Continued from the first post....
Under no circumstances should you eliminate this safety feature on 91 octane pump gas. When it comes to boost limit, just set the limit slightly higher and please leave the boost delay alone. Setting the boost limit @ 300 or maxing it out to 319 simply removes the safety from your ECU in case of an overboost condition. Here is the way I set the boost limit on an Evo running on 91 octane gas:
The boost limit is increased by 10 load points from 3000 to 3500 rpm, 5 load points @ 4000 rpm, left stock between 4500 and 5500 rpm, increased by 5 load points between 6000-6500 rpm, and left stock at 7000 rpm. After 200+ WOT logs in 3rd and 4th gear I have yet to hit the boost limit.
So how did I come up with these load numbers? Is it guesswork? No, it is not. I came up with these numbers by logging the 2 byte load from the ECU. Here is what the log told me about the load (which is close to boost) is on my car:
Notice how the maximum load numbers of these 3rd and 4th gear logs is close to the set boost limit on my car. For example, the 263.8 In the log @ 3500 rpm is very close to the 265 @ 3500 rpm in the boost limit table. Now you can choose to increase them slightly, but I would not go over 270 load on 91 octane gas @ peak boost.
D. Tuning the Air Fuel Ratio (AFR)
AFR refers to how many parts of air are mixed with how many parts of fuel. So an 11:1 AFR means that 11 parts of air are being mixed with 1 part of fuel to create the air/fuel mixture. When your Evo is at idle or when your Evo is at cruising speeds your AFR is around 14.5-14.7:1. This is known as stoichometric or stoich for short. It has been found that the 14.7:1 mixture produces the least amount of emissions. And since cars spend 90% of their time at idle/cruise then that is the number that the manufacturers use to reduce the emissions on their car. It is worth noting that the 14.7:1 AFR does not produce the best gas mileage. The best gas mileage is produced are 15.2:1 AFR.
What AFR produces the best power for gasoline? Gasoline gives the best power when it burns at an AFR of 12.5:1. This is regardless of whether the car is normally aspirated, turbocharged, or supercharged. Some modern turbocharged engines with direct fuel injection can run that lean during WOT operation. The turbocharged Ecotec in the Solstice GXP is such an engine. That engine can boost up to 18 psi, yet it runs at 14:1 AFR at 3500 rpm and tapers down the AFR to 12.5:1 by redline.
So can I run my Evo at 12.5:1 AFR? NO you cannot and should not. The Evo’s engine is about 17 years old, it does not have direct fuel injection, and the combustion chamber is not designed to handle such a lean AFR. Furthermore, the car is running on 91 octane gas. Under high boost and lean conditions, 91 octane gasoline becomes very unstable and can self ignite causing knock and other assorted problems.
So why tune the AFR last? There are two main reasons. First, changing the mivec map has an impact on the AFR. If you tune the AFR before mivec and then you tune mivec, the AFR will change and you will have to do it again. My testing indicates that adding the JDM RS mivec map to an already tuned AFR will make the AFR leaner by about 0.25 points. Not a lot, but something to take into account when you tune.
Second, increasing the boost will also impact your AFR. Why? The higher the boost the higher the load cell that the car will hit in the fuel map. Mitsubishi designed the fuel map to become richer the higher the load cells. So when you up the boost you will hit those higher load cells and the car will run richer. If you tuned your AFR before your boost, then you will have to do it again after you increase the boost. Why do things twice?
Below is an Evo 9 fuel map. I have interposed on it green dots (stock Evo), black dots (Evo with TBE), and red dots (Evo with TBE and boost set @ 22 psi and tapering to 19.xx by redline). You will note that a stock Evo hits lower load cells (200 to 180) than a TBE Evo (220 to 200) and a TBE Evo hits lower load cells than a TBE Evo with boost increase (260 to 230). You will also note that the numbers in the load cells decrease as the load cell increase, i.e., the car becomes richer. It is very important to note that the numbers in the load cells of the fuel map are NOT actual AFR numbers that you will log with a WBO2. Under no condition should you enter the AFR that you logged with your wideband into the fuel map. They are just numbers. The higher the number, the leaner the AFR, and the lower the number the richer the AFR.
So what does the AFR look like on a completely stock Evo? It looks horrible. The Evo runs very rich from the factory. It is tuned very poorly. The AFR falls below 11:1 at 3500 rpm and continues to get richer until it hits 9.5:1 by 7000 rpm. Despite the overly rich AFR, a stock Evo still knocks due to the overly aggressive timing.
Setting your AFR depends to a large extent on the boost and timing that your car is running. When running the boost and timing mentioned in this essay, I generally set the AFR at 12.5-12 during spool up, 11.7-11.5 during peak boost, and then slowly taper the AFR until it hits 11-10.9:1 by redline/cutoff.
So how do I go about editing the fuel map?
Well, first you have to log your AFR. Below is a chart with three back-to-back logs from my Evo 9 with a TBE and K&N drop-in. The map is stock with the exception of extending lean spool from stock (7000 rpm) to 7700 rpm. I will explain the lean spool trick later on in the essay. The boost is also untouched.
First, notice how the AFR became leaner by simply adding a TBE and a drop-in filter. At 3500 rpm, for example, the AFR became leaner by 0.77 points, exactly where we want it to be. Second, the actual AFR is very close to our target AFR right up to 5000 rpm. Beyond that the AFR goes rich again. This will require adjusting the fuel map.
The formula to adjusting the fuel map is very simple. Let us look at the 5500 rpm row and load cell 220 in the log chart above. The Actual AFR (AAFR) is 11.13:1. The fuel map AFR (MAFR) shows a 9.7 number in the 5500 rpm and 220 load cell. Let us assume that we want a desired AFR (DAFR) of 11.4:1 in that load cell. What should the new map AFR (NMAFR) be?
NMAFR=DAFR X MAFR / AAFR
NMAFR=11.4 X 9.7 / 11.13 = 9.9353
So the number that you should enter in the fuel map in the 220 load cell @ 5500 rpm should be 9.9.
To make this easy on yourself, simply create a template in excel with the above formula and use it over and over again. This way you will not have to do any manual calculation. Just plug in the numbers and excel will take care of it. That is what I did and it works like a charm. This method takes the guessing out of AFR tuning and allows you create a very flat and consistent AFR. Below is a 4th gear log with the AFR indicated @ spool up, peak boost and 7000 rpm. It is nice and flat from 3800 rpm to redline. The data is not smoothed. Logworks and Evoscan do not smooth the data. Logworks allows you to smooth the data afterwards. I like the data raw.
E. Reading your knock sensor
An ideal combustion process behaves in the following manner:
1. The air fuel mixture is brought into the combustion chamber. Ideally this mixture should have around 12.5:1 AFR to extract maximum power from gasoline. Given that the Evo engine is about 17 years old, crappy CA gas, and high boost, this ideal is pretty hard to achieve w/o running water or methanol injection. As stated before most amateur tuners that I know run between 11.5-11:1 AFR on an Evo.
2. The intake and exhaust valves close and the spark plug fires. On an Evo 8 a spark plug fires at around 18-21* BTDC by 7000 rpm. On an Evo 9 there is less timing advance with the spark plug firing around 14-16* BTDC by 7000 rpm. Why less timing advance on the Evo 9 than the Evo 8? In part, it is because the Evo 9 is blessed with a better cooled and better flowing cylinder head than the Evo 8. The Evo 9 can run leaner AFRs. Leaner AFRs burn faster up to 12.5:1. Beyond that they burn slower. A faster burning mixture does not require as much timing advance as a slower burning one. I am not saying that the Evo 9 has a leaner AFR from the factory. Far from it. What I am saying that it has the potential to run leaner AFRs and consequently less timing advance.
3. After the spark is fired the burning of the mixture proceeds. It begins at the spark plug and progresses in an orderly fashion across the combustion chamber. It is as if you took a pebble and threw it in a pond and watched the ripples progress outward from where the pebble fell. The burn should be complete with no remaining air-fuel mixture by the end of the combustion process.
In reality combustion sometimes does not progress in an orderly and smooth fashion. Sometimes the air-fuel mixture spontaneously combusts after the spark plug is fired but before the flame front reaches the mixture. This is commonly known as detonation or more commonly knock. Why does that happen? Too much pressure and too much heat combined with the lack of enough octane in the mixture to resist self-combustion. Think of octane as the ability of gasoline to resist self-combustion under pressure and heat. The higher the octane the less likely the gasoline will self-combust under high boost and heat that the Evo is known to generate.
Unfortunately in CA, we are stuck with very poor quality 91 octane gasoline, yet the Evo has a very old combustion chamber design and very high boost. That is a perfect recipe for detonation when you factor in the advanced timing that the car runs from the factory.
When a car knocks, it causes a very sharp pressure spike that is outside the normal shape of a pressure curve during normal combustion. The pressure spike creates a force in the combustion chamber. The structure of the engine pings/rings in reaction to the force generated from the pressure spike. That is where the knock sensor steps in.
The knock sensor is usually connected to the back of the engine block. It is nothing more than a microphone. It reads the noise in Hertz and transmits it to the Evo ECU. The Evo ECU filters that noise using 12 different tables in the rom and decides if the noise is knock. If it is, the ECU sends a signal to the sensors to pull the timing in order to save the engine from further detonation and possibly damage. The knock sensing system is reactive and not pro-active. The timing pull happens after knock is detected and pulls timing to prevent further damage. It does not prevent knock, it tries to limit it after it has happened.
The signal that the ECU spits out is commonly known as “knock sum.” The loggers that we use have the ability to log knock sum. Generally speaking the higher the knock sum the more timing will get pulled, the lower the knock sum the less timing will get pulled. More on that later.
So what sort of damage does knock cause?
If left unchecked, knock can break the spark plugs, the valves, and the rings around the pistons. Second, knock can be very abrasive to the crown of the piston. Pistons on an engine that is suffering from excessive knock will look like as if it has been sandblasted with small pits in the top of the piston. Finally, excessive knock will cause a premature failure of your rod bearings resulting in the very distinctive rod knock sound.
Having said the above about the dangers of knock do not be surprised to know that almost all cars knock. As long as the knock is occasional and moderate
cars can run for thousands of miles with little to no problems. While detonation is not an optimum situation for engine operation, it does not guarantee engine failure.
So how should I deal with knock?
As I briefly mentioned earlier the Evo ECU spits out a parameter known a “Knock Sum.” That parameter is one of the most important to log when tuning your Evo. Evoscan tells us that this parameter can vary from 0 to 50 with 50 as the maximum knock count that the Evo ECU can register.
When tuning your Evo it is advisable to tune timing, fuel, and boost w/o triggering more than 1-2 occasional counts of knock, three at the most. We know for a fact that 3 knock counts pull 1* of timing. I have also logged occasions when 1 knock count pulls 1* of timing.
I tune for 1 to 2 occasional and sporadic counts of knock, three at most. Anything above that is unacceptable. Here is my take on knock:
1. All cars knock on occasion. I have logged an Evo that knocked the first log and then gave me three knock free WOT runs. Generally speaking, the first WOT log that you do tends to be knock prone. You have to do at least three back-to-back logs to make sure that knock is consistent. I do not worry about an occasional log that has knock it. If the knock is transient and does not repeat
, I usually ignore it.
2. Knock is a problem when it is consistent and repetitive
, i.e., it happens every log and at the same point in the rpm range. That is the kind of knock to worry about and work hard to eliminate.
So my Evo has more than 2 counts of knock and the knock is consistent and repetitive. What should I do to eliminate it?
IMO, the biggest cause of knock on an Evo is too much timing advance. Let us take a look at my stock Evo 9 with no tuning. My Evo 9 consistently and repetitively registered 5-6 counts of knock from 5000 rpm on. Below is a chart of a typical 3rd gear WOT run on my Evo 9.
Notice that the timing @ 5224 rpm was 10* and after 6 counts of knock the timing was pulled to 8* by 5500 rpm. 6 counts of knock pulled 2* of timing, in line with our assumption that 3 counts of knock pull 1* of timing.
So what is the ECU telling us to do to combat knock?
We know from MTBT (minimum timing best torque) theory that we should advance the timing until we either stop making power or we see the onset of knock. In this case we clearly see the onset of knock. So what we have to do is pull 2-3* of timing to combat the knock in that rpm range.
Here is the way the log looked after I pulled timing. The boost was almost unchanged and the AFR was slightly leaner in that rpm range. Pulling the timing from 10* to 7* @ 5200 rpm cured the knock in this instance.
Here is another example. The chart below is for an Evo 8 with the following modifications: TBE, O2 housing, 264 cams, Walbro fuel pump, 720 injectors, and Greddy EBC and a tune.
As you can see the car was knocking. The data labels show that it had: 3 counts of knock @ 5000 rpm that pulled 1* of timing, 5 counts of knock @ 5400 rpm that pulled 2* of timing, and 12 counts of knock @ 6250 rpm that pulled 4* of timing. These numbers are very closely in line with the fact that 3 counts of knock pull 1* of timing. By 7500 rpm the timing was @ 13* despite the fact that the timing in the rom map was much higher than that.
So what is the ECU telling us in this instance?
The ECU is telling us not to set timing more than 13* by 7500 rpm. We can basically set the timing at 13* by 7500 rpm and work our way backwards. With this in mind, I set the timing at 2-3* at peak boost and slowly incremented it upwards until the car hit 13* by 7500 rpm. Here is the way the log looked after the timing was changed:
As you can see, the serious knock in the car is gone. What remains are 1-2 counts of knock here and there. You will note that 1 count of knock still pulls 1* of timing as can be seen @ 7000 rpm, but most of time it is the 3 counts of knock that you will have to worry about. On this Evo, any timing increase beyond 13* triggered knock. So I left the timing at 13* at the top end.
And here is the final example.
This one is very interesting because the tuner wanted to force this Evo 9 to run advanced timing and was willing to run the car extremely rich to do it. At peak boost the target AFR was set to 7.4 and the knock prone segment of the power band had a target AFR of 8.5. The AFR curve on this car was only a smidgen better than the AFR on a stock Evo 9. But no matter how rich he ran the car, this Evo still knocked because of the advanced timing. The timing profile was block tuned with 5* from 1500-3500 rpm, 6* from 4000-5000 rpm, 7* @ 5500 rpm, 8* @ 6000 rpm, 11 @ 6500 rpm, 14* @ 7000 rpm and 16* @ 7500 rpm. Take a look at the chart below:
Notice how the Evo’s ECU pulled 3* of timing when it registered 8 counts of knock. Even though the tuner ran the car at 10.3:1 in order to make it accept 11* of timing, the car still knocked and ended up with 8* of timing @ 6500 rpm. So why not run 8* of timing to begin with and run the car safer with a decent AFR? If the ECU is going to pull the timing anyway, then why insist on advancing the timing so much and run the car as rich as stock to boot?
Here is a log form the car after the timing was retarded and the AFR leaned to reasonable levels:
The AFR is 11.13:1 in the knock prone area and the timing was set at 8* which is exactly what the ECU told me that this Evo wanted. The car is running leaner, cleaner and safer. It is also putting just as much power as before.
The above three examples show you why reading the knock sensor when setting the timing is essential. By reading the knock sensor the tuner is able to give the Evo the timing that it wanted rather than the timing that he assumed it wanted.
F. Injector Scaling
The stock injectors on the Evo are rated at 560. Generally speaking, the stock injectors will give you adequate fuel flow with a TBE, an intake, and 19-20 lbs of boost by redline. Once you get cams, then it is advisable to get bigger injectors. If you get a bigger turbo, then definitely get bigger injectors.
If you install bigger injectors, then you will have to scale them properly and use the correct injector voltage latency, otherwise your car will idle poorly and stall on occasion.
So how do you go about scaling your new injectors?
Scaling injectors is a PITA. It involves a lot of trial an error. It is not enough to simply put numbers in the rom tables and simply declare the injectors scaled. You must test and make sure that the scaling is accurate.
Here are the steps I follow when scaling injectors
1. Open Ecuflash and open your rom. Under fuel locate the Injector Scaling table and the Injector Battery Voltage Latency table. On a stock Evo they look like this:
The table on the left refers to the injector value that the ECU is using when making its fuel supply calculations. The number in the table is always smaller than the actual size of the injectors on the engine. For example, the stock injector size is 560, but the number in the table is 513. As a general rule of thumb, enter a number in the table that is 15-20% less than the size of the injectors installed on your Evo.