[moved] ecu getting hot thermal paste

Hi All,

I hope someone can help with below and I am posting in the right section.

My car has a bosch ecu(engine management unit) and once the car has been driven for 20-30 mins and then switched off, the engine will not start again .
My finding from numerous hours of investigation is the ecu gets hot and does not send signal to any of the other ecus needed to switch the car on such as transmission and fuel pump as confirmed by reading the errors. both repoted missing can signal from ecu .

Also when this happens , I can't access the ecu to check error etc .

as a result I get lots of error related to transmission and fuel pump fault which is a false alarm .

Now this happens only when the car is switched off otherwise the car can drive and run for hours and hours if not switched off.


so me thinking it is the ecu , i took it off and opened it . all seems good but it has on it what i refer to as thermal paste so think that is old and need replacing so that it transfers the heat .

Am I correct assuming this ? can i use cpu thermal paste on this circuit board?

can I check anything on the circuit board to verify if the board is gone ?
visually i am not seeing any cuts on the board or anything melting/corrosion.

is that ok ?

thanks

Re: ecu getting hot thermal paste

Can you determine the driver output states to see if one is stuck On?
Either it is an internal failure from voltage spikes and latchup internally of a gate, or stuck pulsing driver externally.

But yes you can use CPU paste if you think there is a large differential temp across the paste.

Check driver ports for Injectors or AIC valve or EVAP solenoid or relays.

Re: ecu getting hot thermal paste

sure, by driver outputs what are you referring to ?

I am no expert with chips and capacitors but can they fail intermittently ?

not sure what stuck pulsing driver externally means ?

what I can see is:

Tranmission reporting : lost signal from ecu
fuel pump : lost CAN and CAN timeout
DSC: lost CAN signal to ECU

ecu is non responsive and can not read the codes . the software interface says unable to communicate .
30 minutes later all good.

is it a paste ? it is pasted on the circuit board at bottom of the chips and then stuck to the metal plate . I am just worried it can conduct electricity and destroy the board.

I have purchased a thermal paste that is non conductive according to manufacture but have to take their word.

I have checked all wires and all look ok and have used contact cleaner on them .

the fact that the ecu shuts down completely for 30 minutes and works with no intervention suggests some sort of heating issue rather than cable or anything else.

I have changed all relays possible and related .

in fact i have changed the ecu relay 3 times and checked all fuses and they have a clear connection .

Not saying you're on the wrong track, but did you look up some of the online auto repair forums? You might find car owners reporting this problem, the cause and the fix.

There may even be a recall for this problem, with a dealer fix at no charge to you.

I have spent two month on this with numerous bmw forums and no one knows what it is .

it is one of those bugger faults.

BradtheRad said:

There may even be a recall for this problem, with a dealer fix at no charge to you.

Click to expand...

Now that the OP has opened the ECU, I'm sure it's no longer a no charge fix even if there was a recall. That is something that should be verified before performing surgery.

Given that the unit has overheated numerous times (unintentional life cycle testing) I would probably consider the controller in that unit has been stressed to the point that it will eventually fail sometime in the future. I would probably consider replacing the unit instead of trying to make it live for what will be an indefinite time.

I'd really hate to be in your shoes driving in the desert and you pull over for a bladder break shut off the car, and can't start the car again as the ECU finally permanently failed.

- - - Updated - - -

When the ECU dies in your BMW with the Broken Micro Within ;-) just hold up a sign with edaboard written on it any I'll give you a lift to the nearest gas station

lol ,


so my questions to experts were, is it possible there is a a micro chip failiure that bare eye can not see.

is intermittent fault possible in circuit boards? what i am confused on is , with computer is it either on or off , 1 or 0 so where does the intermittent comes in ?

when you contact ecu repair people, they say they will put the item on bench and use simulator to test it by tetsing input and output, would that really show the issue if it is intermittent ?

is thermal paste safe to apply then ?

From your description of the problem, if it's overheating like you say, I kind of doubt it's an intermittent fault, depending on the device it might be going into thermal shutdown, and won't "boot" until the junction temperature drops enough to let the micro start up again. Severe thermal cycling like you've been describing will also damage solder joints (especially those RoHS solder joints), besides reducing the life of the components. I really don't think fixing the thermal issue will make the ECU reliable. A really hot day, which the heat sinking can't compensate for could cause the problem to occur again. I doubt the micro in the ECU is meeting it's own specs anymore.

A heat problem is the first thing that comes to mind, since cooling down makes it go away.

The fact that it is intermittent makes it hard to localize. As you suggest, there could be one little wire, that expands from heating, breaks contact, cools down, restores contact, etc.

Tactics to improve heat sinking can help, as you are doing.

There are chill pads you might try, or attachable fans, or peltier units, or heat pumps. The type of thing computer enthusiasts install to cool microprocessor units.

Post #3 had some new info "the software interface says unable to communicate .30 minutes later all good."

Without knowing details, I ask, would that be the time to cool down with some type of software timeout or thermal cool down period?

To test this theory, I might try to change the cool down period with cold spray or a heater to rule out temp.

If software, then some fault condition must exist, detected by slow discharge of some offline voltage.

To test this theory, I might probe for ECU power on the card when switched off during a failure.


Side Note
======
I once had an Audi who's Instrument Panel started to fail shortly after a final cash sale. It would work cold and fail after 30 minutes or so leaving no TACH and other feedback signals off but no ECU functions affected. Garage said it was either the Main Kontroller of the the Instrument Panel board and since there was no COMM on CAN to Instrument panel, there was no diagnosis avail. the fault could be at either end., each well over $1k. It eventually got worse so that it failed at room temp and then permanently and had no obvious loose connections. Fortunately I gave the car away to a son-in law who didn't mind until the car was totalled by another driver.

Sometimes flakey wiring or SMD solder joints can mimic bi-metallic switches with whatever thermal time constants exist and hysteresis. Sometimes Hall Sensors can fail hot.

I kind of think the intermittent idea is incorrect mostly because the ECU works as long as the engine is allowed to continue to run, shutting it off and restarting it won't allow it to cool enough to affect a temperature expansion related intermittent connection.

Besides when it's cold it also starts. It's hard to imagine a connection that works while hot, until it's turned off and then stops working until it's cold. But any thermal monitoring would definitely stop the ECU from starting back up. From what the OP mentioned none of the signals (CAN) that normally come out work when it's dead.

Come to think of it, maybe the problem is with a DC-DC converter, that won't start when the temp is high. Maybe the OP can check voltages in the unit when this problem occurs during a hot starting condition.

My knowlege of circuit board is very very limited but have some knowledge on electricity etc hence not sure how to check voltage at ecu as i also dont have a map of the pins or wiring diagram

With all this in mind , i am going to buy a cold spray so can check my theory and if it is ecu then happy to change it

Would cooling the board down very quickly using spray cause any burst or whatever else?

What kind of spray do you recommend ? Maybe air spray as when you hold it for a while it gets really cold

Freeze spray can cause damage due to the sudden and severe temperature change, so be careful and don't just douse the thing like you're spraying Raid on a cockroach. ;-)

Just make sure if you use freeze spray do only one component at a time for testing. You don't want to overdo it and spray the whole board, otherwise you'll never isolate the problem.

Just a clarification , i am looking to ensure it is the ecu and as soon as it is i will swap it so dont need to narrow down which chip on the ecu but rather knowing it is ecu 100%

So can i spray it on the metal casing while in engine compartment for quick test?

It might work if there is good enough thermal conductivity between the casing and the components inside the ECU. But given your desire to add thermal grease, because it doesn't appear to be making good contact...

ads-ee said:

I kind of think the intermittent idea is incorrect mostly because the ECU works as long as the engine is allowed to continue to run, shutting it off and restarting it won't allow it to cool enough to affect a temperature expansion related intermittent connection.

Click to expand...

I accept your explanation as the most plausible.

When I spoke of it being intermittent, it was in contrast to 100 percent failure.

I imagine a 'jostling wire' would cause the engine to stop unpredictably.

However this problem is more or less predictable. It's not 100 percent failure (which could be diagnosed, followed by 'Remove and Replace'). This is one of those 'on again off again' problems that will drive anybody crazy. Unpredictable. Hard to troubleshoot.

so are we saying it is not possible it can be the ecu ?

what I am trying to find out is, can ecu have intermittent fault that works all the way indefinitely but as soon as the car is off, it stops working ?

stranger12 said:

so are we saying it is not possible it can be the ecu ?

what I am trying to find out is, can ecu have intermittent fault that works all the way indefinitely but as soon as the car is off, it stops working ?

Click to expand...

No, there is just some agreement on the idea that it's likely not some intermitant mechanical fault in the circuit, the jostled wire, bad solder joint, etc.

I still suspect the micro not booting when hot due to some thermal sensor, or the dc-dc converter not powering up when hot.

thanks , is dc to dc convertor on the ecu ?

stranger12 said:

thanks , is dc to dc convertor on the ecu ?

Click to expand...

Would have to be, unless a CAN bus supplies power (don't know never looked at this bus).

12V battery ===> 3.3V ECU ====> magic_smoke

12V batt ===> LDO ====> 3.3v ECU ====> wasted_power

12V batt ===> DC-DC ===> 3.3V ECU ====> happy_system.

LDOs are way to inefficient to go from 12V to 3.3V, so a point of load dc-dc converter(s) probably powers the ECU.