Stoichiometric Air/Fuel Ratio

General Discussion
#1

There are a lot of very smart people, engineers, and tuners on this board so I hope they can help me get my head around this concept. There is a lot of information over on the terminator 03-04 cobra boards which is mainly what I have been reading. What I gather is a cobra with a 100% stock ECU will run lean if switched from pump gas with a stoich of ~14.7 to a race gas such as VP MS109 with a stoich of ~13.4. The stock ECU has a ‘hard coded’ stoich value of pump gas at 14.7 and iterates/adjusts parameters based on sensor readings which is fine when the fuel assumed to have 14.7 actually is fuel that has a stoich of 14.7. However, when running pure VP MS109 the stoich of the race gas in the tank is 13.4, yet the stock ECU uses the now incorrect stoich of 14.7 to do its calculations. The general consensus is that this specific condition (totally stock ECU looking for 14.7 stoich when actually running a lower stoich found in various racing fuels) creates a lean condition and could be very dangerous.

Is this true, why? How dangerous could that be? What if a different race fuel was used that had a stoich of 14 which is more than the 13.4, but still not quite at pump 14.7? Is this concept of stoich a/f ratio generic enough to be applied across all makes/models assuming the ECU is 100% and not tuned?

#2

The ecu bases it’s calculations of lambda as the o2 sensors can’t tell the difference between stoich on race fuel and stoich on pump gas or stoich on e85 for that matter. Stoich is stoich as far as o2 voltage readings are concerned, if that makes sense.
Wideband vs. narrowband o2 sensors also plays a role in how much (if any) adaptation the ecu can make for you.

Where you can really run into trouble is fuel injector calibrations and fuel pump capabilities.
As an example, running .8 lambda(11.76 afr on pump) on wot with pump gas translates into running .85 (12.5 afr on pump) lambda at wot on ms109 without any fueling changes on a narrowband ecu.

#3

Does all of that apply to a 9 year old Ford Mustang jude? He asked about the terminator.

#4

This is a better question for CV as I’m not familiar with the ford ecu’s.
But the basic fundamentals of o2 sensors apply and it would depend on if the car was narrowband or wideband… i suspect it’s narrowband.

#5

Thanks jude, and hopefully CV will chime in at some point. Sticking with the terminator stock ECU for now for purposes of discussion, is the WOT lambda set at .8 by the Ford engineers or is the lambda a result of an AFR target set by Ford engineers? If you have pump gas at 14.7 and a target lambda of .8 (because 1.0 is complete combustion and .8 is “rich” enough for safety) you get the 11.76 AFR ( .8 x 14.7 = 11.76)… OR is the 11.76 AFR set by the Ford engineers and the result is a .8 lambda (target AFR / stoich = 11.76/14.7 = lambda = .80)? If the AFR target is set to 11.76 based on pump gas at WOT resulting in a .8 lambda then why does running race gas with a 13.4 stoich go lean? Is it because an 11.76 AFR target on the basis of a race gas stoich of 13.4 results in a .88 lambda (11.76/13.4 = .88) and that isn’t as rich as a .8 lambda? Is an 11.76 AFR on pump 14.7 stoich different than an 11.76 AFR on race 13.4 stoich… thus comparing the lambdas is a way to normalize(?) across fuels? This is very interesting to me for some reason and I appreciate any and all input.

#6

Lambda is independent of stoich AFR of different fuels so yes it’s somehow a normalization.
As you have already said the ECU uses a fixed reference value (in that case 14.7:1) to calculate the fuel mass needed for a given mass air flow.
As a simplified example: (MAF) 250mg/stk*(reference AFR)0.068027/(Lambda setpoint)1 = (Fuel mass) 17mg/stk

But as there are tolerances in fuels and hardware there are also adaptations that (try) to compensate that effects. These adaptation values (typically one offset and one factor) are learned when the ECU is in closed loop fueling operation (on binary lambda sensor equipped cars only with a lambda setpoint =1). Setpoints for fueling are always given in lambda on “real ecus” :wink:

Fueling calculation then is:
MAF * reference AFR * adaptation factor / lambda setpoint = fuel mass

On MS109 the ECU should adapt a factor of about 1.09 to reach lambda 1 in closed loop operation and therefore compensate the different stoich AFR even when you are not in closed loop as long as the correction factor is applied on all operation points.

So in a perfect word you would not run lean if the car is adapted to the fuel even if you only have a binary sensor. In real world conditions all kind of stuff can throw this off depending on what’s been calibrated.

As you have someone here that works for Ford maybe he knows the exact strategy that is used on that ECU.

#7

No need to repeat what josh amd bob said, just wanted to say holy shit, silentbob is here, for those that do not know, this guy knows his stuff.

Thanks for being here and on n e f, you are a huge asset!

#8

Thank you Silentbob I appreciate your insight. Very helpful. I’m going to spend some time digesting the information you presented.