Let's get technical: Tuning the ECU for mpg

[DrivenByNothing]

My approach to better fuel economy has been mainly to increase fuel efficiency.

Enough background, I'll get down to the meat.

Ok, basically your ECU is watching you. Well, not you, but your driving via sensors. It's taking in information and making educated guesses based on it.

I've been researching Nissan ECU's, but I'll try to keep this as broad as I can.

The main things your ECU is looking at to decide how much fuel to inject are a) RPM b) load. Load is calculated from sensor input (ie O2, MAF, MAP, etc.) Depending on engine conditions, the ECU will run in either open or closed loop mode.

Open loop is entered at WOT and at warm-up. Open loop uses fuel and timing values found in 'maps' in the ECU's memory. For a given load and rpm, the ECU decides the pulse width frequency of the injectors. There are only so many programmed rpm and load values, so when the actual values fall somewhere in between, the ECU's processor interpolates to find a happy medium between two known values. Open loop is more or less static (ie the ECU doesn't care how rich/lean the mixture is because it's not taking input from the O2 sensor.)

Closed loop is your friend. This is the mode in which the ECU utilizes the O2 sensors' voltage(s) to make fuel and timing decisions. Similar to the open loop, there are 'maps' for the ECU to read from for given rpm and load values. However, these values are 'difference' values rather than 'absolute' values. Difference values tell the ECU how much to alter the current setting and in which direction (positive or negative).

Closed loop sounds great, right? Well, it has potential. Ok, that's sort of a pun (voltage=potential difference). 99.999% of EFI cars use a narrow-band O2 sensor. The voltage ranges between 0 and 1 volt and it's probably the least accurate sensor on your car (outside of the 'No Oil Pressure' light I had on my Nissan or a 'Your engine is a pile of molten metal, pull over.' light.) In reality, the sensor is there to protect the engine from running lean. The values aren't accurate enough to clearly separate lean and rich. If it runs way rich, the ECU will usually cut back fuel until it's lean again, richen, lean, etc.

So what's the solution? Wideband. Much more accurate, but also more expensive. VW has begun using wideband O2 sensors in their vehicles. I don't believe they've matched the wideband's tech with the ECU's programming just yet, but it's definitely a step in the right direction.

[GasSavers_Lincoln]

I use the Innovate Motorsports LC1 wideband with my tuner.
It was $199 and I don't think that is bad for a wideband.
http://www.innovatemotorsports.com/products/lc1.php

[Snax]

I disagree with the idea that a narrow band sensor is less accurate. It is in fact very accurate, but over an extremely narrow range. This allows the computer to see conditions that are lean or rich very quickly - but with the caveat that it can't tell how rich or lean.

In other words, it's the difference between finding the vertical center of a bubble vs. the geographic center of a puddle. Both sensors will do it just fine, but the latter will tell you how far you are from it.

That said, I think one of the largest improvements in fuel economy can come from 'throttle pump' tuning. (The amount of fuel dumped in when the accelerator is initially depressed before the oxygen sensor can respond.) A generous amount of throttle pump will tend to make a motor more responsive and zippy to throttle increase, but it often does so by over-richening the mixture. The flip side is that response can fall flat if one attempts to hold it too lean.

[Spencyg]

I'm glad this discussion is taking place because this topic is where everybody seems to fall flat on their face.

No modification can really fool a modern (OBD2) ECU. There are too many checks and balances built into modern engine management to allow a backyard mechanic to make a tweak here and an adjustment there which results in consistent and long-term fuel savings.

That being said, going directly to the source of the trouble and modifying that will get you lots of options in both savings in fuel and increases in power. As has already been mentioned, OBD2 (anything from 1996 on) uses two different management strategies...open loop and closed loop.
Open loop uses a set of fuel curves and ignition curves which control the injector pulse width. The maps are 3 dimensional with the X axis typically being RPM, Y axis being load, and Z axis being either pulse width of the injector (for the fuel map) or ignition advance (ignition curve). Open loop is called upon when the engine is cold or at WOT.
Closed loop uses the oxygen sensors to compare actual oxygen content of the exhaust to a 3rd set of maps. These maps has various A/F ratios for different loads and throttle settings. The maps are referenced much the way the open loop utilizes the maps.
Narrow band vs. wide band is definitely a consideration for those of us who modify the stock injection computers, but as far as one being more efficient in a stock system versus another....you won't find a difference. Both are referencing stock maps to make their decision. ANY OEM wants to avoid warrantee returns as it cuts into the bottom line. We all know a rich engine is a happy engine (from a longevity standpoint), so most engines are programmed to run richer than stoich. My Pathfinder stock fuel curve (closed loop) often times asks the injectors to provide enough to maintain 12.5:1. This is a serious problem from an efficiency standpoint. I don't have any hard data to prove it, but I would bet Honda doesn't run their lean-burn series of engines any leaner than 14.7:1....they just run close to the line, and do it more often.
The code used to program OBD 2 is not universal and is proprietary to each car maker. The big 3 (Ford, GM, Chrysler) have released their programming code for many of their sports lines to tuners...this is why you can buy a handheld tuner for a Mustang, Camero, Corvette, etc. but not one for a Toyota, Nissan, or Honda.
The ONLY way to mess with the stock curves is to either do a piggyback unit or complete standalone. There is an interesting oxygen sensor interrupter made by www.eagle-research.com which supposedly fools the computer into thinking the oxygen sensor is reading rich, and leans out the mixture. I haven't used this at all, but when looking at the way the computer utilizes the oxygen sensor, it appears as though you could potentially do some harm to the engine when running under a high load condition. I would love to hear of some experience with this option.
I am installing a Greddy E-Manage Ultimate system into my truck. This system allows full control over fuel and ignition curves, and actually has two completely different maps which can be toggled between.
Stand alone systems are pretty serious business and I wouldn't personally recommend them for what we're all trying to accomplish here.

Any modifications to the fuel and ignition curves obviously are followed by the risk of detonation and engine damage. Its all well and good to tune the engine so it is running at stoich, but if provided with the proper additional hardware, and engine can run consistently and reliably for an indefinite period of time at upwards of 20:1. This would require water mist injection which should be easily set up....there are kits available for the turbo and supercharged crowd to control their detonation. At 20:1, you would be using roughly 55%-60% of the fuel which would be used at the stock fuel setting....

This is the direction all future modifications to an engine needs to go. Gone are the days of making a tweak with a screwdriver and realizing gains as a result. Its all about computers, sensors, and programs these days....

Spence

[SVOboy]

ECUs have been talked about quite a bit on GS over the year and months we've been around, and I used to do a lot with it...until I realized there really wasn't much to do with it. You can tune and this and that and pull more power out, but on a well designed engine you can't do much unless you've extensively modified things (at least on a honda, there is very little wiggle room for power or economy through ecu modification with stock equipment).

It may be different with a truck of course, but I would be wary of doing anything that will cost too much money expecting magnificent results. And also, I wouldn't run 20:1, poisoning the air is not worth a few mpg...

[JanGeo]

I leaned out be 1980 VW CIS injected engine and ran it 163k miles without problems - stock setting it ran 27mpg dead cold 0 degree winter or 80 degree summer and with some injector mixture adjustments by building up the "air cone" mass air flow sensor the mileage was up into the 37mpg range. Problems when you get into the 18 to one and higher start to show up at highway speeds but around town should not be a problem. It would be a real kick to get the xB to achieve 70-80 mpg by leaning out the A/F ratio on occasion. I think it is time for the dual A/F gauge order to be placed so I can see what is going on. Been feeling a little skipping lately at light throttle as I approace 14,000 miles - btw I now have 10,000+ miles on Synlube and all seems ok.

[Spencyg]

Ah yes....you're referring to Nitrogen Oxides. Though this is a very good point, the use of water injection negates the issue. Nitrogen oxide production increases as combustion temperatures rise. Water injection cools the incoming charge by a huge amount due to its latent heat of vaporization properties. As a result, you can run very lean, very advanced, and your EGT's will be lower than if you were running a stock fuel curve.
To comment on the 18:1 and above ratios being a problem at highway speeds, I think this would be more of an issue if you were to use that oxygen sensor signal modifier I referred to in my previous post. When playing with actual fuel and ignition curves, loads at various speeds (RPM's) can be compensated with a little more fuel (or ignition) if a little extra power is needed at a particular RPM. I think at the end of the day using water injection in conjunction with lean burn AND ignition advance would make for very comfortable cruising power under FE mode...
I agree that modifying an ECU without any other modifications is not very helpful....but as with anything, if you understand the whole system and the effects each variable has on every other variable, you can make significant improvements. Don't assume OEM engineers have designed everything perfectly. I work for an OEM and the design philosophy is based on dollars, not engineering.



Spence

[davidjh72]

Not only does Open Loop occur at startup and WOT, but also at closed throttle (CT? NT=no throttle?) when coasting at zero load.

Say I'm driving along the freeway in my Saturn, let off the gas, in gear, manual transmission, to take a rather long exit ramp. Or downhill. With zero throttle, load drops to zero and the ECU goes into Open Loop. Fuel usage shows well above the idle rate of 0.2 - 0.3 gallons/hour, but not as high as if I still had my foot on the gas. So the ECU is commanding the injectors to either shut off (fuel cut-off on decel - yeah! but SG doesn't know how to report it) or minimal injection to keep fuel trim from not leaning out too much.

[JanGeo]

Yeah we know about that because the ecu continues to calculate the injection of fuel all the time is registers on the SG but all indications are that there is no fuel being burned in my xB because on a rather long hill I see no temperature rise in the engine which does occur at those fuel burn rates normally. So it surbtracts from the trip MPH a little and throws off the gallons burned in the tank - by maybe a tenth of a gallon in 10 gallons 1%!!

[usedgeo]

Well I own an LC1 now. I have not installed it. Can I program one output near stock and one for say 17:1 and switch between the two with a toggle switch. Just thinking about it.