Lumpy as hell at cold.

[Rullingen]

https://photos.app.goo.gl/K4tASXwd48uuRpoz6

This link is to a video I took earlier when the engine had been running for about two minutes. Initially it was worse than this. Reluctant to start, then would not pick up when the throttle was opened, and was sucking loads of air through the IACV. 

Last Autumn I changed the IACV for this reason, but now it seems worse than ever.

I'm guessing this is sensor related. If I could find the OBD2 port I might get some clues as to what's going on, but I can't see one under the dash where it's supposed to be.

Does anyone know what the cause of this problem is? 

Thanks, Rob.

[SM25T]

Sucking air through the IACV ? Does that mean an air leak there ?

[Rullingen]

I don't think so. It sounds like the IACV is opening and shutting somewhat erratically. That's why I changed it last year, I thought it was sending 'bad' position info back to the ECU, but the new one doesn't seem to be any better, Also, the warmer it gets, the better it behaves.

[SM25T]

Does this engine have a lambda sensor in the exhaust ?

[Rullingen]

Yes. It's a stock Caterham k series exhaust.

[SM25T]

Do I recall that the lambda sensor works during the warm up phase. Try unplugging the connector in the cable to it. If the engine runs smoothly ... replace the sensor. Remove it and look at the part number engraved on it. Buy on line.

[Rullingen]

So, the engine should run normally with the lambda sensor disconnected.

In which case; if it runs rough with the lambda sensor connected and well with the lambda sensor disconnected I can assume the lambda sensor is faulty.

Have I got this right?

So will a fault with the lambda sensor only show itself when the engine is cold?

Thanks, Rob.

[SM25T]

1. Yes. 2. Yes. 3. Mine showed up with an apparent loss of power under load when warm. No ... I don't know why !

[revilla]

The IACV doesn't send any position info back to the ECU. It doesn't really operate as a servo, just a pure stepper motor. When you switch off, the ECU simply closes it by enough steps to guarantee that it's jammed up against the closed end, then sends opens it by a number of steps corresponding to the learned idle position ready for the next start. That's all the whirring clicking electronic chicken noises you hear after switching off.

After that, the only feedback to the ECU to tell it whether it needs more or less IACV is the idle speed. It doesn't measure the position of the valve itself, just keeps asking for more or less based on the engine response.

The OBDII port should be dangling on a free bunch of wires somewhere up around (possibly tucked behind) the driver's knee trim panel. The MEMS3 ECUs are very fussy. Leave the car switched off for enough time to ensure the ECU has powered down (can take 10 minute after switch off as in "Rover" mode it stays alive to run the fan to keep the engine temperature down). You can sometimes hear a high pitched whine from the IACV, once this stops and you hear the main relay click off, it's powered down. Or just wait 10 minutes. Then plug in the OBDII scanner and power it on, and only then switch the ignition on to power the ECU up again. It seems the ECU looks to see if there's anything to talk to when it powers up, and after that it stops listening if nothing is found.

The lambda sensor won't be used until it is hot; but that means the lambda sensor is hot, not the engine. Takes about 30 seconds typically to reach operational temperature. If you do get an OBDII scanner with live data working, it will probably report "Open Loop" for the first 30 seconds and then switch to "Closed Loop". This is the point when it starts using the feedback from the lambda sensor.

After that, it will be used at lower throttle setting and loads. At full throttle the ECU ignores it as it wants to run rich to cool the mixture and prevent detonation, and a standard narrow band lambda sensor only gives a lean/rich signal right at the normal mixture point and can't give any information once you go a bit richer or leaner than that. So the ECU falls back on its map information and runs open loop again ignoring the lambda sensor.

Yes the engine should run reasonably well with the lambda sensor unplugged, as the ECU will detect the lambda fault and ignore it, running open loop all the time. But if that does fix the problem, do replace the lambda sensor and don't be tempted to just drive without it, is it could be running lean and the ECU wouldn't know to correct the problem, which can rapidly damage the engine at higher loads.

[Rullingen]

This is all great information, and I thank you all.

If the ECU regulates the IACV according to engine revs, why is mine reving so high and erratically at cold?

Also, if the ECU ignors the O2 sensor for the first 30 seconds of running, and mine behaves particularly badly at first start up, does that not rule out the O2 sensor?

 

Thanks, Rob.

[revilla]

My guesses would be:

• It's not the IACV at fault. The ECU is just fighting to recover the situation using the IACV, which is why you hear so much sucking noise from it.
• I think you've got an air leak, and/or a miscalibration of the TPS. You can get a soaring idle like that if the ECU doesn't realise it's supposed to be idling; it uses different strategies for on idle (tries to achieve a fixed engine RPM, varies the throttle via the IACV and ignition timing to achieve it) and off idle (throttle commands power level, RPM is just whatever that results in). If it doesn't realise the throttle is fully closed, it will not switch to idle strategy. It's easy to reset this on standard MEMS3 ECU (which you have) as rather than adjusting the TPS like you would on other ECUs, you just train the ECU to recognise the voltage TPS gives at idle:

1. Start with the ignition off.
2. Turn the ignition on, but DON'T start the engine.
3. Smoothly depress the throttle to full, then back to fully closed (either pressing the pedal or just turning the throttle quadrant on the TB like you were doing in the video). Do this FIVE times in a row, take about a second to open it and a second to close it.
4. Turn the ignition off.
5. Then turn it back on and start up and see if it's made a difference.

Also check that the wiring to the TPS is OK and that the connector is plugged in properly. If you can get an OBDII scanner working, the TPS should show 3.92% exactly at idle when the calibration is complete and should increase smoothly to a fairly high percentage (quite variable) as you open it.

• If that doesn't fix it, you're looking for an air leak around the inlet manifold next. Somewhere after the throttle body. Check for little vacuum pipes split or come off, any unused vacuum ports that should have a rubber cap on which has split and come off. If in doubt, stick your thumb over it and see if it helps! Not sure on the plastic plenum you have but on the alloy plenum there is once vacuum port with a short line to the fuel pressure regulator, one next to it that is capped off and one up near the throttle body that is also capped off. There's also a big brake servo take off point on the side of the plenum which is unused and should have a plug inserted, make sure that's not leaking. Failing that, leaking around the inlet manifold gasket. You will probably hear a whistling or sucking noise. You can listen around using a tube as a stethoscope to locate any sounds.

It could be a sensor problem, but with an idle that soars as high as that, the engine has got to be getting air from somewhere and if it's not getting through a closed throttle and it's not an IACV issue it must be leaking in somewhere else, so I would look at the above first.

And no, if it is odd straight from a cold start I doubt it's the lambda sensor, but it's such an easy test to unplug it that you might as well try it.

[revilla]

On second thoughts, listening to the video, I doubt it is the TPS calibration. It does actually sound like it's trying to regulate the idle, which means it knows it's idling. Look for air leaks still though.

[revilla]

The more I look at the video the more confused I get.

Was that video taken immediately after a stone cold start?

• 0:00-0:14 you were blipping the throttle a bit and it was returning to idle OK-ish, a bit of hunting on the IACV before it settled. Sounded to me like it initially went too low and then saved it on the IACV each which could mean the throttle stop needs trimming (don't play with that unless you know what you're doing you can seriously mess up driveability). But could just be the way the engine runs when stone cold as everything is trimmed to run nicely when at normal temperature. Throttle response is never going to be perfect stone cold.
• 0:14-0:31 it sound like a fairly healthy fast idle (will be set to idle faster than normal when the engine is cold to smooth it out) to me. Steady idle speed. So hard to tell on videos - what idle speed is it sitting at? Something like 1300? Looks happy enough.
• 0:31 onwards you return to blipping the throttle, the response is a lot lumpier and the idle is all shot. The initial dive in the idle return response has gone, it goes straight to a really high idle speed. What is that, 2300 or something? Something significant has changed between 0:14 and 0:31.
• 0:57 it's chuffing out black smoke. And later it on it bogs down when you throttle it suddenly, like a rich cut.

That does look like it could be lambda to me, after looking carefully. I would guess it went closed loop somewhere around 0:30 and the lambda is giving a false lean read, so the ECU is overfuelling.

Unplug it and see what happens.

[MADMALC]

When you start the car do you normally 'tickle' the accelerator to try and 'catch' the engine as it fires? This action seems to upset the K. It is best to not touch the accelerator at all until the car is running.

Have you tried the ignition on, no start, accelerator slowly to floor and return several times, ignition off? When you restart do not touch the accelerator.

Worked for me and a few others, hope it does for you.