To tune or not to tune....that is the question

[Dan Carter]

This subject may be too involved for the forum, but I thought it might be a good idea to share some knowledge about the tuning process, so I have put this together to get a dialogue started. It may go nowhere, or we may learn a lot from the more experienced among us. I’ve chose like so many, the LS3/480 (GM hot cam), but needed some tuning to make it more street friendly. I used 3 tuners, and I also enrolled in a tuning school for self-education, which led me to compare tuning techniques used by the tuners and the techniques taught by the school house. I’m sharing my very high-level review.

Attached are a series of pictures where I mapped out all the options that you have access to within the ECU as programmed by GM. Mine is a Gen IV ECU E67, and each generation of ECU has different data tables etc. The school house basically focuses on the ECU maintaining a perfect stoic AFR (14.7:1) or the demanded air fuel ratio programmed into the ECU by addressing the short and long-term fuel trims. Our tunes need the three basic interactions to run the engine, fuel, spark, and airflow (sweet spot intersection). While our right foot controls the airflow (monitored by the mass air flow sensor and the throttle position senor), the ECU manipulates the fuel injectors and ignition timing to achieve the demanded stoic value (14.7:1 for gas).

The ECU operates basically in two modes, open loop and closed loop. In open loop, the ECU ignores sensor input and uses the virtual volumetric efficiency (VVE) tables by using coefficients with the Mass Air Flow (MAF) tables to forecast engine run parameters (ouch that hurts my head). The engine uses VVE tables when the engine is outside of normal operating conditions (i.e. cold or erratic airflow etc.). In closed loop, the ECU uses two primary sensors to adjust fuel and timing to achieve the demanded Stoic value. While there are other sensor inputs (see pictures), the two involved the most are the MAF and O2 sensors. The standard O2 sensor in the LS3 is a narrow bad which is limited in scope (see the included picture). It does not provide the ECU any AFR reading, rather a voltage reading from 0 to 1 telling the ECU if the AFR is lean or rich. The ECU then adjusts fuel trying to either lean or richen the engine to achieve demanded stoic value. In school house tuning you focus a lot on trying to achieve a +-3% of the demanded stoic value by adjusting certain fuel and spark tables. They essentially want you to run the car using MAF only (VVE tuning is optional). The logic is to allow the ECU to do more of the fuel and stoic management vice the more static VVE method. Without saying, the school house materials are several hundred pages, so I certainly cannot go over every step. But, I did want to share the mapping of the ECU data files in a hierarchical manner. I laid it out as levels 0, 1 and 2, and if you peek at the attached photos you will note the hierarchical relationship of all three levels with a lot of colors, and a set of numbers following each choice. That number is the number of singular data adjustment selections, or tables that reside below that level 2 selection that can be visited to alter the tune (ie Lvl 2 Idle >> RPM>>Adaptive Idle -7, means seven more selectable options under adaptive idle). If the level two option is highlighted in yellow, it’s an area that the school house may have you select to make changes. If the words are in red, those are areas where my three tuners made changes. In a few cases the tuners and the school house agree, but notice the number of times the tuners and the school house visited non common areas.

Again, this is a huge topic, that perhaps we will share knowledge, even help each other, or not. So I will start a discussion hopefully by asking the first question:

Stoic for pure gas is 14.7:1. Most of us can only procure E10 at the pump, which has a stoic of 14.13:1. Knowing that one tune goal for the school house process is the sum of short and long term fuel trim (STFT + LTFT) is +-3% of stoic for gas, shouldn’t we be adjusting the commanded AFR (the ECU’s target AFR per tables) to 14.13:1 in lieu of 14.7:1? This is about a 4% error to the lean side to start with (not sure the error pushed by the ECU). Which table in the ECU controls the AFR commanded inputs?

 

[Frank Clark]

Let me preface this by saying I have no experience with HP Tuner or tuning GM ECUs (other than having a pro tune previous ones I've had). I do have some experience with Megasquirt and have Innovate O2 equipment in my cars/bike - even on the Sprint Car with Mechanical FI.

O2 sensors do not measure AFR. They measure Lamda (the goofy symbol before the 1.0 under the graphs above). This is contrary to the text above the graphs. The "perfect" Air-Fuel ratio is 1.0 Lamda for any fuel (we run O2 sensors in cars on Race Gas, E85 and Methanol). Keep in mind, the O2 sensor is just reading left over O2 in your exhaust - it does not know how much air & fuel went in, just how much O2 is left.

Here is a very good article that explains it better than I ever could: http://blog.innovatemotorsports.com/lambda-vs-afr-whats-the-difference/

So, when you "say" a car is running 14.7 AFR, then someone did some calculation to get from a Lamda value to that 14.7.

If you are tunning to AFR, then you need to adjust how that AFR is calculated to match the fuel you are using. If you are tunning to Lamda, you can ignore the fuel when it comes time to tune a running engine (Lamda 0.9 is always rich, 1.1 is always lean).

The GM ECU should 'learn' enough to handle E0-E10 by adjusting it's fuel trims (again, far from an expert on GM ECUs). But I did find my out of the box LS7 with GM's Harness+ECU did need some tuning on the dyno. The out of the box GM ECU probably deals with E0-E10 just fine. But there are things in your car GM probably could not address such as your Intake and Exhaust probably being less restrictive than a production car.

$500 spent on a dyno/tuner session is probably worth while most of the just finished SLCs...

 

[Dan Carter]

Good article to help this discussion. For those with the GM ECU and the HP tuner, let me add what I understand about this ECU. In the ECU tune file, the user sets a demanded stoich value (ie GM sets 14.7 for 0% alcohol, recall E10 is 14.13). As the ECU is working and collecting data, its using the demanded AFR as its baseline number to provide a percentage deviation output to the HP scanner as short term fuel trim (STFT) and long term fuel trim (LTFT) deviation which is an average of the STFT and retained in ECU memory. If the ECU is learning, its a slow learner, but its calculating and retaining LTFT to use during future engine runs. My concern is using 14.7 in lieu of 14.13 for E10 which seems to be injecting an upfront lean target error. There is a direct mathematical relationship with lamda and AFR (lamda = AFR observed/AFR Stoich), but the GM and HP tuner software and school house procedure don't use lamda in its tuning process directly (you can collect Eq Ratio error), rather the LTFT percentage outputs. The LTFT values that are collected during a very specific engine run are then used as a corrective multiplicative factor in the MAF tables searching for an illusive -+3% of zero deviation from the demanded/targeted AFR, or shooting for a lamda of 1.0.

PS. I wont tell you how many dyno pulls I've paid for ($$$$), so it was time to get some schooling....lol, cause I got schooled.

 

[Frank Clark]

...recall E85 is 14.13)..

assume you meant E10...

Are the GM O2 sensors even wide band?

 

[Dan Carter]

Good eye Frank, yes I meant E10, corrected. No my GM O2 sensors are narrow band, and provide voltage from 0-1 volt to the ECU, 0.5 volts being stoich. Unfortunately, neither the sensor nor the ECU converts the voltage to an actual AFR we can read, rather provides a percentage deviation from the stoich provided in the programmable data, aka 14.7 for gas etc. The ECU uses that voltage deviation to adjust fuel/spark input trying to get back to stoich. To get wide band, its an external system that I can hook into the HP tuner reader, but the sensor is stuck into the tail pipe unless permanently installed in the exhaust. Without a cross over pipe, I will only get one side at a time, not optimal.

 

[Frank Clark]

That all makes sense. That's what the turner we use does for our GM stuff.

It is my understanding (but needs verification), that all of the GM ECUs past a certain point will support an Ethanol sensor. I've been told you just need to hook it up and set the ECU to use it.

That may help, plus realistically you are probably not consistently getting E10. I'll bet the actual Ethanol content varies by brand and time of year.

Or, switch to Methanol. Then run it eye wateringly rich and tune it to keep the engine temp where you want ... works for the 50yr old tech on the Sprint Car