I have two P codes indicating a CAN issue -- P1637 and P1643.

The problem is intermittent, and the car runs beautifully, until it doesn't. Then it shuts off, and we go to the side of the road. Ignition OFF, then re-start, and all is OK. One other case is that twice I have returned to the warm car after it has been sitting for a while, and it won't start. Only after sitting and retrying for 10 or 15 minutes does it fire right up, asking "yes, what's your problem"? It has once, after the above scenario, started in "Performance is Limited" mode. Took it home right away, then turned it off and restarted. It was ready to go, and seemed unable to understand why I am concerned. Runs like brand new, again.

Has anyone else been treated this way?

This car is a 2006 XK8, has 60K miles on it, and looks like it is still sitting on the showroom floor -- yes, even underneath.
Although there seems to be a major change in the communication networks during this model year, the car appears to follow the drawings for the 2003 series. The problem seems to be a CAN bus issue, with the P1637 and P1643 codes.

After troubleshooting the two indicated faults, I have found that the problem indeed has to do with CAN faults between the ECM and TCM. (As it happens, under the car there is no back, or shell, to the connector to the TCM, set at the back LH side of the transmission. The pins and wires at the back of the connector are therefore exposed to the weather (and winter salt water, etc.). The car has been extensively detailed, including under the car, so I expect that cleaning fluid has been applied to the connector as well.)

I have found that although the paralleled termination resistors measure 60 Ohms, the resistance to either ground or B+ from CAN+ or CAN- are all about 5k Ohms at the Data Link Connector (FC53).

I have traced this to the DSCCM by isolating parts of the bus until, upon removing the connector at the the DSCCM, with the ECM and others in place, the CAN to ground or B+ resistance measurements go high. This pretty much exonerates the TCM connector, although I still do not understand how the open to the air connector backside is OK.


In more detail, I split the bus by disconnecting the Linear Switch Module, so I can look at either the DSCCM/ECM side or the Major Instrument Cluster side, right there at the LS connector. With all the other modules connected, I looked at pins on the LS connector (FC100) this way:

CAN- at pin 2, CAN+ at pin 3, on DSCCM side;

CAN- at pin 7, CAN+ at pin 8, on IC side;

Pin 11 is GND;

Battery disconnected. Then I found


Pins 2 to 3 approx 120 k Ohms (confirms that there is a termination resistor in the DSCCM side.)

Pin 2 to Pin 11 (GND) approx 5k Ohms

Pin 3 to Pin 11 (GND) approx 5k Ohms

Now with the other modules still connected, I disconnected the DSCCM, and again looked at pins on the LS connector:

Pins 2 to 3 are open (confirms that the termination is within the DSCCM itself.)

Pin 2 to Pin 11 (GND) approx 120k Ohms

Pin 3 to Pin 11 (GND) approx 120k Ohms

This set of measurements indicate that the 5k Ohms is within the DSCCM.

These measurements suggest to me that at least one of the output transistors in the CAN driver for the DSCCM has failed in a low resistance mode. I feel that this resistance value might bias the bus such that it is very near the line between working and not working; which could explain how the problem is intermittent.


I am considering either sending the module out for repair or fixing it myself. For me to fix the module, I need to know the part number, or at least the pin-out, for the CAN transceiver chip inside the DSCCM. In the pictures I have seen of the opened module, the part number for the transceiver is not visible.

My questions are:

Will I see the part number if I open the DSCCM module?

Does anyone know what transceiver is used in that module?

Am I full of beans with this study?

 

Sympathize, chased this kind of thing for over a year for my 2001. Especially disconcerting is that my eventual solution likely involved two problems. Your symptoms are exactly mine.

The first was likely a disconnect in the solder joint in the gear illumination module that got worse with increasing heat in the transmission below (occasionally disconnected the network, fixed early on).

The second was a transient disconnect in one of the connectors in the engine fusebox (occasionally turned the ECM off while driving, occasionally did not let the ECM turn on when stopped, I found this by logging all of the CAN commands coming across the bus and noticing that the ECM was missing at those times). I finally located it while standing on the road with the bonnet/hood up, noting that the relays in the driver side fusebox hadn't clicked on after turning the ignition on, and I gave it a not very gentle smack. Everything turned on. All the relays were new since I had already gone down the relay path. Turns out the connector underneath had lost its tab and was almost fully connected. Almost was not good enough in the heat from the engine following the second start of the day, or randomly after the car had reached temperature.

Termination is in the DSCCM, but I'd be surprised if transient resistance bias on either CAN leg is the issue. If it's either on the high side or the low side, the system should actually still be robust to this.

One issue with my advice is that I am very familiar with my own wiring diagram (2001-2002) and not so familiar with yours (2003-2005ish). However, the fuse boxes and power supplies appear to have quite similar arrangements. So, I would suggest you remove and clean the fusebox connectors/relays/fuses for both the driver side and engine management fuse boxes before working on the DSCCM.

Your J gate illumination model is likely not the source of your problem since they were apparently upgraded.

Just curious, are there any other messages in the message center other than "limited performance" when this problem happens?

 

Hi Dale;
Good to hear from you.

I, too, have been fussing with this issue for over a year.

Although I have removed, checked for continuity and replaced each fuse in the Engine box, I have not checked a connector to that box. I will check that; and I will also get into the J gate again to look at that, as you have suggested. Yes, there have been other messages, not consistent, that I have unfortunately not documented. (I am new to this kind of automobile trouble shooting.) By memory, this include loss of DSC and Cruise, Limp-Home, and I think an ABS message.


Even if I do find another fault whose correction will eliminate the problem, it seems to me that the DSC module is in need of repair. The 5k Ohms I have found in the DSCCM aren't a transient feature. I understand that the CAN bus is very resistant to errors in the bias levels, more so in one of the + or - side than the other. I expect that the similarity in the (5k Ohm) measurements between the sides (they are not exactly the same) is because they are connected together by the 120 Ohm termination resistance. That is not large compared to the 5k Ohms, so the resistance measurements would be similar. Altogether this tells me that I could have a failed transistor on only one side of the bus that would put the bus bias (idle) voltage near a threshold that would allow a small variation to enable or disable the bus operation.

Thanks for your reply. This is quite an adventure for me. I will be more diligent in my documentation from now on.


Doug Y

 

An oscilloscope would go a long way to helping diagnose this. Even something like the U-scope from AES Wave https://www.aeswave.com/uScope-Basic-Kit-p9621.html would give you an idea of what is going on. Or not sure how much the Pico non automotive scope (2204A) is over there, but for careful use it's brilliant.
For CAN bus resilience to shorts and at what resistance levels it starts to fall apart take a look here


he also does some good stuff on CAN diag with a meter but it's not really the right tool.

 

So, my basic issue the ignition positive relay in the driver side fusebox and fuse 5. You don't appear to have any functions in that fusebox similar to fuse 5 (originally branched through the ECM and to the coils), but you still have the ignition positive relay, potentially causing the same issues. If you look at Figure 1.1, one leg of the relay runs through the inertia switch (and LF7, more later), which is why I asked about the messages. On the 2001, you get a different message on the message display than if the inertia switch is tripped (i did this on purpose to see if it was the inertia switch.) When the inertia switch is tripped, I got an additional message about brake fluid low that I never got when I was having the issues.

Under the fuse box are four UTA/Lear fuse box connectors in different colors, white, black, green, and blue. Tis the contact in the green one (LF7 for both the 2006 and the 2001) that can cause this issue. You probably want to clean the others since it's dirty under there...

On the DSCCM, one thing maybe I'm misunderstanding. You disconnected the bus in half at the linear switch. With the DSCCM, you measure 120 ohms between CAN leads and 5000 ohms from each to ground (CAN+ may be 120 ohms higher). With DSCCM removed, you measure that same half of the circuit with now open ends to the DSSCM at both pin 2 and pin 3 @ 120 ohms to ground. Based on the ground measurements, how could pin 2 to pin 3 be open and not either shorted to each other or 240 ohms from each other? And how is the 2 to ground or 3 to ground not higher than 5K? I'm sensing a fundamental misunderstanding on my part...

 

Are you actually measuring 120 K for the termination or is it 120 ohms? I thought it was supposed to be 60 or 120 depending on what’s connected.

in any event, your symptoms are classic for a heat induced solder joint failure. As the engine gets hot, and the interior of the car warms up, and electronics are active, weaknesses can appear. there’s a lot of heat generated in the electronics and when you add the additional increase in temperature from operating the car, The heat makes the components in the modules expand but they expand at different rates dud to their CTE’s. Metal leads on electrical parts expand at a different rate than the printed circuit board so if you have a weak solder joint it can crack. That component might make contact when everything is cold but then as it heats up things can move apart. And theres the failure.

Open up the module and get yourself a good microscope, or eye loop or an extremely high powered magnifying glasses and start looking at the solder joints for the components and see if you can find one that looks cracked. The Internet is full of pictures of cracked solder joints so you can see what you should be looking for. You can also see if anything looks overheated like a a component of some sort. Sometimes when connections open current goes in the wrong direction and it causes something else to heat up more than it supposed to and then leave little brown marks but sometimes that just happens and other times it doesn’t happen at all Its just something I look for when I inspected module

 

Hi again, Dale;

You are close to a correct interpretation of my notes. Here is the issue: "With DSCCM removed, you measure that same half of the circuit with now open ends to the DSSCM at both pin 2 and pin 3 @ 120 ohms to ground. " Well, that measurement is 120k Ohms to ground at either pin 2 or pin 3.

That's why I thought the 5k Ohms is(are) in the DSCCM.

I am still studying this problem, but it is pretty cold in my unheated shop right now. Gives me a chance to see to my other projects, but I have not given up. Thanks for your interest -- I am taking you and everyone's advice to heart.

Doug Y

 

Hi Wight8;
That is a great video! Lets me see what the wave forms look like in a real auto setup. I do have a 'scope, and will use this new knowledge to educate my guess about the 5k Ohm question.

Thanks,

Doug Y

 

Hello Aarcuda;

The 120k Ohms is between each of the CAN bus leads and ground, with the DSCCM removed. Since the 120 Ohms is within the module, that step is not measuring the termination. This test tells me that the 5k Ohms, that I think is incorrect, is also within the module. I expect that the high resistance I see has to do with the bias supply for the OFF state of the bus (2.5v) that can be overcome by the CAN output transistors.

I do have a microscope, and you have helped me make up my mind to remove the DSCCM and have a look. It is pretty cold out in my unheated shop, so for now I am hiding out in the office, thinking about the whole thing. I do see what you mean by the intermittancy being a heat and damaged solder joint issue.

Thanks for you participation.

Doug Y

 

Okay, I see, appreciate it.

So, no short to ground in the outside wiring and 5K to ground on the inside. So, now, i'm not clear on why 5K to ground from the CAN pins inside the DSSCM is an issue. This doesn't say that the usual power solder point issue in the DSSCM isn't a problem, but seems like that is not necessarily an issue if pin 2 to pin 3 is 120 Ohms on the bus.

 

Hello Dale, and all of you;

I have removed the Engine fuse box and checked the contacts at the four connectors. There was some corrosion, but not much. I burnished the contacts to get the crust off without ruining the surfaces -- looks like they are tinned. Next cleaned all with contact cleaner and Q tips.
Then I removed the Ignition Positive relay, and cleaned it, too. Also checked it at my workbench for operation. The contacts closed nicely (no load), with a small fraction of an Ohm resistance. Put everything back together, cleared the DTC and went for a ride. Drove 20 miles or so, with no issues.

Maybe it's fixed! I am afraid to try again, lest I be disappointed, but I will, soon. In any case, thank you so much for the tip!

Doug Y

 

Good luck in it remaining so.

 

Well done, and happy motoring....