Forum » Webinar Questions » 083 - Understanding Knock

083 - Understanding Knock

Webinar Questions

Ask questions about webinar lessons here. To see the Previous Webinars for a complete list of archives tuning webinars. 

Page 1 , 2

In case of running on E100, is possible to occur knock or E100 is susceptible only to pre ignition? And in what conditions the knock on E100 occur?

On E100 (or E85 for that matter) knock is pretty unlikely but not impossible. The sort of situation that would cause knock would be very high boost levels coupled with extreme inlet temperature, or potentially a high compression ratio in a boosted application. Our FA20 with 12.5:1 CR for example will knock on E85 if you try hard enough.

Ethanol is however more susceptible to pre ignition than a gasoline based fuel and this tends to get worse with richer mixtures.

Andre, can you put together a webinar on how you analyze knock from the plex knock monitor with an audio program like gold wave. I would love to see how you accomplish this. If I could listen to A saved audio file from my plex while reviewing my data log I would find it very useful. I would like to log data while drifting as it is hard to simulate the load/ temps on the dyno or street. Thank you

If you're going to use software such as Gold Wave to analyse the signal from the knock sensor you'd generally want to do this with the raw signal rather than one that's already been run through a digital signal processor. Essentially you use the software to perform a FFT analysis of the signal which allows you to see the signal amplitude vs frequency. With a sample that contains knock you can then see what frequency knock occurs at and separate it from the background engine noise.

I'll consider a webinar on the topic as it could prove quite useful to many.

Although nothing can beat one of Andre's webinars for clarity and understanding, MoTeC have a really good example of this in the manual for their SKM and OKM manuals.

Thank you both for replying

Hey Andre,

I was wondering if there is a significant difference in knock sound for a hemi head versus a wedge head? I have a ka24e (early USDM 240sx motor, wedge head) that I have been tuning and when I listen with knock headphones it has that rocks on a tin sheet sound even with extremely retarded timing (5* btdc at NA load). I briefly tried turning the ignition up to about 30* BTDC in an effort to induce knock and learn its sound. I can hear a different noise that is lot more like a coin being tapped on a windshield. Is it likely that the "rocks on a tin sheet sound" is just the sounds of a high mileage motor and that the "coin being tapped on a windshield" is the actual knock? Or are they likely both knock, just one more severe than the other?

It should also be noted that my knock sensor is mounted about 2/3 of the way down the side of the block due to there not being a factory knock sensor mounting location.



@Stingy, there are subtle difference to the sound you'll hear on different engines but for a given knock detection system it's always going to sound pretty similar. I'd be very surprised if you're truly seeing knock with 5 deg advance on a N/A 240 motor unless you're running on some absolute garbage fuel. I would follow the technique of advancing the timing at high load and low rpm until you can induce knock and hear it clearly.

Another tip I often use if I'm not 100% sure if I'm hearing knock is to start by retarding the timing 5-10 degrees and see if that removes the sound. I wouldn't however suggest that technique given you're only at 5 deg now. This also assumes that your base timing is currently correct.

Hello Andre,

I am tuning a heavily knock limited 2006 JDM legacy GT Spec B via open source and depending on the stock knock control. I am running pump 90 octane. I would like to know what you consider acceptable limits of ignition retard for fine knock learning because I don't want to keep chasing my tail to have the learning table populated with 0s with each tank of fuel. The IAM is always 1. The A/F learning is within 5% and actual AFR at 10.5 to 10.8 at maximum load (it doesn't like 11.0). I don't see any feedback knock at WOT during tuning.

Hey Brenton, you're really up against it if you're trying to run reasonable boost with such low grade fuel. It's not necessary to have the knock control system completely silent in order to provide a safe tune as some occasional knock is acceptable. Rather than an amount of retard that I find acceptable, I look at where the retard is occurring. If you are consistently getting knock retard occurring at the same point in the ignition table then you should address this. On the other hand if you have the occasional knock event that's not repeatable then I'd be less concerned.

I am already running a very conservative timing map as the IAM was suboptimal on the factory tune with repeatable knock events all over. The factory ignition and fuel maps are also very erratic with spikes and dips which corresponded to areas where knock occurred. I have refined those maps with good results. In this event I have opted to add a little fuel (~5%) as opposed to pulling timing. The HPA courses are the best thing since sliced bread.

The other thing I'll add with any Subaru tuning is that it's important to understand the interaction between the base ignition table, ignition advance table, and IAM (which it sounds like you are on top of). I quite often will reduce the values in the ignition advance table and smooth them and then do my ignition tuning in the base ignition table. I find this gives me a more predictable outcome in terms of final delivered timing and a smoother torque/power curve.

I have a similar approach but I smooth and tune the advance table to retain the factory logic of the AVCS timing maps. There r four base maps, cruise and non-cruise with each having an AVCS Related map while there are only 2 advance maps. This way I still have the protection of timing optimization based on AVCS activity. The factory logic appears to run more timing under cruise during AVCS warm up and less timing under non cruise conditions in the same situation. The additional 5% fuel has contained the knock.

Andre, et. al...

a few questions about knock as it relates to the webinar...

1. assuming an engine is MBT limited in it's naturally aspirated (NA) state is it wrong to assume it's MBT limited in an force-induced (FI)/tubro'd/supercharged state? Or is it more dependent on how much boost / compression is within the cylinder that pushes it towards knock limited?

2. I've heard the 'coins in the glass tray' audio file provided in the tuning that particular sound representative of full-on knock versus the initial onset of knock? If I'm at a point in my ignition timing I'm getting (as best as I can describe) a electrical-static type pulses similar to the aforementioned 'coins in a glass tray' audio but not nearly as intense or as long in duration does that mean I'm at the onset of knock?

3. Since you've posted a audio clip of knock, can you post an audio clip of what a normal operating engine sounds like?

for what it's worth its my tt VQDE around 10psi, around 6k rpm by means of a tapped & amplified my OEM knock sensor signal.

Thank you in advance

Hey James, here's some answers to your questions:

1. I'll use the term 'knock limited' which I define as an engine that begins to suffer from the onset of knock before reaching MBT - i.e the torque continues to increase but the engine begins to knock. If this sort of engine is knock limited in N/A form it will almost certainly be much worse in a F/I form since we are now increasing the heat and pressure in the cylinder.

2. Every engine sounds a little different and it will also depend on what device you're using to listen to the engine since some, like the Plex Knock Monitor, offer filtering that changes the sound profile of the engine. I'd suggest trying to run the engine at low rpm and moderate load and advance the timing sufficiently to generate knock that you can clearly distinguish. This will let you distinguish more clearly what you're hearing on your specific engine and with your specific knock detection system.

3. I don't have any audio of typical engine noise and as above, this will differ from engine to engine and system to system. Have a listen to the samples at the Knock Box website as their samples include low timing periods with no knock, and then advanced timing that generates knock.

Hi Andre,

I am working with an OEM knock control system on the Ford Ecoboost. This platform's knock response control (if you are not already familiar with it) will allow you to set the max advance and retard allowed, the 'steps' of how large of a change each advance makes, and the time between advance events as you rev through the RPM range.

The system seems to be very sensitive, and you can define how much timing you want pulled based on the knock intensity. I took the OE numbers and made them more aggressive just to be on the safe side (pull more timing at the various knock intensities)

My question is that I will see all sorts of different knock events, in my tuning I am basically ensuring that I do not have any events that either A: cause a LARGE removal of timing (anything over 1.5 degrees or so) or B: if I see a reduction followed immediately by another reduction or C: see a trend that is flat or negative - or knock across multiple cylinders at once. I want all cylinders to trend positive advance as we go through RPM.

As it stands right now I am seeing timing advance through the revs, here and there I will see a small event, removing 0.3-0.5 degrees and then it will continue to advance again. I am calling this safe.

When you say you want to eliminate knock, are you stating that even these small events, on one cylinder, and then the timing continue to climb would still be unacceptable - or are you defining safe as no consistent knock across say...all cylinders which would indicate for sure you are over advanced? I am just unsure if by safe I should NEVER see any knock, even small, random events (as in only in one cylinder). Past datalogs I have sent in for revisions for pro tunes have had more knock activity than my maps and they seemed comfortable with it so I just want to get my head straight on what should be deemed safe.



For reference for anyone who would like more info on how the Ecoboost knock system works, these guys have some great articles:

In general it's not actually desirable to eliminate absolutely ALL knock activity. This would result in a tune that is almost certainly overly conservative. What we want to avoid however is consistent and repeated knock that is occurring in the same places of the map. Beyond this though it's also advisable to liste using audio knock detection equipment to ensure that the knock the ECU is logging is actually real.