Ignition Problem?

[wjbertrand]

Nearly three months and 21 pages, and no solution yet? Time to either live with it, take it to the dealer, or sell it because you guys are just grasping at straws now.

Two Honda Dealers and one dyno test later the problem is not solved. If you rode the bike you would not suggest living with it. Selling it may be the only answer but I hate to go that route.

 

[wjbertrand]

I am assuming that after three months, all viable potential causes have been checked and have come up as non- solutions, so at this point, it's safe to say that something this complex and specific is either a figment of the owners imagination,

This is frankly insulting, particularly as you've admitted to not reading the whole thread. If you've nothing constructive or helpful to add, please ignore and refrain from posting on this topic further.

 

[wjbertrand]

Thanks Mark.

 

[wjbertrand]

Based on some conversations I had at WeSTOC, particularly with Brian Browning, I've been running an experiment with STat this week regarding the O2 sensors. I disconnected the battery before leaving for WeSTOC and upon my return, disconnected both O2 sensors before reconnecting the battery. The bike started up immediately as normal, but of course there are #23 & #24 error codes from the disconnected O2 sensors. Brian and I talked about the ECM perhaps needing to re-learn without the O2 sensors in the circuit. My question here is how long should the re-learn process take? Is it mileage or Key-start cycle number related? I've put maybe 300 miles on the bike and a dozen key-start cycles and so far the misfire/stumble problem seems unchanged, i.e. I don't sense any "learning" going on.

Methinks the O2 sensors are actually working fine as since running with them disconnected I've dropped 5-7 MPG. Since using more fuel means a richer mixture in my mind, I'm beginning to question the dynomometer mixture data we obtained a few weeks ago. I think my next step, also suggested by Brian, will be to rig up a propane bottle in my tank bag with a hose feeding into the intake. I'll try dialing in a little more fuel to see what happens to the misfire when I do that. Should help to better define whether this is a lean misfire as suspected, or something else.

At this point, since running with them disconnected seems to have made no difference what so ever, I am not inclined to replace the O2 sensors or try the emulator/eliminators (thanks anyway Adam K.) as I'm pretty much convinced the problem does not lie with them.
__________________

 

[wjbertrand]

Did a little more exploration today and discovered a previously unknown (to me) 5-pin connector. It's located to the rear of the battery behind the starter relay assembly. You can't even see it until you fish it up from the pocket in the fender/battery box. Anyone know what it's for? Could not find it in the manual.


-Jeff

 

[pdfruth]

Did a little more exploration today and discovered a previously unknown (to me) 5-pin connector. It's located to the rear of the battery behind the starter relay assembly. You can't even see it until you fish it up from the pocket in the fender/battery box. Anyone know what it's for? Could not find it in the manual.

Did you happen to take note of wire colors for the wires in that 5-pin connector?
Looking at the wiring diagram for any 5-pin connectors that might be in that vicinity... There is a 5-pin connector that carries the signals to the rear ABS modulator. Wire colors in that 5-pin connector are; O/Bu, W/P, P/W, R/Bl, and Br/Lg

 

[wjbertrand]

Thanks, nope, stupidly I didn't check the colors. This connector doesn't seem to connect to the modulator though. I see a separate connector that plugs directly into the modulator assembly.

Edit: well I think you're right. The connector on the modulator assembly that I saw is likely the 2 or 3 pin connector. Shown in the diagram. I found this connector incompletely seated so was hoping for a "smoking gun". Unlikely to be the case if it's for the ABS system.

-Jeff

 

[wjbertrand]

Many thanks to Jim Henke today for coming over and scoping up my bike. We were able to show that both coils and the fuel injectors are being interrupted when the bike misfires. We were also able to show that the cam and crank sensors are putting out steady signals at the time of the misfire. I attached the screen captures below.

Capture 002 shows the left and right coils and you can seen the interruption when the bike misfires
Capture 003 adds one of the fuel injectors (magenta) and it too can be seen to suffer an interruption in sync with the coils.
Capture 004 adds the crank position sensor and it can be seen that it's output is steady when the misfire occurs
Capture 005 we moved the probe from the injector to the cam position sensor and it too remains steady in output during the misfire episode

At this point, due to the abruptness of the interruption (not to mention the lack of an error code), Jim is thinking it's probably not sensor related. I'd be tempted to blame the ECM except we already eliminated that by swapping it out with Hal's. We're thinking more an more a connection issue either of some signal going to the ECM or some signal coming out of it.

Open to ideas for where to go next?

 

[Blrfl]

Many thanks to Jim Henke today for coming over and scoping up my bike.

Now that's news you can use. Thanks, Jim!

Capture 002 shows the left and right coils and you can seen the interruption when the bike misfires

Probably not related to your problem but interesting: There's a small spike in the cyan trace when the coil is done firing and blue trace rises back to its charged state. It's hard to tell from the pictures if the opposite is happening because the cyan trace extends into the blue trace where it would be. There may be some crosstalk in the wiring and isn't anything to worry about. The injector trace gets a bit of it, too.

Open to ideas for where to go next?

I still think this is being caused by one of the things that induces a no-start. If the sensors are off the table, the injectors are next. They appear to operate the same way as the coils, where the line is held normally high and pulled low when it's time to fire. The coils have to work that way to make spark, but the injectors may do it so the ECM can continuously monitor the circuit and kill the works if one goes south. (I have meetings all day today, but I'll check the manual later and see if I can figure out of that's correct.)

If Jim's willing to spend some more 'scope time with you, I'd instrument each of the injectors and watch for one misbehaving just before the misfire. You'll want to test both the supply and drain (ground) side of each, looking for a difference in the shape of the signals just before it happens. You should see some voltage difference across the injector, but if there's an intermittent open in the injector itself, the drain side will drop to zero before the ECM burps and doesn't fire the plugs/injectors. If you see a full drop of everything, you could have a wiring problem, and it would be worth repeating the test at the ECM end of the wiring harness.

One other idea from off in left field: If the FI indicator is driven by a single line from the ECM, probe that and see if it goes high briefly during the misfires. My thought here is that the lamp itself is incandescent and doesn't illuminate very quickly, so and this might at least give you some idea that the ECM thinks there's a fault. Won't tell you what the fault is, but at least it's another data point.

HTH.

--Mark

 

[wjbertrand]

First a qualifier. I know absolutely nothing about the ECM, throttle bodies or how the coils get the signal to generate a spark on the ST1300. But that's not going to stop me having a look !

I'm assuming the trace showing the coils shows the input to the coils, not the HT 'spark' output.

Yes, that's correct, we're looking at the primary side of the coils.

The output showing the throttle bodies having the same problem at the same time as the coil suggests a fault common to both of them.
Looking at the wiring diagram, the throttle bodies are connected direct to the ECM
The coils are connected to a different output from the ECM

But there does seem to be a lead that is common to both of these (and to other actuators and sensors). It is the Black/White lead which seems to be a +12v supply which is turned on by the Bank Angle relay, activated by the bank angle sensor. I'm looking at that brilliant coloured wiring diagram from David (Anna's Dad). The Black/White lead is a little difficult to follow on the diagram as it seems to loose a black line on one side, but I've compared it with the workshop manual version. That power for the Bank angle relay comes from fuse L, but the trigger wire goes all over the place - including the start/stop switch and the Engine Starter switch.

It would be useful to know if this power line is interrupted at the same time as the coils and throttle bodies switch off momentarily.

If it does, then the cause could be
interruption (bad contacts) in the power line
interruption (bad contacts) in the trigger lines to any of the relays.
Faulty bank angle sensor
Faulty relay contacts
Bouncing relay contacts

But to put the above in context.
It is odd that two devices demonstrate the same fault at the same time
Ive assumed it must be a common power supply or earth problem.
I've looked for a common lead
I've put two and two together and may well have come up with 22 rather than four.

But it is certainly feasible from an electrical point of view.


If you still have access to Jim and his scope, testing the 12v line compared with the coil and injectors should be straight forward to establish this as a possible cause.

Replacing a 12v supply ignoring all relays and switches should be an easy workshop test.

Then work backwards, starting with the bank angle relay (checking power in and out and checking the relay trigger) to establish the cause.

Note that according to the diagram, devices such as the O2 sensors and the Exhaust Air Injection Solenoid Valve are powered from the same line.

Food for thought anyway. I hope it helps.

John

Thanks John, I'll need to re-read your comments (probably a couple of times!) with the wiring diagram in front of me. Not sure I'm following where to test that 12V supply. The BAS comment is interesting, I've not yet visited it but will take a closer look and perhaps jump the connector to see if eliminating it from the circuit helps. I've swapped out the fuel cut relay with another one (there are several identical relays on the bike) just to see what would happen, obviously without effect. The scope results confirm what I was feeling on the road though and that is a complete loss of power momentarily.

 

[wjbertrand]

Now that's news you can use. Thanks, Jim!



Probably not related to your problem but interesting: There's a small spike in the cyan trace when the coil is done firing and blue trace rises back to its charged state. It's hard to tell from the pictures if the opposite is happening because the cyan trace extends into the blue trace where it would be. There may be some crosstalk in the wiring and isn't anything to worry about. The injector trace gets a bit of it, too.



I still think this is being caused by one of the things that induces a no-start. If the sensors are off the table, the injectors are next. They appear to operate the same way as the coils, where the line is held normally high and pulled low when it's time to fire. The coils have to work that way to make spark, but the injectors may do it so the ECM can continuously monitor the circuit and kill the works if one goes south. (I have meetings all day today, but I'll check the manual later and see if I can figure out of that's correct.)

If Jim's willing to spend some more 'scope time with you, I'd instrument each of the injectors and watch for one misbehaving just before the misfire. You'll want to test both the supply and drain (ground) side of each, looking for a difference in the shape of the signals just before it happens. You should see some voltage difference across the injector, but if there's an intermittent open in the injector itself, the drain side will drop to zero before the ECM burps and doesn't fire the plugs/injectors. If you see a full drop of everything, you could have a wiring problem, and it would be worth repeating the test at the ECM end of the wiring harness.

One other idea from off in left field: If the FI indicator is driven by a single line from the ECM, probe that and see if it goes high briefly during the misfires. My thought here is that the lamp itself is incandescent and doesn't illuminate very quickly, so and this might at least give you some idea that the ECM thinks there's a fault. Won't tell you what the fault is, but at least it's another data point.

HTH.

--Mark

Thanks for the analysis Mark. Jim has indicated willingness to have another look once I've got a list of objectives. With respect to the FI light, even if the fault were too short for the lamp to illuminate, wouldn't a code still be set?

So to follow what you're saying about the injectors - if one is faulty it may be causing the ECM to shut down the others? I guess that would have to include shutting down the coils too. You've given me an idea though. If I understand what you're saying correctly, I should be able to test this idea by disconnecting one of the injector connectors. Depending whether the bike keeps running on the other three or not, would prove/disprove that theory? I know (accidentally) that the bike will at least idle with one of the coils disconnected and that takes out two cylinders, so it should be able to run on three injectors.

 

[dwalby]

Hey Jeff,

I've lost track of all the parts swapping, but can't recall if the ignition switch has been ruled out. Given that you've confirmed a power dropout that's affecting multiple elements simultaneously, try hanging one channel of the scope on the ignition switch output.

 

[Blrfl]

With respect to the FI light, even if the fault were too short for the lamp to illuminate, wouldn't a code still be set?

Hopefully that's the case, but it's also possible that fleeting problems might be discarded if the situation improves quickly enough. Only Honda knows. What we don't know is whether we're seeing a stop/start cycle (ECM stops the engine because of a fault, the fault clears up, ECM sees the engine spinning as if the starter was being cranked, starts fueling) or a reset, which happens when the thing gets totally confused or the software hangs up and the watchdog timer forces the issue. Resets are supposed to happen quickly (milliseconds) and you're not supposed to notice. From what I've read about the ECMs in cars, resets are a pretty common occurrence.

So to follow what you're saying about the injectors - if one is faulty it may be causing the ECM to shut down the others?

I'm thinking that it kills everything, just like it would if it lost track of the cam or crank position. We know for sure that a mouse-chewed FI subharnesses causes a no-start and it probably causes a no-run, too. Disconnecting an injector would be a good way to find out. If I were designing it, I'd stop the engine, but I don't work for Honda.

If Jim comes back for a second round, I'd also suggest having another run at the crank and cam sensors from both sides just to make sure you didn't miss a failure from one side. Might also be worth checking at the ECM end, too.

It's great that you got hold of the right instrumentation to test this stuff. Makes it easy to rule out a lot of things.

--Mark

 

[wjbertrand]

Thanks again. Not sure what you mean by crank and cam sensors from both sides?? There is only one of each.


-Jeff

 

[Blrfl]

Thanks again. Not sure what you mean by crank and cam sensors from both sides?? There is only one of each.

Both sides of the sensor. The fuel injector circuits and many of the sensor circuits work like this:

ECM Send ---(A)---> Injector or Sensor ---(B)---> ECM Return​

The ECM will supply current to the sensor or injector on a send line, the sensor will do whatever it does and return a signal on the return line.

Let's say your problem is being caused by a fuel injector whose actuator circuit goes open briefly as you fall through 4,250 RPM because of some mechanical issue. If you're scoping the circuit at the point marked A, you're testing what the ECM is providing to run the sensor whether the injector is there or not. Put the probe at point B and you'll see whatever is being returned to the ECM. If the injector circuit opens, you'll see the voltage drop to zero briefly just before you see the ignition miss. (Again, that's if that's the cause.)

Sensors may work the same way depending on what kind of sensors they are. Some like the ECT are resistive and you'll be able to observe a difference in voltage between A and B while it's operating. The cam and crank are probably monitored by simple inductors or Hall effect sensors and will do the same thing.

--Mark

 

[wjbertrand]

OK, I mistook your comments to mean both banks of cylinders or something. Can you tell I'm over my head with this whole thing? Yeah, I guess the next step is to make sure sensor output is reaching the ECM. I may be wrong but I'm under the impression that the cam and crank sensors generate their own signal and do not require powering from the ECM. They each have only two wire connectors.

 

[st1300r]

FWIW I get 500 ohms across the crank sensor I pulled from my fiasco and the probe tip is magnetic.
Assuming its one of these.



The tone wheel looks like constant pitch but the notches have a pattern. Maybe just a visual for timing?

 

[wjbertrand]

I remember measuring mine but don't remember the result. There's a metal tip but I didn't notice that it was magnetic at all.

 

[Blrfl]

Can you tell I'm over my head with this whole thing?

Maybe, but I'll bet you're learning something new.

Yeah, I guess the next step is to make sure sensor output is reaching the ECM. I may be wrong but I'm under the impression that the cam and crank sensors generate their own signal and do not require powering from the ECM. They each have only two wire connectors.

It depends on the sensor. Hall-effect sensors have three lines, one of which provides power. These are probably plain inductive sensors, meaning they're a permanent magnet and a coil that generate a small signal when something metallic passes by. (More details and a good illustration here). In some applications, a voltage is passed through the sensor and the passing metal disturbs that. This is very likely the case here since the ECM can detect a failure.

Ideally, you'll want to test this at all four interesting points in the circuit: both pins at the ECM and both pins at the sensors and injectors. If you hit all four and one or more shows a signal drop just before the engine stutters, which one it is will help determine whether the fault is in the sensor or the wiring. Come to think of it, most of those sensors also pass through the FI subharness connector, so if you see a failure in a circuit, testing those four additional points (two wires, two sides each) can help narrow down exactly where it is.

There's still the slim possibility that the conditions on the bike are bringing a bug in the ECM firmware to the surface, but all of the other factors have to be ruled out. If the injectors and sensors are provably good, I will be out of ideas.

--Mark

 

[adamk]

There's still the slim possibility that the conditions on the bike are bringing a bug in the ECM firmware to the surface, but all of the other factors have to be ruled out. If the injectors and sensors are provably good, I will be out of ideas.

That's been my gut feeling ever since DynoJet's failure to get a Power Commander to work with my ST1300. Almost like the introduction of the PC created a load on the ECM which manifested itself in a very similar way to what Jeff is experiencing with his bike.

http://koczarski.com/ST1300/PowerCommander/PowerCommander.htm