View Single Post
Old 07-08-2011, 08:34 PM   #28 (permalink)
jfitzpat
EcoModding Lurker
 
Join Date: Jan 2010
Location: Los Angeles, CA
Posts: 54
Thanks: 0
Thanked 20 Times in 8 Posts
Quote:
Originally Posted by EdKiefer View Post
So you have to somehow adjust this or any changes will be reverted, learned back to the ECU O2 closed loop values, which are for most vehicles around 14.7 .
Sorry, I saw this in the email 'best of' today and just had to comment. There are really three main approaches in altering factory tunes:

1. Reflashing
2. Piggy Back
3. ECU Replacement

Reflashing is a broad term. For many ECU's, there are make/model specific parameters that can be tweaked with factory tools and/or ripped off versions of factory tools.

For even more hardcore reflashing, code in the ECU's is functionally altered. For most makes/models there is a an active community in this area, though some makes/models are easier than others. For example, Suburu and Mitsubishi ECUs have been heavily hacked and reflashing hardware is cheap and readily available. On the other hand, vintage Hondas are often reflashed by taking apart the ECU and replacing masked ROM parts with reprogammable memory chips.

Piggy Backs generally use stock ECU's, but then fool them into doing desired things. They do this by replacing sensor signals. If I under or over report MAF or MAP to an ECU, it's response is going to alter. Piggy backs are less flexible, but often perfectly adaquate for typical light weight performance mods. I change the airbox, VE of the engine is changed, so I need just a scootch more fuel...

ECU replacment is sort of the ultimate in control. These range from inexpensive DIY projects, like Megasquirt, to very expensive replacment ECUs. There is a very nifty line out of Australia that uses FPGA technology and a pictoral scripting type language so that you can, effectively, alter virtually anything that the ECU does.

In order for any of these things to be used to get economy, some considering has to go into where fuel is wasted, and why. When a relatively modern vehicle is 'closed loop', it is almost always running very close to lambda 1.0.

People refer to this as 14.7:1, but that is wrong. The actual air fuel ratio to acheive lambda 1.0 is fuel dependant, it moves around a lot just with altering fuel blends. The ECU is not using AFR, but one or more O2 sensors. The sensor itself measures 'equivelency ratio', or 'fi'. Lambda, the scale we use in engine science, is the recipricol of fi. Until the last few years, most cars used 'narrow band' O2 sensors. That is, they only measured 'fi' over a very narrow band. Basically, 'lambda 1.0 = 450 mV'. If you go richer or leaner, the sensor very quickly shoots to about 1V or about 0V.

So, in those vehicles, the only point that the vehicle knows actual lambda with any confidence is lambda 1.0. To understand why this leads to fuel being wasted, we need to understand why this point is being targetted in the first place.

Lambda 1.0 is the 'stoichiometric ratio' for the fuel. That is, it is the optimum mix of air and fuel for the most thermal energy released. This is why it is relatively easy for a narrow band sensor to detect. If combusion is richer, then there are unused HC's in the exhaust, if leaner, excess O2. This sounds really efficient, but it's not. If you want 'best power', you run richer, somewhere around .86 lambda. If you want 'best economy' you run leaner, around 1.05 lambda.

This is because a combusion engine is not a steam engine. It isn't just about how much thermal energy can be released, but how much mechanical advantage can be gained pushing against a piston with expanding heated gases. Running richer than lambda 1.0 also changes the speed and chemistry of the flame front, so peak pressure is moved to a point of more mechanical advantage. And since fuel is plentiful, all O2 gets used. Conversely, going leaner means that all the fuel can get used which, combined with the changing properties of the flame front, still has reasonable mechanical advantage.

The reason that the car runs at 1.0 is purely for emissions. Lambda 1.0 has two important properties. First, it is a point where certain polutants kind of 'bottom out', go away from that point and things like CO and HCs soar. Second, it peak EGT. Peak thermal reaction means hottest exhaust gas. And that heat is needed for a modern cat to work. In fact, cat efficiency plummets very quickly as you move away from peak EGT.

So, much of the time, the vehicle is targetting lambda 1.0, strictly for emissions. But when you stomp on the gas, this is not the place to be. CHTs are fairly high (proportionally) at lambda 1.0 (they peak just rich of stoich, in the .9's), so if you stomp on the gas and stay there, you won't get best power and you'll have high CHTs, lots of fuel, lots of heat and lots of pressure is basically a recipe for detonation. So the ECU runs richer. Best power not only means more power to do whatever it is you stomped on the gas for, it also means lower CHTs.

But, remember 'narrow band' above? An ECU armed only with this sensor doesn't really know for sure where it is running in this case, so factory tunes virtually always error towards 'richer', which costs power, but also brings CHTs down further (by moving peak pressure, not by actually using fuel as some sort of 'spray on coolant').

Performance tuners lean this out (and potentially shift peak pressure with timing as well). They want the lost power back. But this also means better fuel economy for this mode of operation.

Newer vehicles are much more likely to have a 'wideband sensor', or a better idea of exactly where they are running in the .7 to 1.3 range. So those factory tunes generally waste less fuel in those cases. Though there is generally still some waste. The most cutting edge is to use wideband closed loop at targets other than 1.0, but that is another subject.

But people here already avoid that case. So, while they will get savings, it won't be as much as my lead footed wife might get with the same change.

If you don't care about the environment, better economy is not all that hard to acheive. I've done this with a number of vehicles on the dyno and in test environments, like UC Riverside. Basically, I remove the narrow band sensors and replace them with wideband sensors. I then simulate the narrow band signal for the ECU. Think of it this way, the ECU is targetting 450 mV. On a narrow band, this is lambda 1.0, but if I'm reading the exhaust mixture with accuracy at other points and simulating this signal, I can make 450 mV anything I want. If you slowly shift it towards theoretical best economy and let the ECU's adaptive algorythms catch up, you can typically get some clearly measurable fuel savings for same weight, speed, and distance. If you optimize timing to match, there is more savings still.

But emissions get terrible and you potentially destroy your CAT. On my development 'road bed' I still simulate the narrow band signal, but 'tighter' than a real narrow band. That is, I collapse the normal S curve of a typical narrow band. If you track O2 sensor voltage with an OBD-II scanner, you will typically see it oscilalte back and forth, rich/lean/rich/lean as it chases stoich (lambda 1.0). These swings are nec. for cat operation (you need some O2 to burn), but they are a little sloppy. So by making my stock ECU chase a tighter signal, measured emissions actually go down (because the cat spends more time at peak efficiency). There is a tiny fuel savings, but it is so small that it is certainly not worth persuing for its own sake.

When I run with the non stock ECU, I generally run closed loop even in rich operations, which gives the typical driver a lot of savings, but, again, not so much for folks here.

My point in all this is that there are savings to be had here, but the big Kahuna is off limits unless you don't care about emissions. Still, there are some non obvious things that ECUs do that can also be tweaked for savings. Like tweaking shift points to match your driving habits.

Sorry to just jump in (and somewhat rehash), but it is 'my area', so to speak.
  Reply With Quote
The Following 8 Users Say Thank You to jfitzpat For This Useful Post:
California98Civic (07-08-2011), d0sitmatr (07-09-2011), meelis11 (07-13-2011), Phantom (08-11-2011), Piwoslaw (07-11-2011), SDMCF (10-30-2015), todayican (04-25-2012), Xist (10-31-2015)