Race gas?

[arghx7]

If the race fuel does not contain ethanol, fuel economy may go up on account of that. The rest of the engine probably won't take advantage of the additional knock resistance.

And yes, it IS possible for the ECU to estimate the octane of the fuel. There are many stock ECUs that do just that. You can also calculate the MBT timing using an onboard model.

Assuming the BRZ isn't drastically different from previous Subarus, it is using a very simple timing control system which doesn't model flame front propogation or MBT.

[rice_classic]

So on ECU's that have the ability to proactively work in conjunction with knock sensors to "test" ping and advance timing accordingly, how far will they advance?

From what I've been reading the engine will advance itself if you go from 87 to 93 octane and thus take advantage, like has been stated, but will it go beyond the advanced timing map that's optimal for 93 if you put in 100? It it my understanding that's not part of the program. So while and engine with an adaptive ECU like that of the Camry will see a benefit with 93 over 87, it doesn't appear that it's programed to provide benefit for any octane above that without telling it do so. Is this accurate?

[uspspro]

Originally Posted by rice_classic

So on ECU's that have the ability to proactively work in conjunction with knock sensors to "test" ping and advance timing accordingly, how far will they advance?

From what I've been reading the engine will advance itself if you go from 87 to 93 octane and thus take advantage, like has been stated, but will it go beyond the advanced timing map that's optimal for 93 if you put in 100? It it my understanding that's not part of the program. So while and engine with an adaptive ECU like that of the Camry will see a benefit with 93 over 87, it doesn't appear that it's programed to provide benefit for any octane above that without telling it do so. Is this accurate?

You could be right, and I'm sure there are limits to how far it goes.

[7thgear]

Originally Posted by rice_classic

So on ECU's that have the ability to proactively work in conjunction with knock sensors to "test" ping and advance timing accordingly, how far will they advance?

can we assume that it will advance timing until it starts to ping and then back off?

[Jordo!]

It's possible that the ECU will add in some more advance, but it's not even worth trying without a dyno.

Just stick with 93.

[rice_classic]

Originally Posted by 7thgear

can we assume that it will advance timing until it starts to ping and then back off?

I tend to assume too much. I was hoping someone knew the answer. It doesn't seem to be readily google'able.

[uspspro]

Originally Posted by rice_classic

I tend to assume too much. I was hoping someone knew the answer. It doesn't seem to be readily google'able.

Honestly Toyota tends to keep certain info like this under lock and key.

the only way to find out would be to dyno test and data log.

[serialk11r]

Originally Posted by 7thgear

can we assume that it will advance timing until it starts to ping and then back off?

I think that's a bad assumption, the ideal amount of advance for a certain engine speed is fuel dependent, and the only thing about the fuel the ECU can see is if it knocks and what the exhaust temperature looks like. Someone correct me if I'm wrong, but I think the ECU would have a target spark advance map and pull back from there.

[arghx7]

Originally Posted by serialk11r

I think that's a bad assumption, the ideal amount of advance for a certain engine speed is fuel dependent, and the only thing about the fuel the ECU can see is if it knocks and what the exhaust temperature looks like. Someone correct me if I'm wrong, but I think the ECU would have a target spark advance map and pull back from there.

You could broadly categorize the modern strategies as having two types: those with an MBT calculation/lookup, and those without. Subaru traditionally doesn't use an MBT calculation. MBT = Minimum spark advance for Best Torque. The well-known Subaru software doesn't use this kind of calculation but other manufacturers do.

There are different ways to do it and it varies with the manufacturer and how new the software design is. The well-known Subaru software up to now starts with a base timing ignition timing map, and then it has a 3D lookup table that gives authority to advance timing a certain number of degrees based on rpm and load. There is a multiplier that determines how much of this total additional spark advance will be used. Then there's short-term knock sensor feedback and also another learning table that's populated by rpm and load.

Many different manufacturers use some kind of "high octane" and "low octane" map. Mitsubishi does for example. You start on the high octane and interpolate to the low octane table based on knock sensor feedback. There are a number of ways this is implemented across manufacturers. You can accumulate some kind of knock value that counts up and counts down as a timer increments.

The other style of system uses an MBT lookup table or an MBT calculation. Bosch was an early adopter of the fully torque-based engine control systems that also directly implement the concept of MBT.



So in the Motronic 7 system diagram above (late 90s/early 2000s German cars) for example you have an MBT map and then a spark efficiency transfer function. That function tells the ECU how much torque output it's losing by deviating from the MBT value. Knock control is also implemented into the torque request logic.



A lot of Hitachi based control systems implement MBT with an actual simplified MBT calculation. It's based on the concept that, as a general rule, MBT spark advance typically results in the 60% cylinder mass fraction burned occuring between 12-15 degrees ATDC. Working off a basic differential equation you go backwards and fill in a bunch of variables based on sensor signals and constants based on engine geometry.



That results in your real time MBT calculation. Since you can't achieve MBT under normal conditions with pump gas, you have to have compensation maps to push your timing advance curve out to the borderline detonation range. This could involve advancing or retarding in some fixed increment with a counter, or it could use a more sophisticated way to actually model the fuel octane value in RON (Research Octane Number).

So running on unleaded race gas could actually result in more timing advance but only if it's been set up that way. It depends how much authority the ECU has to advance the timing in a given area.

[uspspro]

Originally Posted by arghx7

You could broadly categorize the modern strategies as having two types: those with an MBT calculation/lookup, and those without. Subaru traditionally doesn't use an MBT calculation. MBT = Minimum spark advance for Best Torque. The well-known Subaru software doesn't use this kind of calculation but other manufacturers do.

There are different ways to do it and it varies with the manufacturer and how new the software design is. The well-known Subaru software up to now starts with a base timing ignition timing map, and then it has a 3D lookup table that gives authority to advance timing a certain number of degrees based on rpm and load. There is a multiplier that determines how much of this total additional spark advance will be used. Then there's short-term knock sensor feedback and also another learning table that's populated by rpm and load.

Many different manufacturers use some kind of "high octane" and "low octane" map. Mitsubishi does for example. You start on the high octane and interpolate to the low octane table based on knock sensor feedback. There are a number of ways this is implemented across manufacturers. You can accumulate some kind of knock value that counts up and counts down as a timer increments.

The other style of system uses an MBT lookup table or an MBT calculation. Bosch was an early adopter of the fully torque-based engine control systems that also directly implement the concept of MBT.



So in the Motronic 7 system diagram above (late 90s/early 2000s German cars) for example you have an MBT map and then a spark efficiency transfer function. That function tells the ECU how much torque output it's losing by deviating from the MBT value. Knock control is also implemented into the torque request logic.



A lot of Hitachi based control systems implement MBT with an actual simplified MBT calculation. It's based on the concept that, as a general rule, MBT spark advance typically results in the 60% cylinder mass fraction burned occuring between 12-15 degrees ATDC. Working off a basic differential equation you go backwards and fill in a bunch of variables based on sensor signals and constants based on engine geometry.



That results in your real time MBT calculation. Since you can't achieve MBT under normal conditions with pump gas, you have to have compensation maps to push your timing advance curve out to the borderline detonation range. This could involve advancing or retarding in some fixed increment with a counter, or it could use a more sophisticated way to actually model the fuel octane value in RON (Research Octane Number).

So running on unleaded race gas could actually result in more timing advance but only if it's been set up that way. It depends how much authority the ECU has to advance the timing in a given area.

Awesome post.

[jedibow]

Originally Posted by rice_classic

I know. The ECU has no way of measuring or knowing the AKI rating of the fuel.

Okay, so I think it is time for some education.

Every OBD2 ECU has the ability to track fuel octane. The ECU has a minimum of two fuel maps, High octane and Low octane. Based off of the amount of knock present the ECU either raises the Octane count, or lowers it. What does this mean? Well fuel calculations are based off an interpolation between the two maps, for instance...

If a given load and RPM access a lookup cell in the high octane fuel map and it is calling for a 12.5 AFR, and the corresponding low octane fuel map calls for an 11.5 AFR, based on the octane rating (or more correctly the amount of knock over time the engine has witnessed) the actual cell maybe somewhere between the two, or it no knock has been detected 12.5, or consistent knock 11.5. Remember pulling timing isn't the only way to fight detonation, an increase in fuel lowers the cylinders temperature and decreases knock also. If you don't believe me go pick up a scan tool and log "Fuel Octane"....

[rice_classic]

Originally Posted by jedibow

Okay, so I think it is time for some education.

Every OBD2 ECU has the ability to track fuel octane.

If you had been reading along you would have noticed that I got schooled about this before you got to me. Also, are you SURE every OBD2 ECU does this?

[arghx7]

If you search for 'octane' in the list of universal OBD parameters attached, you won't find any matches. That doesn't mean that a modern ECU doesn't have some sort of way to respond to knock. There is no universal parameter to log knock learning; you'd have to read RAM addresses directly, with an application-specific tool.

As I mentioned above, there are more sophisticated ways knock control can be done. It's not just interpolating between maps intended for high and low octane, although that's what say Mitsubishi has typically done. Up until now at least, Subaru hasn't been doing it in exactly that way. Here is a very good writeup on the Subaru knock-control system, at least up through the current WRX/STi http://www.romraider.com/forum/viewt...hp?f=25&t=1840

[mastertech86]

So in a nut shell, going higher that 93 in octane without a proper tune would be a waste right? With higher octane requires more timing advance amongst other changes because of the resistance to ignite. But the factory ecu can only advance so far before reaching its limits on how much it can change