burn, so the exhaust will light the catalyst. The M112 description of 5 to 10 degrees of warm-up retard is one published example.

Those are the reasons the angle cannot be one number. Finding the number, on a load-holding dyno, up to MBT or up to the knock limit, is written as its own page: Ignition tuning. A measuring spark plug is the instrument that shows whether the pressure peak actually landed where the torque says it should.

Ignition faults on a five-gas analyzer

A five-gas analyzer reads hydrocarbons (HC), carbon monoxide (CO), carbon dioxide (CO2), oxygen (O2) and oxides of nitrogen (NOx). HC is fuel that did not burn. CO is partial combustion and the usual rich marker. CO2 is the product of combustion that finished. O2 is oxygen that was not used. NOx forms when the peak cylinder temperature is high. The directions below are the combustion patterns used in five-gas training. They are not statutory limits, and they describe the gas leaving the cylinder. A lit three-way catalyst then consumes HC, CO and O2 and reduces NOx near stoichiometric, so a tailpipe sample can hide a small ignition fault. Sample ahead of the catalyst when the question is combustion. Use the tailpipe when the question is the catalyst. How to move the angle on purpose, rather than how to recognise a wrong one, is on the ignition-tuning page.

The practical split is between an ignition misfire and a mixture fault. An ignition misfire dumps raw fuel and the oxygen that should have burned it, so HC and O2 rise together while CO stays low or falls and CO2 falls. A rich mixture raises CO and uses the oxygen up, so O2 falls. HC only climbs hard on a rich engine once the mixture is rich enough to misfire. A lean engine that is still firing raises O2 and NOx and drops CO. Once it is lean enough to miss, NOx falls again and HC joins the high O2.

A calculated lambda from the five gases can read lean during an ignition misfire, because of the unused oxygen, even though the injector pulse was right. Chasing that number with fuel hides the ignition fault and makes the firing cylinders rich.

Timing moves NOx first. More advance raises peak temperature and pressure, so NOx rises, and enough advance to fire while the piston is still rising also leaves HC behind and can rattle. A small retard cuts NOx with little harm to HC. A large retard, including a controller that is pulling a lot of knock timing, drops NOx, raises exhaust temperature, and eventually raises HC and drops CO2 because the burn is still going on as the exhaust valve opens. Before condemning a distributor or a timing map, check knock retard, crank and cam correlation, and a coolant sensor that never lets the warm map in.

Weak spark, a wide gap, a fouled cold plug and a failing coil sit on the ignition-misfire row, sometimes only when the engine is loaded. Cylinder pressure under load is what pushes the required voltage over what the coil can still make. A twin-plug cylinder with one plug dead may not reach a full misfire: Mercedes described the second plug as enough to keep the charge burning. The gas change is then a smaller HC rise and a loss of efficiency, until the remaining plug circuit fails too. Unburned fuel and oxygen entering a catalyst together are what overheat it. An ignition misfire is a catalyst fault in the making, not only an emissions number.

Sources

Hardware behaviour follows the cited descriptions. Where a figure belongs to a launch claim, a research engine or one generation of a vehicle, it is attributed there and is not a specification for every later calibration. This wiki has not repeated the measurements on a vehicle.

• Spark plug. History from Lenoir to Honold, construction, typical firing voltage, gap range, surface-discharge plugs, copper-core electrodes, reversed-polarity voltage and plug reading. Accessed 2026-11-11.
• Gas Engine Magazine, the spark plug. Bosch 1902, Champion, and the scale of early plug brands. Accessed 2026-11-11.
• NGK, development of the spark plug. NGK's origins, copper-core plugs in 1965, V-groove in 1979, platinum in 1982 and iridium in 1996. Accessed 2026-11-11.
• NGK, What is a spark plug's heat range? Firing-end window, heat-range step, numbering direction and the modified-engine guidance. Accessed 2026-11-11.
• NGK, Heat range conversion chart. Cross-brand heat-range scale. Accessed 2026-11-11.
• SparkPlugs.com, How do I find a colder or hotter plug? Worked NGK and Bosch part numbers. Accessed 2026-11-11.
• Photonics Spectra, July 2025, a laser spark plug. The stationary gas-engine comparison at 200, 250 and 275 kW, lens damage and window fouling. Accessed 2026-11-11.
• European Commission, LASIG-TWIN. The prototype laser plug and the 2018 four-cylinder petrol test. Accessed 2026-11-11.
• Kistler, measuring spark plug 6118C. Integrated pressure sensor, M14 fitment and 45 kV insulation. Accessed 2026-11-11.
• Kistler datasheet, type 6118C. 0 to 200 bar, overload, natural frequency and calibration interval. Accessed 2026-11-11.
• Saab Direct Ignition. Cassette ignition and ion-current sensing through the plugs. Accessed 2026-11-11.
• CFD simulation of pre-chamber spark-ignition engines, a review. Prechamber volume, jets and lean-burn use. Accessed 2026-11-11.
• M112 engine highlights. A published reproduction of the Mercedes-Benz M112 introduction material, with the site author's own later notes mixed in. Phase-shifted twin plugs, the 5 to 10 degree warm-up retard, and ME 2.0 on the new V6. Accessed 2026-11-11.
• W220 S-Class Encyclopedia, Ignition system. One coil assembly per cylinder on the M112 and M113 installation in that chassis. Accessed 2026-11-11.
• Alfa Romeo Twin Spark engine. Road-car history, plug sizes, and CF1/CF2 versus CF3 coil pairing. Confirm the pairing on the car's wiring diagram. Accessed 2026-11-11.
• Motor, July 1999, distributorless ignition. Wasted-spark pairs, series plugs and reversed polarity. Accessed 2026-11-11.
• Automotive Tech Info, Waste spark. Why the wasted plug is in series, and why a five-cylinder has no wasted-spark partner. Accessed 2026-11-11.
• MegaManual, sequencer coils. The 6000 rpm charge-window example. Accessed 2026-11-11.
• Smog Tech Institute, Understanding 5-gas diagnosis. Directional HC, CO, CO2, O2 and NOx changes. Accessed 2026-11-11.
• Tuning Wiki, Ignition tuning. MBT, knock-limited spark advance and the spark sweep.