Fabia Suddenly Stops, ECU Fuse Replaced, Now Won't Stop At All!

[NeilTM]

3 hours ago, Wino said:

Hi Neil, it was good thanks, as far as I can remember! Yours?

 

I'll have another look at the wiring diagrams to refresh my memory and look at possible causes.

 

It's all a blur!

 

Page 308.

 

Unfortunately, this time I need to solve the crime, otherwise she will want rid, and I can't blame her.  I tried telling her that all cars come equipped with Russian Roulette games now, citing my Doblo with its crankshaft sensor sudden demise the same year, but in truth this is upping the stakes even more unreasonably isn't it?

[NeilTM]

 

I've had a reply from ATP Electronic Developments Ltd, Cannock, which leaves us with a £60 (inc VAT and postage) gamble basically on them finding a fault with the ECU, and us paying them £230 for a repair of ours or reconditioned unit without the fault. They would offer us that anyway even if they didn't find a fault, but that amounts to just throwing a large lump of money at replacing a lump of car in the hope that it cures the fault and it isn't something else. They said on the phone that it is 50/50 whether it is the ECU or the wiring.  What they won't be drawn on is offering any kind of confidence in the unit if their test does not find the fault.  If it does it does, if it doesn't it doesn't but if it doesn't that doesn't mean there isn't an intermittent fault.   I thought their reply might be of interest here anyway, and would be interested in whether others thought it might be worth the punt and time involved to go for their bench test?  Basically if the test comes back with nothing found we have not achieved anything towards reducing the Russian Roulette odds for my daughter, only if THE fault responsible is found can we then reasonably rule out the likelihood of a coincidental short in the wiring as well, which will be a pain to replace.

 

I think the bottom line is that the only way to end THIS particular Russian Roulette game, there are still crankshaft sensors and the like that can dump you with no warning without any go, stop or steering, is to replace both the ECU and the wiring.

 

They did say that with this unit they did not need the ignition key, (they do with some).  Does this imply that any working ECU sporting the same part number known to have come out of a working car with no issues might do instead, or does it mean that from the vehicle information we supply they have the same access to key codes as the manufacturers and their agents would, and that they would thereby programme the ECU to the key?  If anyone knows what the issues are with utilising a scrap ECU of the right number, please advise?

 

Anyway, this was their reply:

 

Me:

Is it going to be an intermittent short to earth somewhere that is
causing the fuse and the diode to blow?

 

ATP:

On inspection of the wiring diagram I would say that the power feed on
ECU pin 23 is your issue (drawing excessive current).

for the diode to fail then there has to be excess current draw through
this circuit with a secondary fault of the fuse failing.

 

Pin 62 Does not draw current through the diode it is upstream of this,
so personally would count this out

 

Me:

Is it possible, and if so how likely is the ECU to be the source, and
how reliably could that be tested?

 

ATP:

ECU'S Can be intermittent Our test rigs simulate the vehicle which
includes loads to test relay outputs / injectors / ignition etc.

 

Me:

Given that the interval between intermittent events was 6 months, could a
2 day test (its actually 50 mins on their rig) for intermittence reliably reveal the intermittence if that is
indeed where it is? A description of why it would, if indeed it would, would be
appreciated in order to feel confidence in the relevance of paying for
this test for this particular fault.

If the unit passed your bench tests, what level of confidence could I
have that the unit was not implicated in the fuse and diode blowing, and that
the cause must be elsewhere such as an intermittent short in the wiring loom?

 

ATP:

Extremely tough question this. The only honest answer is that we fully
load the ecu constantly this can increase the chance of seeing an
intermittent fault. However, this is not always guaranteed. We do monitor current
draw from the ecu amongst other measurements and can monitor for unusually
high figures which can on the car at the time not cause a fault but is
showing that there is a failure mode sitting there hiding ready to show up at
some point.

 

Me:

Do you offer any level of guarantee for a customer's unit that has
passed all your bench tests?

 

ATP:

The guarantee can only be that at the time of the test these were the
particular results we saw.

 

 

[sepulchrave]

1 hour ago, NeilTM said:

They did say that with this unit they did not need the ignition key, (they do with some).  Does this imply that any working ECU sporting the same part number known to have come out of a working car with no issues might do instead, or does it mean that from the vehicle information we supply they have the same access to key codes as the manufacturers and their agents would, and that they would thereby programme the ECU to the key?  If anyone knows what the issues are with utilising a scrap ECU of the right number, please advise.

 

The immobiliser is separate and remote from the PCM, this is why they don't need the key. So, yes, any scrap PCM with the correct part number would work although there is still soft coding work to be done once fitted.

[Breezy_Pete]

I just looked at wiring diagrams for an AZQ (like ours) and BME (like your daughter's) for the relay/ECU connections, and the same ECU pin numbers are used for both. That suggests that the connected circuitry inside may be common to both ECUs, in respect of these pins at least. 

I have a spare AZQ ECU I can have a browse inside, possibly identifying some of the components/circuitry connected to those two pins.  That might suggest some possible (partial) failure modes whereby, for example, a component is taking more current than it should continually (rather than there being an intermittent condition), putting the diode in the relay, or the fuse, under more stress than designed for.

 

All speculative, but I'll dig out the ECU and take it to work tomorrow, should be able to have a look at lunchtime.

 

 

 

[NeilTM]

16 hours ago, sepulchrave said:

 

The immobiliser is separate and remote from the PCM, this is why they don't need the key. So, yes, any scrap PCM with the correct part number would work although there is still soft coding work to be done once fitted.

 

Ah, thanks.  Can you say what would be involved in the 'soft coding'?  This is all new territory to me.

[NeilTM]

16 hours ago, Wino said:

I just looked at wiring diagrams for an AZQ (like ours) and BME (like your daughter's) for the relay/ECU connections, and the same ECU pin numbers are used for both. That suggests that the connected circuitry inside may be common to both ECUs, in respect of these pins at least. 

I have a spare AZQ ECU I can have a browse inside, possibly identifying some of the components/circuitry connected to those two pins.  That might suggest some possible (partial) failure modes whereby, for example, a component is taking more current than it should continually (rather than there being an intermittent condition), putting the diode in the relay, or the fuse, under more stress than designed for.

 

All speculative, but I'll dig out the ECU and take it to work tomorrow, should be able to have a look at lunchtime.

 

That would be marvellous, thank you, generous as ever with your time and help.

 

It occured to me that if I had a look at both sides of the board, there might be something visible, but didn't want to disturb all that mozarella glue too much if I might send it away ;-)  I may have talked myself out of that uninspiring option.

 

Presumably, if we know what the current draw should be, placing the current meter between battery contact and battery post with ignition on ought to measure the current draw?  If we can't find a figure for that maybe someone with an OK car could test theirs, and I could run the same test?  It would be great to find a good contender for the fault that way, and it would enable me to test a secondhand ebay unit as well.  They seem to start at about a tenner, and might be a reasonable prospect if I can speak to the seller and feel some confidence that this was off a vehicle scrapped for other reasons.

 

Hope springs eternal, or 'its not the despair, its the hope I can't stand'!

 

 

 

[Breezy_Pete]

Would have been a waste of time looking at the other side, there's nothing on it.

 

Pin 23 was relatively easy to trace, it comes from the output of a 'Smart' Power FET transistor as shown rather scruffily here:

 

Pin 62 took a bit more probing, but I eventually found it connected to at least a couple of things, one being the big chip in the middle, via a potential divider pair of resistors (ringed in blue) which I guess takes it from nominal 12V down to <5V as an input to that chip (telling it whether the relay is energised or not, at a guess). The other way, towards the red-ringed 10k resistor I haven't traced yet.  The resistor with the blue 0 next to it is a zero ohm link.

Edited 27 June, 2017 by Wino

[Breezy_Pete]

I can only think of two possibilities here at the moment. I have no doubt there are others though.

 

1. That power FET (link to a datasheet) has some internal damage causing it to fail to adequately deal with the switch-off spike that relay coils always generate when they are de-energised.  This could mean that the diode in the relay is getting a voltage spike instead, every time the engine is switched off, gradually damaging it to the point where it fails short-circuit, giving the strange back-feeding of 12V and blowing the fuse.

 

2. The voltage spikes are coming from beyond the ECU/relay system, for example due to an alternator fault.

 

It would be nice to compare an oscilloscope trace of what's happening at pin 23 (or pin1/86a of the relay) during shutdown/ignition off for your daughter's car (with new ECU relay fitted) and A.N. Other like ours. Also a comparison of voltage smoothness across the battery terminals might eliminate or point to possibility number 2.

 

It would be fairly straightforward to change that FET for a new one if the traces looked significantly different.  You could just change the FET anyway, but then you won't really know if you've actually got to the bottom of the matter.

 

Pin 62 appears to be monitoring the presence (and probably magnitude) of the voltage on the relay side of fuse 14.  I can't see how that could be the cause of any trouble.

 

[NeilTM]

1 hour ago, Wino said:

I can only think of two possibilities here at the moment. I have no doubt there are others though.

 

1. That power FET (link to a datasheet) has some internal damage causing it to fail to adequately deal with the switch-off spike that relay coils always generate when they are de-energised.  This could mean that the diode in the relay is getting a voltage spike instead, every time the engine is switched off, gradually damaging it to the point where it fails short-circuit, giving the strange back-feeding of 12V and blowing the fuse.

 

2. The voltage spikes are coming from beyond the ECU/relay system, for example due to an alternator fault.

 

It would be nice to compare an oscilloscope trace of what's happening at pin 23 (or pin1/86a of the relay) during shutdown/ignition off for your daughter's car (with new ECU relay fitted) and A.N. Other like ours. Also a comparison of voltage smoothness across the battery terminals might eliminate or point to possibility number 2.

 

It would be fairly straightforward to change that FET for a new one if the traces looked significantly different.  You could just change the FET anyway, but then you won't really know if you've actually got to the bottom of the matter.

 

Pin 62 appears to be monitoring the presence (and probably magnitude) of the voltage on the relay side of fuse 14.  I can't see how that could be the cause of any trouble.

 

 

Thank you so much for all this Wino, I really appreciate it.  Unfortunately my understanding lags a bit behind my appreciation, its clear you understand electronics a lot more than I do.

 

I don't have an oscilloscope, but I have a brother who does 70 miles away, so that might be a possibility at some point.

 

However, I get that I can certainly test battery voltage smoothness while running, and I have an analog multimeter as well as a digital one if that might be better at showing up a spike?  A couple of questions about that test:

 

Would I need to put the new relay in for that test?  I don't want to risk blowing it if I don't have to, they are more expensive than fuses!

 

Would this just be a test with the engine running, or should it include stopping and starting?

 

How long to test for?  OK, that's 3 questions!

 

I recall my brother saying that these ECUs are incredibly robust, to the extent that they will even survive a reverse polarity jump start mistake, and so should presumably be able to cope with voltage spikes?  Perhaps they do, but it is the relay and fuse which cops it instead of the unit?

 

The Cannock people also thought that pin 62 was not significant to this fault.

 

I'm quite keen on the idea of testing the normal current draw with just the ignition switched on, since if I have understood correctly from what the Cannock testers and yourself are saying, that might be an underlying predisposing condition, and easily tested for if I just know what I should be comparing it with?

 

Its still raining here, so can't do anything immediately.

 

Thanks again,   Neil

[Breezy_Pete]

Only an oscilloscope is going to show the very short spikes involved, I believe. 

 

I have a plan though...The datasheet for that power FET shows you, on the right of the diagram, an internal diode that would tend to catch that switch-off spike when the relay is de-energised. With multimeter set to measure diodes, we could compare readings between pins of that device on our respective ECUs. If different, that would be useful evidence that your FET might be the problem?

[Breezy_Pete]

FET block diagram with diode ringed, and positions arrowed for diode measurement, then an action shot.  With probes swapped, I just see a negative sign come up in front of the same number. 

 

 

 

If you do see a significantly different number, try sliding your meter probes across to the other FET just to the left of the one I'm probing, and see how the readings compare between the two. On the pictured ECU both give the same number.

Edited 28 June, 2017 by Wino
extra check

[NeilTM]

What an amazingly clear and instructive photo, thank you.

 

I've conducted the test and suspect that it does not reveal any anomaly.  both diodes read 650 on my meter which reads out differently from yours.  There is no decimal point in my readout, just 3 digits.  If I short the leads out it reads 2.  As a comparison, I tested the diode in the new relay which reads 634.

 

I would expect that the different numbers in our readings probably reflect comparative inaccuracies between our meters, unless your slightly different ECU has slightly different value components.  The consistency, and being in an expected sort of range seems more significant don't you think?  My meter cost something like £2.50 inc postage from China through ebay, LOL.

 

You haven't said what you thought of the idea to test the current draw of the unit, to compare with A. N. Other?  Is that something you would mind doing?  This is what the Cannock folk said about that:

 

"We do monitor current
draw from the ecu amongst other measurements and can monitor for unusually
high figures which can on the car at the time not cause a fault but is
showing that there is a failure mode sitting there hiding ready to show up at
some point. "

 

I took this to mean that the higher current draw would be permanently present and presumably therefore easier to test for?  The question is, would a current meter between battery post and battery terminal with ignition switch turned on be a useful or possible test?  Beginning on the 10amp range for safety of the meter?  If there's a doubt about that I would be happy to sacrifice my meter to save you overloading yours since they are so cheap and I have 2 of them anyway.  But I don't imagine somehow that they are talking anything like a 10amp draw as the excess, the thing would surely start glowing!

[Breezy_Pete]

Yep, that test gives the FET the all clear in terms of the protection diode, I think.

I would have been surprised if it hadn't, because I think if that diode dies, the first subsequent victim would be the main guts of the FET, rather than anything external, but it was worth a try.

 

Current draw of the ECU is a good thing to investigate, I'll have a think about the easiest way to measure 'just' that later on. I might have a cunning plan for that.

[Breezy_Pete]

On your daughter's car, and my partner's, Fuse 52 supplies just the main power feed to the ECU, and the ignition coilpacks.  I have a butchered blade fuse with wires hanging off it (and no actual fuse link in between), to enable one to measure the current in a circuit at the fusebox.  I can substitute that for fuse 52 on our Fabia, and measure the current used by ECU + coilpacks, then send the thing to you for you to make a comparison.  If I do it tonight, and post it to you tomorrow, there's a fighting chance you might get it on Saturday.

The result would be slightly ambiguous, as a difference might be explained by either an ECU current consumption difference or a coilpack consumption difference, but if they're very similar we could probably rule out both.

[Breezy_Pete]

Well that was quite interesting. Widget fitted in place of fuse 52 (15 Amp).

Ignition on and the ammeter says about 0.35A, start-up (from cold) had a very brief blip up to about 2A, then settling quickly down to about 1A. Danced about a bit between about 0.85 and 1.25A. This is all at idle, sitting on the drive. 

The interesting bit was shutdown. Key right round to off position: Brief blip up to 1.5A as the engine dies, then down to 0.7 for about 10 seconds, then down to 0.35 for another 15 or 20 seconds before finally going to zero.

 

I had been wondering what the 'secondary' feed to the relay coil from pin 23 was all about. Why does the ECU need to continue to arrange power for itself (its pin 3, via the relay) after ignition off? Best guess it's a little like shutting a computer down (well the ECU is one, after all) and it takes a while for it to sort itself out and get in a state where it's ready for the next start-up?

 

A bit surprised how little current the ECU/coilpack combo takes at idle, I guess the coilpacks will take an amount roughly proportional to revs, but I didn't try that.

Picture of 'widget':

 

 

Edited 29 June, 2017 by Wino

[sepulchrave]

Coilpacks, like the injectors are switched to earth via the ECU so you wouldn't see what current they use but like all HT coils the current drain is small since they're all about the voltage.

[NeilTM]

Excellent idea!

 

Not really sure why you couldn't substitute fuse 14 though since that only supplies the ECU does it not?

 

I can make one of your widgets easily tomorrow and try your experiment and see how it compares.

 

Even with the coil packs connected, I would have thought that any significant difference ought to be down to the ECU, but we'll see what we get.

[Breezy_Pete]

10 minutes ago, sepulchrave said:

Coilpacks, like the injectors are switched to earth via the ECU so you wouldn't see what current they use but like all HT coils the current drain is small since they're all about the voltage.

Fuse 52 feeds the constant 12 volts to the primary of each though, as well as supplying the ECU power.

[Breezy_Pete]

11 minutes ago, NeilTM said:

Excellent idea!

 

Not really sure why you couldn't substitute fuse 14 though since that only supplies the ECU does it not?

 

I can make one of your widgets easily tomorrow and try your experiment and see how it compares.

 

Even with the coil packs connected, I would have thought that any significant difference ought to be down to the ECU, but we'll see what we get.

Fuse 14 only supplies the current for the coil of the ECU relay, that's a separate circuit from the one that supplies the ECU's power. We could also compare the currents through our fuse 14s, but that's a mini-blade, so would need another widget.

 

[sepulchrave]

5 minutes ago, Wino said:

Fuse 52 feeds the constant 12 volts to the primary of each though, as well as supplying the ECU power.

 

Ok, you would see it then, but it'd be insignificant compared to the ECU.

[Breezy_Pete]

I guess; just intrigued by why the fuse is 15A and I didn't see much more than 1, except very fleetingly. I guess the digital meter may not show the true magnitudes of the peak currents.

[Breezy_Pete]

 

 

Hopefully this diagram and some of the following words will clarify what I’m trying to describe, for the benefit of those who haven’t seen the current flow diagram that we’re referring to.

 

The blue, red and orange bits are my additions to the circuit diagram, showing some of the internal stuff in the engine ECU, and some of the fuses/destinations that aren’t shown on the original page.

 

There are a few different scenarios, which I’ll try to describe by what’s going on in each of the wire legs,  A-K.

 

1. Key not in ignition, hasn’t been for a few minutes, everything off. No current flow in any part of circuit; but leg F to pin 2 of the ECU relay is at battery voltage. The relay is not energised (switch still in position drawn) so it cannot make a current flow towards leg G.

    

2. Key in ignition, and turned to the position where the dash lights up and does its self-tests. Current is now available via Fuse 14 down leg A, which connects to both the coil of the ECU relay via leg B, and ECU pin 62 via leg C. Leg B to pin 6 of the relay will energise the relay coil (through internal diode), closing the switch contacts and allowing current through the relay from leg F to leg G. A very small current will also flow down leg C, to the ECU pin 62, which connects internally to a voltage measuring circuit consisting of two resistors and an analogue input of the CPU.  The ECU is now powered, through fuse 52 and pin 3, legs G, H, I, and once it has booted up, will be capable of measuring the voltage at junction A74 (A, B, C junction), and, if/when it decides to, supplying current via the FET to the relay coil (pin 23 of ECU to pin 6 of relay, leg D), so it can keep itself fed with power even if the ignition supply via fuse 14 is removed.

   Current is also available now for the coilpacks and EGR solenoid if reqd, via legs J and K (via fuse 17) respectively.

    

    

3. Engine started. All as before, but now current will certainly be flowing down leg J to the coilpacks, and may flow down leg K via fuse 17 as and when EGR operation is required.

    

4. Ignition switched off (shutdown). Current flow down leg A ceases, so current can no longer flow down legs B or C. ECU will detect that voltage at A74 has gone to zero. This will tell it that the driver has requested engine shutdown.  It will keep itself powered, using the FET output through leg D to keep the relay coil energised. When it has shut the engine down cleanly, and done whatever housekeeping, memory-storing or whatever, it will stop this FET from conducting, de-energising the relay, and depriving itself, the coilpacks and the EGR solenoid of any power, until such time as the ignition is switched on again.

    

5. Fault condition – Internal diode in relay has failed short-circuit.  Now in scenario 4, when the ignition is switched off, chaos may ensue, because the FET will now try to power not just the relay coil, but back-feed everything connected to ignition 12V, via fuse 14. Depending on what loads (e.g. cabin fan) are switched on at the time of the ignition switch being turned off, the fuse will last more or less time before blowing. Furthermore, the conduction through the failed diode, down legs B and C to the voltage measuring circuit of the ECU, will mean that it doesn't even know the ignition is switched off at the key.

Edited 30 June, 2017 by Wino

[NeilTM]

The whatdunnit is finally solved and fixed, so no more repeat of the intermittent short causing the relay diode to short circuit and the fuse 14 to blow should be possible.

My immense thanks to Wino for taking so much time and trouble all the way through this difficult diagnosis and testing, but he really came through, not only with the answers, but posted me a couple of test leads on cannibalised fuse blades for testing the circuit, and to see what was happening while running.

Trying to find an intermittent short that has a frequency of every 6 months and 3,900 miles was never a promising prospect, although latterly it not only showed up again immediately after I had conducted lots of running tests, but remained a fairly consistent dead short!  That was quite a bewildering (and convenient!) coincidence, and a wonder that I hadn't just blown another diode and fuse.

My brother realised that the FET in the ECU was a 'smart FET' that could not pass more than about an amp, and so that route for the diode shorting and fuse blowing as Wino described in note 5 above was ruled out.  In fact I think it is probably correct to say that if enough current to short a 10amp fuse had passed through the ECU board it would have left scorch marks and destroyed some components along the way.  I'm thinking that the nature of the event in itself weighed against the likelihood that a visually inspected OK board could have delivered such a current, but the Cannock testers said 50/50 chance of it being wiring or ECU.  Worth a visual inspection first therefore before committing to expensive testing I would say in these sorts of circumstances.

 

By the time that the very slow first class snail had slime trailed its way to Milton Keynes from Saturday to Wednesday with Wino's test lead for measuring the current draw of the ECU and coils, I think we had already worked out that the culprit had to be leg D in Wino's diagram, the thin .35 blue and green wire that connects pin 86a of the relay with pin 23 of the ECU.

Wino supplied me with a link to a video of his test of his wife's Fabia, showing the current draw and pattern of current draw on the multimeter throughout the sequence of ignition on, starting, running, and stopping.  I made my own similar video, and it seemed that the measurements were comparable, and so we concluded that the ECU was functioning as it should, and not drawing excessive current as the ECU testers had posited could be a sign of eventual stress failure.

Wino and my brother having deduced from an analysis of the circuit diagram that only an intermittent short to earth somewhere on leg D could have shorted the diode and blown the fuse, I set about trying to find out where the short was likely to be.  By testing continuity between pin 86a and nearby earth, I was obtaining more or less a dead short.  By testing between pin 23 and nearby earth at the other end of the wire I was getting a resistance of .1 to .2 of an amp, suggesting the short was at the relay end and measuring the resistance of the length of the wire.  This was then confirmed by measuring continuity from one end while waggling the wires at the other.  Nothing from the ECU end of things, but tugging on the wire to the back of the relay socket it became open circuit several times, and shorted, and all points in between, and activity furthest from the relay socket seemed to produce it.  We had found which bit of the wire the short was occurring on, but impossible access behind the relay holder, visually and physically, as well as fruitlessly searching high and low for my endoscope, has robbed us of the possibility we may have got a visual of the actual short, and perhaps some idea as to what could have caused a short in a pristine and never visited deep region of behind the dash against the bulkhead.

I was able to access the back of the relay plug by cutting through enough of the holder in the empty row above, and filing down some empty bay catches to withdraw the holder forwards and gain access to the wire at the back, leaving as long a tail as we could, thanks to my wife's small hands, to solder a new wire to decently, having first passed one through sellotaped to the end of a rod passed through a 5.5mm hole I drilled through the bulkhead next to the panel through which the cableform wires passed through the bulkhead to six connectors on the cabin side.

The length of the wire replaced must be less than a foot, and from two accessible places outside of wrapped cableform, and the new wire takes a straight route through the bulkhead to the back of the fuse holder - all quite ideal really.  I grommeted the hole and added sealant where the wire goes through it. Testing for shorting after cutting the wire also confirmed which side the short was, and subsequent testing revealed no shorting at all.

Had to take the bonnet off and undo the panel behind which the wiper motor is located in order to remove that to gain access to the panel through which all the wires pass.  All straightforward once I had used my large puller to remove the wiper arms which refused to budge any other way.

 

Thanks to all who contributed to making this difficult repair possible, and to be able to be confident that it will not recur.