Tuning lesson: Full 3-Dimensional tables vs 2-D tables vs 2-D+modifier tables.
#1
Addict
Rennlist Member
Rennlist
Small Business Partner
Rennlist Member
Rennlist
Small Business Partner
Thread Starter
Tuning lesson: Full 3-Dimensional tables vs 2-D tables vs 2-D+modifier tables.
There seems to be some misunderstanding on what true 3-dimensional tuning is. So to hopefully make it a bit more transparent, I will explain different methods.
AFAIK, Rogue Tuning is the only one to use full 3-Dimensional tables. And we use them for very good reason...
Part 1
The difference between 2-D and 3-D isn't trivial. For example, the stock DME software uses a 2-D WOT ignition table:
This table clearly has 2-dimensions, the 1st being the RPM, the second being the associated value. This can be easier to digest with a visual representation of the table:
The limitation of this type of tuning is that the tune is static - it has no adaptability. If you are over ~66% throttle, then the DME uses this table, and it doesn't care about actual load. So, for example, if you are at 4000rpm, and 0psi of boost, your ignition timing is ~19°BTDC. And if you are at the same RPM, but 15psi of boost, your ignition is exactly the same: ~19°BTDC. Obviously the load on the motor is not the same for these two situations, so the ignition timing should be different, but it isn't.
What is the solution?
The solution is to change the engine-management strategy to use 3-dimensions. Here is a 3D table:
The added dimension should be immediately obvious. The 1st dimension is LOAD; the second dimension is RPM; the third dimension is the associated value. Again, this is easier to understand with a quick visual:
Now the ignition timing is no longer purely dictated by RPM. With the additional axis of load, we can adjust to the dynamic-load nature of a turbocharged vehicle. Using our example from earlier: 4000rpm and 0psi of boost, the ignition timing dictated is ~42°BTDC. And at 4000rpm and 15psi of boost the ignition timing is ~23.2°BTDC. That is a huge change from the static 2-D table in the first example.
The addition of the 3rd-axis provides much better control and adaptability to changing engine loads and rpms. The end-result is a smoother, more linear power delivery, with multiple "secondary effects" such as economy, emissions, safety, ect.
AFAIK, Rogue Tuning is the only one to use full 3-Dimensional tables. And we use them for very good reason...
Part 1
The difference between 2-D and 3-D isn't trivial. For example, the stock DME software uses a 2-D WOT ignition table:
This table clearly has 2-dimensions, the 1st being the RPM, the second being the associated value. This can be easier to digest with a visual representation of the table:
The limitation of this type of tuning is that the tune is static - it has no adaptability. If you are over ~66% throttle, then the DME uses this table, and it doesn't care about actual load. So, for example, if you are at 4000rpm, and 0psi of boost, your ignition timing is ~19°BTDC. And if you are at the same RPM, but 15psi of boost, your ignition is exactly the same: ~19°BTDC. Obviously the load on the motor is not the same for these two situations, so the ignition timing should be different, but it isn't.
What is the solution?
The solution is to change the engine-management strategy to use 3-dimensions. Here is a 3D table:
The added dimension should be immediately obvious. The 1st dimension is LOAD; the second dimension is RPM; the third dimension is the associated value. Again, this is easier to understand with a quick visual:
Now the ignition timing is no longer purely dictated by RPM. With the additional axis of load, we can adjust to the dynamic-load nature of a turbocharged vehicle. Using our example from earlier: 4000rpm and 0psi of boost, the ignition timing dictated is ~42°BTDC. And at 4000rpm and 15psi of boost the ignition timing is ~23.2°BTDC. That is a huge change from the static 2-D table in the first example.
The addition of the 3rd-axis provides much better control and adaptability to changing engine loads and rpms. The end-result is a smoother, more linear power delivery, with multiple "secondary effects" such as economy, emissions, safety, ect.
#2
Addict
Rennlist Member
Rennlist
Small Business Partner
Rennlist Member
Rennlist
Small Business Partner
Thread Starter
Part 2
Some might say that having a modifier table is a dimension, but that would be an incorrect or misleading statement. A modifier table is simply a scalar. A scalar x vector does not increase the dimension of the vector. (trivial to the math-savvy)
For example..
If we have our 2-D WOT table from before:
Now, if we also have another table:
This table is another 2-D table, referencing boost pressure. This table modifies the value first looked-up in the 2-D WOT table.
Now the system can adjust to boost pressure, albeit a limited adjustment...
Using our example from earlier, at 4000rpm and 0psi of boost, the ignition timing is first dictated by the WOT table, then modified by the value in the Boost table.
So, the WOT table dictates 19.1°BTDC, and the Boost table adjusts it by 4.8°, resulting in a final value of 23.9°BTDC.
Next, at 4000rpm and 15psi of boost, the WOT table gives the same 19.1°BTDC, and this time the Boost table gives -2.0°, resulting in a final value of 17.1°BTDC.
So where is the limitation? Well, unlike 3-D tables, using two 2-D tables, you can only adjust an entire "row" or "column" at a time... For example what if you wanted/needed to reduce the ignition timing at 4000rpm and 15psi of boost? Well you have two options:
1- change the 4000rpm timing value in the WOT table
2- change the 15psi value in the modifier table.
Regardless of what choice you make, by changing a value in either tables will cause a change in more than just the single point at 4000rpm and 15psi.
I.E. if we change the RPM table value @ 4000rpm, then that change will carry through every boost pressure when the motor is at 4000rpm. Conversely, if we change the boost table value @ 15psi, then that change will carry through every RPM when the motor is at that boost pressure.
Because of this limitation, the addition of a modifier table, such as this "boost table" is not considered a true axis. In-fact, the stock DME software has many such modifier tables: AIT modifier, Coolant temp modifier, ect. If one was to count these modifier tables as true axis, then the stock software would be ~5-7dimensional (!), which is quite obviously not the case.
Some might say that having a modifier table is a dimension, but that would be an incorrect or misleading statement. A modifier table is simply a scalar. A scalar x vector does not increase the dimension of the vector. (trivial to the math-savvy)
For example..
If we have our 2-D WOT table from before:
Now, if we also have another table:
This table is another 2-D table, referencing boost pressure. This table modifies the value first looked-up in the 2-D WOT table.
Now the system can adjust to boost pressure, albeit a limited adjustment...
Using our example from earlier, at 4000rpm and 0psi of boost, the ignition timing is first dictated by the WOT table, then modified by the value in the Boost table.
So, the WOT table dictates 19.1°BTDC, and the Boost table adjusts it by 4.8°, resulting in a final value of 23.9°BTDC.
Next, at 4000rpm and 15psi of boost, the WOT table gives the same 19.1°BTDC, and this time the Boost table gives -2.0°, resulting in a final value of 17.1°BTDC.
So where is the limitation? Well, unlike 3-D tables, using two 2-D tables, you can only adjust an entire "row" or "column" at a time... For example what if you wanted/needed to reduce the ignition timing at 4000rpm and 15psi of boost? Well you have two options:
1- change the 4000rpm timing value in the WOT table
2- change the 15psi value in the modifier table.
Regardless of what choice you make, by changing a value in either tables will cause a change in more than just the single point at 4000rpm and 15psi.
I.E. if we change the RPM table value @ 4000rpm, then that change will carry through every boost pressure when the motor is at 4000rpm. Conversely, if we change the boost table value @ 15psi, then that change will carry through every RPM when the motor is at that boost pressure.
Because of this limitation, the addition of a modifier table, such as this "boost table" is not considered a true axis. In-fact, the stock DME software has many such modifier tables: AIT modifier, Coolant temp modifier, ect. If one was to count these modifier tables as true axis, then the stock software would be ~5-7dimensional (!), which is quite obviously not the case.
#3
#4
Addict
Rennlist Member
Rennlist
Small Business Partner
Rennlist Member
Rennlist
Small Business Partner
Thread Starter
A vector multiplied by a scalar will be in the same plane as the original vector (thus the same number of dimensions).
scalar k
vector v = <v1, v2>
kv = <kv1, kv2>
Lets not derail the thread (already). The point of the thread is to show and compare the different engine management strategies regarding table dimension.
#6
Burning Brakes
Trending Topics
#8
Addict
Rennlist Member
Rennlist
Small Business Partner
Rennlist Member
Rennlist
Small Business Partner
Thread Starter
Thanks! Hopefully, people will have a better grasp of whats offered, and their merits/downfalls.
#10
Calculus FTW...
A vector multiplied by a scalar will be in the same plane as the original vector (thus the same number of dimensions).
scalar k
vector v = <v1, v2>
kv = <kv1, kv2>
Lets not derail the thread (already). The point of the thread is to show and compare the different engine management strategies regarding table dimension.
A vector multiplied by a scalar will be in the same plane as the original vector (thus the same number of dimensions).
scalar k
vector v = <v1, v2>
kv = <kv1, kv2>
Lets not derail the thread (already). The point of the thread is to show and compare the different engine management strategies regarding table dimension.
Sorry I even asked!!
Last edited by Bri Bro; 06-15-2012 at 08:25 AM.
#11
The subtlety here being that the vector can happen not to be linear.
If a table is used, this is precisely because we need the "vector" not to be linear. If the "vector" was linear that would not even make sense to use a table but just a product of two values - the two boundaries of said table.
The term "vector" may not be the most appropriate, but it should be meaningful enough to illustrate the concept.
Using the term "soft string" would be more accurate but equally... ropey
#13
Three Wheelin'
I can see that this would be really useful for turbocharged engine due to the variability of boost. On a supercharged engine with a linear boost curve, do you see any advantages of 3D WOT maps?