049 | Setting Up Knock Control - Link G4+/ViPEC
Knock or detonation is the most common cause of engine failure and damage in any high performance engine. In this webinar we will look at what detonation is, what causes it, and how we can use closed loop knock control to protect an engine. For this webinar we will be using the Link G4+ ECU fitted to our Nissan 350Z
- It's Andre from the High Performance Academy, welcome to this webinar where we're going to be looking at setting up and configuring knock control, closed loop knock control on the Link G4+ and Vi-PEC platform of ECUs. Now the information we're going to be looking at in this webinar is going to be applicable to a fairly broad range of ECUs anyway. The knock control strategies across a lot of ECUs are very similar in their functionality and the way they're set up so even if you're not tuning on the Link/Vi-PEC platform, there's still going to be a lot of valuable information in here. As usual we will have some questions at the end of the webinar so if you've got anything that you'd like to know more about or anything I cover during the webinar you'd like me to elaborate on, please feel free to ask that into the chat box and we'll address those at the end of the webinar. Now before we move into setting up the knock control, I just wanted to go back and talk a little bit about what knock is, I find it's always easiest to understand how to set up a function in an ECU if you have a better knowledge of what's actually going on inside the engine and what we're trying to make that function do so understanding knock is really important.
So knock or detonation is an abnormal form of combustion and to understand that we need to have a quick talk about what normal combustion is and how that looks. So most people sort of think that when the spark event occurs, we get an explosion inside the combustion chamber and it's not really the case, what we actually get under normal combustion is quite a slow combustion event and we get a flame front that begins at the spark event and it propagates out through the combustion chamber and as it moves through the combustion chamber it ignites the unburned fuel and air ahead of it. So we get a relatively slow flame front and we get a fairly smooth increase in our pressure inside the cylinder. Now knock or detonation on the other hand is an abnormal type of combustion and that's where pockets of end gas around the outside of the combustion chamber, so pockets of fuel and air, it gets to a point where the temperature inside the combustion chamber is so great that those pockets of fuel and air spontaneously combust. Now this time it is a little bit like a stick of dynamite exploding, we get a very very violent explosion, we get a violent burn and we get large spiking in the pressure inside the cylinder.
We want to stay away from this situation because the pressure spikes that occur with knock or detonation can strip away the boundary layer of gases that protect the cylinder head and the piston from the full heat of combustion. And that lets that full heat of combustion contact the piston so even in light detonation, sustained light detonation, if you're looking at a piston that's come out of an engine that's been suffering from that, it's likely to have a sand blasted appearance to the piston crown. In really extreme situations, we quite often see the whole side of the piston missing, it's like someone's taken to it with a gas torch. At the same time the pressure spiking is acting on the top of the piston and often under knock conditions, we will see maximum cylinder pressure reach or exceed double the sort of pressures we're seeing under normal combustion. Now all of that pressure's being transferred through the piston into the connecting rod and down into the crankshaft and that extreme pressure spiking, if your engine's fragile can be enough to damage components.
Another key point you'll often see in an engine that's been suffering from detonation but perhaps hasn't actually failed is you might see that the big end bearings will start to have a hammered appearance and that's because the pressure spiking is kind of like hitting the top of the piston with a sledge hammer and that can end up having the big end bearings, the conrod bearings actually contact the crankshaft journal and it does damage that way. So the whole thing, it's all a lot of really ugly stuff that we must stay away from any time we're tuning. Now I just want to, before we move on I just want to discuss as well, there's 2 very different situations, we've just discussed knock or detonation. Now often that is confused with another abnormal form of combustion called pre ignition. Now those 2 types of abnormal combustion are very very different, pre ignition is not detonation.
Detonation or knock occurs after the spark event has occurred and it is caused by the pressure and heat inside the combustion chamber following the combustion event, beginning, rising so high that the fuel can no longer, will spontaneously combust. So pre ignition on the other hand occurs before the spark event occurs. So that's a very clear distinction between knock and pre ignition. Pre ignition is actually a much more damaging situation but I'm not going to delve into that today. OK so how can we detect knock or detonation in our engine? So most modern engines now are equipped with a knock sensor.
Or in the case of V configuration engines or inline 6 it's quite often multiple knock sensors. And these are a piezo electric sensor that's bolted somewhere to the engine block and what they do is they respond to noise or vibration being transferred into them through the engine block and they produce a voltage relative to the vibration they're being subjected to. And what we find is that knock occurs at a reasonably specific frequency, often very closely, or not often, it is very closely linked to the diameter of the piston or the diameter of your cylinder's bore. And I'm going to look at that shortly. So that's how the ECU will get a signal for what's going on inside the engine in terms of engine noise and in terms of whether knock or detonation is occurring.
Now there are a couple of different types of sensor and I wanted to just talk about those because the type of sensor that you're using will have quite a large impact on how you set up your knock control strategy. In the aftermarket probably one of the more common sensors that we would use is the Bosch donut style knock sensor. Now that is what's known generically as a wideband knock sensor and what that means is it will respond to a very wide range of frequencies so it's useful for us tuning because we can use it on multiple engines and it's going to work quite well. The other type of sensor that we see in a lot of OE applications is a narrowband sensor. Now a narrowband sensor is often tuned to a certain frequency range and the reason OEs do this is to improve the signal to noise ratio and it's really common on a lot of late model engines, particularly late model Subaru and Mitsubishi to use what's called a second harmonic sensor which is tuned to the second harmonic or second order harmonic of whatever the current knock frequency is for your engine.
I know that's a mouthful so I'm going to explain it. So let's say for example, we'll look at the calculation shortly, let's say for example that the knock frequency for our particular engine works out to be 6 KHz or 6000 Hz. So that's the frequency that the block will resonate at due to a knock event occurring. The problem with that is we'll also have a lot of other noise going on inside the engine and a lot of noise, mechanical noise, pistons contacting the bores, valve train noise, that sort of thing. So the engine is obviously a very noisy place and a lot of that noise generally tends to happen at lower frequencies kind of close to that 6 KHz.
That can make it quite difficult for us to distinctly tell if the engine is knocking or whether it's just background noise. Now the OE manufacturers know this so instead of looking at 6 KHz, they look at the second harmonic frequency which is simply double the initial frequency so 12 KHz and often what we find is that will give a better signal to noise ratio. So what that means is we get a more distinct knock signal relative to the background engine noise. So it's important to understand that a lot of late model cars will be equipped with one of these second order harmonic sensors and that if we're using one of those with an aftermarket ECU we do need to choose a suitable frequency and we're going to look at how to do that shortly. Now if you've got no idea where to start, how do you actually decide on what knock frequency you're going to want to use? OK there's quite a handy calculation that you can use that's going to get you really close to the correct knock frequency for any engine and all you need to know is the bore diameter.
Now this isn't going to be 100% bulletproof but it's definitely going to give you a good place to start when you're tuning your knock control system. So if we know the bore diameter all we need to do is multiply the bore diameter by pi, 3.14 and then we look at that answer, we divide 1800 by that answer and that will give us the approximate knock frequency for the engine so on our laptop screen now we'll just go through that calculation for our Nissan 350z. So we have a bore diameter of 95.5 mm, sorry just to be clear as well this frequency calculation does work in mm, not inches so if you're working in inches you'll need to convert that into mm. So we have 95.5 mm, multiply that by pi which is 3.14 and that gives us 299.87. Now if we divide 1800, and that is simply a constant in this equation so you don't need to know where that's coming from, it's just simply a constant, if we divide that by 299.87 we see that our answer is 6.0 so what that means is the knock frequency for our VQ35 engine is likely to be approximately 6000 Hz or 6 KHz.
Again it's important to understand that calculation works in KHz or thousands of Hz. OK so again if we know that then we know that we can look initially probably at 6 KHz and then if we're not getting a good signal from our knock control system there we can try looking at double that frequency so 12 KHz. In a later webinar what we will look at is some analysis of a knock audio trace using a fast fourier transform to actually look at the specific frequencies that knock is occurring at. But for the moment, for this particular webinar we're just going to focus on that basic calculation. So let's get our engine running and what I'll do is we'll just talk through what our options are in the Link G4+ platform and how this knock control function works.
So if we go to our knock setup and we have a range of parameters here that we need to set up and I'll work through them individually so first of all we have our knock mode and if we double click on this we can turn it off, turn it on or on the older G4 platform if you wanted to use knock control you had to interface with an external digital knock interface. Now the G4+ and Vi-PEC platforms use an onboard digital signal processor so we can use the internal knock. Next we have our frequency channel so that's exactly what we just talked about. So we have the ability to select a range of filters so this simply adjusts the internal digital signal processor that the ECU is using so this signal processor focuses on whatever range of frequencies that we choose here. So we can choose a range of wideband frequencies, so anywhere between 4 and 10 KHz or 10-16 KHz.
Now that will work over a wider range as its name implies but the problem with that is we're also going to be likely to pick up a lot of background noise so wherever possible I would suggest you use a narrowband filter and then we have those narrowband filters here and you can see at the moment I am, I've selected the 12 KHz range which I know works well for this particular engine. Now we've got a gain channel and I know a lot of people have asked me in the past what this gain number means. And the gain number simply is an overall multiplier for that noise signal that's coming from our knock sensor. So what we want to do is set the gain so that we're getting a good range of values from our knock control system. And in general in the Link G4 platform, what we were looking at on our knock numbers, so the numbers that the ECU is actually representing is coming from the knock sensor.
The range for those is between 0 and 1000 and generally what I like to do is choose this gain number here so that at wide open throttle and at maximum RPM so on our rev limiter or at our rev limiter, we're seeing somewhere around about 400 to 500 as our maximum number. Now it's not set in stone, we don't have to have exactly that. What you'll find is if that gain number is too low, what we're going to get is very very low numbers in our knock values, even where knock isn't occurring. That makes it very difficult for us to set a threshold accurately, we don't have as much resolution when we're setting that threshold so for example if we are getting maximum peak value numbers in our knock at 7000 RPM of 20 for example, and we want to set our threshold 10% above that, it's quite a tight threshold. If for example our maximum number was 500 instead of 200 obviously that gives us a slightly better resolution.
So a good rule of thumb is if we're seeing maximum values with no knock occurring of 400 to 500, that's a pretty good place to be. The flipside of that is if our gain is too high and we're getting numbers that go off the limit of 1000 with no knock occurring, obviously it makes it impossible for us to tune the system. OK we've got some ignition trim tables that the ECU will store so the Link system works quite well in my opinion, it's quite smart so what it will do is it will trim timing out of the individual cylinder tables when knock occurs. And when that knock stops occurring, what it will then do is it will advance the timing back into the system and what that means is if we get one random occurrence of knock that may be due to very high intake air temperature or something similar, then when that air temperature reverts back to normal and we don't have knock anymore, the ECU will start feeding that timing back in so we can clear these ignition timing tables, trim tables when the ECU's powered up, or we can turn that function off. Next we have our knock window, now this is the window during the combustion cycle where the ECU will actually be looking for knock to occur.
So it's not looking through the entire engine cycle, it's only looking at a narrow window where it knows to expect knock. So in this case the window starts 10° after our ignition event and it continues for 50° of crankshaft rotation. In my experience with setting up knock control systems on the G4+ and other ECUs, this general size of window, these parameters seem to work fairly well across the board so these are the default values and I haven't seen any need to change them. We then have our ignition retard limit. So this is how much ignition timing we're going to allow the ECU to remove if it detects detonation.
Now generally I suggest setting this relatively conservatively. It's important to note here that the knock control system is not a bandaid for poorly tuning our engine. OK this is a safety backstop in case we get a poor tank of fuel for some reason that's low on octane or something goes outside of our normal bounds of operation and results in knock that we're not expecting to see. So for that reason we shouldn't need a large amount of ignition retard control and generally I'd set this somewhere to around about 5° so if you need more ignition retard than that, then you've probably got bigger problems to address in your main ignition table. We've got the retard gain which defines how aggressively the ignition timing will be removed in response to a knock event.
So this really depends on how far above our knock threshold our knock values are getting. So again we'll see how this responds at the moment, this is just the default value in the 350z package of 0.2° per percent and that's worked reasonably well. Now I said that the G4+ system, the knock control system will also advance the timing back in once the knock event has been removed. And we've got 2 functions here that define how that works, we've got our advance delay, so this is how long the ECU will wait with no knock occurring before the timing will be advanced back so in this case we've got it set to 0.5 seconds. And then the timing itself will be advanced at 5° per second.
Next we have our lockouts for RPM so in this case we're not going to allow the knock control system to work below 1500 RPM or above 7000 so you can set these to whatever you want to set them up to. The system, when it's tuned correctly though is very very accurate and works very well so there's no real need to not have these set up across the whole operation range of your engine. Really the area we're going to most likely want to protect the engine against is knock at high RPM and high load. Likewise we have a low throttle position lockout, so in this case if we're below 5% throttle, the knock control system will not work. OK so that's our basic setup and obviously I've already got some parameters set for this particular car that are going to work reasonably well.
So how do we actually go about setting up the knock control system? First of all, the basic setup that we've just looked at there works in conjunction with our knock target. So this knock target is the threshold of noise values above which the ECU will decide that knock is occurring. So this is our limit, this is our line above which the ECU will start retarding timing and this is the important part for tuning. So what we want to do is set this up correctly when the engine isn't knocking and then we can test to see how well the system is working when we induce some light knock. OK so what we're going to do now, I've got a system set up at the moment which hopefully we're going to be able to let you listen to what knock sounds like when it's occurring.
So to help me test this system I'm just going to have a listen too while I'm running the car through my headset. It might just be worth mentioning we're using the Plex Knock Monitor system here which is what you're going to be listening to. And I've got that set up at the moment, a lot of people I know who use audio knock detection systems have invested quite heavily in aviation style noise cancelling headphones and I've tried those in the past and I've actually found the best results I get are simply with a cheap set of iPhone earbuds and I use a set of grade 5 plus Peltor ear muffs when I'm listening to knock for tuning and I've found that that gives me the best way of actually audibly detecting knock and getting rid of all of that background noise. OK so what I'm going to do is I'm going to start our engine running on the dyno and I'm going to get into 4th gear. And what I'm going to do, when I'm setting up either closed loop knock control or even for that matter when I'm just configuring my own audio knock detection system, what I like to do is induce some light knock at low load and low RPM and that way I can audibly listen to make sure that I'm being able to distinguish between knock and the normal background engine noise.
So right now what you can hear hopefully is just what we would normally hear with the engine running, so this is normal background engine noise, we've got no knock occurring. What I'm going to do is go down to the wide open throttle areas of my ignition table here and what I'm going to do is purposefully select those 2 zones under wide open throttle and I'm going to add 15° of ignition timing to them which should be enough in my experience with this engine to create knock or detonation. Then what I can do is on the dyno I can just open the throttle into those areas and that should lead to knock or detonation occurring which we should be able to hear. And hopefully you've been able to hear that clear distinction between normal engine noise, it's a very very big distinction. Now while I've been running that as well, let's just have a look, this is our data logger on the Link ECU and what I've got it set up at the moment is here we've got the knock level from each of the 6 cylinders being monitored and at the same time we've also got at the bottom our knock threshold so that's our background noise level above which we're going to say that the engine is knocking.
And down here below that we've got our knock trims. So this is a good way of setting up the logger so we can see exactly what's going on with the knock control system so what I'm going to do now you understand what's going on there, I'm going to just go to wide open throttle again and we should see those noise levels in our logger spike. OK and we've seen exactly that, what I'll do is I'll just stop the logger. Now when you're seeing the log trace cycling along like that it can be a little bit hard to see exactly what's going on but if we look at what happened there, we're sitting at 2000 RPM and our normal background noise level is shown here so the values are relatively low, they're in the 10 to 20% numbers. However in this area here where we advance the timing up you can see that we do start seeing those numbers increase, we're actually picking up that knock occurring.
Now at the moment we've got the knock limit set to 0° of ignition advance so that ECU isn't actually able to retard the timing in response to that knock occurring. But just for an example of how effective that is, let's just go back into our knock setup and what I'm going to do is I'm going to give the ignition retard limit, give the ECU some control, I'm going to set the ignition retard limit at 15° and we're going to redo our little demo there and this time we'll hopefully be able to see the ECU pulling timing out in response to that knock event. It's important to also mention right now my knock threshold isn't accurately set up, it's probably a little bit high for this particular RPM so it's not picking up the knock, it's not detecting the knock as quickly as it should. OK so let's just stop the logger there and we'll have a look at what's actually going on in that area. So you can see first of all we've got our noise starts spiking up and it starts spiking up above our knock threshold so that indicates to the ECU that knock is occurring.
And in the ignition retard table, the ignition retard group down below you can see that the ignition timing is being retarded on these individual cylinders in response to those knock events. Now when we've go this set up nicely in steady state we can actually run the engine on the dyno and we can go to full throttle with the ignition timing purposefully slightly over advanced and what we'll see is the engine will begin to knock and we'll instantly see the ignition timing being pulled on those cylinders where knock's occurring and then once the knock is removed, you'll see the ignition timing being advanced back and then obviously knock will start occurring again. So you see the cycle of the ignition timing being advanced and retarded automatically in response to knock. Now in our logger here you can see, if we look at this particular trace here, you can see that initially it pulls quite aggressively 14° of ignition retard but you can see also that straight away that ignition, that knock event is removed and in response to that the ignition timing then steps back up, exactly as we had programmed it to do. OK so that's how the knock system works.
We just lost audio here so I'm just going to talk you through what I was doing in the live webinar. So we were talking about choosing the knock frequency channel and we'd started here with this demonstration with it set to 12 KHz narrowband which was the second order harmonic for the frequency we calculated for our engine so that was 6 KHz. So for the demonstration we've just changed this back to 6 KHz which is what I'm doing right now, so this is the first order harmonic for knock in this engine and then on the dyno I just did a quick test again in steady state at 2000 RPM and the idea behind choosing this frequency is to look at which particular frequency we get the best signal to noise ratio so which frequency shows us most clearly the difference between background noise and knock and we're just bringing the car up to 2000 RPM now and we'll just induce some knock by going to full throttle. And that's what we've got there. That time you would have been audibly able to hear the engine detonating and if we compare to our last log file, first of all just take a quick snapshot of the knock we're seeing, the knock levels we're seeing here, we've only really got this one event here where we've exceeded our knock threshold.
Despite the fact we could audibly hear knock occurring. If we go back to our last log here. And you can see in comparison we've got a much higher knock threshold. So that's how we're deciding whether or not we're going to use that first order or second order harmonic and of course we could also go through and test at a range of different frequencies to see what ones going to actually give us the best result Now I'll also point out that here on my Plex Knock Monitor we have the ability to decide on what frequency we're going to listen to knock. So here you can see I've got it set up on the high pass 5 KHz filter and we can change through to all of the low pass narrowband filters as well.
Now I've found just through my own testing with this particular engine that the 5 KHz high pass filter actually gives a better signal to noise ratio so we can audibly hear the knock a lot clearer. OK so we've look at actually distinguishing knock in our threshold table there. However what we want to do as well is see how to do that under a wide open throttle ramp run. The system is basically the same, what we're trying to do is run the engine under wide open throttle in a no knock strategy so we want to run the engine with no knock occurring. And we want to look at the natural background noise level that we're seeing happening when we've got no knock so obviously as the engine RPM increases, generally there is a trend for the background engine noise to increase.
So we want to see what that signal looks like and then we want to use that to help us program our knock threshold. So we'll look at a run now with no knock occurring and we will see how that works out. Now unfortunately we've got a bit of a glitch in our Plex Knock Monitor and despite a little bit of testing earlier in the day, we find that under high RPM running the Plex actually induces some noise that is audible both through the headset and you'll be able to hear it as well so it's an interference that unfortunately makes it really hard to hear what's actually going on so I'll disconnect that because it's not going to help you, it's only going to serve to confuse. So what I'm going to do is start by removing that 15° that I just put into the ignition table because obviously that will create knock at 2000 RPM. A lot of people would think that if we want to get a noise signal where we've got no knock occurring, that's easy, why don't we set our entire wide open throttle running area of our table to something very retarded like perhaps 4 or 5°.
Now the theory seems sound but what we find is that that background noise threshold that we get is also relative to the amount of power the engine makes and the whole noise signal of an engine, so even if you're audibly listening to an engine running on the dyno, not through knock detection, just listening to it running, you'll hear the engine is very flat when the ignition timing's very retarded and we see exactly the same thing happening with our knock signal so we can't just go and set a very retarded ignition timing, we still need to have the timing set relatively close to our normal running conditions. So I know that the timing that I've got in the engine at the moment is relatively safe, what I'm going to do is just start by highlighting the whole wide open throttle running and I'm just going to take 2° of timing out. Now that's not going to make the engine note go particularly flat but it's going to just safeguard the engine, I know I'm not going to get any knock. Now what we're going to do is log a wide open throttle ramp run and we'll see what that background noise threshold looks like. So let's do our ramp run now.
OK so we've got our ramp run done there, we're not connected up to the dyno for the video side of things because it doesn't really matter what we're interested in, it's the logger so I'll stop the logger now. And now on our graph we can see we've got our individual cylinder knock levels coming through and you can see also this red line here is our background or knock threshold so any time our knock levels exceed this threshold, that's where knock's going to occur. Sorry any time that threshold is exceeded that's when the ECU is going to decide that knock is occurring. OK now this is a run where we haven't had any knock occur and we can use this as a guideline for setting up that background knock threshold. Now one thing I want to talk about is our individual cylinder knock gains.
Remember if we look at our knock setup sheet, we have our gain channel here which I've already talked about a little bit, that sets the overall range of our knock values so if we double this gain value, what it does is it doubles the numbers that we're seeing from our logger and you can see that our peak numbers here are close to 400 which I use as a pretty good guideline for where I want the maximum RPM knock values to be. But what we also can see is a discrepancy in the knock levels from one cylinder to another. And what we will find is that some cylinders possibly due to their proximity to the knock sensors will tend to give a higher knock reading than other cylinders. So if we go into our cylinder setup here, what we can see is we can actually set up the gains for the individual cylinders. So if we've got a cylinder which is for any particular reason giving a much higher signal level than another cylinder we know knock's occurring, we can then go through and adjust that particular cylinder gain to get it in line.
So the idea behind it is under no knock conditions we really want all of the knock levels to be relatively similar and if we've got that occurring it's going to mean that we can set our threshold tighter to that background noise level and that's going to allow the ECU to do a better job of picking up individual cylinders knock occurring. So in this case here you can see the spike which is on the purple line which is cylinder 5. In this case spikes up to 365, it's a little bit higher than the other cylinders and I would probably then reduce the gain on cylinder 5 here which I've got set to 0.8 at the moment so I'll take that down to 0.7. So the idea where we're setting these individual gains is just to even out the noise levels on the different cylinders. So once we've got a no knock run like this logged, what we can do is use it to set our threshold.
And as you can see I've got a threshold set already which is reasonably close. What I can do now is I could probably pull that down a little bit closer. So let's go and do that now, we can go back into our knock threshold table. What I'm going to do just for the purposes of this demonstration is I'm going to highlight the whole table and I'm going to reduce it by 10% by multiplying it by 0.9. I'm going to save that and then I'll go back into my knock.
I'm going to go back into my knock setup and I'm going to give the ECU some ability to retard the timing in response to a knock event. So we'll save all of that. And now what I'm going to do is up at 5500 RPM here, I'm just going to advance the ignition timing by 10° and we'll do another run now and see how well the ECU's able to pick up that knock event occurring. OK so that's worked perfectly, if we look at what we've got on our logger here, remember I just reduced the knock threshold by 10% and you can see now our knock threshold is sitting a lot closer to that background noise level. And you can see that through the main part of that run there we had no knock occurring which is exactly what we'd expected, we haven't changed the ignition timing.
Remember however in that high RPM area of the table I had increased the ignition timing and you can see that the knock level has spiked up massively above our knock threshold there and we've ended up pulling ignition timing out in response to that. So again it's acted exactly as we would expect. Now the other thing that I'll point out as well is, I'll just shut the engine down now. The other thing I'll point out as well is we can allocate under our cylinder setup, a knock trim table for each cylinder. So this is the individual cylinder aspect of it.
The ECU is able to detect which cylinder the knock is occurring on and remove timing just from that particular cylinder. Now this is one of the real advantages with this system is it's not an over the board, overall trim and we will normally find that one or perhaps 2 cylinders in an engine are more knock prone than others and will start knocking earlier. So if we're applying an overall trim to our ignition timing to help keep those particular cylinders happy, then what we're doing is giving away power potentially on the other cylinders which will accept more ignition timing. So the knock control system will only remove timing from those individual cylinders that are knocking and here's where we define which cylinder, which knock control which knock trim table is allocated to which cylinder. Unsurprisingly in this case I've allocated them in their table numbers to their cylinder numbers so it's nice and easy to keep track of those.
And if we go into our knock trim tables, so what I've done there is I've just opened the ECU settings menu, you can see I've just typed KN which starts bringing up everything relative to knock and you can see we have our individual knock trim tables here. So let's click on trim table 1 and if we've run the engine on the dyno and we've got some knock occurring, then what we should see is the result of that knock will be logged in these tables, in this case that knock event was actually, which occurred around about here, was actually removed before the table, before I've shown you this table which is why it's showing 0. Right what I'll do is I will just do another demonstration now, just really quickly on how this knock table, these knock tables work because they are quite a useful function for our tuning while we're on the dyno. So what I'm going to do is again I'll just go back to my ignition table, we'll do this in steady state and what I'm going to do is induce some light knock as we've already looked at and I'm going to show you how we can pick up where that's occurring in our knock control tables. So we'll just get the engine back into 4th gear and I'm going to just trim out that 10°, oh no I'm going to set that all to 10°, doesn't really matter.
What I'm going to do is add 15° to our 2000 RPM running point and we've just got our logger running so I'll increase the throttle opening until we actually get some knock occurring. OK it's not actually going to show me what I want to show you because the knock is being, the ignition retard is being removed too quickly so what I'll do, is just make another change here. Really want to be able to show you this 'cause it is quite useful when we're tuning. We'll just set our decay to 5 seconds for the time being and we'll go back to our logger. And we'll do exactly the same again.
And I'll just add another 10° there just so we can be sure we've got some really aggressive knock occurring. Now hopefully if we go back to our knock tables in cylinder 3 and cylinder 1, cylinder 6. OK perfect, it actually has worked which is great, always pleasing when something works how I want it to. Right so what we can see here is as a result of the knock event that we had occur here it's held these knock values here which we can now see. So we've got 5° being pulled out at 100 kPa and 2000 RPM.
So when we've tuned the car on the dyno, we can use these knock tables to see what knock was occurring and what cylinder's it was occurring on so it's quite a powerful tool. Alright so this brings us really to the end of the lesson and I just want to recap a little bit because there has been a lot that I've discussed here, it is a complex function and I just want to cover over again, reiterate what we've gone through. So first of all, what we've done is we've set up our knock frequency, that's really the important part there and I've given you that calculation to start off and figure out approximately where the knock frequency will be for your engine. So remember that was 1800 divided by pi multiplied by your piston diameter in mm. So in that case we found our base frequency was going to be 6 KHz and we've talked about the fact that a lot of OE engines are now equipped with second harmonic sensors which are tuned to that second order harmonic.
So in that case, double the frequency. So we looked at how easy it is though to adjust that knock frequency in the G4+ and then just simply do a test under steady state at low RPM and look at the signal to noise ratio and decide on which frequency is giving us the best signal to noise ratio. So which frequency is best able to show us knock occurring. So that may not always be at the frequency that we first calculate. OK once we've done that then the technique is to set our knock threshold to be somewhere just a little bit above our background noise level throughout the RPM range and we again talked about how that background noise level will increase with engine RPM so we need to get a sample of what that noise looks like with no knock occurring and set our threshold a little bit above this and when I say a little bit, generally I would start with around about 10% but it's very easy to test, what we want to do is do some runs on the dyno or out on the road and test that when we're audibly hearing no knock occur, that we're never exceeding that knock threshold, that's when we know that we've got that threshold set up safely above our background noise level.
Then what we want to do is induce some light knock and make sure that the ECU is able to correctly identify that knock occurring. So once we've got those 2 out of the way and we've got our knock threshold correctly tuned the ECU's going to be able to accurately and quickly detect knock and identify that above background engine noise. We've also talked a little bit about how we can use both the overall gain channel to control the level of our entire signal and again we want to see that maximum value sitting somewhere in the region of about 400 to 500 at wide open throttle at maximum RPM with no knock occurring. And then we also talked about how we can set the individual cylinder gain levels to even out the noise level that we're seeing from individual cylinders and again when we've got no knock occurring we want all of those levels from all 6 cylinders to be approximately similar so that the ECU again can do a better job of detecting knock on individual cylinders. Right hopefully that's all been reasonably clear and I hope that the demonstration has improved your understanding of how the knock control system works.
Particularly I hope that the audio component has helped you understand what knock sounds like through an audio knock monitor. Right so we'll get into these questions, we've got one from Chris, any chance you could also cover the difference from over advanced and over heating knock? Look to be honest Chris, I don't know if I've really ever found an audible difference from those two. Knock essentially is going to be knock, if your own experience is different, I'm keen to hear about that. Of course the other thing about that as well is when I'm tuning a car on the dyno or on the road, I'm being reasonably careful to ensure that I'm always tuning at a normal operating temperature so I'm not allowing the engine to purposefully overheat. But my own experience would suggest that that knock audibly and graphically on the ECU through the digital signal processor, should look the same whether it's an over advance knock or it's from overheating.
Actually let me clarify that as well, knock is still knock when the engine becomes hotter though, what happens is our combustion chamber piston, our whole engine becomes hotter, remember it's really the heat in the combustion chamber, the heat inside that cylinder that is going to make the engine more prone to knock or detonation so if you've got an engine that's tuned very close to the edge and is knock limited, as the engine gets hotter it may become more likely that it will detonate. I hope that's cleared that up for you. Barry said, also do these different types of knock, overheating or overadvancing show in different areas of the pistons? Eg over advanced on the intake side only? Look again not as far as my own experience shows. Generally the damage we're likely to see on the piston is pretty similar. I've never been able to look at a piston and say hey this has been damaged because it's been overheated and that's caused knock, it's really, it's the same thing occurring.
Again, keen to hear if your own experiences are different to that. Janoo's can I talk about the damage between knock and pre ignition? Yeah OK I will touch on that, I said I'm not really going to talk about pre ignition. Knock can be very damaging. Probably again if you're reasonably new to tuning and you've just watched me purposefully induce knock on our 350z, that might have been a little bit scary for you. It's one of those things we always say we need to stay away from and here I am purposefully making the engine knock.
It's all about understanding the way I'm doing that and particularly at low RPM on a low power engine like our 350z, it's going to be relatively difficult to damage that and obviously I'm also not staying in the throttle, I'm not getting sustained knock there which would damage it. However sustained knock will damage our engine in time and it can, particularly as our specific power level increases it can damage the engine very quickly. The damage we see from pre ignition though is actually much more severe and it happens much quicker so the chance I guess of the engine surviving through pre ignition is generally much lower and the damage you're likely to see will be where you've got a large hole literally melted through the crown of the piston. Deducing the difference between pre ignition and knock can be difficult on an analysis of a failed engine though. When you're looking at a bunch of damaged parts, sometimes it's hard to be able to definitively say exactly what has occurred.
Pre ignition, for most of our performance engines running on pump fuel is normally not huge concern, it's not something we're going to encounter. It's more of an issue when we start moving onto alcohol based fuels where alcohol is a little bit more susceptible to pre ignition compared to a gasoline based fuel so I'm talking methanol and E85. Jano's asked, are ring land failures caused by detonation or pre ignition? Both events could cause broken ring lands. Again probably more likely that you're going to be having an engine encounter knock rather than pre ignition though. Barry's asked, what type of sensor's on the 350z right now? Look the sensors that we've got input into the Link G4+ are just the factory sensors.
My testing there, while I can't say definitively would suggest that they're a second harmonic style sensor and again that is relatively common on a lot of late model engines so probably more common I find than a first harmonic sensor. So that's what we've just seen, I tried it at 6 KHz which is the first harmonic for this engine and we saw a much improved noise level at the second harmonic. Jano's asked, I notice a lot of tuners like to disable the knock sensor below boost on turbocharged cars to avoid false knock. Is it safe to do this or is it common practice for turbocharged cars? Look if your knock threshold table is tuned correctly then there should be no real reason to disable your knock control strategy or turn that off, you shouldn't really need to. However at the same time the knock that you're likely to suffer from under light load, so cruise conditions for example, first of all it's much harder to make the engine knock there and also because the specific power level of the engine under cruise conditions is so low, it's also much less likely that knock would actually result in damage.
Now I'll just touch on that as well, everyone's probably heard a clapped out old car, perhaps an old Corolla that's done a million kilometres and it's going up a hill and the driver's at full throttle and it's sitting at 2000 RPM and you can audibly hear it pinging its head off and for a tuner that's kind of cringeworthy but that car could probably quite happily survive doing that for the next 100,000 km and the reason for that is the point in the engine's operation where that knock's occurring, the specific power level is so low that it's simply not going to do any damage to the engine. Certainly not something we want the engine to be doing, if it's avoidable but at that point it's not going to damage the engine. Barry's asked, is the Plex set to the same frequency as the Link ECU? In this case I'm using the 5 KHz high pass filter so it kind of is, it's looking at frequencies above 5 KHz and what I found which is interesting, I always tend to play and try the different frequencies on the Plex knock monitor to see which one gives me the best audio distinction between knock and background noise and I actually found that the 5 KHz high pass filter gave a better distinction than the 10 KHz high pass filter so I guess what I'm saying is there's no right or wrong way to set that up. I simply go through the different frequencies and see which one sounds the clearest when knock occurs. Jano's asked, when using the stock ECU when do we need to mess with the knock filter? For example if I just changed to forged pistons, do I need to mess with my stock ECU knock filter setting? This is a real problem with the factory ECUs and the answer really is it depends.
Which I know on the face of it might not seem particularly helpful. It depends on whether or not your ECU is still able to accurately pick up noise or whether it is now being triggered, seeing false knock as a result of the increased noise background noise, that is a result of those forged pistons. The reason for forged pistons creating more noise is generally we'll run a slightly wider piston to bore clearance so the pistons tend to be a little bit noisier in operation than a factory cast piston. So the technique, and this is a really important point here, the technique I'm using when I'm setting up knock control is always to confirm what the ECU is showing me in the logging with audio knock detection. I know it's a little bit frustrating but you cannot set up knock control without audibly listening and proving to yourself when the engine is actually knocking.
We want to confirm that when we hear knock, the ECU is properly registering knock and at the same time we want to also confirm that the ECU's not registering knock that we can't audibly hear. So you do need to back up your knock tuning with audio knock detection. So to answer your question, you'd only need to adjust the knock filtering if the ECU is either picking up false knock that you're not audibly hearing, or alternatively it's not picking up knock that you can audibly hear through your knock headphones. So that's how I do that. The factory knock control systems I find for the most part are really well sorted out and will normally handle something like a set of forged pistons without too much trouble.
It's not always the way but generally that sort of a modification doesn't tend to throw the factory knock control strategy. Barry's said, my last experience with a Link G4 Xtreme, I remember turning on per cylinder knock control and used up lots of memory allocation, is that the same with the G4+ platform? No that actually Barry was one of the reasons why Link upgraded to the new processor. They'd simply run out of memory allocation spots and particularly if you're tuning a multi cylinder engine and perhaps you wanted to use some fuel trim tables, it was easier to run out of memory space for additional tables so the G4+ that's definitely not an issue. You can do individual cylinder knock control and also individual cylinder fuel trimming if you wish, that's not an issue. Jano808 said, if you have unequal length headers do you still make the noise gain equal? Yeah the headers really shouldn't affect, I say they shouldn't affect the noise threshold but the reality is lots of things that you change on the engine can affect it but ultimately what we're still doing is tuning those individual cylinder gains until our no knock noise levels from each cylinder are as close to even as possible.
The whole reason, just to reiterate, that we're doing this is because if the no knock levels are consistent, then we can set our noise threshold, our knock threshold closer to that background noise level. If we've got one cylinder that under no knock conditions is consistently putting out a noise level 20 or 30% higher than the other 5 cylinders, we would have to set our threshold above that which means that the ECU will struggle to detect knock on the other cylinders with a lower noise level so that's why we're doing that. Jano's said, I noticed Subaru ignition trim tables always have cylinder 2 and 4 with positive values on the stock ROM, why would they want to advance those 2 cylinders? That's an interesting point you raise and on the FRS or BRZ ECU I actually see exactly the opposite, instead of adding timing into cylinder's 2 and 4, they remove timing from cylinders 1 and 3. The overall net effect is the same, basically what Subaru have found is that on that particular engine design, cylinders 2 and 4 are less prone to knock and you can, they will accept more ignition timing than cylinders 1 and 3. So that's why they've done that and it's again about optimising the power the engine's making by tuning each individual cylinder to MBT or at least as close to the knock threshold as we can get.
Jano's asked, and this is a question I know that you asked before we came on and went live, can you talk about tip in knock and shift knock? OK now tip in knock or shift knock, what that refers to is a knock event that occurs during transient throttle event. So particularly on a sharp acceleration event which you're obviously very likely to see on a fast gear change. Now I can't give you a clear reason as to why that tip in knock can occur. What I would tell you in 15 years of tuning cars, I have only actually had a handful of cars, perhaps as many as 3 or 4 cars where tip in knock was actually an issue. So it's certainly, in my experience, not the norm.
The first thing we want to do is talk about whether that tip in knock that we're seeing is real. So again we need to confirm that with our audio knock detection and make sure that the ECU is picking up real knock. As for why we have an engine suffer from tip in knock, again I can't really give you a clear explanation for that, there's a few theories out there but in my own personal experience because I haven't seen it occur enough, or put it this way, I haven't seen it occur enough for me to really delve into it in any detail. It's not something that you should be needing to worry about for the most part in your day to day tuning. The other thing about tip in knock as well is understanding whether or not it's a real issue for you.
So if you're audibly hearing tip in knock and it's only occurring occasionally then the chances are it's probably not going to have a huge effect on the reliability of your engine. The real key, what we're really wanting to prevent is sustained knock under high RPM wide open throttle conditions. That's the knock that is going to definitely damage your engine. If we're getting occasional 1 or 2 knock events on a tip in then I certainly wouldn't be particularly worried about it. Jano's asked, does upgrading the thermostat or turning on the coolant fan earlier help to reduce knock and does using a colder plug reduce detonation or knock? OK those are some really great questions there and I'll address them individually.
First of all, in terms of the thermostat temperature or the cooling fan. Yes the engine will be more prone to knock or detonation as I've kind of touched on. When the engine's operating temperature is higher. It's all about that combustion temperature. So the trend that we're seeing a lot with a lot of late model cars now is to run a very hot thermostat, perhaps as high as 92 to 95°C, you'll have to convert that into Fahrenheit sorry, my brain doesn't work in imperial.
But that's to aid fuel economy. However that hotter operating temperature does definitely make our engine more knock prone and on engines that are already knock prone, it can make our job tuning the engine a little bit more difficult. This is something I found on our Toyota 86 with the FA20 engine, the stock thermostat on hat is 92° and I dropped that down to an 82° thermostat and found that I did get an improvement in the engine's ability to take timing. One thing, without trying to get too involved with this, one thing that you do want to watch is it's all well and good if your operating temperature on the dyno is perahps 80 or 82°. If however like I see with our Toyota 86, the engine temperature creeps up to 95 or even 100° out on the racetrack, that's the situation we really need to be tuning for and being careful of so you do need to be aware of how the engine coolant temperature may affect the car once it's off the dyno.
Turning on your cooling fan, same sort of effect but it's important to realise when you're out on the road or racetrack at normal driving speeds we're not relying on the cooling fan. The cooling fan's only really going to be useful when the car's stopped or at very low speeds in traffic so it's not really going to be a way or reducing the engine's chance of knocking. OK so in terms of colder plugs. Now this is, there's a lot of misinformation here. What we need to do is choose a plug with a heat range that's relevant to the sort of power level that our engine's making.
Now there's some misinformation around the electrode on the plug beginning to glow hot and cause knock. If that was the case, what we'll actually see is a glowing electrode would cause pre ignition, it would actually ignite that fuel air charge earlier in the combustion cycle before the spark events occurred. My own experience though, if we're getting to that sort of situation, you're very quickly going to end up with the electrode on the spark plug eaten away or completely eroded. So that's not really a huge consideration in our engine's likelihood of detonating. What we do need to do though as I say is make sure that our spark plug is chosen relative to the sort of power levels our engine's making.
For me, I use NGK plugs and on a pump fuel or E85, on a turbocharged engine I would be choosing a plug around about an 8 or perhaps a 9 heat range. If you go too cold on the plug what you will find is that the engine will tend to foul when it's being cold started consistently. Right let's see, we've got one more question here. No we haven't, that's from an earlier session. Alright so that covers our questions and I hope that that's helped improve your knowledge there with how to go about setting up knock control in an ECU particularly obviously our G4+ and Vi-PEC ECUs.
As usual, if you are enjoying our webinars and our courses, I'd urge you again please to head to our Facebook page and if you haven't already, leave us a review. We'd love a 5 star review and that will just help others who are trying to decide whether to get involved with High Performance Academy, it's going to help them make their decision a little bit easier. Alright thanks guys, as usual if you do have any more questions that crop up that you want answered, please ask them on the forum and I'll be happy to answer them there, thanks.