Diagnostics - your opinions please
Discussion
Having suffered an intermittent misfire ever since I've owned the car (it plagued the previous owner too and mystified several reknowned indepndents - understandable given its intermittent nature) I think I've got to the bottom of the problem (if not the solution!).
Basically, regardless of temperature, time, driving conditions, throttle angle, revs, day of the week etc etc the car would suddenly lose power and run 'rough'. It could happen every 5 miles or every 500.
I've checked every wire and cleaned every connector (inc. fuse box and behind dash), replaced all the usual suspects (LT components, HT components, relays, fuses, fuel pump, fuel filter, temp sender), run through all the ECU continuity tests, checked and adjusted the afm as per RPi, the idle, the timing, cleaned the stepper, cleaned the fuel tank vent etc etc. I've even monitored the afm and throttle pot outputs in real time and observed their being OK (even during misfire).
According to the 'GreenV8S test' (ie watch the rev counter during misfire) the LT system is OK. The fuel pump always primes, and I've got a voltmeter across the fuel pump so I know it's getting power.
The way the misfire kicks in, and it's severity meant it couldn't be a dirty injector (unless four injectors all gummed up and ungummed at exactly the same time).
Misfire would last 20-40 seconds and would normally clear on stopping and restarting. It would never occur in the hands of a garage!
Here's the important bit:
On rare occasions I managed to pull up, not stall, and get under the bonnet, I discovered that pulling HT leads or injector connetcors on the n/s bank would make no difference. Pulling the same on the o/s would make it worse. = problem on n/s bank.
Wiring 12v LEDS across each injector bank at the ECU, and watching them during misfire, I expected to see one bank going down - but they both still operated
...then I noticed the n/s bank appeared to be firing in longer pulses than the o/s bank. The n/s bank was getting maximum fuelling at all revs/load, and hence flooding.
This observation was borne out by revving the engine under load through to 4.5 k rpm where it ran cleanly again up to 6k - because the fuelling now (more or less) matched the airflow into the engine. Also a smell of fuel at misfire.
Now I reckon, because the firing of the injectors on EACH BANK is different, the problem must lie with the ECU and not elsewhere. If it were a sensor/signal problem BOTH banks would be affected. If it were a problem in the injector loom or with one or more injectors I'd see either a short or a break (LED on, off or flashing normally - but not MAX pulse duration). Ergo the problem must be with the ECU, and moreover with the hardware unit itself rather than the EPROM.
Is this logical? Do you concur or have I missed something?
Opinions, suggestions and comments very welcome.
Oh and anyone got a spare 4.3 litre ECU? (Or any other kind - eg a 500 you'll sell me (you know who you are
)
Basically, regardless of temperature, time, driving conditions, throttle angle, revs, day of the week etc etc the car would suddenly lose power and run 'rough'. It could happen every 5 miles or every 500.
I've checked every wire and cleaned every connector (inc. fuse box and behind dash), replaced all the usual suspects (LT components, HT components, relays, fuses, fuel pump, fuel filter, temp sender), run through all the ECU continuity tests, checked and adjusted the afm as per RPi, the idle, the timing, cleaned the stepper, cleaned the fuel tank vent etc etc. I've even monitored the afm and throttle pot outputs in real time and observed their being OK (even during misfire).
According to the 'GreenV8S test' (ie watch the rev counter during misfire) the LT system is OK. The fuel pump always primes, and I've got a voltmeter across the fuel pump so I know it's getting power.
The way the misfire kicks in, and it's severity meant it couldn't be a dirty injector (unless four injectors all gummed up and ungummed at exactly the same time).
Misfire would last 20-40 seconds and would normally clear on stopping and restarting. It would never occur in the hands of a garage!
Here's the important bit:
On rare occasions I managed to pull up, not stall, and get under the bonnet, I discovered that pulling HT leads or injector connetcors on the n/s bank would make no difference. Pulling the same on the o/s would make it worse. = problem on n/s bank.
Wiring 12v LEDS across each injector bank at the ECU, and watching them during misfire, I expected to see one bank going down - but they both still operated
...then I noticed the n/s bank appeared to be firing in longer pulses than the o/s bank. The n/s bank was getting maximum fuelling at all revs/load, and hence flooding. This observation was borne out by revving the engine under load through to 4.5 k rpm where it ran cleanly again up to 6k - because the fuelling now (more or less) matched the airflow into the engine. Also a smell of fuel at misfire.
Now I reckon, because the firing of the injectors on EACH BANK is different, the problem must lie with the ECU and not elsewhere. If it were a sensor/signal problem BOTH banks would be affected. If it were a problem in the injector loom or with one or more injectors I'd see either a short or a break (LED on, off or flashing normally - but not MAX pulse duration). Ergo the problem must be with the ECU, and moreover with the hardware unit itself rather than the EPROM.
Is this logical? Do you concur or have I missed something?
Opinions, suggestions and comments very welcome.
Oh and anyone got a spare 4.3 litre ECU? (Or any other kind - eg a 500 you'll sell me (you know who you are
)One question I'd ask is;
Do the injectors fire in pulses still (much longer but still include an off cycle), or just fully open? If fully open, could it be a short somewhere for that bank? I would have thought it very strange for just one bank to go amiss, unless there is a common injector driver circuit for each bank.
Early morning thougts, ignore me if I'm talking tosh!
Do the injectors fire in pulses still (much longer but still include an off cycle), or just fully open? If fully open, could it be a short somewhere for that bank? I would have thought it very strange for just one bank to go amiss, unless there is a common injector driver circuit for each bank.
Early morning thougts, ignore me if I'm talking tosh!
The two banks have separate lambda feedback, maybe the ECU is getting garbage from one lambda sensor? It's even remotely possible that the two sensors have been crossed over (which would make one side go very lean, and the other side very rich). But there's a limit to the amount of correction you can get from lambda feedback, and I don't think it's enough to flood the engine, so this doesn't seem very likely to me.
I think you're right and the symptoms point to an ECU hardware fault. If you haven't checked the ECU earth (normally the n/s/r fuel rail mounting bolt) then do that now - I wouldn't expect an earthing problem to produce such specific symptoms but you never know. Otherwise I think it's time to try a new ECU.
I think you're right and the symptoms point to an ECU hardware fault. If you haven't checked the ECU earth (normally the n/s/r fuel rail mounting bolt) then do that now - I wouldn't expect an earthing problem to produce such specific symptoms but you never know. Otherwise I think it's time to try a new ECU.
virgil: yes the injectors both still fire in pulses - but n/s bank pulses 'longer' (LED brighter due to longer 'on' time - but pulses still distinct) than the other - the other behaving 'normally' (ie as both do when not misfiring). This is why I am steering away from a short/break in the injector loom and towards the ecu still pulsing, but for some reason - ON ONE BANK ONLY - at max fuelling. Therefore I'm assuming a fault in either the n/s injector bank output transistor, or the signal feeding it. Something specific to that bank, and after the fuel map (cos the other bank's injector pulse is OK) anyway.
GreenV8S: no lambdas - it's a precat! Glad you concur though! Have already done the ECU earth, thanks.
GreenV8S: no lambdas - it's a precat! Glad you concur though! Have already done the ECU earth, thanks.
Edited by Griffter on Monday 26th June 10:08
shpub said:
No but the ECU might think it has them, can't find them and then goes into a default cycling mode to help.
Worth checking the tune resistor.
Worth checking the tune resistor.
Interesting...so on a 'cat' car can the ECU provide different fuelling to each bank in 'cat' mode then? I assumed the same fuelling was always provided to each bank, but the seperate lambdas helped tell you which bank there was a problem on.
Tune resistor checks out 'static' but I guess I need to monitor it while running to see if there's an intermittent problem with it.
How have you wired the LEDs on the ECU ?. Guessing here but I imagine the supply goes to each of the injector coils and is then switched to ground by a transistor. As you are looking at the brightness of the LED to guess at the pulse width, is it possible that it is not the pulse length that is changing but the peak voltage to the LED. Presumably you have checked the supply to the bank of injectors is stable but what if one coil is going low impedance and dragging down the supply rail but possibly providing more drive to the LED. Obviously not practical but it would be nice to look at the pulses on a scope to prove the point, or here is an idea if you replace the LEDs (or wire alongside) a small loudspeaker or earphone you may be able to detect if the pulses are louder or longer by the sound.
If it is a common transistor in the ECU for each bank I would guess it to be a FET which being high impedance on the gate, depending on how it is driven could have a resistor pulling the gate back to ground when it is not on. If this is intermittent the FET may stay turned on until the internal capacitance is discharged. So a scout around for dry joints may be in order.
When looking for an intermittent fault on the V8s (no cats) I borrowed a spare ECU from the local Landy people (V8 Discovery I think). The map may be slightly wrong but it proved the point
If it is a common transistor in the ECU for each bank I would guess it to be a FET which being high impedance on the gate, depending on how it is driven could have a resistor pulling the gate back to ground when it is not on. If this is intermittent the FET may stay turned on until the internal capacitance is discharged. So a scout around for dry joints may be in order.
When looking for an intermittent fault on the V8s (no cats) I borrowed a spare ECU from the local Landy people (V8 Discovery I think). The map may be slightly wrong but it proved the point
steve-v8s, thanks.
The LEDs are 12v jobs in parallel with each injector bank. The 12v supply is common to both banks, and the earths are switched (per bank). So although I like your theory, I don't think there's any more than 12v across the injectors. Although I'm judging the difference in pulse length by the injector brightness, you can see the LEDs flickering with the pulses - at idle at least. I'm glad you suggested it though, as now I'm more certain the problem is pulse width, on one bank only = ecu hardware problem.
I don't fully understand the FET/resistor bit, but it sounds feasible!
However before buying a new ecu I'll try running without the tune resistor to test shpub's idea about the ecu looking for and correcting for non-existent lamba sensors too.
The LEDs are 12v jobs in parallel with each injector bank. The 12v supply is common to both banks, and the earths are switched (per bank). So although I like your theory, I don't think there's any more than 12v across the injectors. Although I'm judging the difference in pulse length by the injector brightness, you can see the LEDs flickering with the pulses - at idle at least. I'm glad you suggested it though, as now I'm more certain the problem is pulse width, on one bank only = ecu hardware problem.
I don't fully understand the FET/resistor bit, but it sounds feasible!
However before buying a new ecu I'll try running without the tune resistor to test shpub's idea about the ecu looking for and correcting for non-existent lamba sensors too.
First of all, YHM.
LED brightness is not a particularly good indication of pulsewidth, unless the LEDs are well matched. (unlikely to happen) You could test the theory by swapping them over and seeing if the same trend happens, or else use a DVM that will measure frequency and duty - or even use an oscilloscope. (Maplin have a cheapy that would work)
Have you taken the case off the ECU yet? The problem could be absolutely obvious. I had a blown track on an ECU once that caused havoc - took seconds to find & repair once I'd decided to take the case off. Just observe anti-static precautions!!
LED brightness is not a particularly good indication of pulsewidth, unless the LEDs are well matched. (unlikely to happen) You could test the theory by swapping them over and seeing if the same trend happens, or else use a DVM that will measure frequency and duty - or even use an oscilloscope. (Maplin have a cheapy that would work)
Have you taken the case off the ECU yet? The problem could be absolutely obvious. I had a blown track on an ECU once that caused havoc - took seconds to find & repair once I'd decided to take the case off. Just observe anti-static precautions!!
I find that at low rpm you can get a pretty good judge of the duty cycle from LED brightness. As a sanity check you can move the LED connections, and the actual injector connections, from side to side to see whether the symptoms stay with the physical injector or with the wiring loom, or with the LED.
My observations are based on two things:
1) the fact that when operating 'normally' both LEDS indicate substantially the same brightness and apparent 'pulse width' - ie they look substantially identical.
2) the fact that when misfiring, the n/s bank LED is noticably brighter and (although difficult to discern with the human eye) pulse width appears to be longer, particularly at low rpm, than the o/s bank.
To put it another way, if I only had the LED on the n/s bank, I'm not sure I'd have noticed anything - but having the one on the o/s bank (which behaves the same under normal running OR misfire) to compare it with, it's obvious that under misfire the n/s bank is behaving differently.
It's the relative difference between the two LEDS during misfire as compared to normal running which is leading to my hypothesis. I realise I would get better data with a 'scope etc, but my facilities (and funds) are limited. I do appreciate the comments and suggestions though - thanks to all.
As an aside I think I'm going to wire them in permanently as my passengers love them. 'Knight Rider' etc.
1) the fact that when operating 'normally' both LEDS indicate substantially the same brightness and apparent 'pulse width' - ie they look substantially identical.
2) the fact that when misfiring, the n/s bank LED is noticably brighter and (although difficult to discern with the human eye) pulse width appears to be longer, particularly at low rpm, than the o/s bank.
To put it another way, if I only had the LED on the n/s bank, I'm not sure I'd have noticed anything - but having the one on the o/s bank (which behaves the same under normal running OR misfire) to compare it with, it's obvious that under misfire the n/s bank is behaving differently.
It's the relative difference between the two LEDS during misfire as compared to normal running which is leading to my hypothesis. I realise I would get better data with a 'scope etc, but my facilities (and funds) are limited. I do appreciate the comments and suggestions though - thanks to all.
As an aside I think I'm going to wire them in permanently as my passengers love them. 'Knight Rider' etc.
Gassing Station | Griffith | Top of Page | What's New | My Stuff