Reading Ignition Timing and Knock Logs
Ignition timing is the angle before top dead center the ECU fires the spark, and it moves constantly with rpm, load and boost to keep a margin against knock. This guide covers what a healthy timing trace looks like in a log, what knock retard means when it shows up, how to tell a real knock event from sensor or mechanical noise, and how heat soak and fuel octane change the picture.
All 13 sections
The basics
What it is, what it changes, and whether you need it.
What ignition timing advance is
Ignition timing is the moment the spark plug fires, measured in degrees of crankshaft rotation before the piston reaches top dead center (BTDC). The spark does not happen exactly at the top of the stroke because the fuel-air mixture takes a small but real amount of time to burn, so firing a little early lets cylinder pressure peak just after the piston passes top dead center, which is where it does the most useful work pushing the piston down.
More degrees before top dead center is called more advance; fewer degrees, or firing closer to top dead center, is called more retard. The two words describe the same angle moving in opposite directions, and both show up constantly in normal engine operation, not just when something is wrong.
The driver does not set this angle. The ECU calculates it from a timing map, a table of values indexed by engine speed and load, and adjusts it many times a second as those two things change. See What Does an ECU Tune Do? for how the timing table fits alongside the fuel and boost tables a tune also sets.
How the ECU changes timing with rpm, load and boost
At light load, cylinder pressure and temperature are relatively low, so there is more room before the mixture is at any risk of igniting on its own. The ECU generally runs more advance in this region, since there is little downside and it gets more work out of each burn.
As load rises, whether from more throttle, more rpm, or boost building on a turbocharged or supercharged engine, cylinder pressure and temperature both climb. That is the same condition that makes uncontrolled combustion, knock, more likely, so the ECU pulls timing back to keep a margin against it rather than chasing the same advance it would use at light load.
On a boosted engine this shows up clearly once boost comes in: timing drops from what it was running before boost built, then tends to hold roughly flat once boost is fully built and holding steady, rather than continuing to change through the rest of the pull. The chart below shows this pattern across a full pull from low rpm to redline.
Up the left: Ignition timing (degrees)Across: Engine speed (rpm)
- Ignition timing, actual
- Before boost, the ECU runs more advance at this light load.
- Boost is building, so the ECU pulls timing back as cylinder pressure rises.
- Under full boost, timing holds flat with no further retreat, a sign of a healthy margin.
Sample data drawn to show the pattern, not a measurement of any real car.
The numbers behind this graph
| Engine speed (rpm) | Ignition timing, actual |
|---|---|
| 1,500 | 19 |
| 2,000 | 20 |
| 2,500 | 21 |
| 3,000 | 20 |
| 3,500 | 16 |
| 4,000 | 13 |
| 4,500 | 12 |
| 5,000 | 12 |
| 5,500 | 12 |
| 6,000 | 12 |
| 6,500 | 12 |
| 7,000 | 12 |
What knock is
Knock, also called detonation, happens when some of the unburned mixture ahead of the normal flame front ignites on its own from heat and pressure, instead of being consumed in order by the flame the spark plug started. The result is two pressure waves colliding inside the cylinder instead of one smooth rise, which produces the metallic pinging or knocking sound the condition is named for.
Occasional, brief knock is something most control systems are built to catch and correct. Sustained or repeated knock is what actually risks damage, since the pressure spikes it produces can hammer piston crowns, rings, head gaskets and bearings over time, in severe cases within minutes rather than gradually.
Most engines have a knock sensor, a small sensor bolted to the block that listens for the specific vibration signature knock produces. The ECU uses that signal, along with the rest of its inputs, to decide whether and how much to pull timing back.
What knock retard means in a log
When the ECU's knock control decides it has detected knock, it subtracts a number of degrees from the timing it was otherwise going to command, at that moment, usually at that specific cylinder or bank if the platform can tell them apart. That subtracted amount is knock retard: a correction, shown in a log as its own channel, separate from the commanded timing itself.
Different platforms name and report this channel differently. Some call it knock retard, some call it feedback knock or timing correction, and the exact number that shows up can mean slightly different things from one ECU to the next, which the later sections of this guide come back to.
A small amount of retard that appears once and then clears is a different situation from retard that keeps appearing at the same point in the rev range, pull after pull. Telling those two apart is most of what reading a knock log is actually about, and it is covered in detail further down.
Signs your timing or knock needs attention
A handful of symptoms are worth paying attention to, especially if more than one shows up together or a symptom appears right after a change to the car or a tank of fuel. None of them on their own proves knock is happening, but together they are a reasonable prompt to have the car logged rather than guessed at.
- An audible pinging or knocking sound under load, especially accelerating uphill or under boost
- A check-engine light tied to a knock or knock-retard code
- A loss of power or hesitation under load that was not there before
- The engine running noticeably hotter than usual under the same conditions
- Any of the above starting right after a fuel fill-up, a hot day, or a change to the car
Choosing and fitting
How to pick one, what it needs alongside it, and the usual mistakes.
Reading knock retard in a datalog
A pull, a full run from low rpm to redline at wide-open throttle, is the usual way to check timing and knock together, since it lets every point on the rev range be checked at once rather than a single steady-state reading. See Datalogging: What to Log and Why for the other channels worth logging alongside timing and knock, and How to Read Datalog Graphs for reading AFR, boost and fuel trims the same way.
In a healthy pull, timing follows a smooth curve: more advance at low load, less once boost builds, then roughly flat once boost is fully built, and the knock retard channel sits at or near zero the whole way through. That flat, quiet retard channel is exactly what marker 3 in the chart from the beginner section shows.
A knock event looks different: the retard channel spikes at a specific rpm, and the timing channel drops by about that same amount at the same instant, since timing actual is what remains after retard is subtracted from what was commanded. Recovery is usually gradual rather than instant, the retard fading back toward zero over the following samples as the ECU regains confidence, not snapping back the moment the spike ends.
| Reporting format | What it shows | How to read it |
|---|---|---|
| Degrees of retard | The actual timing pulled at that instant | Read directly: a bigger number is a bigger correction |
| Knock event count | How many times knock was detected in a window | Says nothing about severity alone; check it against the timing channel |
| Raw knock sensor level | An unprocessed signal from the knock sensor | Needs comparing against a baseline for that specific engine before it means anything |
Up the left: Ignition timing (degrees)Up the right: Knock retard (degrees)Across: Engine speed (rpm)
- Ignition timing, actual (left scale)
- Knock retard (right scale)
- Knock retard jumps to 6 degrees, the ECU's response to a detected knock event.
- Timing drops from its usual 12 degrees to 6 at the same rpm, the direct result of that retard.
- Retard fades gradually rather than snapping back, typical of how knock control recovers.
- Timing returns to its pre-knock value once retard clears completely.
Sample data drawn to show the pattern, not a measurement of any real car.
The numbers behind this graph
| Engine speed (rpm) | Ignition timing, actual | Knock retard |
|---|---|---|
| 1,500 | 19 | 0 |
| 2,000 | 20 | 0 |
| 2,500 | 21 | 0 |
| 3,000 | 20 | 0 |
| 3,500 | 16 | 0 |
| 4,000 | 13 | 0 |
| 4,500 | 12 | 0 |
| 5,000 | 12 | 0 |
| 5,500 | 6 | 6 |
| 6,000 | 10 | 2 |
| 6,500 | 12 | 0 |
| 7,000 | 12 | 0 |
Telling real knock from false knock
Not everything a knock sensor picks up is real detonation. Mechanical noise, a loose heat shield, valvetrain clatter, an accessory bracket vibrating at the wrong frequency, can trigger a knock reading that has nothing to do with combustion, and a wiring, connector or grounding issue on the sensor circuit can do the same thing with no mechanical cause at all.
Real knock tends to repeat at the same rpm and load, pull after pull, because it is tied to a genuine condition in the cylinder at that point in the map. False triggers from mechanical or electrical noise are usually the opposite: scattered across different points, or showing up at idle or during deceleration where cylinder pressure is far too low for real detonation to physically occur.
On platforms that report knock per cylinder, real knock often shows consistently on one cylinder or bank, reflecting an actual localized condition there, while a noisy sensor or a mechanical source tends to scatter more randomly across cylinders. Real knock also tends to correlate with something that changes the knock margin: intake air temp climbing across a session, or a switch to lower-octane fuel. A reading with no such correlation, and no repeatable rpm-and-load pattern, is a reasonable one to suspect as noise rather than real detonation.
| Trait | Real knock | False knock (sensor or mechanical noise) |
|---|---|---|
| Repeats at the same rpm and load | Usually yes, pull after pull | Usually no, or appears at random points |
| Which cylinders it shows on | Often one cylinder or bank consistently | Often scattered with no consistent pattern |
| Relationship to intake air temp | Often gets worse as IAT rises | Usually no relationship to IAT |
| Relationship to a fuel change | Often appears or worsens after an octane or fuel change | No relationship to fuel |
| Typical real-world cause | Too little margin for the fuel, boost or heat | A loose heat shield, valvetrain noise, a wiring or ground issue |
Knock learning and adaptive values
Many ECUs keep more than just the instant-by-instant retard applied during a knock event. They also keep an adaptive, or learned, correction that shifts timing at specific rpm-and-load cells over time based on the knock history the ECU has accumulated at that cell, separate from any single event.
How this works varies by ECU and platform, so it is worth understanding for the specific car rather than assuming one behavior applies everywhere. Some systems reset their learned values every time the car is started, some carry them forward across drive cycles, and some let a tuner view or manually clear them as part of diagnosing a knock issue.
The practical effect is that a car which has learned a negative correction at certain cells can keep running reduced timing there even after whatever caused the original knock has been fixed, a fuel tank topped off with the right grade, a heat shield reattached, until that learned value is relearned or cleared. This is one reason a tuner may want to reset adaptive values and re-log a car after resolving a knock cause, rather than assuming the timing snaps straight back on its own.
Common mistakes when reading a knock log
The most common mistake is judging a single reading in isolation instead of checking whether it repeats. A retard that shows once on one pull and never returns is a different situation from one that comes back at the same rpm and load every time, and treating them the same, either by dismissing every reading as noise or by treating every reading as a crisis, leads to the wrong response either way.
A second common mistake is reading knock retard without also checking intake air temp and any recent fuel change, since both are common, ordinary explanations for a pattern that might otherwise look alarming. A retard that only shows up on a hot, late pull in a session, or right after a tank of lower-octane fuel, usually has an explanation that does not require assuming something is mechanically wrong with the engine.
A third is treating a knock event count the same way as a degrees-of-retard reading, when the two formats say different things, as the table above shows. An event count with no corresponding timing change to check it against is much harder to interpret on its own.
In depth
The numbers, the trade-offs and what goes wrong.
Why boost lowers the knock limit
Knock happens when the end-gas, the portion of the mixture still ahead of the advancing flame front, reaches conditions where it ignites on its own before the flame reaches it. Both temperature and pressure in that end-gas push it toward that point, and both rise together as load increases.
Boost raises cylinder pressure directly, and it also raises intake charge temperature through compression heating, the same effect that makes a bicycle pump warm up as it is worked. Higher pressure and higher temperature both push the end-gas closer to autoignition, which is why the same engine has less margin at high boost than it does at light, unboosted load, and why the ECU responds by holding less timing there.
Compression ratio, the intercooler's effectiveness at removing that compression heat, and the fuel's own resistance to knock all shift where that limit actually sits for a given engine. See Intercoolers Explained for how intercooler sizing and design affect how much the intake charge heats up under boost, and the fuel octane section below for the fuel side of the same trade-off.
Intake air temperature and heat soak across pulls
Hotter intake air pushes the end-gas toward the same autoignition condition that higher boost does, for the same underlying reason: more heat in the charge means less margin before it ignites on its own. A pull logged with hot intake air genuinely has less knock margin than the same pull logged with cool intake air, even at the same boost and rpm.
Back-to-back pulls, on a dyno or repeated hot laps at a track, do not always give the intercooler and the rest of the intake tract enough time to fully shed the heat picked up on the previous run. Intake air temp then climbs from one pull to the next across a session, a pattern generally called heat soak, and a tune's timing and knock behavior on the third or fourth pull of a session can look noticeably worse than it did on the first, even though nothing about the car or the tune has actually changed.
This is one reason a single pull is not always enough to judge a tune by. A tune that looks clean on a cool first pull can show real retard by the third pull of the same session purely from accumulated heat, which is a different finding than a tune with too little margin from the start.
Up the left: Ignition timing at redline (degrees)Up the right: Intake air temp at redline (°F)Across: Pull number
- Ignition timing at redline (left scale)
- Intake air temp at redline (right scale)
- On the first pull, intake air temp at redline is already up to 98°F.
- By the second pull, timing at redline has dropped to 13 degrees as the intake charge stays hotter.
- By the third pull, with only a short rest between runs, intake air temp at redline reaches 131°F.
- Timing at redline falls to 9 degrees on the third pull, heat soak eating into the tune's margin.
Sample data drawn to show the pattern, not a measurement of any real car.
The numbers behind this graph
| Pull number | Ignition timing at redline | Intake air temp at redline |
|---|---|---|
| 1 | 16 | 98 |
| 2 | 13 | 119 |
| 3 | 9 | 131 |
Fuel octane and knock margin
Octane rating describes a fuel's resistance to autoignition, not its energy content, and it is not a measure of power on its own. A higher-octane fuel resists igniting on its own under heat and pressure better than a lower-octane one, which is exactly the property that matters for knock margin.
Because of that, a tune written to take advantage of a higher-octane fuel can generally hold more ignition timing, or run more boost, at the same knock margin than the same tune would on a lower-octane fuel. A tune is written for a specific fuel, and running a lower octane than the tune assumes removes margin the tune was counting on, which is why a fuel switch, even a seemingly minor one at the pump, is not something to treat as inconsequential on a boosted, aggressively tuned engine.
E85 sits at the far end of this spectrum: it resists knock considerably better than gasoline, which is part of why a tune built around it can often run more timing or more boost than the same engine on gasoline, alongside the extra fuel flow E85 needs for the same power. See E85 and Flex Fuel Explained for how that fuel switch changes the rest of the tune, not just the knock side of it.
Up the left: Ignition timing (degrees)Across: Engine speed (rpm)
- Timing, 91 octane
- Timing, 93 octane
- Below boost, both lines sit at the same timing, since knock is not a concern at this light load regardless of octane.
- On lower octane fuel, timing is pulled back to 11 degrees at this load, a thinner knock margin.
- On higher octane fuel, timing holds about 3 degrees higher at the same rpm and load.
- The roughly 3-degree gap holds steady from here to redline, since boost and load are no longer increasing.
Sample data drawn to show the pattern, not a measurement of any real car.
The numbers behind this graph
| Engine speed (rpm) | Timing, 91 octane | Timing, 93 octane |
|---|---|---|
| 1,500 | 18 | 18 |
| 2,000 | 19 | 19 |
| 2,500 | 20 | 20 |
| 3,000 | 19 | 19 |
| 3,500 | 14 | 16 |
| 4,000 | 11 | 14 |
| 4,500 | 10 | 13 |
| 5,000 | 10 | 13 |
| 5,500 | 10 | 13 |
| 6,000 | 10 | 13 |
| 6,500 | 10 | 13 |
| 7,000 | 10 | 13 |
What tuners check across a full knock log
Timing and knock retard are rarely read on their own. Cross-referencing them against intake air temp, AFR and fuel trims, and any recent change to the car or its fuel, is usually more informative than reading the knock channel in isolation, since the same retard reading can point at several different causes depending on what else in the log moves with it.
A retard that grows across a session while intake air temp also climbs points at heat soak, as the earlier section covers. A retard that shows on one cylinder only, consistently, points at something localized to that cylinder rather than the tune as a whole, an injector, a spark plug, or a fuel distribution issue worth checking directly rather than adjusting the whole tune's timing table in response.
A retard that starts right after a fuel fill-up is worth checking against what octane the tune was actually written for, since a lower-than-expected octane in the tank is one of the more common, easily overlooked causes of a knock pattern that otherwise looks like it came out of nowhere.
| Pattern in the log | Likely cause | What to check |
|---|---|---|
| Timing pulls at the same rpm and load every pull | Not enough knock margin for the fuel and boost in use | Fuel octane, boost target, and the tune's table at that cell |
| Retard grows pull after pull in one session | Heat soak raising intake air temp faster than it sheds | Intake air temp trend across the pulls, and rest time between them |
| Retard on one cylinder only, every pull | A localized condition on that cylinder | That cylinder's injector, spark plug, and any cross-cylinder fuel trim |
| Retard that starts right after a fuel fill-up | Lower octane or a fuel quality issue in the new tank | What octane the tune was written for, and what is actually in the tank |
Questions people ask
- What does knock retard mean in a datalog?
- Knock retard is the number of degrees the ECU subtracts from commanded ignition timing at the moment it detects knock, shown as its own channel on many platforms. A bigger number means a bigger correction; occasional small retard that recovers is a different situation from repeated retard at the same point.
- Why does ignition timing drop once boost comes in?
- Boost raises cylinder pressure and, through compression heating, charge temperature, both of which push the unburned end-gas closer to igniting on its own instead of being consumed by the normal flame front. The ECU commands less timing under boost to keep a margin against that, then tends to hold it roughly flat once boost is fully built.
- How can I tell if a knock reading is real or just noise?
- Real knock tends to repeat at the same rpm and load pull after pull, and often correlates with rising intake air temp or a recent fuel change. Noise from a sensor or a mechanical source, like a loose heat shield, tends to show at random points or on cylinders with no consistent pattern.
- Is a little knock retard normal?
- Brief, occasional retard that the ECU catches and recovers from happens on some tunes and platforms without it meaning anything is wrong. Retard that repeats at the same rpm and load across multiple pulls is different, and points at too little margin at that specific point in the map.
- What is knock learning, and does it matter?
- Many ECUs keep an adaptive table that shifts timing at specific rpm-and-load cells based on knock history over time, separate from the instant-by-instant retard applied during a single knock event. How it works, and whether and how it resets, varies by ECU, so it is worth understanding for the specific platform rather than assuming one behavior applies everywhere.
- Does higher octane fuel let me run more ignition timing?
- Generally yes. Octane rating describes a fuel's resistance to knock, not its energy content, so a higher-octane fuel typically lets a tune hold more timing advance, or run more boost, at the same knock margin. A tune written for one octane and fuel assumes that fuel is what is actually in the tank.
- Why does timing get worse on the third pull of a dyno session?
- Back-to-back pulls don't always give the intercooler and intake tract enough time to fully cool between runs, so intake air temp climbs pull over pull, a pattern called heat soak. Hotter intake air pushes the engine closer to its knock limit, so the ECU responds with less timing, or more retard, later in a session than on the first pull.
- Can a bad knock sensor cause a false knock reading?
- Yes. A failing sensor, a loose connector, or a wiring or grounding issue can all produce a signal the ECU interprets as knock even with nothing wrong in the cylinder. This kind of false trigger tends to show without the repeatable rpm-and-load pattern, or the correlation with intake air temp or fuel, that real knock usually has.
Read next
- Datalogging: What to Log and Why
A datalog is a recording of what the engine was actually doing under a tune. What channels to log, how to read a pull, and the signs of trouble to look for.
- How to Read Datalog Graphs: AFR, Boost and Fuel
How to read a datalog graph: lining up AFR, boost and fuel pressure through a pull, and what a lean spike usually means.
- What Does an ECU Tune Do?
An ECU tune changes the fuel, timing, boost and limiter tables an engine runs on. What changes, how it is delivered, and how tuners verify it worked.
- E85 and Flex Fuel Explained
What E85 is, why it resists knock better than gasoline, what it needs from the fuel system, and why it needs its own tune.
- Intercoolers Explained
Why compressed air gets hot, how air-to-air and air-to-water intercoolers cool it, the physics behind the heat rise, and how to judge one with IAT logs.
Reviewed 2026-09-25