Closed loop lambda control

[robespierre]

Hi,

I'm using a M800 with ECU Manager 3.51U2 and I'm trying to understand how the Motec closed loop lambda control works. My lambda sensor is successfully configured and reads correctly. I'm trying to use closed loop lambda control to achieve lean burn during idling and cruising and I have adjusted the lambda table accordingly for this purpose. Cells that are not used during idling and cruising are turned off with the value '0' in them.

What I don't get is how the ECU determines closed loop and the factors causing it to cycle in and out of closed loop. While idling, I can see 'LaCtrl1' changing its status between 'on' and 'off' very often and the 'lambda 1' values are quite far off from the 'La1 Aim Value'. For example, if I specified '1.02' in the lambda table, I can see swings between '0.95' to '1.07' even though the car is idling at the desired 1200 RPM pretty solidly and there's not much fluctuations in MAP readings. I have already tuned the fuel map with the lambda control turned off.

How do I adjust the lambda control settings to achieve a more consistent AFR during closed loop lambda control, and how do I control when the M800 uses the closed loop lambda control?

[Martin]

First, this is when lambda will work and when not:

Hysteresis is performed on the sites bordering the off to on transition.
ie. The sites along the border will be :
OFF during the transition to the on value.
ON during the transition to the off value.

Short Term and Long Term Trim values are adjusted to maintain the
correct Lambda. The Total Trim applied is the Short Term Trim plus
the Long Term Trim.

Lambda Control is deactivated in the Fuel Main Screen so that
adjustments may be made without being overridden by Lambda Control.
(Both Short and Long Term Trims are Zeroed)

Lambda Control is suspended during the following conditions.

- The Lambda Table value is zero
- The Lambda Sensor is too cold
- The Injector pulse width is too narrow (<1.5msec)
- Acceleration enrichment is operating
- RPM Limiting is operating
- Traction Control is operating
- Gear Change Ignition Cut is operating
- Overrun Fuel Cut is operating

Note * During suspension the Short Term Trim is slowly decayed to zero
and the Long Term Trim is held constant


You will also need to adjust that rate at which the Lambda control adjustments are made.........

Defines the Lambda Control Calculation Update Rate. Set to allow for sensor
response time.

Units: seconds

Typical: 1.5sec at low duty
Typical: 0.4sec at hi duty

Typical Axis Channel: Fuel Inj Duty Cycle


You will also need to set the step size per update period (the above).............

Dont make it larger than 80%....too large values will make it react faster but it will overshoot easily ...a bit like too much Proportional gain.


Also, make the step sizes smaller and lower RPM and Lower load, (less, slower exhaust gas flow)


Ive find that a small misfire is a nice way to determine the sensor reaction time and how long gas takes to travel down the pipe from combustion chamber to sensor. Set up an Ign Cut or fuel cut and measure the total cut of either one you use and log it fast with the lambda. When you see the IGN or fuel being cut, measure in I2 the time that it takes for from the cut till the sensor sees the lean spike. Itll be different for diffrent RPMs and Load conditions. Im not suggesting you do this for every single load site, but if you have no Idea what numbers to fill in, this is a way of measuring it.

[robespierre]

Thanks for the post! Its very helpful and I now know why it has been cycling in and out of closed loop...my injector pulse width is often under 1.5ms during idling as they are pretty big injectors (1000cc). Is there a way to enable lambda control even if the IPW falls below 1.5ms?

[Martin]

Controlling injectors below 1,5ms is not as predictable as dead time becomes a problem. I would decrease fuel pressure to get the Inj PW up.

Is your Fuel pressure regulator connected to the inlet manifold for a pressure reference?
What engine is this?
Does the engine run fully sequential?

[robespierre]

Yes, my FPR is taking reference from the intake manifold and with the vacuum line disconnected, the FP is set @ 43 PSI. This is a 4G63 engine.

[Martin]

a 4g63 with 1000s....what sort of power do you make? is it still 2L? GT40 about?

If you have trouble keeping the INJ PW above 1.5 at idle id suggest you keep the lambda control off at idle. Im not too sure about motec changing the 1,5ms to a lower value or making it adjustable. You ask them, Ive allready asked too many things and im becoming a pain for asking so many things

[robespierre]

Yes its still a 2L and running a 4088. Hoping the Motec guys see this thread and enlighten me

[RossB]

Martin is correct, you should switch the lambda control off. Apart from the problem with the short injector time I am guessing that you must have a fair amount of valve overlap if you are using injectors like that. It is unlikely that the reading from the lambda sensor is going to be reliable enough to tune with at idle speed let alone be used as a feedback for lambda control.

[xpro]

Mine is the very same, closed loop control is not stable at idle, due to very low injector duty ( im also running 1000cc`s),
it switches on-off and error is coming up, but when cruising and part throttle driving it works very very good,

[IJ.]

Mine is the very same, closed loop control is not stable at idle, due to very low injector duty ( im also running 1000cc`s),
it switches on-off and error is coming up, but when cruising and part throttle driving it works very very good,

Have you tried dropping base fuel pressure a little so you can crank a little more duty cycle in at idle?

I had to do this with 1000's here but there's a fine line where idle quality starts to suffer if you go too low as the injector starts to dribble.