Modifying Closed Loop Cars that Need LOTS More Fuel! Part 1

Back in our January 2005 issue (see Altering Closed Loop Mixtures) we covered how to modify the air/fuel ratio of a car that’s always in closed loop. (‘Closed loop’ means that the output of the oxygen sensor effectively controls the mixtures.) In that case, we used the pictured Simple Voltage Switch kit to disconnect the oxygen sensors once the load exceeded a preset threshold. The resulting mixtures were then tuned with the Digital Fuel Adjuster kit.

But what if you’re dealing with a car that always runs in closed loop but needs lots more fuel flow – for example, you’ve turbo’d the engine? In that case, just disconnecting the oxy sensors is unlikely to give you the result you want – you need a heap more fuel flowing through those injectors. After many weeks of experimentation on a car that has super intelligence in its engine control system (it’s a hybrid Toyota Prius) and needed a lot more fuel at full load (it’s been turbocharged), we’ve come up with an approach which we’re pretty confident will work on any car that normally always stays in closed loop.

How Cars Work in Closed Loop

‘Closed loop’ refers to the ECU following a certain control approach. In closed loop it allows the output of the oxy sensor(s) to correct the pulse widths with which it is triggering the injectors. Most oxy sensors are narrow band; that is, they are not designed to be able to accurately measure air/fuel ratios of say, 12.5:1. The output of the sensor is high when the AFR is richer than 14.7:1, and low when the AFR is leaner than 14.7:1. (For more on oxygen sensors, see The Technology of Oxygen Sensors.)

When the output of the oxy sensor is high (eg 800 millivolts) the ECU leans out the mixtures. When the output of the oxy sensor is low (eg 200 millivolts) the ECU richens the mixture. The result is that in closed loop a car has mixtures (and so an oxy sensor output) that rapidly rises and falls, a bit like a sine wave. The average AFR, however, hovers around stoichiometric, the chemically correct ratio of air to fuel that ensures best combustion. (And also allows the cat converter to work most efficiently.)

The best way to see if your car works in closed loop all of the time, or only at lower loads, is to fit a Mixture Meter – see Smart Mixture Meter, Part 1. The dancing LEDs will soon show you the ECU strategy – if they continually oscillate back and forth, the car is in closed loop. If the LED remains steady, open loop is being used. (The illuminated LED may slowly move, but once a constant load is reached, it will stabilise.)

For more on closed and open loop, see the breakout box at the end of this story.

Yes, we know that’s an awful lot of electronic kits mentioned in the first five paras. However, they really are the cheapest and best way of achieving these outcomes.

Modifying Closed Loop Mixtures

A car that is always in closed loop cannot effectively have its mixtures modified.

(Well, that’s not quite true. If it’s a car that uses a wide-band oxygen sensor, and if the modifier is able to access the ‘target air/fuel ratio’ part of the ECU software maps, the air/fuel ratio can be modified very well. And some very experienced software re-writers can do just that – but for most of us, that’s not a viable approach. Especially if we’re dealing with a one-off project.)

By why can’t the closed loop car have its mixtures modified? Basically, because the car will immediately ‘learn around’ any changes that you make. Increase the fuel pressure and the car will run richer - but only until the ECU learns from the oxy sensor(s) that there’s too much fuel going into the engine. Reduce the output of the airflow meter and the car will run richer - but only for as long as it takes the ECU to make a correction. (And in some cars, that can be just seconds!). 

So when in closed loop (ie the ECU is using the input of the oxy sensor to fine-tune mixtures), the mixtures must always remain as the ECU expects them – stoichiometric or about 14.7:1. 

The above statement looks pretty obvious – but it isn’t. Take for example the fitting of a rising rate fuel pressure regulator on a car that’s been turbo’d. If you allow the fuel pressure to rise while the car is operating normally in closed loop, the injector pulse widths will be pulled back. Why? Because the ECU will see that the engine is running rich (because of the higher fuel pressure) and so will trim the mixtures leaner. And as part of its short-term fuel trim learning, all the injector pulse widths will then be leaned out. (After all, the ECU figures that the fuel pressure reg has just gone mad!)

Don’t see the point? OK, let’s take an example. Say you’re working with a car that runs in closed loop at all revs and loads and a mixture check shows that yes, the air/fuel ratio is at 14.7:1 all the time. You’ve just turbo’d this car and desperately want mixtures much richer than 14.7: when on full boost. Say, 12.5:1 would be nice...

As part of the upgrade, you’ve fitted a rising rate fuel pressure reg that increases fuel pressure disproportionately with boost pressure. (So, as boost pressure rises by 7 psi, fuel pressure might rise by 20 psi – see The Vortech Fuel Management Unit for more on this topic.) This increase in fuel pressure will force more fuel through the standard injectors still being operated at their standard pulse widths, so enriching the air/fuel ratio. You intend switching out the oxy sensors to force the ECU out of closed loop whenever the load is high – and have set that switch threshold at 4 psi boost. (If you switch out of closed loop as soon as there’s any boost pressure at all, you’ll dramatically drop in fuel economy – a well-matched turbo will develop positive boost pressure at quite small throttle openings.)

So now we have the situation where even in closed loop, the fuel pressure is rising. To take an example, say you’re driving on the flat then reach a gradually steepening hill. Initially, the car is in closed loop, there’s no turbo boost and everything is fine. Then you gradually start climbing the hill and the boost pressure gauge shows 1 or 2 psi boost. In response to this, the rising rate fuel pressure reg starts to increase fuel pressure – and so also as a result of this, the ECU reduces pulse with of the injectors, to keep the mixtures at 14.7:1 even at the heightened fuel pressure. The short term ECU fuel trim then learns that the system is over-fuelling, and pulls back all injector pulse widths.

The hill steepens and you reach the point at which you switch out the oxy sensors. But the ECU, which has been learning for the past few seconds that the car is being over-fuelled, is now faced with a dilemma. It’s lost its feedback loop from the oxy sensors, but last thing it knew, the mixtures were too rich. It then decides that its ‘forced open loop‘ mixtures should be leaner than usual.

Aaaagh. Your increased fuel pressure – even in forced open loop – has resulted in mixtures that are now leaner than they were previously, and much leaner than they would be if you went straight from no throttle to full throttle!

As we said above, the trick is that when in closed loop, the mixtures must always remain as the ECU expects them – stoichiometric or about 14.7:1. 

OK, so how do we keep the ECU seeing what it would normally see (and so staying happy in fuel trims), and then suddenly change the fuel flow at just the same time as we switch the oxy sensors out? The trick is to use two independently adjustable fuel pressures.

Modifying Oxy Sensor Signals But why all the hassle? Why not just modify the output of the oxygen sensor(s) to indicate to the ECU that the car is running leaner than it really is? Voila! One richer mixture... Trouble is, modifying the sensor outputs is very difficult, even with a programmable interceptor like the Digital Fuel Adjuster kit that can be configured to work with oxy sensor voltages. Basically, because the sensor’s chemistry makes it switch from high to low (and low to high) as the mixtures pass through stoichiometric, altering the output signal doesn’t change the switching point of the sensor. In practice, narrow band sensor signal modification either results in no change to the mixtures, or the ECU thinks the oxy sensor is stuffed and ignores it. Wide band sensors? That’s another story.

Two Fuel Pressure Systems

Take this scenario. It’s much the same as the one above but with one critical difference.

You’re driving on the flat then reach a gradually steepening hill. Initially, the car is in closed loop, there’s no turbo boost and everything is fine. Then you gradually start climbing the hill and the boost pressure gauge shows 1 or 2 psi boost. However, fuel pressure remains absolutely stock and the oxy sensor closed loop system continues to work normally. In response to the increased airflow, the ECU increases fuel injector pulse widths so as to keep the air/fuel ratio standard.

Then – WHAMMO!!

In one fell swoop you switch out the oxy sensor(s) and increase the fuel pressure! The ECU doesn’t know anything about the increased fuel pressure and so fires the injectors at normal (or even a little above normal) pulse widths. But with your suddenly increased fuel pressure, you can squeeze a heap more fuel through the standard injector opening times.

You have enriched mixtures (the amount of enrichment being best set by adjusting the high load fuel pressure) and the ECU doesn’t know a thing about it... and so cannot learn around it.

So how do you organise a fuel system that can dramatically switch from one (regulated) fuel pressure to a much higher (also regulated) fuel pressure? We’ll show you how next week...

More on Closed Loop 'Closed Loop' is the term given to the ECU behaviour when the oxygen sensor signal is being used to largely control how much fuel the injectors are adding to the intake. In most cars the ECU works in closed loop most of the time - when the car is warmed up and idling, in constant throttle cruise - and so on. With this control strategy the ECU watches the oxygen sensor output and if the mixtures are getting a bit rich, it leans them off. If the mixtures are getting a bit lean, it richens them up. This causes the mixtures to fluctuate rapidly around 14.7:1 air/fuel ratio - what's called the stoichiometric ratio. However, an air/fuel of 14.7 doesn't give the best power, so when you put your foot down, in most cars the ECU forgets all about closed loop and goes instead into an operating system called 'Open Loop'. This just means that it ignores the output of the oxy sensor, instead picking the right amounts of fuel from its internal memories. Typically, with increasing load, the air/fuel ratio outside of closed loop might jump to 13:1, then 12:1 and then even richer still at 10 or 11:1. The final common operating approach is when the injectors are completely stopped - yes, they're actually switched off sometimes even when the car is driving along! This happens on the over-run - you're travelling along at 100 km/h, reach an 80 km/h sign and lift your foot. The ECU will then turn off the injectors until either you reapply the accelerator or the engine speed drops to near idle revs. Like all things, these ideas apply to most cars - not all. Some Porsches, for example, stay in closed loop all the time - even when the mixtures are richer than 14.7:1. In other words, the oxygen sensor (a special one) is used to give mixture feedback to the ECU in all operating conditions. Other cars have a 'lean cruise' system, where on the open highway the mixtures will gradually lean out to say 15 or 16:1, so saving fuel. And as we have discussed in this story, some cars use constant closed loop to keep their mixtures always at stoichiometric (except on the over-run). For example, the Chrysler Turbo Prowler is a car that maintains closed loop 14.7:1 mixtures all of the time. (For more on this engine – Designing a Factory Turbo Engine).

Check Engine Lights? In most cases switching the oxygen sensor(s) out will result in the setting of a fault code in the ECU. Whether this lights the dashboard Check Engine light depends on how often it occurs and the individual ECU fault coding. Unfortunately, we don’t know of a way around this – simulating an oxy sensor signal has proved to be unsuccessful in preventing the ECU knowing that it is has lost the input of the real sensor.