Technical Fuel pump(?) advice sought...

[tjr]

The b-95 is now 29 years old, so I suppose all parts have the right to give up...

Symptoms:
The b-95 has always been eager to start during the two years I've known it, but this season, it is in need of "priming", ie activating the pump 2-3 times with the key (back and forth between OFF and MAR), maybe 50% of the time. The pump will activate every time judging by sound. The "priming" so far gets the engine to start every time.
This morning the engine stalled shortly after take-off (a first, while driving, taking a tight right corner). The engine started up again and completed a 7 minute journey without issues.
Before learning of the "priming" technique and within 36 hours of coming out of winter storage, the b-95 became extremely difficult to start. The first real cold start (0°C, plenty of moisture in the air) was no problem, but the second start-up took three tries and thereafter it took maybe ten tries. During these tries it would try to start maybe three times, and then die.

Additional:
Idle and acceleration is without hiccups as in failing spark(s).
No warning lights turning on (eg check engine, injectors)
Tank over half full
The fuel filter was changed 1-2 years ago (Ridex filter, mounted in the correct direction)
I have no idea of the fuel pump age

Questions:
Anything standing out? My primary suspects are; fuel pump, air leak after MAF, and/or crank shaft position sensor, thanks to @Ilkka. Are any of those more prone to failure? What about the fuel pressure regulator?
If changing the fuel pump, do you order/change the whole setup with housing and level indicator or just the actual pump? If changing the pump I suppose you need a new seal?
Am I missing something?

Thanks,
tjr

 

[jmi]

It is possible that you have original pump. And even more possible that pump is starting to fail, or it is clogged as there might be some debris in the tank. It sounds that there might be some fuel pressure issues as you need to "prime" fuel system. But yes, it might be something else...

When you turn on ignition, fuel pump should run for few seconds to generate pressure and then stop.

Anyway.

Pump itself is quite easy to replace. Or "easy". Remove plastic covers from hood compartment, there is hacth with screws. Open it and pump is under the hatch. Two fuel lines and electrics. Be aware, as disconnecting fuel lines, some fuel Will come. Pump itself is screwed to the tank with 8 to 10 screws (can't remember). And of course there is seal. And pump might be quite hard to get out because seal is quite tight.

Best way is to replace whole setup. It is not that expensive. And of course replace seal as well. Pump housing is hard plastic what will get little brittle over the years. Also it might be little hard to get out from that service hatch, but possible to do. I did replace those ventilation valve seals too, and those can't replace when tank is in the car. So i removed whole tank and did full overhaul. I do have some photos and can try to find them later if needed.

- J

 

[tjr]

Thanks @jmi, much obliged. Yes, I can hear the pump every time I turn the key.

I purchased a pressure tester to no avail (for carburettor engines (range too low)). Now waiting for a 7 bar tester. With it, the service manual and this here video, I'll be checking pump and regulator function:



I tested for air leaks after MAF, none found (ocular and WD40).

The vacuum hose to the pressure regulator was a tad loose, so I zip tie'd it. Don't think it did anything, will replace with a new hose though.

Listening yesterday, when the engine dies, there is no hesitation, it just quits immediately, cold turkey. This makes me lean more towards electrics... So far it only occurs when the engine is cold.

Have a cumbersome MES set-up, but will be running diagnostics, to see if I can become a little wiser...

 

[tjr]

Fuel pump and fuel pressure regulator seem ok. According to the service manual the pressure should be:
3 bar fuel pump on - engine off
2,5 bar fuel pump on - engine on
5-7 bar fuel pump on - engine off - fuel supply to regulator off
The b-95 followed spec.

Connecting MES there were two immobiliser errors registered, I cleared those and ran MES on and off during the day. The engine started happily every time.

I replaced the o-ring between icv-valve and actuator. The valve move freely.

When deciding to do some lambda value logging at both idle and 2000rpm, the engine wouldn't start. I assumed it was an air bubble in the fuel line as I had just closed after the pressure tests. It took three tries before the engine caught.

Once it was running again I checked error codes and it showed: A8 - Electronic key, code not received (fluke/me in a hurry and/or cause?)...

Anyways here are the lambda logs I was hoping to get some feedback on:

How do the values look to you?
If lambda ok, what next, check resistance on sensors or home in on the immobiliser?
Thanks,
tjr

 

[jmi]

Have you checked that key-light in dashboard disappears? Or does is lit when engine dies?

It should turn off few seconds after you turn ignition on. Immobilizer detects key, validates it and tell ECU that it is good to go. If you have problems with it, light stays on and car won't start. I had starting issues because of this. But after cleaning immobilizer antenna cable connector, no any problems since.

If you have these problems. Check that connector and perhaps cable. Antenna is ring around ignition switch. And unit itself next to fuse box. You need to remove lover covers under dashboard driver side.

And bout engine values. For me eyes, those looks ok, but i am not expert on that side. So i could be wrong...

- J

 

[tjr]

Have you checked that key-light in dashboard disappears? Or does is lit when engine dies?

I'm religiously listening and waiting for the fuel pump to finish. In the meantime the key code light will go out.

t should turn off few seconds after you turn ignition on. Immobilizer detects key, validates it and tell ECU that it is good to go. If you have problems with it, light stays on and car won't start. I had starting issues because of this. But after cleaning immobilizer antenna cable connector, no any problems since.

Good to hear.

If you have these problems. Check that connector and perhaps cable. Antenna is ring around ignition switch. And unit itself next to fuse box. You need to remove lover covers under dashboard driver side.

Thanks, will add that to the list.

And bout engine values. For me eyes, those looks ok, but i am not expert on that side. So i could be wrong...

Thanks @jmi

When digging around the internets... there are the Narrowband (0-1V) and the Wideband (0-5V) oxygen sensors. Looking at the values above and in the Service Manual, Chapter 10, it would seem the b-95, being from 1995 has only one oxygen sensor, ant it is of the Narrowband type, located pre-cat.

In a Vauxhall(!) forum, I found this open loop/closed loop explanation:
1. Cold start & warm up phase = open loop - lambda ignored,
2. Idle & part throttle operation when at running temperature = closed loop - lambda is used,
3. Full throttle = open loop - lambda is ignored,
4. Acceleration = open loop - lambda is ignored.

https://www.migweb.co.uk/threads/closed-loop-and-open-loop.54099/

The Service Manual (chapter 10, page 7) seems to say pretty much the same thing in the last section, IF(!) I understand correctly:

I interpret "de - activated" as open loop. IF(!) correct, the b-95 oxygen sensor should be closed loop at warm idle (<70° and >25°C), and thus in contradiction to my pic 3 values.

Also it looks oily around injector 1(?), closest to Cam belt.

BUT I don't see how the above would cause cold start stalling/no start (eg the oxygen sensor should be open loop <25°C).

Oh well, another weekend to spend under the hood/car...

 

[wapper]

When monitoring oxygen sensor readings couple of years ago, I couldn't see any solid relation between engine temp and sensor being open or closed loop. I only did my measuring while being stationary, between 800 and 2000 rpm.
I think it's normal that sometimes correction is not needed (open loop) if sensor output is within allowed boundaries.

There's no way oxygen sensor, or air leaks could cause starting problems. It's still about petrol, or spark, or engine timing/control.

 

[Ilkka]

As wapper mentioned the lambda sensor should not cause starting issue. I'm 99% sure that the Hitachi system will not use the lambda during startup = when the engine and lambda sensors are cold. Or at least when starting the first time.
I could not confirm this from the service manual I sent you earlier with a quick glance but I'm almost certain it's there somewhere.

Does the fuel tank breathe properly = does the fuel cap hiss from under pressure when opening after some driving? The breather valve on top of the fuel tank is known to fail.

 

[tjr]

Thank you kindly for taking the time @wapper and @Ilkka, much appreciated!

There's no way oxygen sensor, or air leaks could cause starting problems. It's still about petrol, or spark, or engine timing/control.

I believe the fuel/fuel delivery can (at the moment) be ruled out, testing showed pressure is as it should, no fluctuation.
At 2000rpm (stationary, no load) I will hear a missed ignition/lack of combustion/stumble every now and then.
Currently:
Warm starts (and running) are problem-free.
The last three cold starts have been a bit different. First attempt, no go, no will to ignite, not in the least, nothing, the starter just turns (the engine). Second attempt, it starts right up, no hesitation, and thereafter it keeps on running evenly (remarkably). This made me suspect the immobiliser once again, but no error codes logged in the ECU.
At the moment it's an all or nothing issue.

When monitoring oxygen sensor readings couple of years ago, I couldn't see any solid relation between engine temp and sensor being open or closed loop. I only did my measuring while being stationary, between 800 and 2000 rpm.
I think it's normal that sometimes correction is not needed (open loop) if sensor output is within allowed boundaries.

I believe I'm seeing similar results (also stationary, also idle and 2000rpm), contradictory to the service manual it seems. I've had MES graph running upon start-up. With the key in MAR position the Oxygen sensor is put in open loop. Quickly (seconds) after start-up (engine for sure <25°C) the sensor enters closed loop(!) then back to open loop after maybe 10-30sec and thereafter stays open. As soon as I start increasing rpm (fairly slowly, ie throttle <70°) the sensor goes into closed loop, and stays closed. When returning to idle it stays closed for a while and then goes into open loop.

As wapper mentioned the lambda sensor should not cause starting issue. I'm 99% sure that the Hitachi system will not use the lambda during startup = when the engine and lambda sensors are cold. Or at least when starting the first time.
I could not confirm this from the service manual I sent you earlier with a quick glance but I'm almost certain it's there somewhere.

Yes, I need to read up on cold starts and what sensors the ECU uses in connection with them.

Does the fuel tank breathe properly = does the fuel cap hiss from under pressure when opening after some driving? The breather valve on top of the fuel tank is known to fail.

I should check. Would that cause a problem with a cold start?

OT, I was surprised how simple it was to disassemble the injection rail and that only two screws (one on each end) are able to hold all four injectors in place. On the flip side with eight gaskets (two per injector), it is quite easy to get them pinched. The lower gasket on No. four injector was incorrectly seated (without causing a fuel leak). I did not have the right gaskets so I ended up doing a (much needed) clean-up around the injectors and the metal to gasket surfaces and then re-using the old gaskets (for the time being). When putting it all together I managed to get the upper o-ring on No. two pinched resulting in a fine fuel mist when pressure testing.

Thanks again,
tjr

 

[wapper]

First attempt, no go, no will to ignite, not in the least, nothing, the starter just turns (the engine). Second attempt, it starts right up, no hesitation, and thereafter it keeps on running evenly (remarkably)

This is indeed strange. Perhaps your pump is not able to start the first time, but will start and function normally next times?
Also maybe I would pull out one spark plug and check if there's a spark when starting cold. Just to be sure.

 

[Ilkka]

Sometimes a failing ignition coil can start to give trouble once warmed up. I have not experienced this with fiat engines, but twice with a 2-stroke engine.
They give sufficient spark when cold or engine running, but not when trying to start a warm engine.

I don't think that this is your likely problem, but might as well try and change your coils and spark plugs since they are quite cheap.

 

[tjr]

Thanks gents, it's so valuable to bounce one's own (reg. incorrect) theories, and getting insightful feedback!

Today, the b-95 started on the first try, although the idle was supposedly (the daughter reporting) a tad wobbly to start out with. No stalling this time either (same as the previous three cold starts). Perhaps a factor; semi-odd (though not uncommon for the season) Southern Swedish weather over the weekend has gone from "winter to summer" (from frost in the mornings and low single digit temps to nearly 20°C).

Perhaps your pump is not able to start the first time, but will start and function normally next times?

It was the first suspect, but has shown no suspicious behaviour, the pump seems up to snuff (both according to sound (key OFF to MAR) and) pressure test wise... (3bar, 2,5bar, 7bar and holding pressure for 60sec (and beyond)).

Also maybe I would pull out one spark plug and check if there's a spark when starting cold. Just to be sure.

Sometimes a failing ignition coil can start to give trouble once warmed up. I have not experienced this with fiat engines, but twice with a 2-stroke engine.
They give sufficient spark when cold or engine running, but not when trying to start a warm engine.

I don't think that this is your likely problem, but might as well try and change your coils and spark plugs since they are quite cheap.

Putting it up on the list. This winter I did mess with both coils and spark plugs (in connection with the tappets, variator and oil duct D inspection and clean).

Heat and cold does make a difference..., isn't it more common the other way around though? When things heat up and materials expand, contact is lost.
(OT I had a Golf II with a known fuel pump (faulty solder) issue, so on French Autoroute, in the summer heat, the engine would regularly die, and we had to take pump cool down breaks (before giving in and driving to the nearest repair shop))

The b-95 symptoms have changed to the better without me doing anything significant (it started improving before the weather improved though).

It's still about petrol, or spark, or engine timing/control.

Agreed, my current prime suspect of the three is engine timing/control and close second is spark. I wonder if the ECU is not getting half-garbled (maybe semi intermittent) sensor feedback (and/or odd results upon adaption):
- open loop Oxygen sensor, idle is smooth (the best I've known it (but coming from faulty timing))
- closed loop Oxygen sensor, (the ECU is supposed to alter injection lengths and timing, idle becomes a little less even
- target warm engine idle 850rpm is reached only once (in the above) otherwise idle is consistently higher

It bugs me that I have found no one culprit...

Depending on the evolution of the issue, I'll hold off until coming "MOT". In any case I'll report back on the developments.

Thanks,
tjr

 

[tjr]

Had another cold start, where the b-95 started on the first attempt

Call it an anomaly and not problem solved, but still on a positive note…

Summary as of now (info - not advice); no single effort has solved the issue, but what I’ve done so far is:
1) checked fuel pump pressure and function of fuel pressure regulator (FPR) according to service manual
2) checked for vacuum leaks, replaced loose-ish hose between FPR and throttle housing
3) checked for air leaks in rubber hose/baffle connecting MAF with throttle housing
4) cleaned throttle housing on “the MAF side” of the butterfly valve - slight gunk
5) checked ICV for movement
6) replaced flat-ish o-ring between ICV-actuator/solenoid(?) and valve
7) checked and cleaned seating around fuel injectors (upper and lower)
8) cleaned and dielectric greased fuel injector electrical connections - slight oxidation
9) dielectric greased oxygen sensor electrical connection, and an additional currently unknown (sensor?) electrical connection (zip-tied together with the O2-sensor one).

I will/hope to get to the other two rear engine sensor connections when addressing the leaky gearbox/drive-shaft seal and completely loose intake support (w two out of three bolts missing)… Double-

 

[StevePD]

Hi, any update on your starting problem. Has it continued to behave itself? Regards SteveD

 

[tjr]

Hi @StevePD,
Semi-behaving, the lump will not always start on first try though if not it'll at least start on second (currently...).
Time's up for MOT..., so will have to (fearfully) face the music whilst draggin my feet!
Best,
tjr

 

[tjr]

Update:
4 mo. later, all is pretty much the same, start on 2nd if not 1st, and a tad lumpy when warm.

MOT (SE version) came and (the b-95) passed (2nd attempt (not engine related tho (another thread))):
CO idle 0,0 (0,5)
CO 2500 rpm 0,0 (0,3)
HC 10 (200)
Lambda 1,03 (0,97-1,03)

The (barchetta happy) technician commented on the high Lambda value (values above 1,0 indicates a lean fuel/air ratio ie too much oxygen), and suspected a small exhaust leak on the last muffler for giving a possibly false lean value.

Will be fixing the leak, changing coils and perhaps the plugs this fall/winter.

 

[tjr]

Further:
Taking a fresh look at the b-95 O2 logs in #4 and with the information from https://www.innova.com/blogs/fix-ad...y functioning oxygen sensor,(up to 0.9 volts) concerning Narrowband 0-1V Oxygen sensors:

"02 Sensor

A properly functioning oxygen sensor will show a rapidly fluctuating output voltage between approximately 0.1 and 1.0 volts. When the air-fuel mixture is perfect, the O2 voltage value will be exactly 0.45V. As the mixture becomes rich, the voltage will increase (up to 0.9 volts). If the fuel mixture is burning lean, there will be more oxygen in the exhaust and the voltage will drop (as low as 0.1 volts). If the oxygen sensor detects a drop in the mixture, it may trigger a check engine light.

The CPU uses the O2 sensor's output voltage to adjust the air/fuel mixture in the engine. Also, you may examine the state of the catalytic converter using live data from the O2 sensors.

• Rich Air/Fuel Mixture: The signal is high (0.7 to 0.9 volts) when the air/fuel mixture is rich (little O2 in the exhaust).
• Lean Air/Fuel Mixture: The signal is low (0.1 to 0.3 volts) when the air/fuel mixture is lean (Too much 02 in the exhaust)."

The O2 sensor indicates a (true) lean air/fuel mix when in open loop 0.13-0.27V. In closed loop with the ECU adapting the air/fuel mixture, the O2 sensor values show 0.45V and above, thereby the mixture becomes perfect to rich (better for the engine)...

A new fuel tank leak reared its ugly face as of late (when full, sitting in the sun). Will check the breather valves as mentioned by @Ilkka in #8.

I have troubles understanding how bad coils and/or spark plugs would affect the lambda values without affecting neither the MOT CO nor HC values. Does it have to do with how these tests are performed? Anyone willing to aid me in clearing my brain fog?

Best,
tjr

 

[Ilkka]

I wish I could give you a straightforward answer on where the problem is, but here's some thoughts.
-I'm 99% sure the 1995 barchetta has only one lambda sensor and it's located in the exhaust manifold.
-A leak in the muffler may cause the lambda to be on the lean side, but always assume the measured value is correct until a proven cause is found.
-if the mixture is lean, then the cause would be either too little fuel, or too much air, of one of the values (MAF value, lambda value, fuel pressure value) that the ECU is using to keep the mixture in balance is off due to dirty sensor, corroded connector or similar.
-since you have measured the fuel pressure the fuel pump seems ok. One of the injectors of their seals might be bad?
-if the MAF works, only an air leak between the MAF and cylinder could introduce more air and make the mixture lean. So check again for cracks, gaskets etc. in rubber hoses and intake manifold or injector seals.
-also check for evidence of exhaust manifold / gasket leaks that might affect the lambda sensor value that in term would cause the ECU to adjust the mixture
-high CO would mean rich mixture and HC unburned fuel spark issues when running. Weak spark during cranking would probably not cause HC.

I would try a new lambda sensor if there's no leaks anywhere between MAF and the lambda sensor. The starting problems and reduced performance sound like lambda/air leak issues and lambdas can start to give trouble without failing completely if they are dirty or clogged.

 

[tjr]

Thanks as ever @Ilkka!

-I'm 99% sure the 1995 barchetta has only one lambda sensor and it's located in the exhaust manifold.

Correct (at least on this engine/vehicle).

-A leak in the muffler may cause the lambda to be on the lean side, but always assume the measured value is correct until a proven cause is found.

Yes, I do think the values are correct. Both the logs above and the MOT equipment indicate lean mix.
It is interesting that when the ECU includes the values from the lambda/O2-sensor (see logs above, closed loop) and adapts to them, the air/fuel mix is corrected.

-if the mixture is lean, then the cause would be either too little fuel, or too much air, of one of the values (MAF value, lambda value, fuel pressure value) that the ECU is using to keep the mixture in balance is off due to dirty sensor, corroded connector or similar.

-since you have measured the fuel pressure the fuel pump seems ok. One of the injectors of their seals might be bad?

One of the injectors two seals was pinched, but I believe I got it sorted. Not sure where to get replacements though...

-if the MAF works, only an air leak between the MAF and cylinder could introduce more air and make the mixture lean. So check again for cracks, gaskets etc. in rubber hoses and intake manifold or injector seals.

Thanks, will do.

-also check for evidence of exhaust manifold / gasket leaks that might affect the lambda sensor value that in term would cause the ECU to adjust the mixture

Thanks, will do.

-high CO would mean rich mixture and HC unburned fuel spark issues when running. Weak spark during cranking would probably not cause HC.

CO and HC were acceptable at MOT with a warm engine.

I would try a new lambda sensor if there's no leaks anywhere between MAF and the lambda sensor. The starting problems and reduced performance sound like lambda/air leak issues and lambdas can start to give trouble without failing completely if they are dirty or clogged.

Thanks again, all the best,
tjr