Finding More Power With Knock Control on Pump Fuel 98RON

Following on from our recent Redblock tuning posts, we’ve spent more time analysing the knock profile of this 16-valve engine. The goal here is to share what we’ve learned and hopefully help others running petrol (gasoline) extract more power safely without relying on guesswork. Knock control is part science and part practical tuning. 

After leaving the dyno, something kept bothering me. The audible knock feedback through the Link KnockBlock was unusually noisy. Most engines give you a very clear distinction between knock and no knock, but this one generated so much mechanical noise the audio feedback didn’t match what I expected, and I wasn’t comfortable leaving the ignition map where it was.

Part of that concern came from the very low ignition numbers we ended up with on the dyno. Pulling timing helps avoid knock by reducing peak cylinder pressure, but it also pushes a lot of wasted heat into the exhaust system. That hurts efficiency and loads the engine bay with extra heat, not ideal when heat-sensitive components are close to the manifold. My uncertainty was the motivation to investigate the knock behaviour further.

Understanding Knock and Knock Frequency

Knock can be caused by many things, hot spots, chamber design, piston shape, poor fuel distribution, or simply exceeding the fuels limit for cylinder pressure. 

But the sound of knock that the sensor actually detects, is essentially a high-frequency ringing of the combustion chamber. This ringing correlates strongly with cylinder bore size. Think of it like the size of a bell.

That means we can calculate the engine’s fundamental knock frequency using:

fknock≈1800/π×bore (mm)

With this engines pistons (96.5 mm bore), this engine’s fundamental knock frequency is roughly 5.9 kHz, with harmonics at ~12 kHz and ~18 kHz.

This is important, because no matter how noisy the engine is mechanically, true knock energy always centres around this frequency and its harmonics. Everything else is just background noise. 

Why the Original Knock Setup Wasn’t Working

I initially configured the ECU to listen at the second harmonic (12 kHz), assuming it would isolate knock away from mechanical noise (from previous engine experience, this has worked well). Instead, this Redblock engine produced a huge amount of noise in that range. The knock signal in the PC logs was erratic and inconsistent, forcing me to set very high thresholds just to prevent false triggers. It was impossible to tell which noises were real knock and which were just normal harmonics.

The audio knock detection suffered from the same issue. What sounded like knock at 3000–4500 rpm now appears to have been nothing more than mechanical resonance.

Working Around Haltech’s Logging Limitation

Haltech’s knock spectrogram is an excellent tool, but it’s not included in PC logs. The ECU only logs knock at the single frequency you ask for, not the whole spectrum.

To work around this, I used the Windows Xbox Game Bar (Win + G) and recorded the live spectrogram while simultaneously doing a Haltech PC log. My process for each run was:

1.     Start a normal Haltech PC log

2.     Open the Sensors → Knock page

3.     Begin a Windows screen recording

4.     Perform several 3rd/4th gear pulls

5.     Stop and save both files

This gave me knock frequency data and engine logs. Reviewing them side-by-side was the breakthrough: I could finally see what the knock sensor was hearing across the whole range, not just at one fixed frequency.

Finding the Correct Knock Frequency

The first logs showed why 12 kHz didn’t work: the Signal noise was everywhere, with constant spikes and no meaningful correlation to load or timing.

When I moved calibration down to the first harmonic (around 5.9 kHz), everything changed. The noise floor dropped, and knock events became clearly visible. By deliberately inducing knock, I confirmed that the ECU could consistently detect knock as distinct red bands clustered around 6 kHz.

At that point, I locked the knock frequency to 5950 Hz, and the signal immediately became cleaner and more reliable than anything I saw on the dyno.

Using the Spectrogram to Build the Timing Map

With the knock frequency sorted, I went back to tuning. Over several hours on a quiet road, I performed 3rd and 4th gear acceleration tests and gradually re-introduced timing while watching the spectrogram and the knock signal at 5950hz.

This visual approach proved far more accurate than audio detection. When you’re driving, your brain filters out subtle cues. Reviewing knock visually after each pull is clearer, cleaner, and gives you more time to think.

By stepping in timing slowly, I could visually see:

·       Normal operation

·       Higher combustion noise as timing advanced

·       The point where true knock began appearing

·       Too much timing at low RPM causing knock events

Once you see what too much looks like, its straightforward to find the safe limit.

Before & After Timing Maps

The dyno timing map was extremely conservative because I was led astray with the audio knock system and the haltech knock system was listening in the wrong place. After correcting the knock frequency and tuning on the street, the new map carries significantly more timing in the areas the engine actually wants it.

below are the before and after maps for reference 

The engine now runs cleaner through the rpm range, and timing is several degrees shy of where knock begins. The improvement in efficiency is noticeable. I would still consider this conservative timing, for reference I was seeing ocasional knock around 17 degrees at 4000rpm and 11.6psi which gives me roughly a 4 degree margin. Some tuners will run right to the limit and let the knock control work often, personally I like it as a safety net but not to operate much

Given how much timing was added, I expect a meaningful power increase at this boost level over the original 400 hp dyno result, based on how the engine behaved during the dyno session. Now with better understanding I’m sure it could take some more boost. I’ll post updated numbers if it goes back on a dyno.

I am also tempted to test an octane booster in this engine on a dyno 

Final Thoughts on Knock Control

The Haltech logs show a stable, predictable knock signal. If any cylinder crosses the 20 dB threshold I have set (consistant with the red zones on the spectograph). Haltech immediately pulls timing from that specific cylinder. It also learns long-term per-cylinder trims, reducing timing in any cylinder that tends to approach knock earlier than the others.

normal behaviour shown here

The knock-control system is now behaving as expected, the engine is safer, and more efficient because of it.

Hopefully this process helps others tuning 16v Redblocks on pump fuel. You don’t need hours of dyno time to set up knock control properly with a quality ECU this is achievable on the street. 

Link ECU and Maxx Ecu also offer similar per-cylinder knock learning at a similar price point, Worth considering if you’re choosing a management system.