Hard to Diagnose Problem Solved

[MYTraveler]

My boat has twin Cummins QSM 11s, with Twin Disc gears. One day, with no prior hint of any similar problem, I went to start up the boat. Port side acted normally -- the SmartCraft Display went through the POST with no problem then indicated 0 RPM. Pushing the station select button on the gear controls indicated power on the port side and that the gears were in neutral. All normal. Starboard side also went through POST without a problem, but the smart craft display showed a dash, instead of a zero for RPM, and it also showed an "Invalid Data Link" error. The twin disc did not engage on that side -- no power light and no neutral light on the starboard side. Maretron showed 0 RPM (and other data) for the port side, but no data on the starboard side.

I confirmed that the starboard engine was getting 12v (13.1, actually), and I checked all the fuses and found no problems. I checked all the connections and everything looked good. Being the original owner, I knew that I had never experienced anything like this before and I was confident that no other work had caused the problem (no work of any sort had been done since the last time, a few days earlier, when everything worked normally).

To rule out a problem with the ECM, I connected the starboard engine to the port side wiring harness -- everything was normal. Similarly, the error persisted when connecting the starboard side harness to the port engine. So, I ruled out a problem with the engine's ECM (and all of the wiring on the engine).

At that point, I was pretty well stumped. The starboard side problem could be at any of the 6 stations, or any of the wiring between. I was completely stumped, and couldn't even think of a good way to diagnose or further isolate the problem. Getting a Cummins tech wasn't easy, and after his computer couldn't get a reading from the starboard engine, he was stumped too. I wasn't surprised that he couldn't get data out of the starboard engine since Maretron wasn't getting data.

Searching for info on line was not fruitful, and even the Cummins tech support people were of little help -- I coudln't talk to anyone who had seen this problem before, although there were higher levels of factory support available to certified Cummins techs (I was actually considering trying to get that certification).

I concluded that three must be a problem with the boat's J1939 network, which would account for the "invalid Data Link" error I was getting, as well as Maretron's failure to get any data from that engine. I further inferred that the data error was cause by one of the data wires being either broken or shorted, but I didn't know how to find the problem in the wiring, since it could be anywhere.

My Cummins tech finally get an insight from Cummins factory support (escalated up a couple levels), and that insight allowed me to find and fix the problem today. This had been going on for a few months, so it is quite a relief.

So, in case this helps someone else in the future, I will share the problem, but first I thought some of you might enjoy the challenge of figuring out the problem or even devising a practical solution to isolating it.

I should add that no part of the wiring is exposed to the elements, all is factory original Cummins wiring (which is encased protective plastic conduit), and no visible portion exhibits any signs of damage, wear, exposure to fluids, etc.

 

[DDW]

Can't the Cummins tech connect his computer (or you could do this with a smartcraft display) closer and closer to the engine until a good connection is made, then regress backwards? Since the engine ECM and wiring harness was known good.

Problems like that make you wish for an all mechanical diesel...

 

[High Wire]

My guess is a loss of power to the data network, wherever it begins. At the on/off key maybe. It’s a shame that today’s engines don’t have a “get home” mode that only powers the injectors, fuel pump, and run/stop solenoids.

 

[dhays]

Rick, I’m looking forward to your solution.

I have a QSB 5.9l engine. I have a Smartcraft display in the pilothouse and a lower tech gauge cluster on the flybridge.

Most of the time, when I start up the engine I get a 0 RPM at both stations. The SmartCraft will display temps but not rpm or fuel consumption. After running for 15-20 minutes, the display behaves normally. The Engine always runs great.

If we are out cruising and so running the engine daily, often it will start working right away. I’ve checked every connection and can’t find a problem.

 

[O C Diver]

I had a similar problem on my John Deere 4045. All of a sudden it just quit while I was cruising Lake Superior. Thankfully it was a calm day on the Lake and I had cellphone reception. Called my dealer back in Maryland. After going through some diagnosis using the motors MFD, we were able to determine it was the cable between the computer and the fuel injection pump. Both ends are fancy connectors with oring seals. His suggestion was to remove and reinstall the connector as the installing wipes the pins when they go together. The one on the pump didn't solve the problem. The massive one on the computer did. The second time it happened a month later, the connector pins got a spray of contact cleaner and corrosion guard. Problem hasn't happened again.

So that's my guess.

Ted

 

[MYTraveler]

Can't the Cummins tech connect his computer (or you could do this with a smartcraft display) closer and closer to the engine until a good connection is made, then regress backwards? Since the engine ECM and wiring harness was known good.

Problems like that make you wish for an all mechanical diesel...

He connected his computer to a port for such purpose directly on the engine, but the computer was unable to send or receive any data. Cummins factory (level one) suggested he return and try connecting to the other engine's port, to verify that the problem was not with his computer (other engine connected fine, computer was fine). Note that the engine remained connected to the boat's wiring harness, and through that to the smart Craft Displays, throttles, one of switches, etc. The ECM gets its power through that harness, so disconnecting the harness, while leaving the diagnostic computer connected, will deprive the ECM of power. Good thought though.

 

[MYTraveler]

My guess is a loss of power to the data network, wherever it begins. At the on/off key maybe. It’s a shame that today’s engines don’t have a “get home” mode that only powers the injectors, fuel pump, and run/stop solenoids.

Good thought, but not in this case.

 

[MYTraveler]

Rick, I’m looking forward to your solution.

I have a QSB 5.9l engine. I have a Smartcraft display in the pilothouse and a lower tech gauge cluster on the flybridge.

Most of the time, when I start up the engine I get a 0 RPM at both stations. The SmartCraft will display temps but not rpm or fuel consumption. After running for 15-20 minutes, the display behaves normally. The Engine always runs great.

If we are out cruising and so running the engine daily, often it will start working right away. I’ve checked every connection and can’t find a problem.

I will PM you the answer, in case it is of immediate use to you, but as long as others are interested in playing 20 questions, I will not post publicly.

 

[MYTraveler]

I had a similar problem on my John Deere 4045. All of a sudden it just quit while I was cruising Lake Superior. Thankfully it was a calm day on the Lake and I had cellphone reception. Called my dealer back in Maryland. After going through some diagnosis using the motors MFD, we were able to determine it was the cable between the computer and the fuel injection pump. Both ends are fancy connectors with oring seals. His suggestion was to remove and reinstall the connector as the installing wipes the pins when they go together. The one on the pump didn't solve the problem. The massive one on the computer did. The second time it happened a month later, the connector pins got a spray of contact cleaner and corrosion guard. Problem hasn't happened again.

So that's my guess.

Ted

It could have been, but wasn't, at least wasn't that cable. Remember, I concluded it is a wiring problem somewhere, I just couldn't figure out where (and the wiring is complex, connecting 6 helm stations, etc.). Thanks for your thoughts - Rick

 

[DDW]

Note that the engine remained connected to the boat's wiring harness, and through that to the smart Craft Displays, throttles, one of switches, etc. The ECM gets its power through that harness, so disconnecting the harness, while leaving the diagnostic computer connected, will deprive the ECM of power. Good thought though.

That is a problem with their test setup. The possibility of faults or induced noise in a large connecting harness must be removed in order to debug. Either that or you need to get down and dirty with an oscilloscope and a lot of knowledge of the serial protocol.

One step I'd have taken is to remove and reconnect all of the connectors in the harness I could get to as these are always suspects. Then I'd try disconnecting anything that was not required to run. But the most efficient technique is usually to go back to what works (engine alone) then add back pieces until it breaks.

An amusing and marginally related experience: flying the sailplane out of a tower controlled airport, outbound the front end for the VHF COM quits communicating with the COM black box on its serial link and it is stuck on the wrong frequency (not the tower). I return to the airport and call the tower on the cell phone. They say proceed to pattern altitude and IF we give you a green light from the tower you are cleared to land. I reminded them that I was in a glider, but it didn't seem to sink in.

 

[rgano]

Could you unplug stations from the network one at a time to help isolate the issue? This kind of issue is why I chopped up the cables and literally threw the whole ZF Micro-Commander out of my boat right after I bought it. Morse cables I get! Way too simple a boat to have that sort of thing installed.

 

[MYTraveler]

Could you unplug stations from the network one at a time to help isolate the issue? This kind of issue is why I chopped up the cables and literally threw the whole ZF Micro-Commander out of my boat right after I bought it. Morse cables I get! Way too simple a boat to have that sort of thing installed.

I would think so, but the wiring is so complicated that without a good schematic specific to my boat's layout, I think it would take the understanding of someone very familiar with how these things are wired up. Knowing now what the problem was, I can tell you that disconnecting all of the stations and trying them one by one would not have solved this problem, so, ironically, not having that schematic probably saved me a bunch of effort and the risk of putting things back together wrong (or damaging connections in the process).

 

[ABfish]

I have no idea what the problem is here, but I'll quote my favorite heavy truck mechanic:

"Start with the grounds, son! Start at the batteries and work your way to the engine."

 

[MYTraveler]

I have no idea what the problem is here, but I'll quote my favorite heavy truck mechanic:

"Start with the grounds, son! Start at the batteries and work your way to the engine."

Sounds like great advice. In this case, there is a constant 13.1v to the engine, including under load. But, the ECM is only energized when the key is turned on. That also energizes the smart craft display and the display does go through, and passes, the power on self test. From that, I inferred (it turns out correctly, though my logic may have been flawed) that the ECM is getting adequate power.

 

[gsholz]

Since the error is a data link error, I'd first check datalink wire resistance at different stations. Should be 60 Ohms with two 120 Ohm resistors at either end (assuming it is based on CAN bus). If you disconnect a module, the impedance should be 120 Ohms on the wire.

Then I'd put a scope on the data link wires. Data packet voltages should range from 1.5-3.5V. If the data packets are distorted, I'd start disconnecting stations until the data transmission cleared up. This should isolate the fault. Put the scope on the working engine data wires to verify what a known good signal looks like. Typically the cause would be a power/ground issue on one module or a defective datalink transceiver in one module.

 

[Lepke]

Just one more problem with electronic diesels.
On my Detroits, I push start buttons, gauges come to life, move the shift levers and go...

 

[MYTraveler]

Since the error is a data link error, I'd first check datalink wire resistance at different stations. Should be 60 Ohms with two 120 Ohm resistors at either end (assuming it is based on CAN bus). If you disconnect a module, the impedance should be 120 Ohms on the wire.

Then I'd put a scope on the data link wires. Data packet voltages should range from 1.5-3.5V. If the data packets are distorted, I'd start disconnecting stations until the data transmission cleared up. This should isolate the fault. Put the scope on the working engine data wires to verify what a known good signal looks like. Typically the cause would be a power/ground issue on one module or a defective datalink transceiver in one module.

I didn't scope it (primarily because I was looking for a pinout, though I can see that wouldn't have been a necessity), but knowing now what the problem was, I don't think it would have helped as there was no data, not just corrupted data. I will add that I am surprised that the problem resulted in no data, especially since that is by design.

 

[twistedtree]

I'd guess it was something on the J1939 bus that was messing it up. Could be the Twin Disk throttle control (depends on how the throttle input is provided), the SmartCraft display, or your maretron converter. Or maybe a bad connection to any of the above.

 

[MYTraveler]

I'd guess it was something on the J1939 bus that was messing it up. Could be the Twin Disk throttle control (depends on how the throttle input is provided), the SmartCraft display, or your maretron converter. Or maybe a bad connection to any of the above.

I believe you are going the right direction (I don't know whether the wiring/component that failed is on the J1939 bus, but I suspect it is), but the question becomes how to isolate/identify that component, vs the bus itself, and if the bus itself, what part of the bus. I did disconnect the Maretron converter to rule it out.

 

[gsholz]

Data on these serial buses comes in packets of zeros and ones with voltages ranging from 1.5 to 3.5V. You do not need to decode the packets to diagnose failures. Anything that disturbs the signals significantly would lead to no communication. Then it is just a matter of disconnecting modules on the bus until the data signals are restored as long as you have at least one good talker on the bus.

To identify the differential data wires (CAN bus high and low) look for wires that are twisted together to eliminate common mode interference.

Any good tech should be familiar with the above. Any car after 1995 is CAN bus based.

 

[MYTraveler]

Data on these serial buses comes in packets of zeros and ones with voltages ranging from 1.5 to 3.5V. You do not need to decode the packets to diagnose failures. Anything that disturbs the signals significantly would lead to no communication. Then it is just a matter of disconnecting modules on the bus until the data signals are restored as long as you have at least one good talker on the bus.

To identify the differential data wires (CAN bus high and low) look for wires that are twisted together to eliminate common mode interference.

Any good tech should be familiar with the above. Any car after 1995 is CAN bus based.

In this particular case, that approach would not identify the problem. When I reveal the exact cause of the problem, which is what makes it so hard to diagnose, you will see why.

 

[tpbrady]

If all else seems to be ok look for an external factor. It’s something external to the network that is corrupting the data. It is probably close to the wiring harness and could be an unshielded data cable from another network, a power supply of some kind, or some other active electrical component.

 

[mickand]

I'm getting bored - give us the answer!!!

 

[MYTraveler]

The short answer is that the flybridge starboard side stop button failed. That button is a normally CLOSED (backwards from most, in my experience) momentary switch. Pushing it is supposed to open a circuit thereby shutting down the ECM. This button was always OPEN, as if the stop button were continuously pressed.

Especially difficult to diagnose because, even if I had disconnected everything (ie, all of the start and stop buttons, both sets of displays and all 6 throttle controls), the symptoms (no data from ECM) would have been the same. In retrospect, knowing the pinout (which I couldn't get) would have allowed me to confirm continuity between those pins, and the lack of continuity would have pointed me straight to the stop buttons. Anyway, I disconnected the starboard side switch and wired its leads together. Engine starts right up with full data. New switches on order.

Ironically, the stop button on the port side (of fly bride) also failed, but it failed to always OPEN, even when pressed. As a result, it didn't stop the engine from starting, but it would not have shut down the engine. I never noticed because I always start and stop from the PH. For both of those buttons to fail is surprising to me. They are both original equipment Cole Hersee, well protected from the elements on the back side, and fitted with rubber covers on the front side. There is no visible damage, water intrusion, etc., on the rear (or front).

Thanks for playing, and good guesses all around.

 

[twistedtree]

Well, I never would have guessed that one.

 

[MYTraveler]

Well, I never would have guessed that one.

Thank you, that is reassuring, especially coming from you. And I figured it is useful information in case anyone else gets stuck with something even similar. What I really needed was a pinout.

 

[BevinB]

May I ask what type and year your boat is sir?

 

[rslifkin]

Thank you, that is reassuring, especially coming from you. And I figured it is useful information in case anyone else gets stuck with something even similar. What I really needed was a pinout.

Definitely a weird one. And it supports my idea that the most important thing about any electronic system is having access to the tools and information needed to diagnose it.

 

[High Wire]

Did you mean to say the port side failed always closed instead of open? Failure open would be like the starboard motor. Or am I confused?

 

[MYTraveler]

Did you mean to say the port side failed always closed instead of open? Failure open would be like the starboard motor. Or am I confused?

Exactly. Both switches bad. Starboard stop switch was always open (as if pressed to stop the engine) and port was always closed' even when button was pressed, so it could not be used to stop the port engine. Note that the bad port side switch had nothing to do with my problem on the starboard side.