I've been having very intermittent ECM CAN issues where the ECM disappears from the CAN bus for about 3 seconds, only when it's hot outside, typically after the car has been on a while, then turned off for some period of time (e.g. 20 minutes), then turned back on at which point it happens within 10 minutes. At my house, the limited performance mode (ABS, TRAC off, etc) happens repeatably at about a mile down the residential streets, before I hit the highway under the right conditions (i.e. over 95 F (35 C), top down, AC off or ~1 hour with the car off in the heat, restart after driving a while). Typical codes are U2500 (ABS CAN error), P1797 (TCM complaining about the ECM CAN). Occasionally the ECM ends up complaining that it can't talk to the TCM (P1797 again), but I have not seen that happen on the CAN transmissions. There are occasionally other random codes, all about the ECM disappearing from the CAN stream. I put this down to an ECM malfunction/reset.

I've had a CAN analyzer on the car twice when this happens and the same thing happens, both the ECM CAN token and all the other ECM IDs disappear from the CAN stream for 3 seconds.

So, hypothesis, it's the ECM getting too hot and either resetting or partially resetting from the backup CPU. I do have some vague evidence for this. Once, in early Aug, it happened repeatedly. Ambient was ~38-40 C, I had been trying to get the car hot to test the hypothesis, no internal AC, top down, and it happened four times within 15 minutes on the highway. Apart from the threat to life and limb from being run over in reduced performance mode (my countermeasure is to run for 30 seconds in reduced performance, shift to neutral, turn car off and on, which works great except for the threat of rear end collisions), the repeated reduced performance mode was impeding my trip to the junkyard, so I put the top up, blasted the AC and it did not happen again that day. And bonus, my son seemed happier on the drive for some reason. Note, occasionally the car stalls too, which I take as the entire ECM shutting down.

Two things that seem to be important in this, first, the ECM architecture appears to have a backup CPU that can take over and reset if the primary CPU has trouble. I haven't been able to find the logic for this published anywhere. Second, the ECM and TCM are cooled by a fan that pulls from the passenger compartment floor and vents back into the floor at another location. This fan was not operating when I got the car, the connector had corroded. Fixed the corrosion, fixed the fan. However, the reduced performance mode and codes still happened after I fixed the fan.

This weekend, I also replaced the original fan (LJB2412BA) with the new part (LJB2412BB) that appears to have slightly larger flow (from ~26 cubic feet/min (CFM) to somewhere north of 26 CFM). While I was doing this, I put a thermocouple on the ECM and pulled it through the grommet in the firewall into the passenger compartment so I can now monitor the temperature on the surface of the ECM directly.

I then tried the 'stress test', but ambient was only 32-33 C since it's September. The codes/reduced performance did not happen even though I drove the car, sat the car idling in the afternoon sun for about 45 minutes, drove on the highway for a bit and brought it back to 'cool down' by turning the engine off. So, 'fixed' by the new fan? Perhaps, perhaps not.

I did notice that my hypothesis seemed to have some support (see graph below)
1. The surface of the ECM got to about 60 C when idling in the sun with no AC (top was up for this).
2. Original plan was to reach highway speeds for about half an hour with the AC off, with the top up to see what peak temperature it could reach. Unfortunately, my personal ECM did not tolerate the cabin temperatures and lack of air flow so in lieu of passing out at 75 mph, I opened the windows. The ECM surface temperature increased for a few minutes before I opened the windows, but went down to about 56 C after that at highway speeds with the cabin airflow, I presume. Since I could never tolerate these conditions for long, I think this is beyond worst case anyway, for me.
3. Interestingly, the ECM temperature increased about 5 C in the 10 minutes after I turned the car off. This is presumably engine temperature equilibration with the isolated ECM/TCM box with no box fan on.
4. There is a long, slow decline in ECM surface temperature over 6-8 hours with ~26 C ambient.
5. It is clear, and a clever design, that the XK8 engine retains heat over long timespans reducing thermal stresses. This can lead to substantial increases in ECM temperature while the engine is off.
6. Even the peak surface temperature of ~65 C is well below the operational design limit of 85 C and survival limit of 100 C. It may be hotter on the board or in other locations than where I put the thermocouple, however. Whether it can happen under hotter conditions remains to be seen, but since the engine temperature is upwards of 100 C, I'm skeptical that the difference between ambient temperature of 32 C and 35 C or higher matters much. The ECM/TCM box is thermally open to the floor of the passenger compartment which limit peak temperatures. Since I didn't do this with the old fan, and have no interest in putting it back to see if I can 'unfix' things, I just have a thermal floor that doesn't cause the problem I've seen.

I'm also replacing the ECM in the next few weeks to alleviate the concern that the ECM was potentially damaged after years of broken ECM/TCM fan before I bought the car. I'm hoping that the problem is fixed, but I'm not clear that I will have the problem until it's midsummer again, which is fine by me...


ECM surface temperature measured by a thermocouple mounted to the lower surface of the ECM. Begins with car idling in the sun for about an hour after on/off driving for about 45 minutes.

 

Dale,

Fabulous write up. Learnt something new here. Never even knew there was a fan there.

 

David,

Appreciate it, also, the saving grace is that it doesn't get to 40 C here, much.

 

Was able to get the surface ECM temperature up to about 70 C by, running at highway speeds for about a half an hour, with stops of about a half an hour between, then sitting the car outside with:

1. top down in direct sunlight,
2. the heat on max aimed at the floor
3. 36 C (96 F ish) ambient temperature (record for 9/12 here today).

No CAN codes when driven, and I am quite certain there would have been codes under these conditions before I replaced the ECM/TCM box fan. Since #2 is quite ridiculous anyway, perhaps new and slightly improved fan has fixed my intermittent but longstanding CAN problem.

 

Just a quick addendum, I have a black interior, so she was mighty toasty when I drove her after she sat idling in the sun for an hour with the heat blasting. Not recommended. Wondered if there would be lasting damage...

 

Hopefully, the end of this particular saga.

I've been waiting a while for a local independent to swap out my ECM (LNG1410BD/103) for a 'new to me' ECM (LNG1410BE/103) because of this random 'ECM turning off in the middle of high speed driving' behavior that frightens my wife and daughter. I've established that it is definitively the ECM rebooting and it is somehow related to temperature. I was a bit reticent about doing it myself because I had a virtual machine copy of SDD/IDS, and I didn't fully trust it.

Now, I have a standalone XP machine that knows nothing about networks, has the network card turned off, with a native copy of SDD/IDS, and I felt brave enough to do it myself. Based on motorcarman's (Bob's) repeated advice around here on the immo setup, it seemed simple enough...

... and it certainly was. The whole operation took about 10 minutes, including me putting my thermocouple back on the ECM to check the temperature. It will take longer to reteach the car to idle and work the windows properly. Hopefully everything is resolved.

So, thanks to Bob for this and everything else he has contributed around here. I owe you beer.