Idle hunt

[Bitsyncmaster]

I had two users of my idle ECU not able to stop idle hunt with high lift long duration cams. But they could stop it using the stock idle ECU. So my car has an idle hunt at 3 minuets after a cold start (lambda starts working closed loop) and it pretty much stops at 8 minuets. So I put a stock idle ECU in my car (I shorted the idle switch signal since I removed my idle switch) and it got about the same idle hunt at 4 minuets and pretty much stopped at 9 minuets.

When I was running a wideband (heated O2) it never hunted. The wideband went closed loop 15 seconds after cold start.

I also tried setting my wideband to slower response and that resulted in very bad idle hunting all the time.

So one problem causing idle hunt is the lambda swinging the mixture a lot when the O2 sensor is not fully warmed up. Also if the O2 sensor is resulting in slow response.

I think I will investigate the lambda ECU as to what starts closed loop. I assume after the O2 produces a set voltage it starts closed loop. I will see if I can adjust that voltage a little higher. But I think the real fix for this problem is to use a heated O2 sensor. I have about 5 lambda ECUs and will see if some set that starting voltage higher then others. There is not a lot to work with since the O2 sensor only produces about 0.50 volts around the AFR of 14.7 volts.

I'm guessing the hot cams affect the O2 running temps but don't know why the stock ECU cured that.

Of course a sticky idle motor can be another cause of hunting.

[Bitsyncmaster]

This data is for people that understand how a lambda system works. Basically the narrow band O2 sensor is high impedance (floating) when cold. When the O2 warms up it's impedance drops and provides a voltage above 0.54 volts if the mixture is rich and below 0.54 volts if the mixture is lean. It takes about 300 ms for the sensor to swing from full rich (0.9 volts) to full lean (0.1 volts).

My testing is driving the lambda ECU and a resistor load for the frequency valve output at 13.50 volts.

Unit #1
With the O2 sensor input floating the voltage on that signal line is 0.54 volts measured with my 10 M_ohm impedance DVM. So the lambda ECU is holding a dwell of 45.6 deg. until the O2 sensor has warmed up and it's impedance drops enough to change that floating voltage.

1) Adjusting my input voltage from 0.54 to 0.57 volts still holds the 45.6 deg. At 0.58 volts the dwell swings to 11.2 deg.
2) Adjusting from 0.58 to 0.57 volts still holds 11.2 deg but dropping to 0.56 volts the dwell swings to 45.6 deg.
3) Adjusting from 0.56 to 0.51 volts still holds 45.6 deg but dropping to 0.50 volts the dwell swings to 86.7 deg.
4) Adjusting from 0.50 to 0.52 volts still holds 86.7 deg but raising to 0.52 volts the dwell swings to 45.6.

Conclusion:
This first ECU has hysteresis. Not sure why they used hysteresis because that would make the dwell swing a lot (mixture swing a lot) which is what we see when setting the dwell value. I would think setting above 0.54 volts swing dwell low and below 0.54 volts swing dwell high would result in more stable mixture and maybe cause less hunting. When I was running my wideband I did set it's output to swing from 1.0 volts to 0.3 volts which may be why I got no hunting with that unit.

I will test some more ECUs to see if they all act the same way.

[Bitsyncmaster]

Well I tested all 6 of my spare ECUs and they all have about the same hysteresis.

Three stock ECUs pn 0280 800 023 and one black 0280 800 050 have floating voltage of 0.54 volts.

Two black 0280 800 052 have floating voltage of 0.48 volts.

I have another stock unit in my car not bench tested yet. I may try one of those with the lower float voltage to see what changes with hunting.

[Bitsyncmaster]

I looked at the circuit of the lambda ECU and I now see why they had to add the hysteresis. The circuit is all analog and they had to spread the range (add hysteresis) to compensate for the variables of the analog circuit. I know I could design a better circuit with a micro which would eliminate hunt caused by the lambda swinging but you can fix this using a wideband driving the stock lambda ECU.

The wideband would also let you adjust mixture to most any AFR you like. When I was running wideband I has a switch to set either 14.7 AFR or 12.0 AFR.

[Michael]

Dave, I have a stage II (with aggressive cams) and my problems are the same. With your unit, idle hunt is prevalent but with the stock control unit, it's smooth as silk hot or cold so I just kept what works.

[Bitsyncmaster]

As always I can't stop thinking about this. I may do some R&D with the lambda ECU. Like I said it would be very easy to do the ECU with a micro, one op-amp and a MOSFET. I did see the OEM unit is driving the ground of the FV though a resistor so it must be the FV should not take the full 12 volts.

I wonder if the mixture is swinging excessively with hot cams and the stock lambda ECU.

[Bitsyncmaster]

I thought of a simple way to test if removing the hysteresis will cure hunting. I can wire up one of my fuel gauge PCBs which has the micro and op-amp. I will add an analog switch so I can drive the stock ECU O2 input with a float, 3.3 volts or 0 volts from micro I/O pins. I will just mount that added PCB inside the stock lambda ECU.

This way I keep the stock rate of change of the duty cycle so all it does is remove the hysteresis. The micro can produce software controlled duty cycle with software but that would be a later effort. Using the micro to produce the duty cycle can fine tune the lambda if the first test does not cure hunting.

So I ordered some analog switch ICs and have to write some simple software to finish this first test.

[Bitsyncmaster]

While waiting on my chip to arrive, I did some more testing of the stock lambda ECU. When the O2 voltage goes above the threshold the dwell slowly (about one second) creeps to max dwell but when the O2 drops to the threshold it almost instantly goes to 45 deg. Same going below the threshold takes about one second to creep to min dwell.

So I feed a square wave signal generator into the lambda and it went to max and stayed there with a 0.1 Hz frequency. If I played with the frequency a little I could find a spot where it would swing slowly from max to min duty. So I'm not sure my simple test will work driving 0 or 3.3 volts into the input. But that sweet spot of frequency is what I want it to do.

I think that fast jump to 45 deg is what is causing idle hunting.

Edit:
On second thought it should work driving a low or high voltage because that is what the wideband unit did. I think I set the wideband to drive 1 volt above set AFR and 0.2 volts below set AFR.

[Elvis]

Well I tested all 6 of my spare ECUs and they all have about the same hysteresis.

Three stock ECUs pn 0280 800 023 and one black 0280 800 050 have floating voltage of 0.54 volts.

Two black 0280 800 052 have floating voltage of 0.48 volts.

I have another stock unit in my car not bench tested yet. I may try one of those with the lower float voltage to see what changes with hunting.

Hi Dave,

it can't just have a hysteresis of 0,0x V. Lambda wire has no GND cable, you have voltage drops on the engine
and other wires and connectors depending of other loads that would always disturb it.

Elvis & 6548

[Bitsyncmaster]

Hi Dave,

it can't just have a hysteresis of 0,0x V. Lambda wire has no GND cable, you have voltage drops on the engine
and other wires and connectors depending of other loads that would always disturb it.

Elvis & 6548

They did the hysteresis so they could tell when the O2 sensor was not warmed up. They used a resistor network to bias the input at 0.54 volts. So around that voltage they force the dwell to 45 deg. since the sensor is high impedance when cold. So the hysteresis is needed to keep that fixed dwell until it warms up.

What I'm going to do is pull the input up to 3.3 volts with a 200 Kohm. Then my software reads the voltage and if above 2.5 volts I float the signal to the stock lambda ECU. If voltage is between 0.50 volts and 2.5 volts I will drive 3.3 volts to the stock lambda ECU. If voltage is below 0.50 volts I will drive 0 volts to the stock lambda ECU.

Now it would be very easy for my micro to drive the FV with software controlled duty cycle. It could even adjust it via the amount of O2 voltage offset. But going to do the simple testing first.