K Series cutting out

[SM25T]

Detective Revill on the case ....

[revilla]

Right, this car is driving me nuts! Ideas please?

I got a fuel pressure gauge onto it. Fuel pressure is unusually HIGH but that's to be expected. The fuel pressure regulator on the rail has been modified (squashed) to raise the pressure and a further external adjustable regulator has been added to raise the pressure further. I'm told this was done because it was running lean at the top end, but now it's left it very rich at idle and low load which is no surprise. The fuel pressure does drop off a little bit with increasing RPM but it stays well above the standard 3 Bar anyway so it's not getting starved of fuel. But ...

The engine cutouts correspond exactly with sharp downward spikes in fuel pressure. The pressure drops abruptly each time the engine cuts out and recovers as the engine kicks back in. See here: 

Just to eliminate any possibility that these were pressure pulses in the manifold coupling to the fuel pressure regulator, I took the vacuum line from the pressure regulator off the manifold and plugged it. The results were pretty much the same. See here: 

I also put a vacuum gauge on the manifold port and there was some fluctuation with the cut outs but only what could be attributed to the abrupt RPM changes.

So what was causing the fuel pressure to spike downwards? Next thing I put the scope on the supply to the fuel pump where the engine loom joins the vehicle loom.

Aha! So the fuel pump supply is dropping out in sync with the engine cutouts. Why was that happening? Here's a trace from the ECU output to driver the fuel pump relay:

So the fuel pump doesn't shut down due to a loose supply connection or similar, it shuts down because the ECU shuts it down. So the ECU is intermittently shutting down both the ignition coil drive and the fuel pump in bursts. Now the only two reasons I can think it would do that would be:

1. It lost the crank sensor signal.
2. It lost a power supply.

So I scoped the crank sensor signal AT THE ECU CONNECTOR to see exactly what the ECU was seeing as it was misfiring in case anything was coupling from the ignition into the wiring:

Everything looks fine there. A perfectly normal pattern, no distortion, no noise, no problems around the missing teeth.

I also scoped the main ECU supply and the ignition switched supply, with the earth of the scope connected directly to the ECU ground at the ECU connector, to see exactly what the ECU was seeing in terms of power supplies. Other than the ignition noise noted previously, everything was rock solid right through the misfires.

So on a hunch I set about trying to identify why, after running it up to the point where it started misfiring, it wouldn't idle for a few seconds afterwards. It wasn't running the fuel pressure down, it wasn't running the battery voltage down, so what was it running down when misfiring that then left it unable to idle without throttle?

Answer ... the Idle Air Control Valve. If I got it warmed up and settled into an idle, the unplugged the IACV, I could run it hard into the misfire zone for as long as I liked and when I released the throttle it returned happily to idle.

So I put the scope on one phase of the IACV motor and watched the signals as the ECU worked it.

When you turn the ignition on you get a very short burst of activity. When you turn the ignition off you get a much longer burst of activity as the ECU recycles the IACV, first closing it to establish a known position, then opening it a number of steps to set the engine ready for the next start (that's all the clicking and whirring you hear after switching off). Here's a recycle pattern:

Then I started the engine and watched the IACV trace. At idle there were occasional bursts as the ECU managed the idle speed, as expected. As soon as I raised the RPM to bring it out of idle strategy, the IACV stopped being driven by the ECU completely, which also made sense. But as soon as I pushed it to the point where it misfired I saw this:

Which looks to me like the ECU recycling the IACV!

So what looks to be happening is, at a certain RPM, the ECU is suddenly reacting as though the ignition switch has been turned off; it's cutting the ignition, cutting the fuel pump and recycling the IACV. It seems to recover very quickly and the engine starts firing again, but the IACV completes it's recycle (I know this is the behaviour of MEMS3; if a voltage dropout on the ignition sense triggers an IACV recycle with the engine running, the ECU continues to complete the recycle and the engine won't run without throttle for a few seconds).

But why? The supplies looked clean throughout, including the ignition sense and the crank sensor signal was clean.

Just for completeness I scoped the TPS signal. The trace below is with the engine off (as it would rev its guts out if I did this whilst running) but the signal whilst running looks very similar other than for the usual earth noise present throughout the system noted before. This is a couple of slow swings from idle to WOT and back and all looks fine:

I guess the next thing I need to do is set the scope up to trigger off any fast transients on the the ECU ignition switched supply and see if I can catch any very brief spikes that could be causing it to think it had been switched off?

Anyone else got any other ideas?

[aerobod - near CYYC]

I would say that trying to get diagnostics software to run with the ECU to correlate with the scope output would probably be a good step. Supposedly if it is an EU2 MEMS 2J ECU, this software should work: https://androidcommunicationapp.wordpress.com/2015/11/01/mems_diag-rover-mems-1-3-1-6-1-9-ecu-diagnostics-software/ and perhaps more useful info here: https://www.pistonheads.com/gassing/topic.asp?h=0&f=101&t=1253898

[elie boone]

Just a thought, if one of the injectors or 2 or 3 injectors get "stuck" due to the high fuel pressure would the ECU see this ? and momentary cut the power therefore recycle the IACV ? i think 2 fuel regulators is just 1 to many and where is the second one plumbed in ? i hope not in the return line.

[revilla]

@aerobod I've already got a T300 that will communicate with MEMS1.9. The only codes it logs are 21 and 41, "Crank Sensor" and "Loss of Sync". Initially this led me down the route of looking at the crank sensor signal but I have since discovered that other MEMS1.9 ECUs seem to log these two codes whenever you shut them down so it seems to be a red herring. Odd behaviour I know but I've seen Land Rover Freelander ECUs doing this repeatably. I really can't see anything odd about the crank sensor signal measured at the ECU connector whilst it is misfiring.

@elie boone I don't the think the ECU would detect a stuck injector. The voltage and current waveform would be identical except for a very subtle "hump" as the pintle moved.

But yes the fuel pressure regulator setup is odd. The adjustable regulator is indeed downstream of the standard regulator in the return line. It should really have the stock regulator blanked off with an adapter with a takeoff to the adjustable regulator. I didn't do it, I just found it that way. One way or another it's not working too badly, the fuel pressure is being fairly well regulated at just under 4 bar. Odd, but I don't think relevant to the problem here.

[revilla]

@Ian McKenzie I guess your suggestion is the next thing to try, wiring the ECU directly back to the battery to see if it makes the ECU stable, just in case there are transient conditions I'm missing.

[CageyH]

As a completely off the wall suggestion, is the fuel filler an aero style one?

Have you tried running it with the filler cap off to see if there is a vacuum forming in the tank, which will affect the fuel pressure.

[revilla]

Yes it is and yes I have, today. No effect. But I had a pressure gauge in the fuel line and the pressure was holding up around 4 bar right though the misfire, other than very short spikes which I proved were when the fuel pump was shut down by the ECU. So there was absolutely no sign of fuel starvation and nothing to support the idea of a vacuum, blockage etc.

[Ian Macquarie]

Does the car have an immobiliser connected to the ECU?

If so could there be a problem with the link so that the ECU thinks the immobiliser is armed and therefore shuts down the ECU?

Ian

[garybee]

Where are you probing for the voltage being supplied to the ECU? Are you measuring the power and ground at the ECU connector?

[revilla]

@Ian Macquarie - Yes there is an immobiliser by the K Series immobilisation doesn't work like that. The ECU will look for a coded signal from the immobiliser before allowing the engine to start but once running it doesn't look for an Immobiliser signal, you can unplug it and it will keep running.

@Garybee, yes probing at the ECU connector. Power supplies measured relative to the ECU ground pin at the connector with back probing pins. ECU ground pin also checked relative to the battery -ve, no significant voltage difference.

[garybee]

I expected so but just wanted to check you weren't measuring the voltage diff between the ECU power and the battery post.  You're clearly too clever for that but it's easy to get focussed on something and miss a fundamental. 

I had an ECU failure a couple of years ago that caused erratic operation of the IACV.  This was only when running open loop on warm-up though.  This happened immediately after I'd removed the loom from the bulkhead area.  I'd missed the earth connection near the ECU and it must have caused an internal failure.  My point is if you've verified the inputs and the outputs are erratic maybe the next step is to swap out the ECU.  This was easy with my EU3 1.6 academy car of course but might not be with an EU2 supersport.  

[garybee]

Didn't you make a crank sensor simulator recently?  Can you hook up the powers and grounds on the bench then feed it an increasing crank signal whilst monitoring the IACV pin?

[revilla]

@garybee - Already swapped ECU, frustratingly it made no difference. I'd really convinced myself it was an internal ECU fault at one point but two ECUs with same fault is a bit of a stretch. So it must be something external to the ECU that it's seeing and I'm not!

 

Yes I did make a simulator but unfortunately it only generated a signal corresponding to idle speed as that's all I needed for testing immobilisation etc.

[revilla]

PS: it's not a fixed RPM at which it kicks in, it actually seems to be getting worse. It was originally showing at 6500rpm (limiter is close to 8000) but today the cutouts were starting around 4000rpm.

[garybee]

Very surprised it's not the ECU after what you've shown but agree...almost sod all chance of 2 duff ECUs.

Apologies if this has already been answered but...

Does it occur if you run the engine with various sensors unplugged?  Can you run with the minimum connected, see if it still happens then introduce sensors until it starts (assuming not present with minimum inputs connected)?

Edit: PS, I don't like this approach but without knowing more about the IACV strategy it might be the best way.

[Ian Mackenzie]

I'll need to have a look through the latest developments.

I always remember a conversation I had with my dad on digital v analogue oscilloscopes....he was always adamant about an analogue scope to see real signal versus the digital interpretation (I'm proud of my dad as he was a chief design engineer working in aerospace, so his advice meant a lot to me...) It looks like a USB scope you are using so you may not be seeing all the signal....of course just another theory...

I'll try and have a look over your last posts tomorrow...

Cheers

Ian

[revilla]

@garybee - I haven't explicitly tried that approach but to be honest, the EU2 non-VVC didn't actually have many sensors! We've had the lambda sensor off (and swapped it because it seemed to be reading rich all the time rather than cycling, but I think that's just because it is rich all the time!). Had the IAT and Coolant Temp off (and checked the resistances of those to be sure they were reading sensibly). MAP is integral to the ECU with a vacuum hose direct to a port on the ECU case. It doesn't have a cam sensor, doesn't run sequential. And isn't at all happy with the TPS off.

@Ian Mackenzie - Absolutely, and I'd love a high bandwidth analogue scope, but can't really justify the pennies for one! Yes USB DSO-2090, 49Mhz bandwidth so would have to be a pretty sharp transient for me not to be able to see it if I'm looking for it in the right way, and I'd be surprised if the ECU inputs weren't filtered to reject frequencies that high, but who knows.

Will set the scope up for single sweep triggering off a negative edge with a threshold around say 10V on the supplies with no HF rejection and see if I can catch and  short record any short excursions.

[garybee]

The remaining items wouldn't be anything I'd suspect anyway.

At this point I would be inclined to cut the car's supply and earths out of the equation by temporarily splicing into the loom to supply good powers and grounds with temporary separate cables.

[revilla]

• At work today, can't do any more digging for now but looking back to post #48 where I posted this:

My money is now on the noise spikes on the ECU earth. The ECU is earthed through the engine to the chassis near the bulkhead at the rear left side of the engine bay. The chassis earths to the engine block via a strap at the front right of the engine bay, and the engine then earth back to the battery at the rear left of the engine bay again, right next to the ECU. So the earth path is a big loop, effectively a one turn coil, and sitting right in the middle of it are the HT ignition circuits forming an RF transformer.

The positive supplies don't route round this loop so any spikes coupled onto the earth won't be coupled onto the supplies and will appear differentially at the ECU. And any positive spike on the earth will then be seen by the ECU as a negative spike on the supply lines as the ECU takes the earth it sees as the definitive 0V.

Next thing I'll try is earthing the ECU straight back to the battery. If that doesn't work I'll splice in some supplies straight from the battery too.

For information here is what I believe is the correct engine loom wiring diagram for this car:

The ECU has the following connections:

• Diagnostics (won't be anything to do with this I'm sure).
• Output drivers for Main Relay, Lambda Heater Relay, Fuel Pump Relay, Charcoal Canister, Injectors, Ignition Coil, IACV Stepper (unlikely to be causing this, more likely to be affected by the issue than to cause it).
• Immobiliser code (the ECU doesn't require an immobiliser signal once cleared to run so this won't be it).
• Lambda Sensor (just gives a steady voltage as it runs so rich, tried disconnected, makes no difference).

Which only leaves:

• Power Supply, Earth and Ignition Sense.
• Crank Sensor (scoped signal, all looks good).
• Throttle Position Sensor (scoped signal, all looks good).
• Inlet Air Temperature (won't be this I'm sure, slow signal, tested, tried disconnected).
• Coolant Temperature (won't be this I'm sure, slow signal, tested, tried disconnected).

[revilla]

Well I wired a good earth with 2mm2 cable straight back from the ECU to the battery and it made no difference whatsoever, just slowly raise the revs and around 4000 it just quits, then repeated stop-start with the IACV closing. Will try the supply wires next. Raining now though.

[elie boone]

Revilla, the fuel pressure in the video shows 5 bar and even up to almost 5,40 ish bar hence my thought on 1 or more injectors failing to deliver a correct spray pattern. 

[revilla]

Right I'm on its tail now! At last something makes sense. I wired the ignition switched supply and the main relay switched supply directly back to battery and the problem was gone, it revved clean to the redline. I then put the original main relay switched supply wire back and left the ignition sense wired back to the battery and the problem came straight back. So it's the main relay controlled supply line to the ECU that's causing the problem. I need to have a good look at that wire again when it's misfiring. It's either going to be those ignition ramps superimposed on the supply through resistance, a loose connection (doubt it it's too repeatable at the same RPM), vibration resonating in a faulty relay in the MFRU, something like that. But now I know it's a supply voltage issue on that one line I know what I'm looking for and its just a case of hunting it down step by step. Finally rules out fuel pressure, crank sensor signal, TPS and all the possibilities.

[Ian Mackenzie]

Good news.....well done!

[revilla]

If only this rain would stop I might be able to fix it! Having to work on the car outside and every time it turns dry and I get set up it starts raining again.

Anyway, here's the ECU supply voltage relative to the battery +ve terminal (so in an idea world this would just a flat line at 0V but any resistance will lead to a voltage drop when you draw current through it, and that's what I've graphed directly here):

Not exactly clean! The downward ramps are caused by the ignition coil current, and there are clear transient spikes as the spark fires. I then took exactly the same measurement at the connector between the engine loom and the main vehicle loom and it looked pretty much identical, so the fault didn't like it was in the engine loom.

To be honest I'm surprised that the ECU isn't robust enough to be immune to that level noise as the voltage is staying up at close to 12V minimum. It could be the positive spikes causing it though. Just to be sure it really was the supply noise I disconnected the spade connector between the engine and chassis looms (and the male spade fell off the wire, looked like it was crimped with pliers and so I replaced it, bit it wasn't actually causing this) and fed the engine loom directly off the battery. Everything was happy and it revved cleanly to the redline again.

So one way or another it's the supply from the chassis loom to the engine loom that's at fault. It's either the resistance-induced voltage noise or the ignition spikes coupling in inductively.

Next thing I'll look at the voltage at the big junction point on the back of the starter to see if it's something around the master switch or further downstream through the regular chassis loom wiring or fusebox.

Somehow I just can't bring myself to be on my hands and knees in the rain today though!