Calling Crasher-I'm desperate now!

[bobfrankish]

I really need to get my car fixed, I have been messing around with this fault for over 2 years and I have to sell the car. I have tried all the common things to no avail. Reminder of fault; car starts first time every time from cold, runs for some 15 to 30 mins then cuts out and wont start again until it has cooled down for at least 30 mins. It never had this fault until I changed the old (BERU) coil which had definitely failed, fit it to the car and it won't start at all. I have just fitted a new Hella coil instead of the Delphi one, fault is exactly the same. ECU relay is a genuine VW German part and has been renewed, and am ruling this out because it also supplies power to the MAF sensor, and I have connected my meter to the supply side of the MAF and it stays solid at 14V right up to and after the engine dies. Crank sensor has been changed and fault is exactly the same, so can't see 2 sensors failing in same way. Tools I have at my disposal are; top quality multimeter, top quality oscilloscope, laptop with VCDS lite installed. Just tried another key which is original VW one, exactly the same. I think perfect steady voltage at MAF rules out ignition switch. If it was your car, what would you try next? Look at ECU outputs possibly?

[mickeybo]

Two of the very most common problems on this is a faulty starter when warm Or a crank sensor goes open circuit when but starts when it cools down . can you post up your engine code or reg no

[bobfrankish]

Hi, thanks for reply. Engine is ADY, reg. N143 GCA. Crasher thinks car is a '95, could be right was first registered Feb '96. Can't imagine it's a starter, no problem with car cranking hot or cold. It is on its' second crank sensor and fault is exactly the same. Thing is, you can duplicate this fault as many times as you want. I must have let it idle for 30 mins or so until it cuts out at least 50 times now over the last 2 years, and it always starts first time again once it has cooled down. I am 70 years old now and have changed many impulse sensors, I have never known 2 sensors be constantly intermittent like this, they normally falter then fail totally. I will try it again today and meter the sensor's resistance as soon as the engine cuts out just to double check. Also, VCDS never registers sensor DTC.

[Crasher]

It is a 96 model built on the 8-07-95 and registered 13-02-96. The previous version with the 2E engine would register a faulty crank speed immediately the engine was not running, after this and still to this day it is very unusual fo a G28 to come up as being faulty even when it is. The way I would do this is set the engine at fast idle and graph as many signals as possible such as rpm, ECU voltage and wait for it to cut and view which drops out first; a faster idle, say 2000rpm, gives a second or so longer of run down to help view signal drop out. Which relays have you replaced and with which make, part number and case colour?

[bobfrankish]

Hello again Crasher! I must get this sorted before I die! Ok, relays 30 and 167 have been replaced, both were from VW, both have sort of dirty/off-white cases and say made in Germany, part nos. are 165 906 381 and 191 906 383C. They were replaced several years ago, (but car has only covered 30K miles in 14 years), to try and cure an intermittent no start fault which was crank sensor wiring insulation had crumbled away and was shorting together. I may have mentioned to you before that my whole life's job was a component level electronics engineer and designer working in the audio and lighting industry. I have had the cases off the relays and the contacts are just like new and all the solder joints in the ECU relay are perfect. I am not completely up to speed with VCDS lite, but I will connect to the car now and see what I can deduce (if anything!) Thanks again mate.

[Crasher]

I must get this sorted before I die!

I have got my bug and Golf 1, a customer coupe quattro, a customer Golf 1 1.8T and a VR6 turbo Golf 2 ohh and our old Golf 2 16v 2.2 to finish before I can lie down and die and at nearly 60, time is running out!

if the G28 is new and of a wood brand and you are sure this is spark not fuel then watching and logging some data should pin point what is dropping out… the ignition switch used to be a common failure. If VCDS drops out of coms as it cuts, that is a clue and if it keeps running that also helps pinpoint it.

[bobfrankish]

OK. I only have shareware version of VCDS lite which appears to only have 8 measuring blocks which is not a massive amount of use I don't think. Car cut out very quickly this time though, only ran for about 10 mins. and coolant temp. got up to about 58 deg C. It is now low on fuel, so I will dump some more fuel in tomorrow and take another look. Also it has just started bleeding snowing again here!

[bobfrankish]

Wow! That's a lot of cars. The project I need to get finished is my 1998 BMW 840Ci which I have been working on for 5 years on and off. Ill health has meant I haven't done anything with it for over 2 years. I also have a 1989 BMW 535I that I need to finish. I restored the Golf about 6 years ago. First G28 was from GSF, cannot remember the make, but never had any trouble with it. Replacement is a Delphi, which I would not recommend anyone buy as body is 0.4mm smaller than previous sensor so leaks oil badly. Had to fit chunkier o-ring to get it to seal, works electrically though. Fault was identical with previous sensor. As you know, they are a bit of a pig for an old man like me to fit, so don't really fancy buying yet another to try. I will scope it tomorrow, weather permitting, make sure it is outputting when engine is turning.IMG-20191001-WA0005.jpg

[bobfrankish]

Today's update: car only ran for maybe a minute before it cut out. Assuming it is the same thing making it cut out as stopping it from firing, I took a plug out and checked for a spark whilst cranking - no spark. Checked the crank sensor with a meter (resistance check) and it is completely open circuit even on 2 megohm scale. Question , is G28 sensor magnetic type or hall effect type? Ok, answered my own question, got to be Hall effect, just measured it again on the meter, totally open circuit any pins any direction, plugged it back in, car started ran for about 10 mins and has cut out again. Fed up now. Just given up again now. Have had VCDS connected when car cut out, and it did not lose communication. No fault codes were stored. So, tomorrow I will connect my scope up to the ECU input drive to the coil and monitor what happens there, then maybe scope G28 and see how that behaves. I will also connect LED's up to all the relevant power feeds so that I can see whether any supplies disappear. How easy would it be to disable the immobiliser just to rule that out?

[Crasher]

The factory immobiliser is built into the ECU and cannot cut the engine whilst it is running, it may however have an aftermarket system and if so, the variables are enormous.