W211 CAN Bus problems
 

Edit to add: central gateway (search optimization)

I have had my W211 for about a year and a half and it has been great so far. Until it started dying on my son for no apparent reason earlier this summer.

Back in June my son was driving the car from Clemson to Charleston, on his way to his girlfriends, late in the evening of course, and he calls me in a panic. Multiple warning lights and messages (ABS light and loss of ABS/ESP message was one I remember), dome light flashing on and off, a/c blowing hot and cold, etc. He gets the 'red battery' symbol so I am thinking alternator. He pulls off the highway and I get him towed back to Columbia.

Pull diagnostics with my Xentry and nearly every module in the car has multiple stored faults for CAN failure to talk to other modules in the car. Both power train and body buses have problems. And messages about CAN bus shorted and CAN bus shutdown. 8 pages of printout. Of course everything is a lower case f which means fault came and went and is not present now, (capital F means fault is active right now).

So I dig into the CAN bus distributors and can't find any corrosion or loose connectors. But I did find under the right dash that the CAN bundle was laying against the aluminum foot plate that protects the electronics boxes in the footwell. There was a slight chafe on one of the CAN wires. So I de-pinned that connector, heat shrunk the chafed wire, and put back together. And zip tied the distributor and bundle in a better location below the dash where it won't hit that aluminum plate anymore.

Before I purchased the car (private sale in spring of 2018) the previous owner had an indy shop replace the oil cooler seals. Well they did not replace the o-rings and gaskets on the intake tubing in the front of the engine. This caused soot to blow all over everything up front and the alternator was caked in black crud. I thought well I better go ahead and replace this thing anyway, and put all new gaskets in the intake tract, and clean this mess up, before giving the car back to my son.

So after getting the car back together it drove fine for about a month until my son decided to go visit his girlfriend again. This time he got there ok and was getting ready to head back home, and only this time it refused to crank. No spin at all. Another tow back to the house.

Pulled codes again, same sort of errors, not an identical list but many similar as the first time. And of course the car started fine once it got back to the house. Tow truck driver confirmed he could not start it at the far end, and in fact could not get it out of P, and had to use the emergency release to get it up on the flatbed.

Spoke to my friendly service advisor at the dealer who used to be a tech and he's real good about letting me bounce ideas off him. He said check for water intrusion and the worst 2 places on a W211 are the rear SAM in the trunk and the ZGW under the dash. He said the trunk water comes from bad trunk gasket (duh) and the ZGW caused by plugged sunroof drains leaking water down the a-pillar into the wiring behind the dash. The ZGW faces up under the knee cover and the wiring harness wicks water drips down into the connectors.

First went to the trunk and pulled the left liner to see if the rear SAM was getting wet from a trunk leak. Everything was completely bone dry with zero evidence of any kind of wet back there. Pulled the plug with the body CAN wires and everything looked pristine.

Next I dropped the knee cover on the drivers side to reveal the ZGW. It is a little gray box about the size of a deck of cards. When I pulled the plugs out I saw evidence of corrosion. I popped the module off the knee cover and opened it up. Sure enough there was corrosion inside on the plug pins going into the circuit board, mostly on pins 7 and 8 and 9 and 10. Pins 7 and 8 are the engine CAN and 9 and 10 are the body CAN. Lots of green grunge growing between the pins.

I got some isopropyl alcohol (91%), wooden toothpicks, and cotton swabs and was able to clean up all the corrosion. Flushed everything out really well with alcohol and then let it dry overnight. I put the module back together and put it back in and the car started right up.

Took the car out and ran several errands so went through about 6 or 7 start/stop cycles and everything running good.

Tomorrow I will pull the dash area apart and verify the source of the water leak.

Just wanted to pass this along, if you have a W211 that is having CAN bus issues check the ZGW module.

As I read your post I was wanting to ask if heavy rain was involved. I have had similar problems after a car wash or heavy rain. My water problems

Caig offers excellent contact cleanses and conditioners

A day of drying has cured my problems. Thanks for the excellent information

Nice write up....thanks for the pioneering leg work.

That's excellent diagnostic work, well done and I hope the problem is resolved.

Please refresh my memory, as I recall the central gateway module is responsible for pulling diagnostic information from all the assorted ECM's in the vehicle and making the resultant data available for ODBII access? I think it also is responsible as a access point and protocol for passing data from one CAN to another correct?

It seems a little strange and disturbing to see how disrupting the CAN architecture on modern MB's can apparently totally disable the car. I realize that data transfer and sharing is essential for many/most of the sub systems in play but it just seems like the hardware and software of the systems should be able to fail in less catastrophic way if data communication is disrupted. I think the engineering term is something like 'graceful degradation' or something like that?

It's been a few years now but I remember someone that had a DCM in a W210 fail and it took down the entire interior CAN network. Seems like if a central gateway has issues it can disable every CAN in the car?

There are 4 modules on a W211 that connect the engine and body CAN buses together. The ZGW is the primary connection and the only one that can pass packets between the two networks. There are 3 other modules have double connection to the CAN buses, so they can read and write each bus, but cannot bridge data between them. I reverse engineered the wiring diagrams and made a one page tree diagram of the networking in a W211 and then found the actual diagram in the Star Finder utility. But now I am on the road and can’t recall what they are (maybe instrument cluster, key switch, and front SAM?)

CAN is meant to be able to run in a degraded mode if you short or open one of the two lines. The data is passed using differential signaling, which is a fancy way of saying that each line has a copy of the data, one is positive biased and the other is negative biased. This results in high EMI tolerance as any induced voltage will affect both lines equally.

But if you dead short the CAN then you are pretty much out of luck. This is what finally happened at the second failure, the green grunge built up to the point where it completely shut down the network.

I agree that it is crazy that a short in the CAN bus will pretty much shut things down.

W210 has both CAN buses and they meet at the key switch instead of a gateway module. The W211 (and other models of that same era) introduced the ZGW as a formal bridge between the two systems and also this is where the vehicle coding is held. I also think this is where they store the fault data on CAN bus health. There is a huge Freescale processor chip in this little tiny package.

They did a couple dumb things on the design in my opinion however. First I would have put the 2 buses on different connector plugs for isolation (and at least not on adjacent pins!). Also I would have located this module differently. Having it face up with the wire bundle leading into it from above is just asking for trouble with wicking water leaks into the connector. Especially under the dash where leaks can happen from the sunroof or windshield.

I didn’t get a chance to look for a source of the water leak this weekend but will try again next weekend. At least it’s parked out of the weather at the airport now.

I remember that the CAN bridge point was the ignition switch in the W210 chassis, always seemed like a strange interface point but I guess there must have been some reason to locate it there.

Agree totally on the placement of the ZGW model and the location of the CAN pins on the connector. Is there enough room on the wiring harness to re-orient the module with the connector facing down?

All CAN modules should have a fail-safe backup mode that implements a plan if all communications are lost with the CAN. At the very least that plan should include an error message in the module that clearly states that CAN communications have ben totally disrupted and a fail-safe mode was entered. I guess you can kind infer this when every module in the car issues 'the-sky-is-falling' general error messages but this kind of response kind leads me to believe that the software people never considered the possibility of this kind of failure.

Whenever I see something like this I keep thinking about how Elon and Tesla are going to deal with these kinds of issues in their 100% self driving cars that will be on the road in the millions by the end of 2020. It seems like an incredibly difficult hardware/network/software problem. Multiple redundant networks would be a good starting point I'd guess.

There are silly things on nearly all vehicles, Benz provides an isolating connector - some manufacturers dont like to spend that 2 dollars. So you get to fight with entire jumbo connectors which have the terminating pieces.

And its really irritating when they make one module the gateway too (e.g. some mazda use TCU as gateway to the slower body bus)

I noticed another clue:

For about the last 6 months, whenever I engaged the cruise control, the 'halo' around the speedometer would be intermittent. I chalked this up to the LCD going bad in the cluster.

Well ever since I cleaned the contacts in the CGW, the halo has been steady whenever the cruise is engaged.

Not 100% sure if this is a definite correlation but if you are getting this symptom on your W211 I would check your CGW to make sure it's not getting funk between the pins. This needs a T25 to take the lower cover under the drivers side dash down (3 screws).

I would recommend disconnecting the battery first before unplugging the CGW as it sets a ton of events in the system.

Still haven't gotten a chance to look for the water leak but it's been parked under cover since I discovered the problem.

Got some time to look at the car this weekend.

First tried gently pouring water into the sunroof trough at all 4 corners, the water flowed easily out to each wheel well with no puddling whatsoever. (Car sitting level).

Next I parked it engine downhill, and poured more water in the trough at the front. Still draining correctly up front.

Took the a-pillar cover off and it appears the drain tube is running through the internal part of the a-pillar (not visible from what you see with the cover removed, the only thing you see is the air bag squib).

Took out the speedometer cluster and felt around back there and could not detect any wetness.

Closed everything up and took it to the automatic car wash. Pulled the lower dash cover immediately afterward, and everything was completely bone dry.

Anyone have any clue how the water could have been migrating down the harness?

Maybe some condensation from the air conditioning system.

When I have been traveling on a long road trip with heavy rain forecast I have taped over the vents in the hood (about 5 or 6 inches from the windshield) and not had any problems. Not certain if that is the area you were referring to as the "trough".

I have had all the warning lights come on after washing the car and figured water got in from the hood vents. I could be wrong.

Good luck and thanks for the updates.

So the fun continues with this car.

On Wednesday I get a call from my son who informs me the red battery is showing on the display. It's afternoon time so he's heading home from his college intern job (about 30 miles one way). I am on I-90 heading west from Quincy to Seattle WA going over the Snoqualmie pass...

He pulls off the interstate at the exit where he lives, and limps into a gas station, and calls me. The car's system are collapsing, no A/C, warnings up and down, transmission stuck in 2nd gear, no acceleration.

I tell him to pull up the system voltage (3x reset button with mph showing and key in 1st position). 9.something volts.:eek:

How long has that light been on? Since this morning when I got into it to go to work.:mad:

Car cranks but it collapses the voltage too much so it bogs down and dies.

He gets it jumped and limps it home. Able to get a ride with one of his room mates (who all work at the same place...) Thursday and Friday.

So yesterday I gather up every tool I can think of and go the 2 hours up to Greenville. Put the meter on the battery, it had self regenerated to the low 11 V. But as soon as I hit the starter it was rrrrp rrrp rr nothing. Jump started the car with my car and got it running again. When I pulled off the jumper cables voltage crashed down to the low 10s. No alternator output confirmed.

Pull the alternator and take it to the local NAPA. They gave me another over the counter.

After 2 hours of disassembly and 2 hours of reassembly got everything back together. In the mean time I had disconnected the battery from the car and threw a 10 A charger on it. By the time we were ready to try to start it, the voltage was back in the low 12s.


Car fired right up and had a ton of codes. Unfortunately the serial cable for my Xentry multiplexer had failed so I had one on order from eBay coming from Gungdong Province on the slow boat not expected until later this month.

So I ordered an iCarsoft II so I would at least have something to take with me to troubleshoot. I wasn't able to pull any codes because the car voltage so low it would not let the iCarsoft stay up.

So once I got the alternator in there were a ton of codes for low voltage. Got all those cleared and drove the car back to Columbia (I left him my E350 again, he's driven it more than I have since I bought it in April...) with one eye on the system voltage display. Did fine, never got below 14.xx V the whole way.

When I got home the package from China had arrived. I hooked up Xentry and pulled codes and found the code for LIN bus failure to communicate with the alternator, current and stored. Cleared it and it came right back.

About to go out and double check I plugged in the connector but I suspect something else is up.

Anyone know more about the behavior when the LIN bus is lost to the alternator, does it go into some sort of default mode where it still charges but does not communicate?

I suspect I got another lemon from NAPA with a bad LIN bus communication chip this time...yay I get to pull the alternator on the car again.

So the LIN bus plug was connected. The LIN wire has approx. 8 V dc on it with the switch in run position but engine not running. That tells me there is continuity to the rest of the system.

There are three nodes on the LIN bus, the engine controller N3/9 pin 88 (the master) and two slaves, N14/3 pin 12 (the glow plug relay), and G2 pin 1 (the alternator voltage regulator). The wire is tapped at Z9 which is just below the connector for the glow plug relay down in the harness. So the first wire segment runs from N3/9 to Z9, the second segment from Z9 to N14/3, and the third segment from Z9 down to the alternator.

Since I am getting 8 V at the alternator connector (with it disconnected from the alternator), and the glow plug relay is working (it's preglowing properly - cold started fine this morning, and no other codes exist for glow plug relay) it's safe to assume that the LIN bus signal is physically getting to the alternator from the ECU.

It is error 2263 in Xentry. The following comes up when you try to run the alternator test. What is most interesting is that it comes up in the quick test as an 'i' (event) not a 'f' (stored fault) or 'F' (current fault).

Alternator diagnosis
Possible cause and remedy :
- Check electrical line between component N3/9 (CDI control unit) and component G2 (generator) for contact and routing and (or) line breaks.
- Test wiring from component N3/9 (CDI control unit) to component G2 (generator) with an ohmmeter for short circuit (positive or negative) and repair, if necessary.
i Connect test wiring harness with socket box with component N3/9 (CDI control unit) not connected.
Other possible cause of fault :
- The alternator regulator is defective
End of test

Going to order a genuine Mercedes remanufactured alternator. Tired of playing the merry go round with NAPA refurbished units. Hopefully with them selling me 2 lemons they will give me my money back. They are $650 instead of $200 but well worth my time and peace of mind.

Seems like a reasonable diagnosis to me, why is the alternator a networked device? Are they doing some sort of trickery to bias the alternator charging scheme to periods of deceleration or something? What is the text associated with the 2263 error?

Alternator
 

Going to order a genuine Mercedes remanufactured alternator. Tired of playing the merry go round with NAPA refurbished units. Hopefully with them selling me 2 lemons they will give me my money back. They are $650 instead of $200 but well worth my time and peace of mind.[/QUOTE]

Having dabbled with my two W211s on can bus issues the Alternator has to be the backbone of reliability. Another option to going OEM is a Bosch New Unit. I recently ordered one from Pelican and it would be the same as a new MB unit which would likely be be a remanufactured.

I am dealing with a front passenger side SAM unit defective right now and some of these modules are starting to be become rare, the one I need is not in the USA and none in stock in Germany. Supposed to be 4 weeks. The bummer on these newer gen tech cars is if your SAM units/ECM units go there is no subsitute for not getting new modules. You can find used ones, but dealers will not program used parts and unless you have a really old version of the Xentry there is no offline coding available.