K Series cutting out

[chris whitlow]

Right, here’s an update.

I’ve tried my ECU in another car (thankyou Giles Dixon) and there wasn’t a problem. We did try the ECU from the other car in mine but my immobiliser wasn’t having any of it.

We systematically change the plugs, rotor arm and ht leads but it still has the same issue.

Having spent 2.5 hrs yesterday evening swapping parts, waggling wires and disconnecting various bits we gave up.

On the drive home it was playing up again but at slightly lower revs. It became virtually undriveable at any revs so I turned the ignition off while driving, coasted fir a while, and then turned it back on. This changed nothing and it still continued to run roughly.

However, when I stopped completely and turned off the battery isolating switch, on restart, it ran fine again as long as I didn’t give it too many revs.

So it appears as if the ECU gets confused by a bad signal and isolating the battery resets the ecu.

Next I’m going to try taking a new feed from the crank sensor directly to the ECU.

Has anyone got any other suggestions?

Thanks,

Chris

[revilla]

Just a thought ... Are you sure the fault doesn't lie with the battery isolating switch itself? Just wondering if vibration could be causing an intermittent connection and it was the act of physically moving the switch that improved things? I find it hard to believe that the ECU gets into some persistent confused state from a bad signal as they are designed to be robust and fail-safe.

[chris whitlow]

When the problem occurs all the other electrics stay on, gauges, headlights, etc. so I don’t think it’s the isolating switch but I’ll have a look. Thanks.

[chris whitlow]

Today I swapped the MFRU but that made no difference. I also disconnected the two brown wires from the isolating switch, the one for the ignition, and joined them together. As well as this I bypassed the loom and put wires directly from the crank sensor to the ECU. None of these solved the problem either.

Could it be the immobiliser malfunctioning?

Does anyone know how to disconnect it from the ECU so that the ECU will work independently? 

 

[rj]

I would be more than surprised if it was the immobiliser. It simply tells the ECU that it's OK to keep running via serial communication.

The crank sensor is possible, as is the wiring between it and the ECU.
 

[rj]

I would be more than surprised if it was the immobiliser. It simply tells the ECU that it's OK to keep running via serial communication.

The crank sensor is possible, as is the wiring between it and the ECU.
 

[revilla]

You can't disconnect the immobiliser without having the ECU programming changed. It's actually more of a "mobiliser" than an "immobiliser". Once disarmed it repeatedly broadcasts a code word that the ECU looks for as permission to start. If you disconnect it, no code word, no start.

But the MEMS ECUs only require the code word once to start. Once the decision has been made, and the  engine has started, you can unplug the immobiliser and nothing will happen. The engine will keep running. So it really doesn't make sense for it to be the immobiliser causing this.

It's very odd, and very frustrating isn't it?

[revilla]

Something just comes to mind that I've seen before, but I can't remember where it was...

Check that the wires on the back of your alternator are all secure. There was a case where the main battery wire was loose and vibration caused it make and break connection. This caused the alternator voltage regulator to produce nasty voltage spikes that brought the ECU down.

The ECUs are designed to recover rapidly from fault conditions so what you are seeing could be the ECU repeatedly resetting and recovering.

It's another long shot but I've we've run out of likely sounding explanations now!

Actually there's another thought...

The alternator voltage regulator itself could conceivably do this. The alternator's "natural" output will rise with RPM and if the voltage regulator isn't containing it properly, it could end up putting out either a higher voltage than the ECU would like (probably not so likely as the battery will act as quite a hefty ballast) or noise spikes on the supply lines.

Since you seem to be happy to try things out to get the bottom of it, it might be worth just slipping the alternator belt off and seeing what happens? Don't do it the other way around and run the engine with the belt on but the wiring disconnected or you will damage the alternator.

If you get completely stuck with this, I'd be happy to meet up and get my oscilloscope on it for a proper look at all the various signals and supplies to see exactly what is happen electrically. It ought to be fairly obvious what is happening on the scope of it is electrical in nature. Supply voltages, spikes, sensors going crazy, dirty crank sensor signal, coil packs or injectors misbehaving should all be visible. It's a bit of a journey for one or the other of us but I've done worse!

[chris whitlow]

Thanks for your input Andrew.

I’ll check the alternator connections and remove the belt as you’ve suggested.

One thing that puzzles me about the immobiliser, as an aside, is that mine would work when fitted to Giles’ car but his wouldn’t work in mine.

Thanks for the offer to get your oscilloscope on it. If the alternator checks don’t help I’m more than happy to come to you if you’re prepared to spend the time investigating the problem, much appreciated.

[elie boone]

If there would be a fault at the master switch 4 pin part then the lights and everything else would still have live 12V but the ign would be cut off or isn't the ECU connected to the back of the master switch. Another possibility could be the inertia switch, you can easily bypass it by connecting the wires together.

[revilla]

That suggests that his ECU has been programmed for "free run" where it ignores the immobiliser, yours hasn't. So his would work with your immobiliser as it would not be looking for the code, your wouldn't work with his immobiliser as it would be sending the wrong code word.

[rj]

As the ECU will run in another car it sounds like the IMMO is off. (The other ECU wouldn't in Chris' car)

[revilla]

Ah yes ... I read it that Chris tested by swapping the immobiliser not the ECU! Same idea though. And makes it even less likely to be anything to do with the immobiliser.

[rj]

Just my thought

 

[anthonym]

Any chance it’s the famous dreadful connector on the purple wire?

that fails slowly.. doesn’t it feed the ECU? Or something.

[revilla]

Chris and his car spent the day in my garage with an oscilloscope hooked up to it testing the patience of the neighbours at 6500rpm (luckily I think they all went out for the day).

In the morning, as always happens, the car decided to run perfectly. It would run faultlessly to the rev limiter every time. So we set about just generally testing things and looking for anything odd. All of the ECU power supplies and earths were good with not too much noise - very little in the case of the earth, about half a volt on the positive supply, mostly consisting of inverted ignition coil and injector coil current waveforms as these currents were causing voltage drops through the small resistance on the supply side in the loom, but all to be expected and nothing untoward.

We then had a look at the lambda waveform, not that the lambda would have directly caused the issue - and what we found wasn't normal. The lambda wasn't cycling at all, it was just indicating a high voltage / rich condition continuously. We swapped the lambda sensor, same result. The sensor did show it going lean on the overrun after revving it up, so we were convinced the sensor was working and telling the truth, that the engine was very rich at idle. The colour of the exhaust seemed to confirm that.

I then looked at the ignition primary voltage and current waveforms. Initially everything was normal:

But then after a while we started to see misfire events which were typical of carbon fouling of the plugs. Here you can see the energy stored in the coil drains away through a leakage path rather than forming an arc across the spark plug gap. This corresponded with the onset of a misfire when revving the engine.

So we thought we were onto something - running rich, fouling the plugs. We shut the engine down in the middle of a misfire episode and removed the plugs and they were black and sooty, and one of them had a cracked insulator.

So we replaced the plugs with a brand new clean set and the fault was still there, straight away, even though the ignition waveforms now all looked normal.

By now it was doing the full routine and it was clearly more than a plug misfire, it was as though the whole ECU was shutting down and starting up again in rapid succession. The tacho was dropping immediately to zero as it cut out. So we decided that the rich mixture and plug issue were a problem to look at later, but not what we were looking for today (Chris has an adjustable fuel pressure regulator and was told it was running lean on the rollers; my guess would be that in the past it has had the fuel pressure increased to cure a lean mixture at the top which could not be mapped out due to running an EU2 MEMS 1.9 ECU, and that had left it running very rich lower down, beyond the ability of the ECU to trim it).

Chris's ECU was an EU2 MEMS 1.9 Supersport which does not support OBDII diagnostics, but I hooked up my T300 Key Programmer to it and was able to access (both read and clear) the diagnostic fault codes (but not live sensor data). After the fault occurred, the ECU logged the following:

And just on one occasion:

... although we never saw that one again. I'm guessing it was logging an issue with the MAP sensor integrated into the ECU, which could have been related to the rich running, but as I say we never say it again and we were getting the engine into conditions where it was running quite badly at times so it could have been a red herring.

So now we had a clear pointer to a crank sensor problem, with the ECU the losing sync with the flywheel, at which point it would shut down ignition and injection and the tacho would stop reading as it driven directly off the coil. So I hooked up the oscilloscope to the crank sensor inputs to the ECU and had a look at the signal. At lower RPM it was a nice clean signal, however as the RPM was increased it did distort a bit; the falling voltages corresponding to the "missing teeth" on the flywheel started to develop a "hump" in the middle, and the onset of the misfire seem to correspond to when this crossed the zero volt level - I know from previous work I've done on the crank sensor waveforms that the ECU seems to simply detect zero voltage crossings, and so this extra crossing would effectively fill in the missing tooth and I was not surprised to see the ECU losing sync.

Here is the waveform just at the onset of the misfire (I accidentally saved the data rather than the image, so I've graphed in Excel but it is just an oscilloscope capture like the others above):

The flywheel looked like a standard Rover cast iron one, the teeth of the trigger pattern looked fine, the crank sensor looked OK with no swarf on the pin (and has been swapped since the problem started without curing it).

The only explanation we could think of was crosstalk of some other signal into the crank sensor wiring, so I set about making up a new separate screened cable for the crank sensor, we popped the two crank sensor pins out of the ECU connector and added the new cable, earthed the screen at the ECU and and fired it up again.

This time the crank sensor signal was perfect and clean (I think we had the vertical scale inverted here which is why it appears upside down compared to the previous trace):

... right up to the point where the ECU started shutting down again! Although we were absolutely convinced we had found it and had a plausible explanation for everything that was happening, cleaning up the crank sensor signal with new cabling made no difference at all.

From there, now we had it reliably misbehaving, we rechecked all of the ECU supples and grounds and everything was still fine.

So the situation is that the ECU is losing sync with the crank sensor signal, logging fault codes for the crank sensor and loss of sync, and therefore not knowing the engine timing it is shutting down the ignition (if it doesn't know what position the engine is in it doesn't know when to fire the spark), even when we could see an absolutely clean signal from the crank sensor measured at the ECU end of the wiring and even when the ECU power supplies and grounds looked perfectly fine right through the event. The engine then loses speed and the ECU regains sync, starts firing again and as soon as the RPM increases sync is lost and the cycle repeats.

As far as I can see, this can only be a faulty ECU at this point. It seems to be unable to process the crank sensor signal correctly above a certain frequency. Once the RPM exceeds a certain limit it just drops out of sync even with a clean signal.

Chris had previously tried his ECU on another car without issue, but given that for the first three hours of our testing it didn't put a foot wrong, I don't think that was really conclusive.

I think the next step is to try to swap the ECU and immobiliser set and see if the fault recurs. Beyond that I'm a bit out of ideas!

[Ian Mackenzie]

Hi there

By chance I've just had very similar symptoms, though I'm running an Emerald ECU....

I had a misfire at around 6-7K with loss of crank sync leading to full ignition / injector cut across all cylinders. After lots of checking, similar to you. My culprit was a poor / noisy+12V supply, once I'd wired the supply for the ECU direct to the battery (with inline fuse) the misfire disappeared. I've not had time to do a re-wire yet, but I've got my cause identified.

Best regards

Ian

[revilla]

Ian, I have the car here at my house for a good looking at. Thanks for the info, I will give all of that a good checking over.

[Ian Mackenzie]

I've just been looking at this a little more....

From what I can see the main power feed to the ECU comes via the MFRU, if the main relay in the MFRU is failing it may give some supply issues to the ECU....

If you have another MFRU about try swapping this and see what happens.

Cheers

Ian

[revilla]

Thanks. I do have a spare MFRU but I will have an oscilloscope on all of the earths, supplies and signals to the ECU so should be able to see any problems. I'll report back what I find when I get the time to give it a thorough going over.

[Ian Mackenzie]

Thanks, it will be interesting to see what you find, I didn't get to the stage of a scope on my car (mainly because I don't have one at present...). Good luck!

[chris whitlow]

Thanks for your help Ian.

The MFRU is one of the many things I’ve swapped and that, along with everything else, had no effect.

[revilla]

Ian, it could well be something along the lines of what you suggested ...

I managed to have a quick play with the car tonight. To be honest the thing that limited my time was the fact that it was running so badly, was alost impossible to keep it idling, and it was backfiring loudly from the exhaust quite regularly so there was a limit on how much of that I could expose the neighbours too. The "misfire" cut in/out events were also starting as low as 3000rpm. It was running like an absolute dog!

First of all a subjective description of what it was doing. Running rough at idle, but as you increased the throttle it would start to run a bit better and pick up. But as you increased the RPM it started to cut in and out, but the odd thing was it slowly needed more and more throttle to maintain the RPM. Something was deteriorating as it kept running. It felt almost as if somebody was letting the clutch up slowly (which clearly they weren't). But if you kept pushing it to keep it revving, afterwards it wouldn't idle without a lot of throttle. And if you gave it just enough throttle to keep it going, it slowly recovered, needing less and less throttle to get it to idle again. This was consistent with what we saw when we were playing with it before.

Does this ring a bell with anyone? It made me wonder about fuel starvation, with the pressure in the rail dropping off when running the engine at high RPM and then recovering when idling at low demand. Maybe not, and I did see a lot of electrical oddities described below but that's the kind of feel it had. Just in case there is more than one problem it might be worth putting an inline pressure gauge in the fuel supply line just to check that the pressure holds up if I can't fix it electrically.

Next I had an oscilloscope on it and did see some pretty odd things, although I wasn't able to follow them through to a conclusion due to time and noise factors.

First of all, the crank sensor signal, which had been the focus of investigations:

That was captured whilst it was actually cutting in and out violently. The highlighted portion represents one complete rotation of the flywheel. The signal is clean and strong (I was using a 20:1 attenuator to protect my scope so that's about 40V peak to peak), the missing tooth pattern is correct. There's no significant noise, no distortion, no additional zero-crossings. No reason at all to suspect that the crank sensor signal is defective, at least at the sensor end of the wiring. I still need to probe it at the ECU end just to convince myself there's nothing coupling into the wiring as I'm sure we did see something odd before, but for now it looks normal to me.

Next, ignition primary waveforms. I didn't have the current clamp on it this time, so it's voltage only. And it's inverted due to the way I had the wiring hooked up. One thing I did see was an awful lot of ignition misfire events as shown below, where the coil drive looked reasonably normal but it didn't actually spark. The coil current just decayed away. Oddly it decayed away more quickly than it would have done if it was sparking, suggesting there was a discharge path that didn't involve jumping a spark gap. Fouled plugs? We swapped the plugs last time it was here after seeing similar things and that didn't make much difference. Coil fault? Internal breakdown of the coil insulation? Possible, but unlikely as Chris confirmed this evening that he has replaced the coil, distributor and leads whilst attempting to fix this.

The next thing I noticed was that there seemed to be a wiring resistance issue somewhere in the ignition coil primary circuit. If you look at the waveform above, the upward jump is the the ignition driver in the ECU turning on and pulling the switched side of the coil down to earth. A steady voltage across an inductor causes the current to ramp up linearly, so at this point the current through the coil starts to ramp up. But you can see that as the current increases, the voltage across the coil falls off. This is resistance in the drive circuit. Because I was taking a double-ended measurement across the coil rather than one side relative to earth I can't tell yet whether it was in the supply or earth side; but the earth side is wired straight back to the ECU so to be an earth side fault it would have to be in that one wire from the ECU pin to the coil, or internal to the ECU, or a problem with the ECU earth itself. I did scope the ECU earth and there was no sign of ignition waveform ramps on it, and we've had two ECUs on it with the same fault, so it only leaves that one wire, or a supply side resistance issue. Again I will have to probe the waveforms at the ECU end of the wiring next, and track it down to the earth or supply side by measuring the two sides of the coil relative to a good ground.

In the waveform below, which oddly contained a burst of repetitive misfires, I scaled it up and measured the voltage drop in the coil drive. It's about 120mV with again a 20:1 attenuator on the scope, so 2.4 volts dropped at coil as the current rises to 6 amps, which is a lot. I didn't have a current clamp on it tonight but when we looked at it last time, the coil current was peaking at about 6 amps, so that's around 0.4 ohms resistance in the supply somewhere. It's also interesting to note that even at the start of the pulse, the voltage delivered across the coil is only about 11V when it should be very close to battery voltage (usually just the voltage drop of a saturated transistor so 0.1V-ish below battery voltage) and this falls to around 9V just before it sparks (or tries to). So as well as a resistance issue the background voltage looks very low. Something else to look into again. Battery on the way out? Surprised the alternator wasn't pulling it up a lot higher than that. I'd better have a good look at the battery voltage next time around. But it was noticeable that the car was behaving a lot worse today than when I last saw it, and it's been left standing, so if the battery voltage has fallen and there's a resistance-driven voltage drop compounding that, the two of them together could be causing a real problem. If this is all supply-side, then the ECU could be seeing a similar voltage waveform depending on where the resistance is, and if the ECU supply is fluctuating between 11V and 9V it won't be happy!

But when it was actually cutting in and out, some of the ignition waveforms looked completely abnormal.

Look at this trace:

Unfortunately this was a one-off and I'm not sure what the RPM was when it happened.

It starts off with a normal-looking ignition pulse, this time with a clear spark, but oddly a period of time with no pulses before it. But shortly after the ECU driver turns on again, the current starts to ramp up again as seen by the falling voltage, but the driver remains on for an extended period leading to a higher current, then the driver seems to sit and hold at about half voltage, obviously with a decent current maintained in the coil because when it does shut off again it fires another spark. But that bit between the two sparks is completely wrong in a way that smacks of an ECU problem, and given that we've swapped ECUs without fixing the issue that suggests the ECU is just being thrown off by something external, power supply problems being the most likely.

So more investigation of the ECU supply voltages is in order.

One other thing I did notice was a lot of radio frequency spikes on the earth lines. Very short duration with ringing as shown below. I'm not sure how normal or otherwise these are. They are high amplitude but very short duration, the spike below is about 30V peak to peak and swings positive and negative but dies away within a small fraction of a microsecond. I'm pretty sure they are ignition switching noise coupling across into the wiring. They were present on earths everywhere, measured relative to the battery -ve. Even measuring at the other end of the earth strap to the engine I was picking them up and they just got stronger the further away you got from the battery. I should imagine that the internal power supply system of the ECU is well enough filtered to eliminate these as it will be designed for rugged operation.

TPS signal looked OK, even when it was misfiring. A steady 5V supply on one end, earth clean other than for the above at the other end and a steady signal from the wiper that increased with increasing throttle. No obvious noise.

More investigations another day!

[revilla]

PS: Just went to put the car away for the night, stuck a voltmeter across the battery and saw a very strong 12.7V. That suggests the alternator has been charging it too, so I reckon there will be around 14V at the battery whilst running. So why only 11V initial drive to the coil at zero current, falling off to 9V at 6A? Defintiely something odd to look into.

[revilla]

OK so I scoped the supply voltage to the ECU:

Not as bad as I expected. Yes you can see ignition coil ramps on it due to wiring resistance but it's only fluctuating between 11.6 and about 13 volts so nothing the ECU shouldn't cope with.

The odd thing is, I was thinking about my earlier idea on fuel pressure. The more you rev it, and the longer you hold it at high RPM, the worse it runs for a while afterwards, takes a few seconds to recover. And I may be imagining this but if I pinch the high pressure fuel line between finger and thumb, I'm sure I can actually feel it going softer as you rev it, and it then pushes your fingers apart again as it recovers afterwards.

I may just be seeing what I'm wanting to see, but I'm beginning to wonder if what I'm seeing electrically is a bit of "tired loom" syndrome but not the only thing going on here. I think I need to get an inline fuel pressure gauge on it and just see if that's doing what is expected (if anything the fuel pressure should increase as you rev it up due to the vacuum compensation). I'll also try to get a current clamp on the fuel pump to see if it looks healthy or if I'm seeing missing commutator segments.

But that theory isn't backed up by the signal from the lambda sensor which seems to be showing it running rich all the time (and it smells like it too!). And I'm not sure how it would fit with the ECU apparently stopping firing the coil, unless the ECU is sensing a misfire?

May be more than one fault.

The mystery deepens.