Vauxhall Vectra Auto | Intermittent non-start

Customer’s description

The customer reported a clicking noise from under the bonnet when attempting to start the engine (the engine failed to crank).

Technical description

Verifying the customer complaint is an essential step in the diagnostic process but is also quite often a time-consuming task with no success. On this occasion, the customer's complaint was verified. The engine failed to crank and there was a clicking noise from the starter motor when we held the ignition key in the crank position. The warning lights in the instrument panel would also flicker in unison with the clicking from the starter motor. When we applied a boost pack to the vehicle battery, the engine would occasionally crank and start as normal, accompanied by a screeching noise from the starter motor.  

N.B. In contradiction to the above, there were several occasions where the engine would crank and start as normal without support from a battery boost pack (i.e. there was no apparent fault).
  
The videos below demonstrate the symptoms described.
 

Diagnosis 

With the customer's complaint verified, we confirmed the Vehicle’s ID and Specification. 
Confirmation of specification is of the utmost importance when it comes to diagnosis. Owners often get tempted to modify their vehicle with fashionable accessories that lack the fundamental quality control and engineering that was intended for the vehicle. The customer confirmed that no such accessories had been installed.  

The Customer Interview follows the 4 “targeted” open question principle below, to help establish facts from fiction:

1.    How long has the problem been evident?
No previous experience of a non-crank/start event or any engine running problems 
2.    When did you first notice the problem?
The symptom occurred when attempting to start the vehicle after it had been standing for over 6 months (MOT was due).
3.    Has any work been carried out on the vehicle recently?
No work had been carried out before the non-crank event.
4.    When do you experience the problem?
When attempting to start the vehicle, once started the engine would continue to run. 
 
Quite often, the customer interview leads to a “probable” diagnosis. My initial thoughts turned to low battery voltage (parasitic drain) or 12 V battery failure since the vehicle has been stationary for several months and applying a booster improves the non-cranking issue. 

The Basic Inspection confirmed no fluid leakages, no visible signs of damage to hoses, connections or wiring harnesses, and no sign of accident repair.

A vehicle scan of all on-board control units revealed several fault codes which were saved then erased:

Steering: 
C1550 Control unit Malfunction
U2105 Communication with engine control unit Communication faulty

Brake control unit: 
U2105 Communication with engine control unit Communication faulty

Steering column switch module:
U2105 Communication with engine control unit Communication faulty

Instrument cluster:
U2105 Communication with engine control unit No communication

Before diving in, we took a step back and checked for Technical Bulletins (Recalls & Campaigns etc.). This check is paramount. The most efficient route to qualify the presence of known issues, fixes, etc., is to use the manufacturer technical portal combined with the vehicle's chassis number. 
While this may seem obvious, there are many avenues you can take to obtain free technical bulletins and so-called “known fixes”! I have learned that while the entire description of such obtained data matches the exact symptom you are experiencing, the information obtained may well be for other markets i.e., not applicable to UK specification vehicles. (Please don’t fall into this trap as I have with previous vehicles.)

Possible causes

• Battery and charging system (inc. parasitic drain) 
• Cranking control circuit 
• Starter motor assembly (Screeching noise upon cranking)
• CAN network communication error 
• Engine management system input error

The action plan

The action plan was predominately governed by accessibility, probability and cost.
We quickly eliminated the battery and charging system as we confirmed them to be OK.
Based on the acquired scan data, you cannot ignore numerous ECU’s reporting loss of communication with the Engine ECU. Again, it was tempting to dive into CAN network diagnosis but I felt as though we could be working blind! To understand how such CAN errors could influence cranking, we needed to interpret how the starter motor was controlled and this was where we focused our attention, referring to the relevant technical information.  

To recap

1.    The engine fails to crank intermittently (Battery voltage and performance verified) 
2.    The engine will occasionally crank with a battery boost pack installed 
3.    Intermittently the engine will crank normally (no boost pack installed) 
4.    Instrumentation/ engine MIL flickers in unison with “clicking” starter motor  
5.    Multiple fault codes stored referring to a loss of CAN communication with Engine ECU 

The cranking circuit relied on the engine ECU to ground pin 85 of the starter motor relay, the image below (courtesy of ALLDATA) helped clarify the circuit.
 

From ALLDATA

Let me explain the diagram above and the missing circuits: 
1.    The ignition switch supplies the gear position switch with a 12 V supply when held in the cranking position.
2.    If the gear position switch is in either Park or Neutral position, the supply from the ignition switch is allowed to pass onto pin 86 of the starter motor relay.
3.    The starter relay (pin 85) is then grounded via the engine ECU (closing the relay contacts) and cranking can commence.

So, what was causing the starter motor to repeatedly “click” and the engine failing to crank? The wiring diagram below (courtesy of Autodata) will assist with the descriptions included in the following waveforms, where we caught the failure.
 

From Autodata

Here we have captured the engine ECU failing to keep starter relay pin 85 grounded, and pulsing the ground path instead (see Channel E).

The result, therefore, is the rapid energizing and de-energizing of the starter relay and, of course, the engaging/disengaging of the starter motor as a direct consequence.

Since we had eight channels to acquire data with, we could eliminate numerous components in one single capture. Below we have zoomed in on just one of the “pulsing” events of the starter motor relay.
 

When we referred to the above captures and wiring diagrams we could conclude the following:
•    The engine ECU did not lose any of the 3 power supplies to pins 15, 16, and 47 during intermittent cranking, our ignition/engine control relays and circuits were, therefore, OK.
•    That also meant that the fuse boxes and the terminals “housing” the above relays had to be OK.
•    The signal from the ignition key to the starter relay pin 86, via the gear position switch, was OK.
•    The power supply to starter relay pin 30 was OK.  
•    The starter relay was OK, as it was simply responding to the pulsed ground of Pin 85.
•    The starter motor did momentarily energise based on the inrush current captured on Channel F.

All in all, a set of worthy test results that allowed our focus to firmly shift to the engine ECU and our numerous CAN codes reporting a loss of communication (we are not working blind).

It is always comforting to know the diagnostic path you have chosen is the right one and by this, I mean validation of the test results above. We knew the starter relay was losing the ground path on terminal 85 and what better way to qualify this scenario than using a bypass test?

With the ignition on, we removed the starter motor relay and joined terminals 30 to 87 in the engine bay fuse box. The vehicle cranked and started every time (accompanied by a screech from the starter motor). This test qualified, beyond any doubt, that we were justified in pursuing the ground path of the starter motor relay via the engine ECU.

While we have tested the ECU power supplies what about the main ECU grounding?   
 
Given that the engine ECU was bolted to the engine (and in turn acquires an earth path via the mounting bolts), we carried out a volt drop test (on the ground path) to make sure that the ECU was earthed correctly.
N.B. for the volt drop testing, we used a 4425A to take advantage of the floating inputs. You can find more information on floating inputs here.

During the starter motor peak inrush above (Channel D), we captured an average volt drop of 785 mV. I initially thought that was high, however, the addition of a ground “jump lead” did not change this value. We also knew that the engine ECU was capable of running the engine during the relay bypass test, which removed any doubt surrounding external earthing. (i.e., ECU aluminium casing to chassis ground).

Armed with the confidence generated by the previous tests, we had to test the CAN network at the engine ECU under the fault condition. (Remember multiple ECUs were reporting a loss of communication with the engine ECU.)

Below, we capture CAN Lo during the intermittent cranking symptom accompanied by a simultaneous loss of communication to the connected scan tool (monitoring engine ECU live data) and the flickering of the engine MIL.

Note how the voltage drops to approximately 8 volts during the momentary operation of the starter motor (Channels E and F). This is to be expected during the initial inrush of the starter motor current. However, this is repeated at approximately 300 ms intervals, which generates huge amounts of noise in the engine control circuits (Channel E).

While these voltages appear to remain at an expected level during the “inrush” at our external measurement points, they also supply other actuators (e.g. the throttle motor) via internal electronics in the engine ECU. (Refer to A177 in the Autodata wiring diagram).

It was noted that, periodically, during intermittent cranking, the throttle motor would appear to “whine” which could suggest a cyclic control of the power supply from inside the engine ECU.
Note the back EMF present on Channel E which is characteristic of stop/start motor assemblies (think of inductive spikes from fuel injector coils).  

Channel G revealed an increased voltage drop (2 V) at ECU pin 31 (which is responsible for grounding the engine control relay), accompanied by some very strange behaviour of our CAN Lo signal on Channel H which appeared to momentarily switch from CAN Lo to CAN Hi!
It was at this precise point that communication was lost with the scan tool.

Post intermittent cranking, we checked the fault codes again and all the previously erased fault codes had returned with multiple ECUs reporting a loss of communication with the engine ECU.

Below, we analysed the CAN network at the engine ECU in further detail during the fault condition.

Note how the CAN noise level fell silent immediately after the momentary inrush of current into the starter motor. During this “noise-free” period of CAN activity, the engine ECU is offline until the noise level increases again after approximately 160 ms. The noise level occurs as the engine ECU momentarily appears online and begins to run through the throttle motor initialisation procedure.

This is characteristic of an engine ECU that has woken from sleep while preparing for the engine to start and is often heard when turning on the ignition of many vehicles (engine off).

Before I move on here, I must mention the fault-tolerant CAN in action in the right-hand capture above. Note how the raw signal (CAN Hi and CAN Lo) is distorted, yet the differential signal (Math channel C-D) remains perfectly intact. You can find more information on fault-tolerant CAN in the following forum post.

Given that our engine ECU was falling offline, we could use the serial decoding feature in PicoScope to select the precise area of CAN data we wanted to decode via the “decode between rulers” option. You can find more details and an accompanying video here.

The decoding allowed for comparisons of CAN IDs before and during the fault condition (ECU on/offline) where we identified the missing CAN IDs associated with the engine ECU, reinforcing our diagnosis.
 
In the screenshot below we captured several CAN IDs with the ECU online before the fault condition.

In the following screenshot, we captured several CAN IDs with the ECU offline during the fault condition.

We exported the data in the serial decode table as a CSV file, which gives you the ability to apply additional filtering of each CAN ID using Excel. You can find more information about this technique in the following forum post.

The following is the full list of IDs that were missing before and after the fault condition.

To determine if the missing IDs are entirely relevant to the engine ECU, we carried out the same procedure again without introducing the cranking fault. We captured CAN IDs with the ignition on and the engine ECU connected, then repeated the capture with the ignition on and the engine ECU disconnected. Now we could compare the missing IDs identified during this deliberate “offline” test with the missing IDs during the true fault condition.

A word to the wise when using this technique: If the ECU you are disconnecting contains a termination resistor, you will need a 120-ohm resistor to link CAN Hi to CAN Lo. This is necessary to ensure the integrity of the CAN network. Be aware that not all missing IDs originate from the engine ECU as some IDs from other ECUs may be absent because the engine ECU is offline.    

To cut a long story short, we identified and confirmed several missing CAN IDs as related to the engine ECU and so we have gathered enough evidence to decide that this component needed to be replaced.

It was at this point (when I realised the cost of a new engine ECU), that I had a moment of self-doubt, especially given the value of the car and the additional work required for the annual MOT inspection.

So, when in doubt, shout for help. This is where I turned to the collective knowledge of our automotive forum. This is not the first time I have needed a shoulder to cry on and I urge anyone who is either struggling or looking for support, to use this forum for what it is worth. If you post asking for help, please provide as much information as you can about the vehicle under test and the diagnosis process already carried out. It is also great if you can include psdata files with correctly labelled channels or good quality screenshots (ideally not a photograph of a laptop screen). The best advice I can give is to imagine that you are helping a colleague in the scenario above. What information would you require to provide additional support?  
 
To say I struck gold with my forum post would be an understatement. Thanks to Liviu who is the Obi-Wan Kenobi (legendary Jedi Master for Vauxhall-Opel), I was able to proceed further with additional checks surrounding intermittent cranking.

If we look at the wiring diagram for this engine management, the only visible “request” for cranking is via pin 85 of the starter relay or the CAN network.

Thanks to Liviu, I realised I had not taken into consideration the crank signal from the Column Integration Module (CIM)! In addition, there was a concern raised about the throttle motor and control circuit with this particular generation of Vectra. Driven by Liviu's experience, I decided to investigate both. I had this inner fear that replacing the engine ECU and not the throttle body would result in premature failure of the replacement engine ECU.

In the screenshot below, we captured the entire circuit of the throttle body during initial activation. Note the instability of the PWM signal at ignition on, which appeared to have no ground reference and “float”. (Remember that the ECU also failed to ground the starter motor relay.) This “instability” appeared to improve approximately 12 seconds after the initialization/calibration of the throttle!

A greater concern was the sporadic throttle motor current, which appeared to fluctuate between –6 and +6 A. You can see this huge fluctuation during the initialization of the throttle in the following screenshot:

In the screenshot below, we could see the repeated initialization/calibration of the throttle while holding the ignition lock/switch in the cranking position. This appeared to be the Engine ECU continually resetting.
Note that the CIM signal (ECU pin 64) remained at 0 V, as expected when cranking. We, therefore, concluded that the CIM signal was correct.

So, what does this all mean? Is the engine ECU at fault or is the throttle motor faulty causing consequential damage to the ECU? Given the sporadic current flow through the throttle motor, we removed the assembly, and dismantled and inspected the commutator.
The video below confirmed the commutator to be serviceable with no burning, shorting, score markings and sufficient brush length. The rotor windings were also confirmed serviceable for resistance and insulation along with the mechanical operation of the throttle butterfly and associated gears.

To recap 

•    The engine ECU PWM control of the throttle motor appeared to “float”.
•    Increased volt drop across engine ECU internal ground path (2 V – 0.785 V = 1.215 V)
•    There was a loss of communication between the engine ECU and the CAN network.
•    All relevant inputs to the engine ECU were present and correct when the fault occurred.
•    There was disruption to the CAN Hi and CAN Lo formation during the fault condition.
•    The engine ECU failed to continually ground starter relay pin 85 during cranking.

I hoped that we had carried out enough testing to qualify the replacement of the engine ECU. The decision was taken (rightly or wrongly) to install a second-hand ECU and a new starter motor (given the screeching during operation).

After the installation of these components, the vehicle cranked, ran and drove as normal accompanied by a sigh of relief from all involved. However, it was not over yet! Later that same evening, Kevin Ives sent a stomach-churning message: “The fault has returned”.   

Based on the logical process taken during diagnosis, what more could we have done and what could I have missed?

Sure enough, when I returned to the vehicle, we found identical symptoms and so the diagnosis began all over again. But who should be paying for it? This is the risk you take with used parts and I was grateful that I was not relying on a bonus system for my salary.

I am not going to bore you with a repeat of the above diagnosis, other than to confirm how our second-hand ECU performed identically to the original unit.

The doubt that can creep in at times like these is overwhelming, I guess it's human nature. I am convinced the diagnosis is correct and the only “variable” is our second-hand component. After much deliberation, we replaced the ECU again, with another second-hand unit, and the vehicle cranked, ran and drove as normal.

To be sure that our fault had cleared, we subjected the engine to multiple cranking events and prolonged road tests in an attempt to reproduce the symptom. I am more than happy to say that after weeks of testing, the intermittent cranking issue was resolved (see capture below).   

The keen-eyed reader would notice that while the engine now cranked correctly, and pin 85 of the starter motor relay was being held to ground as expected, our cranking time periodically increased!

The waveform above reveals one good cranking event of around 1.2 seconds and one prolonged event of 3 seconds. There were times that cranking would continue beyond 5 seconds and the engine did not attempt to start! If reading this case study is making you think “when will it end?”, you can only imagine what I was feeling before handing the vehicle back to the customer.  
Could we have another faulty engine ECU, but with a different symptom?

Looking closely at the waveform above, there is a clue in the prolonged cranking current as to why this engine will not start and the following forum post covers this in greater detail.

The waveform below zooms into the cranking current along with the inductive crankshaft sensor signal:

Note the prolonged cranking with very odd current peaks during the rotation of the starter motor. The current appeared to be switching on/off with periodic gaps where no current was flowing. For every revolution of the starter motor, the current was switching on and off approximately 23 times, which equates to the same number of segments on the commutator!
 
Thanks again to Liviu, I now know that this is often the case with new starter motors where the “bedding in” process between the brush and commutator is incomplete. From the included image we can see the uneven patina/film coating about the surface of the commutator which I must confess I have never experienced or fallen foul of until now!

It would appear to be a typical phenomenon with reconditioned starter motors for several reasons, such as the “final finish” of the brushes and commutator, material properties, environmental use, brush spring tension and alignment.
 
Moving onto prolonged cranking, note the condition of the crankshaft signal during the starter motor operation. The inductive crankshaft sensor signal had been adversely affected by the electromagnetic interference (EMI) emitted from our new starter motor! Such interference may “consume” our inductive crank signal via the wiring harness where there is a capacitive link between the main starter motor positive cable and adjacent wiring, or indeed, via the engine bay environment itself. The irony in all of this is that refitting the original starter motor cured the problem! 

Results / Confirmation of repair

Below we have a “clean” crank and start event, with minimal disruption to the crankshaft sensor signal and seamless current flow through the starter motor (cranking time 784.6 ms).
Note the stability of the CAN signal throughout (i.e. no loss of commutation) and the reduced noise level on channel A (pin 87 engine control relay). 
 

Below, you can see the operation of the throttle during initialisation, start-up, WOT, idle and ignition off. Note how the current flow still seems sporadic yet the throttle control is stable. It is difficult to determine whether this could improve further still with the new engine ECU, however, at this stage, I was happy to hand over the vehicle to the customer.

List All Parts Fitted

Two engine ECUs and a Starter motor (the original starter motor was refitted). 

Additional comments

Why the delayed stater motor inrush? 
I want us to go back to the first waveform in this case study. I was intrigued by the behaviour of the starter motor inrush current, which clearly showed current flowing into the starter motor after the starter relay had de-energized! Following discussions with Liviu, Kevin Ives and the Pico team, we suspect the delayed current flow was due to a combination of the electromagnetic and mechanical inertia of the starter relay and the inductance of both the starter motor and relay windings. This is food for thought and one for further discussion (perhaps another experiment too) at a later time.
 
Why does the battery boost overcome the fault condition?
Early in the case study, I described how installing a battery booster to the vehicle allowed the vehicle to crank as normal. If we take the fault condition of the ECU (failure to ground pin 85 of the starter relay), we can assume that either the control circuit of this relay (inside the engine ECU) or associated PCB components are subjected to insufficient voltage caused by excessive internal resistance. Such events may either shut down specific areas of the ECU or indeed momentarily close and reboot the CPU. To that end, adding a battery booster may overcome the offending volt drop within the ECU, allowing normal operation during cranking.
Why did the engine MIL flicker in unison with the starter motor operation?  
I originally thought that the flickering instruments and engine MIL were due to the volt drop alone. However, with hindsight, a combination of the engine ECU momentarily falling offline would result in no operation of the engine MIL (controlled by the engine ECU) and this of course would impact the CAN messaging to the instrument panel which has the potential to simultaneously interrupt the operation of other warning lights. 

To conclude:
This case study is certainly not unique, as there are thousands of ageing vehicles that are simply too good to scrap but too expensive to repair. The temptation to use pattern and second-hand parts is a natural one, given the costs of tooling, training, test equipment and labour. This begs the question: “Where do you draw the line?”. The majority of customers will pay for an initial one hour of diagnosis (AKA the Golden Hour), but as we have described above how do you quote for unforeseen events like these?
Perhaps the answer is to steer clear or to not get involved with second-hand/pattern parts in the first place? However, if we are dealing with a lifelong customer (often with more than one vehicle), the obligation to help can be overwhelming and cloud sensible judgment. Having an open and honest customer rapport, which includes regular updates and explanations of what could go wrong at each twist and turn of the diagnosis, can help manage their expectations. (Not to mention the reliability of any repair.)  
From my perspective, I have learned so much from this study and as with all diagnoses, there has to be a logical procedure to follow that includes various levels of diagnostic intrusion accompanied by costs. Just how those costs are managed requires a skill all of its own.

Many thanks to Kevin Ives at Ives Garage and Liviu for their support throughout this case, which spanned from December 2020 to April 2021.