  |
|||
 |
|||
| |
Fuelling
Correct fuelling is extremely important on n/a engines, but on turbos it's critical. Get it wrong and you could lose the engine before you know it.
With a n/a engine we're talking accelerated wear like losing the bores early if it's running too rich, and general overheating consequences when it's running too lean. A few clicks and pops in between.
In contrast, a turbo running lean under boost lives on borrowed time. Keep it at full boost while lean, and there's overheating and detonation within seconds. If you're lucky the head gasket will give way and you'll have to slow down either you like it or not. If your luck's out then a piston or two will have melted, like they've been attacked by a blowtorch. The melted aluminium may land anywhere - on the chambers, the valves, the exhaust turbine...you get the picture.
Flashback --- let's start from the beginning:
The engine burns a mixture of air and fuel, but not any mixture. If it's one gram of air and one gram of fuel it won't burn, because it's too 'rich', i.e. it's got too much fuel, it's rich on fuel. If it's got 1Kg of air and 1 gram of fuel it won't burn either, because it's too 'lean', i.e. it's lean on fuel, this one is practically all air!
It has been found experimentally that the ideal mixture, the one where these two elements are perfectly balanced is 14.7:1 by weight, i.e. 14.7grams of air and 1gram of fuel. Or 14.7lbs of air and 1lb of fuel. This is called 'stoichiometric', meaning in Greek measure of the elements.
.
Knowing that it's ideal is very nice, but ideal for what exactly? Ideal for max power, ideal for max economy, ideal for minimum emissions, what?
...You've guessed it, it's emissions. Save the whales, hug a tree etc...
A stoichiometric mixture will burn and produce the least amount of toxic goodies, say hydrocarbons and CO. As it gets leaner it produces less CO and more hydrocarbons, as it gets richer more CO and less of the others. Stoich is a fine balance, where even more nasty stuff of a different nature gets produced, but the catalyst takes care of that. It's the point where all oxygen has been used up, and there is no excess fuel lying about.
Mixtures leaner that stoich (15:1 ~ 18:1) provide better fuel economy (unsurprisingly, since they contain less fuel!) and sometimes we do want our fuelling to be this way, especially during deceleration, or light throttle openings (cruising). But put your foot down at the strip, and if you're running 18:1 you ain't going anywhere. Acceleration needs richer than stoic, 14:1 ~ 13:1, sometimes even 12.5:1. A turbo under low boost needs to be around 13:1 minimum, and under full boost richer than that. HOW MUCH richer depends on chargecooling, ignition mapping and other factors.
Two extreme examples: a light-pressure turbo with a kick-arse chargecooler that brings the chargetemps down to ambient all the time, with perfect ignition mapping and well-designed combustion chamber, running 100-octane fuel would not need to run richer than 13.5:1 under full boost. In fact, it would best if it didn't, because it wouldn't make any more power and the extra fuel would be a waste. The same engine running TWICE the stock boost, on 95-octane fuel, crappy intercooling, fudged ignition mapping and overtaxed cooling system, would run best at around 11:1, or even 10:1. This is 20-30% more fuel, mainly to patch up the cooling issues. In the trade it's called a 'fuel shower' as the extra fuel droplets absorb all that heat from the red-hot valves, pistons and compressed air. It also affects the speed of burn, but heat absorption is the main reason. It wastes fuel, and it's a botch (because the specific heat of fuel is not that high), but it's simple and it works.
There are two kinds of oxygen sensors: narrow-band and wide-band. The first ones typically have 2 or 3 wires and they are found as OEM equipment on catalyst-equipped cars. The other ones are more costly, have more wires, and are usually found in Rolling Roads, Tuning shops (serious ones) etc. You can buy WB sensors from the internet nowadays from several sources, but beware of the displays. They might need a 'translation' stage to display within the 0-1v range, and even then the linearity has to be verified. They also need to be recalibrated every so often, as they can get contaminated and lose their accuracy. Most cars paying for RR time are tuned and tend to run rich and even sometimes use leaded fuel. Both of these poison the sensors pretty quickly. Ask a RR operator when was the last time he had the sensor recalibrated, if you want a cheap laugh!
Lambda (or Oxygen) sensors really are air batteries, creating voltage depending on the oxygen content of their (filthy) environment. That's why they slowly deteriorate with age and use, and after a few years they are slow and inaccurate. It's also why you should never hook the sensor to a multimeter and try to measure its resistance. The tiny current from the multimeter might kill the sensor dead. Here is more info on lambda sensors (local copy here)
AFR gauges simply display the output from the sensor, so they can't really be blamed for erroneous results. An exception would be if the gauge's impedance would be low enough to affect the signal going to the ECU - high quality gauges from reputable companies wouldn't have this problem.
We have to remember that the purpose of the OEM lambda sensor is not to produce a lightshow for the driver, but to provide feedback to the ECU about the exhaust gases being slightly leaner or richer than stoich. This is vital, if the cat is to operate properly and not self-destruct. Rich mixtures can coat it with carbon deposits and kill it, and lean mixtures will cause it to overheat and die.
Therefore, the OEM sensor is only accurate around stoich (14.7:1) and wildly inaccurate beyond that point. For example, a 9:1 mixture will register as 'full rich', as will a 10:1, along with a 12:1. Similarly, an 16:1 will appear as 'full lean', same as 20:1. Not only is the useful band narrow, but accuracy drops sharply the further away the reading is from stoich. It's also very temperature-sensitive. The ECU refuses to listen to the sensor until the engine has reached operating temp, and there's a reason for that. While the sensor is cold, rich appears richer, and lean leaner (depicting our society). As it heats up, the useful range 'shrinks', and the same voltage now relates to a different AFR!
Note how CO and NOx are at their highest at stoich (but the cat takes care of them). Also note the almost linear relationship of CO in the 'rich' section. That's why CO emissions at full boost are a good indicator of how rich the engine runs.
Two caveats then - the readings are not linear, and only worth something when the sensor is hot. Even so, the sensor has to be in good operating order (hard to tell just be looking/measuring it). Even so, the most accuracy we would expect would be in the range between 13.5:1 ~ 15.5:1 with not much linearity either.
This lack of linearity can be easier seen in this graph from the Autometer site:
This gauge has 6 green LEDs, to indicate 'rich'. A 750mV output from the sensor will register as one green LED, and for a fully warmed-up sensor that means AFR = 14.5. Two green LEDs would result from 800mv, or AFR= 13.3 and three green LEDs AFR = 12. Due to the lack of linearity it's highly unlikely that such a gauge will ever show 4 green LEDs (900mV), certainly not at operating temps. I have verified these measurements with a Wide-Band sensor operating in parallel. However after 20K miles that accuracy had gone out of the window. The sensor was not even accurate around 14.7:1, it was showing leaner than reality, fooling the ECU into dumping more fuel than necessary. Not good for a clean engine without carbon deposits.
The red LEDs standing for 'lean' are even more tightly packed, more like a lean indicator (and not gauge of how much)
For tuning and ECU-mapping purposes, a narrow-band sensor is nowhere good enough - but for troubleshooting after the car has been setup, it's brilliant. A narrow band sensor cannot tell you if the engine runs rich enough, but it can tell you if it runs too lean. Tuned turbos shouldn't be without one, as running lean under boost can cost dearly.
The graph below displays this particular shortcoming of narrow-band sensors, compared to a wideband. See how evenly the wideband spreads from very rich (10:1) to very lean (20:1). In comparison the narrowband struggles between 11.5 and 16.5, with usable resolution only between 13.5:1 and 15.5:1
Finally some more technical take on narrowband sensors (taken directly from an old, long and extremely interesting thread of the GTR forum: (if the link is dead click here)
The oxygen sensor utilises the "Nernst effect." Nernst's law measures the voltage between two materials in close contact, one of which is a known constant. When the sensor reaches a temp. 640degrees F ±25F, it produces a voltage which is interpreted by the ECU and adjusts the fuelling to suit.
The outer shell of the sensor is Zirconium Oxide, this material is very sensitive to contamination, think of a nice piece of polished alloy and how easily it will absorb the oil excreted by you skin, well ZircOx will take up all those additives in petrol, but for the most part they don't effect it too badly, but Optimax and all those Octane Boosters do contain contaminents and if over applied or applied constantly (as in using Optimax continually) the voltage engendered is not as it should be, Citric acid cleans ZircOx perfectly, the acid actually cleans down into this 'rough' surface, we are talking very small scale here, so to the hand it is quite smooth.
Lead from fuel screws them, Phosphorous (Optimax, and most OBs') does the same only slowly. If your turbo has ever blown the seals and blown smoke out the back, the phosphorous in that burn out will kill it fairly quickly, you should always replace the sensor if this has ever happened.
Wide Band sensors used to be a bit rough (for non-professional use) and expensive.
Now they're coming down on price, and getting more reliable too (self-calibrating) mainly because the new generation of engines will have to stay closed-loop even at full throttle. The new Golfs have a WB as standard fitment, for example.
Late 2004 a new breed of WB has come out based on a variant of this sensor.
Replacement sensors are cheap, too (what you'd pay for a pattern narrowband!)
Standard dimensions, fits nicely in standard pods. Nothing like the previous generation of WideBands, that needed the exhaust to come out, drill and weld a bung, then use a laptop for the display.
Narrowband on the left, Wideband on the right
This plug has to pass through the firewall. Driving cables through a slot in the firewall is hard enough, but this beast is a challenge.
John Holmes had an easier time perforating young and innocent actresses.
The idea is to cut the smallest possible slot otherwise toxic fumes from the engine bay will be entering the cabin. Tight passage.
Detail shows where to look for the output selector. It's not easy to spot, you need a bright torch and good eyesight. Use a micro-flat screwdriver to turn it fully clockwise for the narrowband emulation (to keep the ECU happy!)
Every time the ignition is ON, the controller inside the gauge will restart and should display briefly P04 or position 4 to verify this.
Currently they are sold at competitive prices (considering rip-off Britain). They could be had for less from the States, but adding postage and (perhaps) duty would bring it close to this UK price.The one shown also has a narrowband simulation, so if the fittings have the same threads, you simply swap the existing one and keep the ECU happy.
For SupraTT specific info click here:
For C20LET specific info click here:
Here's another link on the AFR subject
Here is even more on ideal AFRs from a Toyota tuning site (local copy here)
Using EGTs vs AFRs for tuning - advantages and pitfalls (local copy here)
More on tuning based on EGTs in the exhaust section
Inexpensive wideband sensors also here and here
Well-priced new std oxygen sensors here
...on to "Gotchas & Tips"